neurosurgical operative atlas volume 1

i

NEUROSURGICALOPERATIVE ATLAS

Volume 1

ii

AANS PUBLICATIONS COMMITTEE

Editors

SETTI S. RENGACHARY, M.D.Professor and Chief

Section of Neurological SurgeryUniversity of Missouri at Kansas City

Kansas City, Missouri

ROBERT H. WILKINS, M.D.Professor and Chief

Division of NeurosurgeryDuke University Medical Center

Durham, North Carolina

NEUROSURGICALOPERATIVE ATLAS

Volume 1

THE AMERICAN ASSOCIATION OF NEUROLOGICAL SURGEONS

iii

Editor: Carol-Lynn BrownAssociate Editor: Marjorie Kidd Keating, Victoria M. VaughnCopy Editor: Janet KrejciDesigner: Dan PfistererIllustration Planner: Wayne HubbelProduction Coordinator: Raymond E. Reter

Copyright © 1991, 1992, 1993

The American Association of Neurological SurgeonsChicago, Illinois

All rights reserved. This book is protected by copyright. No part of thisbook may be reproduced in any form or by any means, including photo-copying, or utilized by any information storage and retrieval systemwithout written permission from the copyright owner.

This publication is published under the auspices of the PublicationsCommittee of The American Association of Neurological Surgeons(AANS). However, this should not be construed as indicating endorse-ment or approval of the views presented, by the AANS, or by its com-mittees, commissions, affiliates, or staff.

Printed in the United States of America

Library of Congress Cataloging-in-Publication DataNeurosurgical operative atlas / AANS Publications Committee ;editors, Setti S. Rengachary, Robert H. Wilkins.

p. cm.Includes index.ISBN 0-683-07234-X (v. 1)I. Nervous system—Surgery—Atlases. I. Rengachary, Setti S.

II. Wilkins, Robert H. II. AANS Publications Committee.[DNLM: I. Nervous System—Surgery—Atlases. WL

17 N494]RD593.N43 1991617.4’8’00222—dc20DNLM/DLCfor Library of Congress 90-14551

CIP

91 92 93 94 951 2 3 4 5 6 7 8 9 10

iv

I am honored to write this foreword to The American As-sociation of Neurological Surgeons’ Neurosurgical Opera-tive Atlas. The list of operations is impressive and coversalmost every detail of neurosurgery. The authors selectedto present these operations are even more impressive, rep-resenting, as they do, outstanding individuals in the UnitedStates and Canada who are respected and admired by ev-eryone in the medical profession.

This is not the first effort the former Harvey CushingSociety has made in this field-more than 25 years ago theBoard of Editors of the Journal of Neurosurgery agreedthat it was important to publish a section, entitled Neuro-surgical Techniques, of select operative drawings withbrief explanatory text that would be published as fasciclesover the ensuing years.

Most of us in neurosurgery at that time had dependedon the volume devoted to the nervous system in Bancroftand Pilcher’s Surgical Treatment, the responsibility forwhich had been that of Cobb Pilcher, then at Vanderbilt.Although Surgical Treatment was not specifically a “tech-niques” book, it had much technique in it and was widelyused. The editorial board of the Journal of Neurosurgeryhoped that Neurosurgical Techniques would serve as amore up-to-date version of that volume of Surgical Treat-ment.

Emphasis in Neurosurgical Techniques was given tothe artists’ depictions of established, safe techniques. Theeditors of the Journal realized that a procedure might bedone successfully in more than one way, but at least onegood and safe technique was to be described, and it wasassumed that the more skilled and experienced surgeonswould utilize other methods they found suitable.

The basis of the decision to focus on the drawingsand to have relatively little associated text was a previousatlas, the Atlas of Surgical Operations (1). Many of theeditors of the Journal had become familiar with this atlasduring their general surgical training. Mildred Codding,who had been the artist for Harvey Cushing, had devel-oped an effective technique of drawing the stages of op-erative procedures that had been and remains an effectiveteaching aid.

The fascicles of Neurosurgical Techniques in theJournal of Neurosurgery were to be bound together but,due to changes in publishers, this was never done. Fortu-nately, some of the plates were preserved and given tome, as editor of the fascicles. Three of these are shown toprovide a historical perspective.

The first two figures illustrate a technique by Dr.

Foreword

James Greenwood of Houston, Texas (2), who pioneeredthe complete removal of ependymomas in the cervicalspinal cord and developed unique bipolar Bovie forcepswith suction at the tips. Use of his Bovie forceps en-abled him to carry out detailed dissections with magni-fying lenses before the advent of the operating micro-scope. Figure 1 shows the spinal cord split, the tumorbeing removed, and a line of cleavage being developed;Figure 2 shows a further removal with the suction bipo-lar forceps in the field.

Figure 3 shows a cervical fusion technique (3) and ispublished to give due credit to Mr. George Lynch, theartist who approved of and often redrew illustrations thatwere published in Neurosurgical Techniques. Figure 3 il-lustrates the final stages of cervical fusion for dislocationof C-3, C-4, using bone and wire for the fusion process.This technique was used before the development of theacrylic fusion technique, which came to replace the bonegrafting procedure.

As was expressed in the final paragraph of the intro-duction to Neurosurgical Techniques (4), neurosurgery wasevolving rapidly, and often the procedure described had notoriginated with the person who wrote about it. Then, as now,surgery was a combination of new knowledge with old. Sinceknowledge is freely shared among surgeons nationally andinternationally, original techniques and ideas are passedfrom teacher to student, from surgeon to surgeon, and

Figure 1.

© 1991 The American Association of Neurological Surgeons

v

FOREWO R D

Figure 2.

Figure 3.altered with changing developments and experience. Aprocedure well-established today may seem naive or use-less a few years hence. One can make valid judgmentsonly on the basis of data that become available. Thus, itis my great pleasure to introduce this ambitious projectby Dr. Wilkins and Dr. Rengachary. Not only will it up-date neurosurgical techniques, but it is also a further de-velopment of the effort made by the editors of the Jour-nal of Neurosurgery 25 years ago. We hope it will surpassthat effort.

EBEN ALEXANDER, JR.

References1. Cutler EC, Zollinger R. Atlas of surgical operations. Illus

trated by Mildred B. Codding. New York: Macmillan, 1939.2. Greenwood J, Jr. Surgical removal of intramedullary tu mors.

J Neurosurg 1967;26:275-282.3. Alexander E, Jr, Davis CH, Jr, Forsyth HF. Reduction and

fusion of fracture dislocations of the cervical spine. JNeurosurg 1967;27:587-591.

4. Alexander E, Jr. Neurosurgical techniques introduction. JNeurosurg 1966;24:817-819.



vi

Preface

Man has always had an innate urge to depict his activitiesin drawings, as the paintings of cave dwellers would at-test. Surgical atlases perhaps represent a formalized ver-sion of such an urge; the atlases, in addition to document-ing the work, have instructional value as well-being ableto teach generations of trainees the craft of surgery aspracticed by the masters of the trade. Although electronicimages have greatly advanced the instructional process,printed artwork remains the backbone of the media forteaching.

There has been a perception among all neurosurgeonsfor some time that a contemporary atlas in neurosurgeryis due. To fill this void, the Publications Committee ofthe American Association of Neurological Surgeons hasundertaken the task of producing a comprehensive atlas.The atlas will be published at bimonthly intervals in theform of fascicles containing up to six operations each. Toallow timely publication, topics are included in a randomfashion.

The Neurosurgical Operative Atlas, we believe, isunique in several respects. It is comprehensive. Whencompleted, it will contain descriptions of up to two hun-dred operative procedures. It is multiauthored. Given thecomplexity of the field and the explosive advances in tech-niques, it is impossible for any one individual to be skilledenough to describe the entire spectrum of techniques au-thoritatively. In many instances, the chosen authors arethose who developed a technique originally or have usedthe technique so extensively as to be an authority on it.

Many illustrations are depicted in full color despite thehigh costs involved in preparing the artwork and print-ing. In many instances where there is more than one wayto approach a problem, two different authors have beenrequested to write on the same subject so that the readerwill benefit from knowing alternative surgical techniques.

The atlas has been possible in large measure due tothe efforts and sacrifices of the contributing authors. Inaddition to sharing their knowledge and expertise, theyhave incurred large expenses in getting the artwork done;they have spent long hours with their illustrators to achievethe accurate and esthetically pleasing depiction of theprocedures. One can also see the spectrum of artistic tal-ent that made this work possible.

We thank George T. Tindall, M.D., for forming thePublications Committee; Eben Alexander, Jr., M.D., forpreparing the Foreword; Carol-Lynn Brown and MarjorieKeating of the Williams & Wilkins Company, and CarlH. Hauber and Gabrielle J. Loring of The American As-sociation of Neurological Surgeons for coordinating vari-ous phases of the project; Sherylyn Cockroft and GloriaK. Wilkins for secretarial help; members of the Publica-tions Committee for innumerable suggestions; and DianeAbeloff for overseeing the entire artwork.

SETTI S. RENGACHARYROBERT H. WILKINS

vii

Contributors

AANS Publications Committee

ROBERT H. WILKINS, M.D., ChairmanMICHAEL L. J. APUZZO, M.D.ISSAM A. AWAD, M.D.DANIEL L. BARROW, M.D.EDWARD C. BENZEL, M.D.

PAUL R. COOPER, M.D.OLIVER D. GRIN, JR., M.D.SETTI S. RENGACHARY, M.D.THORALF M. SUNDT, JR., M.D.EDWARD TARLOV, M.D.

CHAD D. ABERNATHEY, M.D.NeurosurgeonDepartment of NeurosurgeryIowa Medical ClinicsCedar Rapids, Iowa

A. LELAND ALBRIGHT, M.D.Associate ProfessorDepartment of NeurosurgeryUniversity of Pittsburgh School of MedicinePittsburgh, Pennsylvania

OSSAMA AL-MEFTY, M.D.Professor of NeurosurgeryUniversity of Mississippi Medical CenterJackson, Mississippi

RICHARD P. ANDERSON, M.D.Chief ResidentDepartment of NeurosurgeryWest Virginia University HospitalsMorgantown, West Virginia

EHUD ARBIT, M.D.Associate Attending SurgeonNeurosurgery ServiceDepartment of SurgeryMemorial Sloan-Kettering Cancer CenterNew York, New York

ISSAM A. AWAD, M.D.Vice ChairmanDepartment of Neurological SurgeryThe Cleveland Clinic FoundationCleveland, Ohio

DANIEL L. BARROW, M.D.Associate ProfessorDepartment of NeurosurgeryEmory University School of MedicineAtlanta, Georgia

JOSHUA B. BEDERSON, M.D.ResidentDepartment of NeurosurgeryUniversity of California, San FranciscoSan Francisco, California

WILLIAM O. BELL, M.D.Associate ProfessorDepartment of NeurosurgeryBowman Gray School of Medicine of Wake Forest UniversityWinston-Salem, North Carolina

EDWARD C. BENZEL, M.D.Professor and ChiefDivision of NeurosurgeryUniversity of New Mexico School of MedicineAlbuquerque, New Mexico

DERALD E. BRACKMANN, M.D.Clinical Professor of OtolaryngologyUniversity of Southern CaliforniaSchool of MedicineLos Angeles, California

MICHAEL N. BUCCI, M.D.ResidentSection of NeurosurgeryUniversity of Michigan Medical CenterAnn Arbor, Michigan

KIM J. BURCHIEL, M.D.Professor and HeadDivision of NeurosurgeryOregon Health Sciences UniversityPortland, Oregon

viii

CONTRIBUTORS

PAUL J. CAMARATA, M.D.ResidentDepartment of NeurosurgeryUniversity of MinnesotaMinneapolis, Minnesota

CHRISTOPHER E. CLARE, M.D.Chief ResidentDepartment of NeurosurgeryEmory University School of MedicineAtlanta, Georgia

RALPH B. CLOWARD, M.D.Clinical ProfessorDepartment of NeurosurgeryJohn A. Burns School of MedicineUniversity of HawaiiHonolulu, Hawaii

CURTIS A. DICKMAN, M.D.Chief ResidentDivision of Neurological SurgeryBarrow Neurological InstitutePhoenix, Arizona

DONALD F. DOHN, M.D.ChairmanDepartment of Neurological SurgeryThe Cleveland Clinic FloridaFort Lauderdale, Florida

JAMES FICK, M.D.ResidentDepartment of NeurosurgeryUniversity of Cincinnati Medical CenterMayfield Neurological InstituteCincinnati, Ohio

EUGENE S. FLAMM, M.D.Charles Harrison Frazier Professor and ChairmanDivision of NeurosurgeryUniversity of Pennsylvania School of MedicinePhiladelphia, Pennsylvania

EDDY GARRIDO, M.D.Private PracticeLancaster, Pennsylvania

SARAH J. GASKILL, M.D.ResidentDivision of NeurosurgeryDuke University Medical CenterDurham, North Carolina

FRED H. GEISLER, M.D., PH.D.Clinical Assistant ProfessorDepartment of SurgeryDivision of NeurosurgeryThe Shock Trauma Center of the MarylandInstitute for Emergency Medical ServiceSystems and the University of MarylandBaltimore, MarylandChief of NeurosurgeryDepartment of NeurosurgeryPatuxent Medical GroupColumbia, Maryland

ATUL GOEL, M.CH.FellowCenter for Cranial Base SurgeryDepartment of NeurosurgeryUniversity of Pittsburgh School of MedicinePittsburgh, Pennsylvania

JAMES T. GOODRICH, M.D., PH.D.Director, Division of Pediatric NeurosurgeryLeo Davidoff Department of Neurological SurgeryAlbert Einstein College of MedicineMontefiore Medical CenterBronx, New York

DAVID J. GOWER, M.D.Assistant ProfessorDivision of NeurosurgeryUniversity of Oklahoma Health Sciences CenterOklahoma City, Oklahoma

STEPHEN J. HAINES, M.D.Associate ProfessorDivision of Pediatric NeurosurgeryDepartment of NeurosurgeryUniversity of MinnesotaMinneapolis, Minnesota

CRAIG D. HALL, M.D.Institute of Plastic and Reconstructive SurgeryAlbert Einstein College of MedicineMontefiore Medical CenterBronx, New York

SAMUEL J. HASSENBUSCH, M.D., PH.D.Section HeadNeuro-Pharmacologic Oncology and Pain ManagementThe Cleveland Clinic FoundationCleveland, Ohio

ix

PATRICK W. HITCHON, M.D.ProfessorDivision of NeurosurgeryThe University of Iowa Hospitals and Clinics andDepartment of Veterans Affairs Medical CenterIowa City, Iowa

JULIAN T. HOFF, M.D.Professor and ChairSection of NeurosurgeryUniversity of Michigan Medical CenterAnn Arbor, Michigan

HAROLD J. HOFFMAN, M.D.Professor of SurgeryDivision of NeurosurgeryUniversity of Toronto;Chief of NeurosurgeryThe Hospital for Sick ChildrenToronto, Ontario, Canada

ROBERT S. HOOD, M.D.Clinical Assistant ProfessorDivision of NeurosurgeryUniversity of Utah College of MedicineSalt Lake City, Utah

EDGAR M. HOUSEPIAN, M.D.ProfessorDepartment of Clinical Neurological SurgeryCollege of Physicians & SurgeonsColumbia University;Attending NeurosurgeonNew York Neurological InstituteColumbia-Presbyterian Medical CenterNew York, New York

JOHN A. JANE, M.D.Professor and ChairmanDepartment of NeurosurgeryUniversity of Virginia Health Sciences CenterCharlottesville, Virginia

HOWARD H. KAUFMAN, M.D.Professor and ChairmanDepartment of NeurosurgeryWest Virginia University HospitalsMorgantown, West Virginia

PATRICK J. KELLY. M.D.ProfessorDepartment of Neurologic SurgeryMayo Medical SchoolRochester, Minnesota

LEE KESTERSON, M.D.Assistant ProfessorDivision of NeurosurgeryUniversity of New Mexico School of MedicineAlbuquerque, New Mexico

PHYO KIM, M.D.ResidentDepartment of Neurologic SurgeryMayo Clinic/Mayo Medical SchoolRochester, Minnesota

MILAM E. LEAVENS, M.D.Associate Professor and ChiefDepartment of NeurosurgeryThe University of Texas M.D. AndersonCancer CenterHouston, Texas

MICHEL F. LEVESQUE, M.D.Division of NeurosurgeryUCLA Center for the Health Sciences andSchool of MedicineLos Angeles, California

ARTHUR E. MARLIN, M.D.Chief, Pediatric NeurosurgeryVice Chairman, Pediatric SurgerySanta Rosa Children’s HospitalSan Antonio, Texas

MARK MAY, M.D.Clinical ProfessorDepartment of Otolaryngology-Head andNeck SurgeryUniversity of Pittsburgh School of Medicine;DirectorSinus Surgery and Facial Paralysis CenterShadyside HospitalPittsburgh, Pennsylvania

DAVID C. McCULLOUGH, M.D.ChairmanDepartment of Neurological SurgeryChildren’s HospitalWashington, D.C.

DENNIS E. McDONNELL, M.D.Associate ProfessorDepartment of SurgerySection of NeurosurgeryMedical College of GeorgiaAugusta, Georgia

CONTRIBUTORS

x

CONTRIBUTORS

ARNOLD H. MENEZES, M.D.Professor and Vice ChairmanDivision of NeurosurgeryThe University of Iowa Hospitals and ClinicsIowa City, Iowa

FREDRIC B. MEYER, M.D.ConsultantDepartment of Neurologic SurgeryMayo ClinicRochester, Minnesota

FOAD NAHAI, M.D.Professor of SurgeryDepartment of Plastic and Reconstructive SurgeryEmory University School of MedicineAtlanta, Georgia

W. JERRY OAKES, M.D.Associate ProfessorDivision of NeurosurgeryDepartment of Surgery;Assistant ProfessorDepartment of PediatricsDuke University Medical CenterDurham, North Carolina

BURTON M. ONOFRIO, M.D.ProfessorDepartment of Neurologic SurgeryMayo Clinic/Mayo Medical SchoolRochester, Minnesota

JOHN A. PERSING, M.D.ProfessorDepartment of Neurosurgery;Associate Professor and Vice-ChairmanDepartment of Plastic SurgeryUniversity of Virginia Health Sciences CenterCharlottesville, Virginia

JOSEPH H. PIATT, JR., M.D.Head, Pediatric Neurosurgery SectionDivision of NeurosurgeryOregon Health Sciences UniversityPortland, Oregon

PREM K. PILLAY, M.D.ConsultantSingapore General HospitalSingaporeFellowDepartment of Neurological SurgeryThe Cleveland Clinic FoundationCleveland, Ohio

JOSEPH RANSOHOFF, M.D.Professor and ChairmanDepartment of NeurosurgeryNew York University Medical CenterNew York, New York

MICHAEL P. SCHENK, M.S.Director of Medical IllustrationUniversity of Mississippi Medical CenterJackson, Mississippi

SYDNEY S. SCHOCHET, M.D.Professor of Pathology, Neurology, andNeurosurgeryDepartment of Pathology (Neuropathology)West Virginia University HospitalsMorgantown, West Virginia

LALIGAM N. SEKHAR, M.D.Associate ProfessorCenter for Cranial Base SurgeryDepartment of NeurosurgeryUniversity of Pittsburgh School of MedicinePittsburgh, Pennsylvania

JATIN SHAH, M.D.Attending SurgeonHead and Neck ServiceDepartment of SurgeryMemorial Sloan-Kettering Cancer Center;ProfessorCornell University Medical CollegeNew York, New York

ROBERT R. SMITH, M.D.Professor and ChairmanDepartment of NeurosurgeryUniversity of Mississippi Medical CenterJackson, Mississippi

STEVEN M. SOBOL, M.D.Private PracticeDecatur, Illinois

VOLKER K. H. SONNTAG, M.D.Vice ChairmanDivision of Neurological SurgeryBarrow Neurological InstitutePhoenix, ArizonaClinical Associate ProfessorDivision of NeurosurgeryDepartment of SurgeryUniversity of Arizona College of MedicineTucson, Arizona

xi

CONTRIBUTORS

THORALF M. SUNDT, JR., M.D.Professor and ChairmanDepartment of Neurologic SurgeryMayo Medical SchoolRochester, Minnesota

EDWARD TARLOV, M.D.Department of NeurosurgeryLahey Clinic Medical CenterBurlington, Massachusetts

JOHN M. TEW, JR., M.D.Professor and DirectorDepartment of NeurosurgeryUniversity of Cincinnati Medical CenterMayfield Neurological InstituteCincinnati, Ohio

GEORGE T. TINDALL, M.D.Professor and ChairmanDepartment of NeurosurgeryEmory University School of MedicineAtlanta, Georgia

TADANORI TOMITA, M.D.Associate Professor of Surgery (Neurosurgery)Northwestern University Medical School;Assistant HeadDivision of Pediatric NeurosurgeryChildren’s Memorial HospitalChicago, Illinois

VINCENT C. TRAYNELIS, M.D.Assistant ProfessorDivision of NeurosurgeryThe University of Iowa Hospitals and Clinics and Department of Veterans Affairs Medical CenterIowa City, Iowa

ROBERT H. WILKINS, M.D.Professor and ChiefDivision of NeurosurgeryDuke University Medical CenterDurham, North Carolina

CHARLES B. WILSON, M.D.Tong-Po Kan Professor and ChairmanDepartment of NeurosurgeryUniversity of California, San FranciscoSan Francisco, California

ERIC J. WOODARD, M.D,Assistant ProfessorDepartment of NeurosurgeryEmory University School of MedicineAtlanta, Georgia

RONALD F. YOUNG, M.D.Professor and ChiefDivision of Neurological SurgeryUniversity of California, IrvineIrvine, California

xii

Contents

Foreword / ivPreface / vi

Contributors / vii

Volume 1

OPTIC GLIOMAS / 1Edgar M. Housepian, M.D.

FIBROUS DYSPLASIA INVOLVING THE CRANI0FACIAL SKELETON /James T. Goodrich, M.D., Ph.D.

Craig D. Hall, M.D.

DEPRESSED SKULL FRACTURE IN ADULTS /Fred H. Geisler, M.D., Ph.D.

CERVICAL HEMILAMINECTOMY FOR EXCISION OF A HERNIATED DISC /Robert H. Wilkins, M.D.

Sarah J. Gaskill, M.D.

LATERAL SPHENOID WING MENINGIOMA /Joseph Ransohoff, M.D

SELECTIVE MICROSURGICAL VESTIBULAR NERVE SECTION FOR INTRACTABLEMÉNIÈRE’S SYNDROME /

Edward Tarlov, M.D.

CHIARI MALFORMATIONS AND SYRINGOHYDROMYELIA IN CHILDREN /W. Jerry Oakes, M.D.

CAROTID BODY TUMORS /Fredric B. Meyer, M.D.

Thoralf M. Sundt, Jr., M.D.

OLFACTORY GROOVE MENINGIOMAS /Joshua B. Bederson, M.D.Charles B. Wilson, M.D.

CEREBRAL ANEURYSMS AT THE BIFURCATIONOF THE INTERNAL CAROTID ARTERY /

Eugene S. Flamm, M.D.

xiii

CONTENTS

TREATMENT OF UNILATERAL OR BILATERAL CORONAL SYNOSTOSIS /John A. Persing, M.D.

John A. Jane, M.D.

CONVEXITY MENINGIOMA /Sarah J. Gaskill, M.D.

Robert H. Wilkins, M.D.

OCCIPITAL LOBECTOMY /Milam E. Leavens, M.D.

SPINAL MENINGIOMAS /Michael N. Bucci, M.D.

Julian T. Hoff, M.D.

PERCUTANEOUS TRIGEMINAL GLYCEROL RHIZOTOMY /Ronald F. Young, M.D.

LUMBAR HEMILAMINECTOMY FOR EXCISION OF A HERNIATED DISC /Patrick W. Hitchon, M.D.

Vincent C. Traynelis, M.D.

TRANSORAL SURGERY FOR CRANIOVERTEBRAL JUNCTION ANOMALIES /Arnold H. Menezes, M.D.

ANTEROLATERAL CERVICAL APPROACHTO THE CRANIOVERTEBRAL JUNCTION /

Dennis E. McDonnell, M.D.

CORRECTION OF MALPOSITION OF THE ORBITS /John A. Persing, M.D.

REMOVAL OF CERVICAL OSSIFIED POSTERIOR LONGITUDINALLIGAMENT AT SINGLE AND MULTIPLE LEVELS /

Ralph B. Cloward, M.D.

TECHNIQUE OF VENTRICULOSTOMY /Joseph H. Piatt, Jr., M.D.

Kim J. Burchiel, M.D.

CEREBELLAR MEDULLOBLASTOMA /Arthur E. Marlin, M.D.Sarah J. Gaskill, M.D.

SHUNTING OF A POSTTRAUMATIC SYRINX /David J. Gower, M.D.

DIRECT SURGICAL TREATMENT OF VEIN OF GALEN MALFORMATIONS /Harold J. Hoffman, M.D.

xiv

CONTENTS

SPINAL NERVE SCHWANNOMA /Phyo Kim, M.D.

Burton M. Onofrio, M.D.

COMBINED CRANIOFACIAL RESECTION FOR ANTERIOR SKULL BASE TUMORS /Ehud Arbit, M.D.Jatin Shah, M.D.

DIAGNOSTIC OPEN BRAIN AND MENINGEAL BIOPSY /Richard P. Anderson, M.D.Howard H. Kaufman, M.D.Sydney S. Schochet, M.D.

VENTRICULOPERITONEAL SHUNTING /David C. McCullough, M.D.

VENTRICULOATRIAL SHUNTING /Paul J. Camarata, M.D.Stephen J. Haines. M.D.

EXCISION OF ACOUSTIC NEUROMAS BY THE MIDDLE FOSSA APPROACH /Derald E. Brackmann, M.D.

UPPER THORACIC SYMPATHECTOMY BY A POSTERIOR MIDLINE APPROACH /Prem K. Pillay, M.D.Issam A. Awad, M.D.Donald F. Dohn, M.D.

CAROTID ENDARTERECTOMY /Daniel L. Barrow, M.D.

Christopher E. Clare, M.D.

TRANSSPHENOIDAL EXCISION OF MACROADENOMASOF THE PITUITARY GLAND /

George T. Tindall, M.D.Eric J. Woodard, M.D.

Daniel L. Barrow, M.D.

COMPUTER-DIRECTED STEREOTACTIC RESECTION OF BRAIN TUMORS /Patrick J. Kelly, M.D.

SAGITTAL SYNOSTOSIS /A. Leland Albright, M.D.

GLOSSOPHARYNGEAL RHIZOTOMY /Burton M. Onofrio, M.D.

OCCIPITOCERVICAL AND HIGH CERVICAL STABILIZATION /Volker K. H. Sonntag, M.D.

Curtis A. Dickman, M.D.

xv

CONTENTS

PETROCLIVAL MENINGIOMAS /Ossama Al-Mefty, M.D.Michael P. Schenk, M.S.Robert R. Smith, M.D.

FACIAL REANIMATION WITHOUT THE FACIAL NERVE /Mark May, M.D.

Steven M. Sobol, M.D.

OMENTAL AND MUSCULOCUTANEOUS FREE FLAPS FOR COVERAGE OFCOMPLICATED NEUROSURGICAL WOUNDS /

Daniel L. Barrow, M.D.Foad Nahai, M.D.

REPAIR OF “GROWING” SKULL FRACTURE /Tadanori Tomita, M.D.

OCCIPITAL ENCEPHALOCELES /William O. Bell, M.D.

FORAMEN MAGNUM MENINGIOMAS AND SCHWANNOMAS:POSTERIOR APPROACH /Chad D. Abernathey, M.D.Burton M. Onofrio, M.D.

PENETRATING WOUNDS OF THE SPINE /Edward C. Benzel, M.D.

PERCUTANEOUS RADIOFREQUENCY RHIZOLYSISFOR TRIGEMINAL NEURALGIA /

James Fick, M.D.John M. Tew, Jr., M.D.

EXTENDED COSTOTRANSVERSECTOMY /Eddy Garrido, M.D.

SURGICAL RESECTION OF POSTERIOR FOSSA EPIDERMOIDAND DERMOID CYSTS /

Lee Kesterson, M.D.

LUQUE ROD SEGMENTAL SPINAL INSTRUMENTATION /Edward C. Benzel, M.D.

EN BLOC ANTERIOR TEMPORAL LOBECTOMYFOR TEMPOROLIMBIC EPILEPSY /

Michel F. Levesque, M.D.

CINGULOTOMY FOR INTRACTABLE PAIN USING STEREOTAXIS GUIDED BYMAGNETIC RESONANCE IMAGING /

Samuel J. Hassenbusch, M.D., Ph.D.Prem K. Pillay, M.D.

xvi

CONTENTS

CEREBELLAR ASTROCYTOMAS /A. Leland Albright, M.D.

EXTREME LATERAL LUMBAR DISC HERNIATION /Robert S. Hood, M.D.

TENTORIAL MENINGIOMAS /Laligam N. Sekhar, M.D.

Atul Goel, M.Ch.

Index /

1

OPTIC GLIOMASEDGAR M. HOUSEPIAN, M.D.

INTRODUCTIONThe primary indication for surgical treatment of opticglioma is disfiguring proptosis and progressive visualloss in a patient with unilateral optic nerve tumor. Whenindicated, tumor resection should extend from the scleralmargin to the chiasm to reduce the possibility of leav-ing residual tumor cells in the cut stump of the opticnerve. The following section will describe in detail theimportant steps in performing the transcranial orbitalapproach required to achieve this end. Although surgi-cal cure is a primary objective of this procedure, a sec-ondary objective Is preservation of a normal appearingglobe and normal extraocular movements, thus ensur-ing a good cosmetic result. To achieve this end, the neu-rosurgeon must be familiar with the regional anatomy.Orbital exploration is somewhat alien to neurosurgeryin that it does not involve the principles of intraaxialcranial surgery and there is no extraaxial space to visu-alize the important nerves, arteries, and veins thattraverse the orbit because of their protective investmentin orbital fat Thus, before describing clinical indicationsand surgical technique this section will briefly revieworbital surgical anatomy (Figs. 1-4).

SURGICAL ANATOMYThe orbit is a pear-shaped structure with the optic canalrepresenting the stem at the apex. The 5- to 10mm-longoptic canal enters medial to the anterior clinoid process.The lateral wall of the optic canal is formed by one of thetwo roots of the lesser wing of the sphenoid which alsoserves as the medial margin of the superior orbital fis-sure. In conforming to the shape of the optic nerve, theoptic canal is a 4 × 6-mm horizontal elliptical cavity at itspoint of entry, is circular in its midportion, and enters theorbit as a vertical ellipse. The lateral bony margins of theorbit are formed by the greater wing of the sphenoid andthe fronto-sphenoidal process of the zygomatic bone. Theorbital roof (floor of the anterior cranial fossa) is formedby the frontal bone, and the orbital plate of the maxillaforms both the floor of the orbit and the roof of the max-illary sinus. The medial wall of the orbit, covered by the


thin lamina papyracea, is bordered by the ethmoid andsphenoid sinuses. The frontal sinus lies in the medial por-tion of the superior orbital rim.

The cranial dura is redundant at the cranial end ofthe optic canal where it forms the falciform ligament. Itthen traverses the canal where the outer layer is continu-ous with the periorbita, while the inner layer continues asthe optic nerve sheath within the orbit. The intracranialsubarachnoid space is continuous with the subarachnoidspace of the optic nerve, although it is partially obliter-ated at the annulus of Zinn were the pial surface of theoptic nerve is fused to this fibrous annular band superi-orly and medially.

Six of the seven extraocular muscles take their ori-gin from the fibrous annulus of Zinn which fuses with theoptic nerve dura at the apical end of the optic canal (Fig.1). The lateral portion of the annulus of Zinn is formedby the two heads of the lateral rectus muscle. These loopwidely around the superior orbital fissure forming theboundaries of the so-called oculomotor foramen.

There are two routes of entry from the cranial cavityto the orbit: the optic canal and the superior orbital fis-sure. The ophthalmic artery (Fig. 2) enters the orbit withthe optic nerve and lies in a split layer of dura. within thecanal. Approximately 1 cm after entering, it gives off thecentral retinal artery which obliquely perforates the duraand courses to the retina within the optic nerve. The oph-thalmic artery then curves medially and superiorly overthe optic nerve, giving off two long posterior ciliary ar-teries and six or eight short posterior ciliary branchesbefore anastomosing with the external carotid circulation.Small branches of the ophthalmic artery supply the ocu-lar muscles at their origin near the apex of the orbit. Oc-clusion of the central retinal artery will result in visualloss, whereas ophthalmic artery occlusion frequently doesnot because of the rich external collateral supply.

The orbital veins become confluent and drain pri-marily through the superior orbital vein which exitsthrough the superior orbital fissure; there is a smallerdraining component through the inferior ophthalmicvein exiting through the inferior orbital fissure. The su-perior ophthalmic vein is the major venous drainagechannel, and occlusion of this vessel along its

2

NEUROSURGICAL OPERATIVE ATLAS, VOL. 1

Figure 1. Six of the seven extraocular muscles take their originfrom the fibrous annulus of Zinn which is fused medially with theoptic nerve dura at the apical end of the optic canal. In order toremove the optic nerve in one piece from the globe to the chiasm,the annulus of Zinn must be opened and the nerve sharp-dissectedfrom the medial dura. This necessitates sectioning of the levator

origin which arises from the medial aspect of the annulus. Theorigin must be resutured to prevent ptosis. The two long heads ofthe lateral rectus muscle loop widely around the superior orbitalfissure and define the oculomotor foramen. Those structures thattraverse the superior orbital fissure through the oculomotor fora-men thus lie directly within the muscle cone.


3

HOUSEPIAN : OPTIC GLIOMAS

Figure 2. The ophthalmic artery enters the orbit through theoptic canal and lies between the split layers of dura. The centralretinal artery is given off approximately 1 cm from the point ofentry of the ophthalmic artery which then loops over the opticnerve and anastomoses widely with the external carotid arterial

circulation. Multiple short perforating vessels supply the retina.The major venous drainage of the orbit is through the superiororbital fissure and the cavernous sinus. There is relatively minoranastomosis with the inferior ophthalmic vein draining throughthe inferior orbital fissure.


4


Figure 3. The fourth, first division fifth, and lacrimal nerves enterthe superior orbital fissure in that order. The frontalis nerve (VI)serves as a landmark identifying the region of the levator andsuperior rectus muscles and is visible through the thin periorbita.The trochlear, frontalis, and lacrimal nerves lie within the orbit butare outside of the muscle cone. Those nerves that enter the orbitthrough the region of the oculomotor foramen lie directly withinthe muscle cone. The superior division of the third nerve inner-vates the levator and superior rectus muscles; the nasociliary nerve

crosses over the optic nerve before exiting the orbit and sendsbranches to the laterally placed ciliary ganglion which gives offtwo long and several short ciliary nerves. The nasociliary nervealso directly gives off six or eight short ciliary nerves. The sixthnerve enters between the superior and inferior divisions of thethird nerve and innervates the lateral rectus muscle. The inferiordivision of the third nerve supplies the inferior oblique and inferiorrectus muscles and crosses under the optic nerve to supply themedial rectus muscle.


5


course or at the cavernous sinus frequently results inproptosis and chemosis.

Masses external to the periorbita may produce prop-tosis without affecting extraocular movement or vision,whereas tumors within the muscle cone, particularly thosecrowding the apex, frequently limit extraocular motilityand encroach upon vision. Structures traversing the supe-rior orbital fissure outside the oculomotor foramen will, ofcourse, be found within the periorbita. but outside themuscle cone. Thus, in order of entrance, the trochlear nerve,the frontalis branch of the fifth nerve, the lacrimal nerve,and the exiting ophthalmic vein are found within theperiorbita but outside of the muscle cone (Fig. 3). Withinthe oculomotor foramen, in order of passage from abovedownward, are found the superior division of the third nervecurving upward and medially to innervate the superior rec-tus and levator muscles, the nasociliary branch of the oph-thalmic nerve crossing forward and medial over the opticnerve, the sixth nerve innervating the lateral rectus muscle,and the inferior division of the third nerve which crossesbeneath the optic nerve to reach both the medial and infe-rior rectus and inferior oblique muscles. The ciliary gan-glion, which lies lateral to the optic nerve, receives pregan-glionic parasympathetic fibers from the oculomotor nerveand sends its postganglionic fibers to the sphincter of theiris and the ciliary muscle by the several long and shortciliary nerves. It is difficult to totally eradicate lesions thatinvolve the optic canal or superior orbital fissure dura with-out injury to those structures which traverse them. Tumorsthat invade the oculomotor foramen are particularly diffi-cult to eradicate for this reason. A review of this regionalanatomy further illustrates that a direct approach to the opticnerve between the levator and medial rectus muscles canbe safely achieved with the least chance of injury to theimportant nerve supply to any of the extraocular muscles.

One should also be cognizant of the fact that the levatormuscle origin lies superior and medial to the superior rectusmuscle. In view of the fusion of the annulus of Zinn with thedura of the optic nerve at this location, it is necessary toelectively section the levator origin to prevent its tearing uponopening the optic nerve dura. After this attachment is sharplydissected free and the optic nerve is removed from the canal,the levator origin is resutured to the annulus.

PATIENT SELECTIONThe term optic glioma as applied to all astrocytomas

of the optic nerves and chiasm would better be classifiedinto tumors affecting a single optic nerve, tumors affectingthe chiasm alone or the chiasm and one or both optic nerves,and those bilateral, multicentric optic nerve sheath glio-mas affecting the optic nerves but sparing the chiasm.

Each of these presentations requires a different man-agement plan. In addition, the differential diagnosis oftumors confined to one or more optic nerve sheaths issimply glioma versus meningioma, whereas a chiasmalglioma maybe difficult to differentiate from a hypotha-lamic glioma. The presence or absence of neurofibroma-tosis is of significance but does not always differentiateglioma from meningioma. Neurofibromatosis is associ-ated with a high proportion of single optic nerve gliomasand a very low percentage of chiasmal gliomas. In a se-ries of 135 cases of optic glioma, 100% of those multi-centric bilateral optic nerve sheath tumors were associ-ated with neurofibromatosis. This raises the question ofpossibly different causation and natural history whenglioma occurs with or without neurofibromatosis.

The availability of high-resolution imaging tech-niques further helps differentiate those orbital tumors thatarise within the optic nerve from those that simply dis-place the optic nerve. Multicentric tumors and chiasmalinvolvement also can be clearly defined.

Thus, current indications for surgery in optic gliomaare limited to those patients with single nerve involve-ment who have poor vision and disfiguring proptosis, inwhom the prime object of surgery is resection of thetumor-bearing optic nerve from the globe to the chiasm,or exploration for biopsy in those cases of chiasmalglioma in which the diagnosis is in doubt. Children withmulticentric optic glioma and patients with chiasmalglioma are currently followed for signs of visual change.Irradiation is recommended only if there is progressivevisual loss. The rare case of exophytic chiasmal gliomahas benefited from subtotal resection of tumor followedby radiotherapy.

The diagnosis of optic glioma is easy to make in achild with proptosis, pallor or gliosis of the optic nervehead, enlargement of the optic canal, and neurofibroma-tosis. In an older age group the differential diagnosis maybe more difficult because of the prevalence of meningiomaof the optic nerve. In the classical case of meningioma,there is progressive visual loss without proptosis, andoptociliary shunting is pathognomonic for meningioma.Furthermore, distinctive computed tomography (CT) pat-terns differentiate gliomas and meningiomas. The CTappearance of the nerve in glioma is one of a massivelyswollen optic nerve with clear-cut margins; there may bekinking and buckling of the optic nerve as well. The com-mon CT pattern for meningioma is a narrow, diffuselyenlarged nerve with polar expansion at the apex. Unlikegliomas, there may also be calcification within the tu-mor. The optic nerve appears as a lucent shadow in thecenter of optic nerve sheath meningioma but a

6


Figure 4. This drawing illustrates the relation of the extraocularmuscles, the optic nerve and superior orbital fissure, the cranial andautonomic nerves, the course of the ophthalmic artery, and the

major venous drainage of the orbit. These very fine structures areenveloped in and protected by orbital fat.


7


similar appearance may occasionally be seen in gliomasassociated with neurofibromatosis.

PREOPERATIVE PREPARATIONIt is no longer necessary to visualize the regional circula-tion by angiography as part of the preoperative workupfor patients with optic glioma. There are thus no specialpreoperative studies or procedures required, and the an-esthetic techniques used are common to all major neuro-surgical procedures. Blood for replacement should becross-matched and available but is rarely necessary. Spi-nal drainage is not required during the operative proce-dure as opening of the chiasmatic cistern readily allowsfree escape of cerebrospinal fluid with ventricular emp-tying, resulting in the development of an operating fieldrequiring little or no retraction. An osmotic diuretic isused, however, to reduce the intraorbital pressure and tofacilitate dissection within the periorbita in the presenceof tumor. The patient or his family should be instructedthat there is a potential for injury to one or both olfactorynerves with resulting alteration of smell and sometimestaste, especially if osmotic diuretics are used. The epidu-ral approach to the orbit minimizes the risk of tearing theolfactory nerve. Similarly, the possibility of permanentptosis or loss of extraocular muscle function due to in-jury to the third or sixth nerve must be understood preop-

eratively. Finally, injury to the sympathetic nerve supplyor circulation to the globe may result in significant iritiswhich could result in the loss of the globe.

Intraoperative antibiotics and high-dose steroids areused and the latter continued for four or five days untilthe peak edema phase has passed.

There is no need for neurophysiological monitoring.The operating microscope is of less help in this procedurebecause the marked obliquity of the surgical field makes itdifficult to maintain focus at higher magnifications.

In summary, the primary objective of transcranialorbital resection of optic glioma is a surgical cure; a sec-ondary objective is maintenance of a normal appearingand moving globe to ensure a good cosmetic result.

SURGICAL TECHNIQUEThe operating table is placed in a low Fowler’s positionwith the patient supine and the head directly midline(Fig. 5). When dealing with optic nerve tumor by thetranscranial route, we prefer an osteoplastic bone flap(Fig. 6); there seems to be little advantage to elevating afree bone flap, which includes the orbital rim, becauseit affords no additional exposure at the apex. A coronalincision is marked and infiltrated with 1% Xylocainewith epinephrine, and the scalp and galea are incised asone. Michele clips or the equivalent are used at the

Figure 5. A coronal incision and low medial frontal flap is the preferred approach to the medial orbit.


8


Figure 6. An osteoplastic, medial frontal flap is illustrated. It isnot necessary to resect the orbital rim because it gives little addi-tional access to the orbital apex. If the frontal sinus is entered it isexenterated, the mucosa rolled, and impacted into the ostium, anda flap of pericranium sutured to the dura for repair. The frontalis

nerve is visualized through the thin periorbita after the orbit isunroofed. Prior to the epidural approach, an intradural inspectionof the optic nerve allows opening of the chiasmatic cistern, releaseof cerebrospinal fluid with emptying of the ventricles, and thedevelopment of an adequate operative field.

scalp margin for hemostasis and the scalp and galea arereflected together over a flap roll. The subgaleal dissec-tion is brought no closer than 1 cm from the orbital rim tominimize postoperative lid edema. Pericranium,temporalis muscle, and fascia are then incised with a cut-ting needle. A four-hole osteoplastic craniotomy is startedwith a performator and a burr and interconnected with aGigli saw. The anterior medial trephination is made justat the top of the frontal sinus and an attempt is made tokeep the mucosa intact when the sinus is unroofed. Theanterior lateral trephination is made just lateral to the tem-poral ridge. Tenting sutures are placed posteriorly andlaterally but not anteriorly, as an epidural approach willlater be made to the anterior fossa. Gelfoam is placed atthe dural margin.

If the frontal sinus is entered it must be carefully re-paired by exenterating the mucosa and rolling and im-pacting it into the ostium. A small piece of muscle is placedover this, and Gelfoam is used to cover this. A flap ofpericranium is then pulled downward and sutured to thedura to secure the repair.

A linear horizontal incision is made in the dura, andusing a small malleable retractor the perichiasmatic cis-tern is visualized and entered to allow escape of cere-brospinal fluid. This is removed by suction over cottonoidand gentle retraction applied over a protective sponge(Bicol) until an adequate field is developed.

This initial intradural approach allows for both in-spection of the intracranial optic nerve and identificationof the site of the optic canal which is difficult tolocate from the epidural approach when unroofing the


9


Figure 7. When the periorbita is opened medial to the frontalisnerve and Ievator muscle, a direct trajectory is made to the tumor-bearing optic nerve. Malleable refractors and cottonoids are used

to hold the field. The fine neural structures within the orbit areprotected by the orbital fat and no attempt should be made todissect orbital fat.

orbit. If the optic nerve does not appear abnormal, sec-tioning of the optic nerve would be deferred until tumorpathology is proved at the time of orbital exploration. Inthe more usual instance where the intracranial optic nerveis grossly abnormal, it is simply cut with scissors at thechiasmal margin perpendicular to its long axis. There isno virtue in leaving a stump of optic nerve because tumorcells may remain within it.

Once these steps have been accomplished, the retrac-tor is withdrawn and an epidural approach made along thefloor of the anterior fossa. Cottonoids are used to hold thefield and the retractor is advanced to the orbital apex. Or-bital unroofing is started with a diamond high-speed burrand enlarged with small double-action mastoid or Leksellrongeurs and orbital micropunches. These miniaturizedrongeurs are similar to Kerrison rongeurs but are scaled

down in size to accommodate the eggshell-thin bone of theorbital roof and optic canal. At times it is necessary torevisualized the intradural optic nerve in order to extendthe epidural orbital unroofing into the optic canal.

After the orbit and optic canal are unroofed and theperiorbita exposed, the whitish frontalis nerve can be seenas the outstanding landmark showing the location of the leva-tor and superior rectus muscles. The latter are usuallyblanched by orbital pathology and are difficult to visualize.

The periorbita must be incised medial to the leva-tor and superior rectus muscle complex, followingwhich an approach is made directly to the optic nervethrough orbital fat (Fig. 7). No attempt must be madeto dissect the fine structures protected by the orbitalfat. Indeed, small cottonoids are used to develop andhold the surgical field at the tumor capsule; three


10


Figure 8. Once the tumor is identified, three malleable refractors are required to hold the field.

Figure 9. Thetumor-bearing optic nerve is then doubly clamped with long mosquito forceps just behind the globe.



11


Figure 10. The optic nerve is sectioned between the forcepsand the proximal clamp is left as a handle for further dissection

of the apical portion of the optic nerve and tumor.

narrow malleable refractors are bent to retract the levatorand superior rectus muscles laterally, the medial rectusand superior oblique muscles medially, and orbital fat an-teriorly toward the globe (Fig. 8).

If the tumor is large, the capsule maybe incised andthe tumor gutted to allow collapse and delivery throughthe relatively small orbital roof defect. In smaller tumorsthe tumor-bearing optic nerve is doubly clamped directlybehind the globe and sectioned between clamps (Fig. 9).This spares injury to the sclera. The proximal clamp isthen left on the cut optic nerve and is used as a handleand attention is directed toward the apical region (Fig. 10).

In young children, where the structures are small anddissection difficult, it is sufficient to section the apicaloptic nerve at the annulus and then to resect the opticnerve intracranially at the cranial end of the optic canaland remove the tumor in two pieces. The intracanalicularportion may be simply coagulated over a nerve hook with-out fear of recurrence. This is entirely unacceptable tech-nique when dealing with optic nerve sheath meningiomabecause the intracanalicular residual may recur.

In order to remove the optic nerve bearing tumor in

one piece from the globe to the chiasmal margin, it is neces-sary to continue the linear periorbital incision through thecanalicular dura (Fig. 11A); in so doing, the levator origin issectioned because of its medial origin at the annulus (Fig.11B). The optic nerve is then sharp-dissected from its tetherat the annulus of Zinn and the specimen is removed in onepiece. If there is bleeding from the ophthalmic artery at theoptic canal it is simply bipolar coagulated.

It is imperative that the levator origin be reattached tothe annulus with a fine figure of eight suture (5-0 or 6-0)(Fig. 11C). The fourth nerve may or may not be visualizedbut cannot be spared, and, as indicated, there is little func-tional or cosmetic consequence of fourth nerve section ina blind eye. By sweeping orbital fat directly on the tumorcapsule and protecting it with cottonoids, the medial ap-proach to the optic nerve is relatively safe and affordsaccess without injury to the nerve supply to the levator orsuperior rectus muscles above or the inferior and medialrectus muscles below the optic nerve.

The cruciate incision in the periorbita, is closed withone or several fine sutures and Gelfoam is placed over theroof defect. Although perhaps not mandatory,


12


Figure 11. A, the periorbital incision is carried directly throughthe apical dura, annulus of Zinn, and intracanalicular dura. If notalready sectioned at the chiasm, the optic nerve is once againvisualized intradurally and sectioned with long fine scissors close tothe chiasm. B, in doing so, the origin of the levator muscle is

sectioned. At this point the entire nerve may be sharp-dissectedfree from the region of the annulus and removed as one specimen.C, once the tumor has been removed, the levator muscle must besutured to its origin at the annulus and the periorbita closed withone or more fine sutures.


13


titanium mesh screen is used to bridge this defect to pre-vent the possibility of pulsating proptosis. This does notinterfere with magnetic resonance imaging or CT scan-ning. It is extremely important that the orbital roof defectbe simply bridged and not totally covered, and that themesh must be bent to reform the arched roof of the orbit.Any flat repair, as with acrylic, will reduce the orbital vol-ume and result in postoperative proptosis. The orbitalbridge need not be sutured in place; it is simply coveredwith Gelfoam. The reexpanding brain will reposition thedura against it and hold it in place. The dura is then closedin a watertight fashion using 4-0 silk. Gelfoam is placedover the dural suture line, a single epidural Hemovac drainis placed through a posterior trephination, and separatebilateral subgaleal drains are placed through separate stabwounds at the temporal regions bilaterally. Titanium meshbuttons are used to cover the medial-most trephinations.If the temporalis muscle and fascia are meticulously closedat the temporal ridge, no artificial covering is needed forthe anterolateral trephination. The pericranium is usuallyclosed with interrupted 3-0 Vicryl; 3-0 Vicryl is also usedto close the galea and interrupted 4-0 silk is used to closethe scalp. A firm dressing is applied and should not beremoved for four or five days to prevent the collection ofsubgaleal fluid in the frontal region.

After a head dressing is applied it is wise to perform

a temporary tarsorrhaphy to allow the placement of a sepa-rate pressure dressing over the eye. Should there be markedswelling or postoperative hemorrhage within the orbit, thecornea under pressure is endangered unless the lids areclosed in this fashion. The pressure dressing, which shouldbe removed and replaced each day, is maintained until thethird or fourth day. Once swelling has subsided, the tempo-rary tarsorrhaphy may be removed. Perioperative and post-operative anticonvulsant therapy is used for three or fourmonths after this exploratory procedure.

SUMMARYWhen optic nerve resection is indicated for tumor, thetranscranial approach should be the primary approach toavoid leaving residual tumor at the apex and within theoptic canal. The primary objective of transcranial orbitalexploration for optic nerve glioma is resection of tumor withmaintenance of the globe and with an excellent cosmeticresult. Complications may be minimized by being familiarwith the regional anatomy, by approaching the optic nervebetween the levator and superior oblique muscles, by avoid-ance of dissection of orbital fat and its contents, by devel-oping the field directly on the tumor capsule in the orbit,and by sectioning and resuturing the origin of the levatormuscle when the annulus of Zinn must be opened to re-move the tumor in one specimen from globe to chiasm.

141

FIBROUS DYSPLASIA INVOLVINGTHE CRANIOFACIAL SKELETON

JAMES T. GOODRICH, M.D., PH.D.CRAIG D. HALL, M.D.

INTRODUCTIONThis chapter will deal with fibrous dysplasia of the cranio-facial complex, in particular the forehead, orbital rim, lat-eral and medial orbital walls, the orbital roof, and the opticforamen. The discussion will involve the “worst case sce-nario,” assuming that, if the surgeon can handle this typeof case, the simpler cases will be easier to treat.

Fibrous dysplasia can involve the calvarium and anyof the upper facial bones. Its etiology is unknown but thepathology involves a replacement of normal bone with afibro-osseous matrix. The surgical principle involves re-moving all of the dysplastic bone (or as much as possible)and replacing it with normal calvarial bone harvested fromother parts of the head. Fibrous dysplasia. can be of asimple type called monostotic, where only one bone unitis involved, or polyostotic, where two or more bones areinvolved. In this chapter we will deal with the more com-plicated polyostotic type.

The most common presenting complaints in fibrousdysplasia of the craniofacial complex are proptosis (Figs.1 and 2), diplopia and headaches, and, in severe cases,progressive blindness due to optic nerve compression.

An x-ray film of the skull will show a sclerotic massexpanding the calvarial and orbital bones. The radiolo-gists typically describe a “ground glass” appearance.There will also be sclerosis or even a cystic appearance tothe bone. It is not uncommon to see complete obliterationof the frontal and nasal sinuses. The proptosis is second-ary to the orbital fibrous dysplasia compressing the globeand forcing the eye forward. As a result of this, an earlypresenting complaint can be diplopia.

The principle behind the surgical treatment of fi-brous dysplasia of the craniofacial complex is three-fold: 1) Since neural compression is common, particu-larly of the optic nerve, decompression of the nerve is


essential. 2) Removal of all dysplastic bone is essential, asany residual can form a new dysplastic center. 3) Use ofthe patient’s own bone for grafts to achieve a satisfactorycosmetic result is preferred.

At the Montefiore Medical Center we have elected todo the reconstruction with calvarial bone which is mem-branous, because we have found that this significantlylessens the risk of resorption which occasionally occurswith rib (endochrondral bone) grafts placed in the cranio-facial region. Another advantage of using calvarial boneis the reduction in operative exposure. This techniquealso avoids the complications that can occur with rib har-vesting, such as pneumothorax and chest wall pain.

PREOPERATIVE EVALUATIONAll patients should have x-ray films of the skull in theroutine views to document the extent of dysplastic in-volvement of the skull and surrounding orbital and nasalstructures. Computed tomography scanning with bonewindows in the axial and coronal views is also performed.If three-dimensional reconstruction is available, it can beextremely helpful in determining preoperatively the amountof bony removal that will be required. We have not foundmagnetic resonance imaging to be helpful, so we do notuse it routinely.

If the optic nerve is compressed, we routinely do vi-sual acuity and visual field testing to have baseline val-ues. Damage to the optic apparatus and to the nervessupplying the extraocular muscles are the most signifi-cant complications to be avoided. Subtle damage may al-ready have occurred preoperatively, and it is best to docu-ment this prior to any surgical intervention.

Since an extensive resection can involve the fron-tal and paranasal sinuses we culture the nasal pas-sageways to look for virulent organisms. If any aredetected, the patient is placed on an appropriate anti-biotic coverage 24 hours before surgery. We routinelystart an anti-staphylococcal antibiotic at the time of

15

Figures 1 and 2. Frontal and lateral view of a patient with orbitalproptosis secondary to fibrous dysplasia. Typical proptosis is evi-dent with fibrous dysplasia involving the right orbital unit including

rim, lateral, and medial walls. As a result, the eye is pushed forwardand downward. Interestingly, the only visual symptom was doublevision; the visual acuity was normal.

anesthetic induction in the operating room. Because thesurgical manipulations are extradural, we do not routinelyuse anticonvulsant medications.

PREPARATION FOR OPERATIONFibrous dysplastic bone can be and usually is, highlyvascular. As a result, the blood loss in these procedurescan be quite high. We routinely plan for a blood loss of 3to 5 units. If the family is cooperative, we ask for pedigreeblood donations from the family members one week inadvance. If available, a “cell saver” unit can rescue up to50% of the patient’s lost blood volume. Because of therisk of extensive blood loss, all patients require at leasttwo large-bore intravenous lines of 16 gauge or larger. Ifthere is any history of cardiac or pulmonary problems, weroutinely put in a central venous pressure line. An arterialline is mandatory for monitoring blood gases, hematocrit,electrolytes, etc. during the procedure. We request anosmostic diuresis, usually with mannitol (0.5 g/kg), at thetime of anesthetic induction. A spinal drainage system isplaced in the lumbar region to assist in cerebrospinal fluid(CSF) withdrawal and brain relaxation. Because of the ex-tensive exposure and brain relaxation that will be needed(remembering that surgical exposure back to the optic fo-ramen is often necessary), every effort at relaxation mustbe done to reduce retraction pressure on the frontal lobes.A simple removal of 35 to 50 ml of CSF can cause a dra-matic relaxation of the frontal lobes.

The use of steroids is always an issue in these typesof cases. On our service we do not routinely use ste-

roids as part of our preoperative management. If there isevidence of postoperative brain or optic nerve edema, thepatient will be placed on dexamethasone at that time.

OPERATIVE POSITIONINGThe patient is placed in the supine position with the headresting on a cerebellar (horseshoe) headrest (Fig. 3). Thehead is placed in a slightly extended, brow-up position.Rigid fixation devices like a Mayfield clamp are specifi-cally avoided, as the surgeon will need to move the head(usually never more than 10 to 15 degrees); this flexibilitycan prove to be very useful.

We also reverse the table so that the head of thepatient is at the foot end of the operating table. This al-lows the surgeon and his assistant to sit with their kneescomfortably under the table and not obstructed by thetable pedestal or foot unit.

Anesthesia equipment is placed on the side oppositethe lesion and parallel to the table. Routine orotracheal intu-bation is performed. All lines are run off to the side of theanesthesia unit. The operating surgeon is placed at thehead of the patient with the assistant to the side. The nursecomes in over the patient’s abdomen but is positioned nohigher than the mid-thoracic region. This allows the sur-geon to be able to move around to see the patient’s facefully for cosmetic evaluation. For this reason we also avoidthe use of bulky overhead tables, such as the Fallon table.

AR of the patients have bilateral tarsorrhaphies priorto formal draping. This prevents unintentional injury tothe globe and cornea.

GOODRICH AND HALL : FIBROUS DYSPLASIA INVOLVING THE CRANIOFACIAL SKELETON


16


Figure 3. Schematic showing the location of the surgical and anesthesia teams.


17

SURGICAL DRAPING TECHNIQUESThe head is fully shaved and draped for a bicoronal inci-sion. In addition, both eyes with their tarsorrhaphies mustbe visible. The facial drape is placed over the nose andnares, but well below the lower orbital rim. This allows theeyes to be visualized during the reconstruction. The restof the draping can be done according to the surgeon’spreference. An important additional point is to keep thedrapes reasonably loose, so that the head can be moved.

We routinely run all our suction lines, cautery cords,etc. past the foot of the patient. As both surgeons aresitting, this allows easy mobility of the chair; i.e., they arenot rolling over the cords and tubes.

Because the operative site is usually copiously irri-gated during the procedure, it is important to have water-proof outer drapes. Some of the newer drapes have largeplastic bags for fluid collection; we have found these tobe quite useful.

OPERATIVE TECHNIQUE

Skin IncisionBecause an extensive exposure of the calvarium and or-bits is required, we routinely use a bicoronal incision fromtragus to tragus. The incision must be well behind thehairline, both for cosmetic closure and to allow for a largepericranial flap that can be used in the subsequent repair.

Flap ElevationA full thickness flap is turned following the standardsubgaleal plane. It is important to leave the pericraniumintact. This is then elevated as a second, separate layer.The flaps are carried down to the orbital rim to the level ofthe supraorbital nerve and artery. These are frequentlyencased in a small notch of bone. This notch can be openedwith a small Kerrison rongeur or osteotome. It is easier toelevate the artery and nerve with the pericranial layer. It isimportant to preserve these structures or t here will beanesthesia in the forehead postoperatively. The flap mustalso expose the entire belly of the temporalis muscle andthe zygomatic arch. In the midportion of the face the nasalsuture should be fully exposed. Using the small periostealdissector or a Penfield dissector it is possible to comeunder the orbital rim and dissect it safely back approxi-mately I to 2 cm. The temporalis muscle has to be elevatedas a unit. Starting at its squamosal insertion, it is elevatedusing a Bovie electrocautery with a fine needle tip. Thedissection is carried out in such a fashion that thetemporalis muscle will be elevated from the zygoma backto the ear, fully exposing the pterional “keyhole.”

CraniotomyThe craniotomy is carried out to incorporate all of thedysplastic bone in the removal. It is easiest to do thefrontal craniotomy by first taking out a forehead bone flapthat encompasses as much of the forehead dysplasia aspossible (labeled A in Fig. 4). This provides the windowwhich will allow exposure to the orbital roof and walls. Weprefer to use a high-speed drill system with a craniotome(e.g., Midas Rex with a B-1 footplate) as this gives a speedybone removal, thereby decreasing blood loss. We nextelevate the frontal lobe with gentle retraction to see howfar into the orbital roof the dysplastic bone extends. Then,by further dissecting under the orbital roof, the dysplasticportion can be completely visualized (Fig. 5). There is usu-ally extensive blood supply crossing these planes, so thebleeding can be quite copious. Keep plenty of Avitineand Gelfoam available for packing in these spaces to con-trol the oozing. Once the limits of the dysplastic bonehave been determined and the brain is adequately relaxedand retracted, we proceed with the bone resection. Usinga combination of osteotomes and a small cutting burr, likethe Midas Rex C-1 attachment, the roof is removed as aunit (Fig. 5). It is helpful to have the assistant place amalleable retractor under the orbital roof. This will preventthe drill or osteotome from damaging the orbital contents.On occasion, the dysplasia can go back to the clinoidsand orbital foramen. In these cases, the entire roof mustbe removed (Fig. 5). A small diamond burr on a highspeeddrill unit is the best method for removing this part of thebone. It allows the surgeon to remove the bone withoutinjury to the underlying structures. Once this is completed,attention is turned to the lateral orbital wall and zygoma(labeled B in Fig. 4). This portion of the procedure can bedone quite easily. The only important points are to haveadequate exposure of the zygomatic arch and a good dis-section of the orbit. The lateral canthal ligament must besectioned and then reattached at the end of the proce-dure. Doing this prior to the medial part will allow easymobilization of the eye and surrounding structures withminimal trauma.

Next, attention is turned to the most difficult phase-resecting the medial nasal structures (labeled C in Fig. 4).By removing the orbital roof and lateral orbital wall, thesurgeon now has some mobility and freedom in moving theeye. If the dysplastic bone involves the nasal bone andmedial orbital wall, the medial canthal ligament must be cut.The assistant then retracts the eye laterally, and the bone isremoved with an osteotome and fine cutting burr. The fron-tal sinuses are usually occluded with bone, which can com-plicate matters. If the sinuses are not occluded, thefrontal sinus can be entered and used as an operating


18


space within which to work. Once all the dysplastic boneis removed, the reconstruction is started.

Calvarial Bone HarvestingBy using a bicoronal skin flap, a large amount of normalcalvarial bone is exposed. Once the surgeon has resected thedysplastic bone and determined how much bone is neededto reconstruct the defect, a craniotomy is performed on theopposite calvarium (labeled D in Fig. 4). Remember that the

most useful bone is over the convexity, where the diploë iswell formed. In the squamosal area, the bone thins out and ishard to split. The bone is taken to a sterile table set up next tothe operating field. Using a combination of small osteotomes,a fine cutting tip like a Midas Rex C-1, and a reciprocatingsaw, the bone is split along the diploic space. Copious irriga-tion is essential, because the bone must not be allowed toheat up; this would lead to dead bone and subsequent ne-crosis. Once the bone has been split, the inner table of the

Figure 4. Frontal view showing four-piece bone removal. A, frontalbone maximally involved with fibrous dysplasia; B, lateral orbital wall;

C, medial orbital wall. The orbital wall roof which is also removed isnot shown in this drawing. D, graft from the opposite calvarium.


19

calvarium is placed back in the harvest site. The outertable, because of its smooth contours, is used as the re-constructing bone.

Crainiofacial ReconstructionThe reconstruction is done in the reverse order fromthe resection. The medial orbital wall is constructed

first and wired or plated into position (labeled C in Fig.6). The nasal bone and cribriform plate are usually themost solid structures to work with. The medial canthalligament also has to be reattached, which can be doneeasily through a small drill hole. Next, a piece of bone isfashioned to form the orbital roof. This is an importantstructure which must be solidly placed (Fig.

Figure 5. Schematic drawing showing the frontal fossa after removal of the dysplastic orbital roof anddecompression of the optic nerve.



20


8). If it is not, subsequent proptosis of the eye canoccur due to downward pressure of the frontal lobe.The bone used to reconstruct the lateral orbital wall isattached to the roof with either wires or miniplates(labeled A in Fig. 6). The squamosal portion of thetemporal bone can also act as an excellent place toanchor this bone. The orbital rim is then fashioned

and attached medially to the nasal unit and oppositeorbital rim (labeled B in Fig. 6). This is the key cos-metic unit and must be perfectly placed to avoid facialasymmetry. The rest of the craniotomy is then closedin a mosaic fashion using the remaining pieces of bone.Miniplates have proved to be extremely useful in sta-bilizing these various bone units.

Figures 6 and 7. Schematic drawing and intraoperative view show-ing bone grafts in position.


21

Figure 8. Schematic drawing showing the split thickness calvarial bone graft in position in the orbitalroof region.



22

Repair of Frontal SinusOne of the most devastating postoperative complicationsis infection arising from the sinus. If the frontal sinus isnot obliterated by dysplasic bone, it must be cleaned andexenterated of mucosal lining. We routinely cover the si-nus with the pericranial flap to isolate it from the epiduralspace. The same principle applies to the other paranasalsinuses if they are violated.

Pericranial TissueThe pericranial tissue is a most useful repair structure. Itnot only provides additional vascularity to the bone but italso helps smooth out the rough contours of the bonethat has been harvested and used as grafts. Therefore, wemake every effort to use this structure and place it backinto its natural anatomical position

Temporalis MuscleTo prevent a postoperative depressed concavity over thetemporal unit, the temporalis muscle is laid back into posi-tion. Sometimes a relaxing incision must be made posteri-orly to allow the muscle to be advanced forward to coverthe keyhole and to be reattached to the zygoma. This iscritical or there will be a significant depression over thisregion postoperatively.

ClosureThe closure is done in a routine fashion. Hemostasismust be meticulous because of the amount of dead

space that can form. A drain to light suction is placed forat least 24 to 48 hours. A fluid collection next to sinusspaces can lead to a devastating postoperative infection.Scalp closure is done in a routine fashion closing both thesubgaleal and skin layers.

POSTOPERATIVE CAREWe routinely place the patients on antibiotics to coverskin organisms and possible nasal contaminants for atleast 72 hours. The risk of osteomyelitis is high and canbe quite devastating to the patient, so every attemptmust be made to avoid it. There may be significant peri-orbital swelling postoperatively; ice packs are applied tothe eye and periorbital regions for symptomatic relief. Ifthere is significant swelling at the end of the operation,we ordinarily leave the tarsorrhaphy in place for abouttwo days. Intensive care for at least 48 hours is manda-tory with close monitoring for hemodynamic changesfrom excessive blood loss and for the development of anepidural hematoma.

The surgeon must always be attentive to postopera-tive CSF leaks. If any dural tears have occurred they mustbe repaired meticulously. Should a postoperative CSF leakoccur, then placement of a lumbar CSF drain may be nec-essary to divert the fluid. These usually need to be left inplace for five to seven days. However, close attention todural tears and verifying dural integrity by asking the an-esthesiologist to perform Valsalva’s maneuver at the endof the case should prevent this problem from occurring.


23

DEPRESSED SKULL FRACTUREIN ADULTS

FRED H. GEISLER, M.D., PH.D.

25

DEFINITIONDepressed skull fractures may occur after head trauma.By definition, a depressed skull fracture is a fracture inwhich the outer table of one or more of the fracture edgesis below the level of the inner table of the surroundingintact skull. The greatest bone depression can occur ei-ther at the interface of the fracture with the intact skull ornear the center of several fracture fragments that are dis-placed inward. A depressed fracture results when the im-pact energy is applied over a small contact area. Typicalexamples include: assaults with a hammer, club, or pipe;sports injuries caused by a hockey stick, golf club, or golfball; or a motor vehicle accident in which the victim’s headstrikes either the interior of the car or an object outside thecar as he/she is thrown from the vehicle.

PREOPERATIVE EVALUATIONMany patients with depressed skull fractures experienceinitial loss of consciousness and varying degrees of neu-rologic deficit. Recovery is determined by the extent ofthe brain injury. However, one quarter of the patients ex-perience neither loss of consciousness nor neurologicdeficit and another quarter experience only a brief loss ofconsciousness. The diagnosis of a depressed skull frac-ture is often made on routine skull films by noting eitheran area of double density (indicative of overlying bonefragments) or the presence of comminuted or circular frac-tures. However, the full extent and the depth of the frac-tured fragments are rarely appreciated on these studies.Physical examination of patients with depressed skull frac-tures is difficult because of scalp mobility and swelling.Scalp mobility can result in nonalignment of the scalplaceration and the depressed skull fracture. Traumatic scalpswelling obscures the step-off at the bony edges, pre-venting accurate assessment of the extent of skull defor-mity for the first few days.

A computed tomographic (CT) head scan is the di-agnostic method of choice. When the image display

windows are adjusted to optimize bony detail, they dis-play the position, extent, and number of fractures as wellas the presence and depth of depression. With the imag-ing windows set to optimize intracranial contents, thesame CT scan also allows an assessment of the underly-ing brain for contusion or hematoma from superficialpenetration of bone fragments, small bone fragments orforeign bodies within the brain substance, and other trau-matic intracranial lesions. Occasionally, coronal CT im-ages through fractures near the vertex of the head orextending into the skull base are used to supplement thestandard CT images because the depth of a depressionis more accurately measured on CT images perpendicu-lar to the depression. Differences between the amount ofdural laceration and cortical damage among similar CTscan images occur because the position of the brain,bone fragments, and remaining intact skull at the time ofthe scan may differ from the actual maximal depth of thefragments at the time of impact. The brain, for example,may undergo linear or rotational movement and the skullcan deform temporarily during the impact. Immediatelyafter the injury, the depressed bone fragments tend torebound partially or spring back as the deformed skullattempts to resume its anatomic shape. Also, only crudeestimates of the sharpness of the bone edges can bemade from the CT images because the limit of resolutionis 1 mm.

RATIONALE FOR SURGICAL TREATMENTAlthough a focal neurologic deficit from the altered cortexdirectly under a depressed bone fracture occasionally im-proves after elevation of the bone fragments (presumablyby increasing local cortical blood flow), this procedure usu-ally produces no neurologic change, implying that corticaldamage is inflicted at the time of impact. The brain dysfunc-tion generally undergoes a neurologic recovery phase ofseveral weeks to months, similar to that following a strokeor a head injury without a depressed fracture. The inci-dence of epilepsy following a depressed skull fracture isapparently determined by the cortical damage at the© 1991 The American Association of Neurological Surgeons

24


Figure 1. Three different positions used in the management ofpatients with depressed skull fractures are illustrated along with adiagrammatic representation of CT images through the depressedskull fracture. The position is individualized for each patient withthe majority of the surgical area at or near the horizontal plane. Theskin incision and location of the midline are marked before draping.A, a patient with an anterior frontal depressed skull fracture posi-tioned supine in the horseshoe head holder with the neck flexed 10degrees in preparation for a bicoronal scalp incision. B, a patientwith a posterior frontal depressed skull fracture held in the three-

point pin skull holder positioned with the head rotated to place themajority of the depressed skull fracture near the horizontal plane.Note the pad under the patient’s right shoulder to lessen the rotationof the neck and possible cerebral venous outflow impairment. C, apatient with a posterior parietal/occipital depressed skull fractureheld in the three-point pin skull holder in the prone position. A 10-to 15-degree rotation of the head can be used to place the surgicalarea exactly in the horizontal plane. If the neck is rotated, referenceto the skin mark at the midline prevents inadvertent extension ofthe exploration to encroach or cross the midline.


25

GEISLER : DEPRESSED SKULL FRACTURE

time of impact since it is not altered by the elevation of thefragments. Thus, the treatment of depressed skull frac-tures is based not on initiating neurologic recovery orpreventing epilepsy but, rather, on correction of a cos-metic deformity as well as on prevention of infection inopen fractures. The treatment of an individual depressedskull fracture depends on the presence or absence of: 1)cosmetic deformity and its extent; 2) scalp laceration; 3)dural laceration; 4) contusion or laceration of the underly-ing brain; 5) extension of the fracture over a venous orparanasal sinus; and 6) coexistence of other traumatic in-tracranial lesions, including epidural, subdural, and in-tracerebral hematomas, cerebral contusions, midline shift,and ventricular compression.

In closed depressed fractures, the indication for sur-gery is usually cosmetic, with the procedure performed onan elective basis in the first few days after the trauma whenthe patient is cleared for elective anesthesia. The locationof the depression, thickness of the scalp, and patient’s bodyimage perspective are critical to the surgical decision. Theforehead is the area where the greatest cosmetic deformityoccurs. Depressions of 3 mm or more result in deformitiesfor which most patients request correction. When the or-bital rim is involved or the scalp is thin, repair of smallerdepressions is often requested. Exploration is more urgentin a patient with a large closed depressed fracture where theradiologic appearance suggests dural laceration, brain pen-etration, mass effect, or underlying epidural, subdural, orintracerebral hematoma. The hematoma is evacuated, thedura is repaired, and the bone fragments are replaced andwired in anatomic position.

A compound depressed fracture represents a neu-rosurgical emergency because of the risk of bacterialinfection of the contents of the cranial cavity. The initialsurgery is performed within 24 hours and usually withinthe first 12 hours after the accident. The major objectivesof the surgery are to: 1) remove contaminated bone frag-ments and foreign material (hair, cloth, dirt, etc.) in thescalp wound, between the bone fragments, and in thecortex; 2) debride devitalized scalp, dura, and brain; and3) provide a watertight closure of the dura. Removingdevitalized tissues and contaminated material is essen-tial to reduce the incidence of infection. Often foreignmaterial or hair wedged between bone fragments is notvisualized through the overlying scalp incision. Thus,simple wound irrigation and closure may be inadequatefor the debridement of the foreign material. Dural closureis essential to prevent the leakage of cerebrospinal fluidfrom the wound and brain herniation into the fracturearea. Dural closure also acts as a bacterial barrier, pre-venting the intracranial spread of infection from the scalp.

Cosmetic correction is performed during the initial sur-gery only if considered safe; otherwise, a cranial defectis left and the cosmetic repair is performed later. Themajor reasons to postpone the cosmetic repair are pre-clusion of additional anesthesia by major head injury ormultitrauma, gross contamination of wounds where thebone fragments cannot be adequately cleaned, and asurgical delay of more than 24 hours. Both the final neu-rologic deficit and the incidence of epilepsy are de-creased by this management plan because it reduces thepotential for intracranial infection and intracerebral he-matoma.

OTHER PREOPERATIVE CONSIDERATIONS Preopera-tively, patients receive a single dose of steroids (methyl-prednisolone sodium succinate, 250 mg, I.V.); no antibioticsare used in open fractures, whereas one dose of a prophy-lactic antibiotic (such as nafcillin sodium, 1 g, I.V., or vanco-mycin hydrochloride, 500 mg, I.V.) is used just prior to skinincision in closed fractures. Grossly contaminated materialin the scalp wounds and representative bone fragmentsrecovered from the cortex are sent for culture in appropriatemedium. The culture results are used to guide antibiotictherapy if an infection occurs following thorough debride-ment and irrigation during the surgery.

SURGICAL TECHNIQUEThe patient’s head is held in place either by the Mayfieldthree-point head holder or the Mayfield horseshoe. Whenusing the three-point head holder, the surgeon must avoidapplying the pins at the fracture sites. If a pin is posi-tioned into a fracture line, it will not only be mechanicallyunstable but may even extend the fracture or cause dis-placement of fragments as the head holder is applied. Thepatient is positioned with the depression in a horizontalplane. Thus, patients with depressions on the foreheadare positioned supine, whereas those with depressions inthe occipital area are positioned prone (Fig. 1). The drap-ing should allow for extension of the incision if additionalexposure is required during the exploration. An exampleof the surgical treatment of an open depressed skull frac-ture with intact dura. and a bone cranioplasty is shown inFigure 2, A-E, and with dural penetration and cortical lac-eration in Figure 3, A-K.

In an elective elevation of a closed depressed frac-ture for cosmetic improvement, the surgeon can useone of several standard scalp incisions behind thehairline. The exact size and placement is determinedby: 1) exposing the intact skull for at least 2 cmcircumferentially around the depressed fragment, in-cluding the larger inner table splinters; and 2) planning to

26



27

Figure 2. The treatment of an open depressed skull fracture withintact dura is shown here. In this example, the large bone fragmentswere reassembled and used in a bone cranioplasty. A, a right frontalscalp wound with an underlying depressed fracture is present. Thelocation of the bicoronal scalp incision is noted in the figure. B, notethe continuation of the scalp wound on the inside of the scalp flapand multiple small pieces of fractured bone in the superior-posteriorarea of the depressed skull fracture being debrided in this example. C,the large bone fragments are removed using gentle force in an out-ward direction without angulation. Angulation of the fragments orthe application of force while removing them can lever them into orpull them through the dura or cortex. A sharp fragment of innertable not in the visual field can cause either a new or extension of theoriginal dural or cortical laceration if not removed gently. Note thatthe large fragments are numbered and kept in relative position sothat they can be easily reassembled in their anatomic position. If thefragments cannot be easily removed or if no portion of the fracture

exposes the dura after the loose fragments are debrided, then a burrhole and circumferential slot craniectomy technique is used, as illus-trated in Figure 3. D, the bone fragments are reassembled and firmlywired together in anatomtc position to form a bone cranioplasty.These wire twists are all bent flat against the inner table. Note thedural tacking sutures circumferentially fixing the dura to the intactskull. The bone assembly is fixed to the skull with circumferentialwires. E, the wires fixing the bone assembly to the skull are twistedand bent to lie between the bone assembly and the intact skull. Thepositions of the twisted wires prevent all sharp ends from inadvert-ently poking the inner aspect of the scalp in normal relative scalpand skull motion with its unpleasant sensation. At this point, thescalp laceration is debrided to normal appearing and vascularizededges and then the galeal layer is sutured from inside. The scalpincision is closed in two layers. Finally, the superficial portion of thescalp laceration is further debrided if necessary and closed with cuta-neous sutures.


28



29

Figure 3. This illustrates the treatment of an open depressedskull fracture with dural penetration and brain laceration treatedwith surgical intervention that includes dural and cortical debride-ment, dural repair with a graft, and craniectomy. A, the scalpdebridement lines are centered on the scalp closure line to pro-vide both adequate initial debridement and to allow wound closureafter scalp mobilization with minimal tension. B, an initial burrhole is placed slightly beyond the edge of the depressed skullfracture to expose normal dura and allow identification of thenormal anatomic relationship of inner table contacting dura. C,the burr hole is enlarged toward the edge of the depressed skullfracture and then extended into a slot craniectomy startingcircumferentially in both directions around the edge of the de-pressed skull fracture as visualized on the outer table. Note the

separation of the bone fragments from the dura as the slot craniec-tomy is proceeding. This prevents inadvertent dural laceration ordebridement with the Kerrison punch at the edge of the craniec-tomy. The central impacted bone fragments are sequentially un-locked by this procedure. D, the bone fragments are locked to-gether in this example because the hard outer table of one fragmentwas impacted into the soft diploic space of adjacent fragments.Note hair wedged between the bone fragments. The sharp edge ofa fracture fragment has lacerated the underlying dura and corticalsurface. E, the depressed fragments that have been unlocked areremoved with a gently outward nontwisting motion. F, the dura isdebrided to remove shredded and contaminated dura near thedural laceration. Note the small blood clot coating the corticalsurface indicating an underlying cortical laceration.


30



31

Figure 3. (continued) G, the inferior end of the scalp incision isextended to expose the temporalis fascia for harvesting a portion asa dural graft. H, the pulped and contaminated cerebral cortex isdebrided gently with normal saline irrigation and suction. Hemosta-sis is obtained with bipolar coagulation on the cortex. I, the dura is

closed watertight with interrupted or running dural sutures. J, afterdural closure, the wound is copiously irrigated with antibiotic solu-tion. K, the scalp wound is closed by manually sliding the skin edgestogether to eliminate the gap produced by the scalp debridement andthen by sewing in two layers with galeal and cutaneous sutures.


32

provide pericranium or temporal fascia for a dural graft incase it is needed to repair the dura. The curvilinear “S”incision (Fig. 3, A) and the flap (Fig. 2, A and B) are the twovarieties of scalp incisions. The S incision is useful in 1- to6-cm diameter depressions behind the hairline because itis easily extended should more exposure be required. Aflap incision must be made of adequate size initially be-cause it cannot be enlarged without adding a “T” to theincision, with its potential healing problems. In cases inwhich an underlying intracranial lesion requires surgicalattention, a large frontotemporal-parietal scalp flap is madethat allows exposure of both the intracranial lesion andthe depressed skull fracture.

The scalp laceration associated with a compound de-pressed skull fracture is usually stellate or contains an areaof contused/devitalized tissue. These areas require debri-dement to normal scalp to prevent breakdown of the cover-ing over the depressed fracture site. If a breakdown occurs,it usually cannot be treated with local care measures; re-moval of the bone, skin debridement/rotation, and delayedcranioplasty may often be necessary. The skin flap isplanned with consideration of scalp closure strategy afterdebridement. The S incision allows local debridement byincreasing the width of its center and then sliding the twosides together (Fig. 3, A and B). Either the flap incision hasto be very large to allow for debridement and suturing in itscenter (for example, a bicoronal skin flap for a foreheadlaceration with the debridement and repair of the foreheadscalp (as shown in Fig. 2, A and B), or else the debrided areamust be on one of the edges so that the two sides of theincision can be slid together. When intact, the pericraniumis carefully opened and saved for harvest later as a duragraft or closure over the fracture site for an additional bar-rier protecting the bone cranioplasty and the brain from ascalp infection. However, if the pericranium is shreddedand or contaminated, then it is debrided.

Once the depressed skull fracture is exposed with amargin of 2 cm circumferentially around it, a decision as tohow to remove the bone fragments is made based on 1)how firmly the bone chips are wedged together and withthe surrounding skull and 2) the amount of angulation ortwisting necessary to remove the individual fragments.The surgeon must consider the likelihood that the visual-ized fragments are only the top portion of a far larger frag-ment with razor-sharp edges already in or directly on topof the cortex.

Occasionally, the fragments are loose and are easilyremoved from the wound in a straight linear fashion out-ward from the brain (Fig. 2, B and C). These fragmentsare not wedged in place and do not tamponade vesselssince they contain no compressive forces. More com-monly, however, the bone fragments are wedged firmly

in place as a result of the mechanical configuration ofthe diploic bone of the skull: the outer table locks eitherinto the diploic space or under the inner table; the innertable can lock under itself (Fig. 3D). The diploic layer issoft and compressible when compared to the corticaltables. During the impact, the diploic layer adjacent tothe fracture edges undergoes compression and remodel-ing, filling the spaces between the locked tables andfirmly fixing them in place. In most cases, a burr hole toestablish the normal skull/dura relationship (Fig. 3C) ismade just adjacent to the depressed area in the outertable away from the areas shown on the CT to have ra-dial splinters of fractures in the inner table and also awayfrom the venous sinuses. The burr hole is then extendedinto a small slot craniectomy with a Kerrison punch (2- or3-mm, 40- or 90-degree upbiting). To accomplish this, thedura is first stripped from the skull for a few millimeters inthe direction of the planned slot craniectomy using asmooth dissector such as a Penfield No. 3 or No. 4. Thenthe overlying bone is removed with the Kerrison punch,taking care not to inadvertently place the lower jaw ofthe Kerrison punch into the subdural space and debridedura (Fig. 3C. The skull/dural interface is carefully fol-lowed to the edge of the fracture of the outer table. Oc-casionally, the skull/dural relationship is initially visual-ized between fracture fragments and the strip craniectomycan be started there. The outer table limit of the fractureis then followed circumferentially in both directions witha Kerrison punch. As this circumferential, thin, ring-shaped craniectomy cut is being completed, pieces ofthe fractured bone will become separated from the otherpieces and the intact skull, allowing atraumatic removal.When the circumferential cut has been completed withthe Kerrison punch where needed and all bone fragmentsin the center of the depressed fracture have been re-moved, the skull edges are all inspected for remainingsplinters of the inner table and dural lacerations. Re-maining splinters of bone under the skull edge are de-brided with a Kerrison punch, reaching under the edge.Care is taken to preserve the largest possible fragmentsof bone and to note carefully the interrelationship of thefragments if a repair of the cranial defect is to be per-formed at the initial procedure (Fig. 2C) This procedureis not necessary if the operative goal is to leave a craniec-tomy, planning a delayed secondary cosmetic repair; inthese cases, the depressed fragments can be reduced insize for atraumatic removal with a rongeur (Leksell orSmithPetersen) or a Kerrison punch.

Depressed skull fractures located over a venous si-nus or extending into the frontal air sinus require specialhandling. The surgical elevation of fractures over avenous sinus may involve massive blood loss if a


33

depressed fragment has been plugging a sinus tear. Un-less grossly contaminated with foreign material, present-ing as a major cosmetic deformity, or causing intracranialhypertension secondary to sinus occlusion, such frac-tures are managed with scalp debridement and copiousirrigation, followed by serial postoperative CT scans forsigns of brain abscess for at least a year. If this fracturerequires debridement and elevation, then four burr holes(two proximal and two distal, one on each side of the si-nus) are made with the fracture area in the center of therectangle and then connected with a cranial saw. The en-tire craniotomy flap with the depressed fracture within isthen elevated. Massive bleeding can occur with this el-evation until proximal and distal control of the sinus isobtained. If the fracture involves the frontal sinus, thesinus is cranialized by first stripping its mucous mem-brane completely and removing the posterior wall of thesinus and then plugging the ostium with the remainingmucous membrane and applying cautery. Bone chips har-vested from intact noncontaminated skull in the peripheryof the surgical field or a muscle graft from the temporalismuscle are then laid over the ostium.

After the bone fragments have been completely re-moved, the dura is fully inspected. If the dura is intact, it isnot opened unless the CT scan indicated an intracerebral orsubdural lesion requiring further operative therapy. Shred-ded dura is debrided (Fig. 3F). Small dural lacerations areenlarged to allow inspection of the cortex. Superficial he-matomas on the cortex and debris are removed with gentleirrigation and suction. Brain that has been pulped by theindriven fragments is removed with gentle suction untilnormal brain is exposed and the fragments and debris areremoved (Fig. 3H). Gentle irrigation with normal saline via a10-ml syringe with a No. 18 angiocatheter with the tip posi-tioned in the depth of a cortical laceration can often deliverdeep bone chips and debris, minimizing the amount of cor-tical resection. Bipolar coagulation is used for hemostasis.The dura is then closed watertight using 4-0 silk or braided

nylon (Nurolon) suture with a dural graft, if necessary, ofpericranium or temporal fascia. After the dura is closed, thewound is irrigated with copious quantities of bacitracin(50,000 units in 1 liter of normal saline irrigation). This anti-biotic irrigation is used for mechanical debridement of theskull and scalp before scalp closure.

In most cases, the large bone fragments are saved forthe cranioplasty at the end of the operation (Fig. 2, C, D,and E). These fragments are first washed in 10% povi-done-iodine (Betadine) solution to remove hair, dirt, for-eign debris, and devitalized tissue and are then rinsedthoroughly in normal saline. They are then reassembledinto the shape of the skull before the fracture to resemblea craniotomy flap. They are wired together with stainlesssteel wire (3-0, No. 24, or No. 26 monofilament surgicalsuture) with the twists on the inside of the bone and flushwith the inner table. Dural tacking sutures of 4-0 silk orbraided nylon are used to hold the dura. to the skull edges.In larger repairs a central dural tacking suture is also used.This is inserted by first making a single pass through thedura directly below two holes in the bone assembly. Thetwo ends of this central dural stitch are then passedthrough these two holes and tied to fix the center of thedura. to the bone assembly after it has been secured infinal position. The reconstructed craniotomy flap is nowwired into place at several circumferential points. The scalpis closed with absorbable suture (2-0 Polyglactin 910) inthe galea layer and skin staples or 3-0 nylon suture toclose the cutaneous layer. A Jackson-Pratt drain is usedin the subgaleal space for 24 hours in those cases wherefracture lines or contused scalp does not allow completehemostasis with cautery to be obtained before closure.Postoperative x-ray films of the skull and a CT scan of thehead with both soft tissue and bone imaging display win-dows are obtained as a base line for comparison with thesame studies obtained several months or years later whichare used for the assessment of bone incorporation/ab-sorption and the cranial shape.


34

CERVICAL HEMILAMINECTOMYFOR EXCISION OF A HERNIATED DISC

ROBERT H. WILKINS, M.D.SARAH J. GASKILL, M.D.

INTRODUCTIONA cervical disc herniation usually occurs in a posterolat-eral direction, compressing the ipsilateral exiting nerveroot (Fig. 1). The most commonly affected cervical disc isthe one between the C-6 and C-7 vertebrae, and this C6-7disc herniation compresses the C-7 nerve root.

The patient with a suspected cervical disc herniationusually presents with pain in the neck that extends intoone upper extremity in a radicular fashion. Loss of neuro-logical function appropriate to the affected nerve root mayalso be present. Such a patient is managed first by rest,analgesics, a muscle relaxant, and/or cervical traction. Theusual criteria for proceeding with further diagnostic testsand therapy are: 1) absence of improvement with the mea-sures just mentioned, 2) significant weakness or markedhypesthesia in an important area (e.g., the dominant thumband index finger), or 3) evidence of myelopathy.

PREOPERATIVE CONSIDERATIONSPlain x-ray films of the cervical spine are helpful in assess-ing the presence and degree of cervical spondylosis and inidentifying another cause of neck and arm pain such as ametastatic carcinoma. In our opinion, magnetic resonanceimaging is not as sensitive in the identification of a cervicaldisc herniation as is cervical myelography followed by com-puted tomography scanning, so we ordinarily obtain thelatter studies to verify the clinical diagnosis.

The usual posterolateral cervical disc herniation canbe exposed either through an anterior approach, whichinvolves removing the intervertebral disc, or through aposterior approach, which involves removing lateral por-tions of -two adjacent laminae and the medial portion ofthe facet joint. If a disc herniation has occurred with adirect posterior (central) vector, and is causing myelopa-thy, the anterior approach is preferred because the sur-geon can remove the herniated disc without manipulating

the spinal cord (and possibly increasing the myelopathy).However, for the more common posterolateral disc hernia-tion that is causing a radiculopathy and no myelopathy,we prefer the posterior approach because of its simplicity:it does not involve risk of injury to the anterior structuresof the neck, such as the esophagus and the ipsilateralrecurrent laryngeal nerve; it does not involve bone graft-ing or a second surgical incision; and, in our hands, ittakes less time than an anterior cervical discectomy andthe patient recovers more quickly.

The posterior approach to a cervical herniated discthrough a partial hemilaminectomy can be performed withthe patient in a prone, lateral, or sitting position. We preferthe sitting position because there is less venous conges-tion, the anatomical alignment of the spine is easy for thesurgeon to visualize mentally after the patient is draped,and the blood and irrigation solution run out of the expo-sure rather than pooling within it. However, when the sit-ting position is used, we think that Doppler monitoringshould be performed for venous air embolism and that anintra-atrial catheter should be placed before the operationis begun, to permit the aspiration of any air that mightenter the right atrium through the venous system. In actu-ality, venous air embolism seldom occurs during opera-tions for cervical disc herniation, although it is commonduring posterior fossa operations in the sitting position.We may be overly cautious in using the Doppler monitorand the intra-atrial catheter.

SURGICAL TECHNIQUEThe patient usually arrives in the anesthesia induc-tion room wearing elastic antiembolism stockingswhich were ordered the previous day as part of theroutine preoperative orders. If not, the patient’s lowerextremities are wrapped with elastic wraps to preventthe pooling (and thrombosis) of venous blood in thelower extremities during the operation and in the im-mediate postoperative period. A restraining strap isplaced across the patient to prevent a fall off the oper-© 1991 The American Association of Neurological Surgeons

35

ating table. A catheter is inserted into an upper extremityvein and intravenous fluids are begun. Prophylactic an-tibiotics are also given. We follow the Malis regimen,giving 1 g of vancomycin intravenously in 250 ml of 5%dextrose in distilled water or in 0.25 N saline over at leastone hour and giving 80 mg of tobramycin intramuscu-larly. In addition, streptomycin is added to the irrigatingfluid used during the operation, at a concentration of50 µLg/ml.

An intra-atrial catheter is inserted, usually via thebasilic or cephalic vein. When these veins cannot be usedsuccessfully, the catheter is inserted via the internal jugu-lar vein. Its position is verified by a portable x-ray film ofthe chest. A vascular catheter is also placed in the radialartery to permit the direct monitoring of arterial blood pres-sure. A urinary catheter is usually not inserted because ofthe relatively short duration of the operation.

Figure 1. Posterolaterally herniated cervical disc compressing a nerve root.

After the induction of anesthesia, the patient’s eyesare protected by ophthalmic ointment and eyelid tapes oradhesive plastic eyelid covers. A three-point head clampis applied, and the operating table is adjusted such thatthe patient comes into a sitting position, with a bolsterunder the buttocks. Pillows are placed beneath the kneesso the hips and knees are each flexed to about 100 to 120degrees. The patient’s arms are usually placed in the lap,and the head clamp is fixed to the table with the patient’shead in straight alignment and flexed somewhat forward.The normal headrest is removed from the table to exposethe posterior surface of the neck and upper thorax. Care istaken to prevent direct pressure against any superficialnerve, such as the u1nar nerve at the elbow and the com-mon peroneal nerve at the knee.

The posterior surface of the neck is prepared withantiseptic solutions and dried. The spinous processes

WILKINS AND GASKILL : CERVICAL HERNIATED DISC EXCISION


36


Figure 2. The skin incision (dotted line) used to expose a herniated disc between the C-6 and C-7 vertebrae.

are palpated by the surgeon to aid in planning the inci-sion. The spinous process of the seventh cervical ver-tebra is ordinarily the most prominent, which can beverified on the lateral x-ray film that was made duringthe patient’s diagnostic workup. The line of the pro-posed incision is marked on the skin (Fig. 2). The op-erative area is then draped as a sterile field, using, alongwith the towels and drapes, an adhesive transparentplastic skin covering. Just before the operation is be-gun, a portable x-ray machine is positioned on the sameside of the table as the anesthesiologist, to permit alateral localization film to be made after the ipsilaterallaminae are exposed.

The incision is carried through the skin and subcuta-neous tissues down to the posterior aspects of the spinousprocesses. The fascia is divided along the side of the

spinous processes with cutting cautery, and the ipsilat-eral paravertebral muscles are stripped away from thespines and laminae by subperiosteal dissection using aperiosteal elevator. Adherent muscle strands and tendonsare divided with curved Mayo scissors. A Williams orScoville retractor is then inserted to maintain the exposureof the laminae and the facet joint (Fig. 3A).

A metallic marker, such as a No. 4 Penfield dissector,is placed at the facet joint and a lateral x-ray film is made toverify the level. If the marked joint is the correct one, thesurgeon proceeds; otherwise the surgeon enlarges theexposure to arrive at the proper facet joint.

With a cup curette, soft tissue is removed from thelateral aspects of the appropriate laminae and from thefacet joint. The inferior edge of the superior lamina is ex-posed in this way, and a portion of it is removed


37

Figure 3. A, the paravertebral muscles have been stripped awayfrom the laminae and spinous processes of C-6 and C-7 and a self-retaining retractor has been inserted. The area within the dottedkeyhole line represents the extent of bone removal. B, a portion

of the inferior edge of the superior lamina is removed with Kerrisonrongeurs. C, the medial aspect of the facet joint is drilled awayusing a diamond burr. D, herniated disc material is removed fromunder the axilla of the C-7 nerve root.

WILKINS AND GASKILL : CERVICAL HERNIATED DISC EXCISION


38

with Kerrison rongeurs (Fig. 3B). The superior rim of theinferior lamina is removed laterally in similar fashion.

The medial aspect of the facet joint is removed alongthe exiting nerve root, to expose its posterior surface. Thiscan be done with a small Kerrison rongeur, but if the nerveroot is already compressed within the intervertebral fora-men it maybe compressed further by the insertion of thefootplate of the rongeur. Thus, it is safer to expose thenerve root with a diamond burr (Fig. 3C), using constantirrigation during the drilling to avoid thermal injury to thenerve. The bone is drilled down to a thin remaining shell,which is then removed with a small cup curette.

The ligamentum flavum is removed to expose the lat-eral aspect of the dura mater within the spinal canal andthe medial aspect of the nerve root. The epidural veinsmay be coagulated with bipolar current (with care taken toavoid thermal or electrical injury to the nerve root andspinal cord) or may be compressed temporarily bycottonoid pledgets with radiopaque markers (having at-tached strings that extend out of the wound as a reminderto the surgeon to remove them before the closure).

The disc herniation is usually best approached un-der the axilla of the nerve root (Fig. 3D). With gentlesuperomedial retraction by an instrument such as a No.4 Penfield dissector, the herniated disc material is ex-posed. If the posterior longitudinal ligament is still in-tact over the disc protrusion, it should be incised witha No. 11 scalpel blade. The disc herniation is removedwith a small pituitary rongeur. Additional disc fragmentscan sometimes be squeezed into view by pressing for-ward on the posterior longitudinal ligament with a bluntnerve hook or small dissector. These fragments are thenremoved as well. The disc space itself is not entered.Before closure, the area anterior to the nerve root andthe adjacent dura mater is palpated with a blunt nervehook or a small dissector to verify that all obtainabledisc fragments have been removed and that the nerveroot has been decompressed (i.e., is slack). The sur-geon must be careful to avoid vigorous or prolongedmanipulation of the nerve root, which may injure the

root and result in unnecessary postoperative pain orneurological dysfunction. Occasionally the disc her-niation is best approached at the shoulder of the nerveroot rather than the axilla; the steps in exposure andremoval of the disc fragments are essentially the same.With either route the surgeon must be certain that theprotruding material is the disc herniation. At times, thenerve root will be found to be in two parallel parts; therisk is that the surgeon may retract one part and incisethe other, thinking it is the disc herniation.

We usually cover the exposed dura and nerve rootwith absorbable gelatin sponge, but this step can be omit-ted. The muscles are reapproximated to the interspinousligament or ligaments with interrupted sutures, and thefascia is closed with similar sutures. After closure of thesubcutaneous tissues and skin, a sterile dressing is ap-plied and the patient’s head holder is detached from itssupport. The operating table is flattened to bring the pa-tient again into a supine position. The head holder is re-moved from the patient’s head, and the pin puncture sitesare covered with an antibiotic ointment. The anesthetic isreversed. The eye covers are removed. The patient is ex-tubated and sent to the recovery room. The intra-arterialand intravenous catheters and the leg wraps or stockingsare removed subsequently, as appropriate.

POSTOPERATIVE COURSEAfter the anesthetic has worn off, the patient may be outof bed as comfort permits. Medication is given as neededto provide adequate pain relief. We ordinarily do not rec-ommend the use of a cervical collar. The patient is encour-aged to begin restoring the range of neck movement tonormal. An exercise program is initiated to reverse anyresidual upper extremity weakness. The patient is dis-charged from the hospital when sufficient comfort hasbeen achieved, usually on about the fifth postoperativeday. The skin sutures are removed before discharge. Thepatient increases activity gradually at home and returns inone month for reevaluation. Ordinarily, the patient is re-leased to return to work at that time.


39

LATERAL SPHENOIDWING MENINGIOMA

JOSEPH RANSOHOFF, M.D.

PATIENT SELECTIONA lateral sphenoid wing meningioma may grow to a largesize before it causes any neurological symptoms. Thistumor bridges the sphenoid wing and can grow symmetri-cally in the anterior and middle fossae or more toward oneor the other. As the tumor enlarges it gradually displacesthe adjacent cortex and at the same time the knee of themiddle cerebral artery and only rarely surrounds or en-cases this vessel. The blood supply to these tumors ismainly from the extracranial circulation, including middlemeningeal and superficial temporal arteries.

The symptomatology may be unilateral headaches,often misinterpreted as migraine, with radiation to the fore-head and homolateral eye, and seizures which may be ofthe focal motor or the partial complex type. Very late in thecourse of the disease, signs of increased intracranial pres-sure and/or intellectual deterioration may occur.

The decision to undertake surgical removal of thelateral sphenoid wing meningioma is based on the size ofthe tumor, the patient’s age, the patient’s general medi-cal condition, the presence or absence of cerebral edema,and shift of intracranial structures. Whereas there is noalternative method of management other than surgicalexcision and the use of anticonvulsants to control sei-zures, relatively small tumors in patients over age 70 canbe safely observed with serial computed tomography(CT) or magnetic resonance imaging (MRI) scans. Therisks of surgical intervention are postoperative hemor-rhage and medical complications which in a healthy indi-vidual should be no more than 5%. The specific risk inthese tumors relates to potential damage to the middlecerebral artery with subsequent neurologic deficit. Ra-diation therapy in these globular tumors is not indicatedexcept for en plaque residual tumor following surgery.The value of “radiosurgery,” either with gamma knife ormodified linear accelerator, in small tumors in poor-riskpatients remains to be evaluated.

DIAGNOSTIC EVALUATION ANDPREOPERATIVE CONSIDERATIONSCT and MRI scans without and with contrast enhance-ment are clearly the major diagnostic tools in the initialevaluation of these patients prior to surgery. The enhancedMRI study will not only demonstrate the bulk of the tumorbut also the degree of dural involvement which may ex-tend beyond the limits of the globular mass and can be ofgreat importance in planning for a total surgical removal.Furthermore, the displacement and/or involvement of themiddle cerebral artery can generally be seen on the MRIscan and gives a clear indication as to the surgical risksrelated to damage of this vessel.

Cerebral angiography is of great assistance in carry-ing out surgery in view of the fact that the major bloodsupply to this tumor arises almost solely from the extrac-ranial circulation (Fig. 1A), which lends itself ideally tosuperselective catheterization and preoperative emboliza-tion (Fig. 1B). A postembolization. MRI scan as well as apostembolization. CT scan can demonstrate the degree ofblood supply of the tumor arising from the intracranialcirculation and aid in operative planning as well as in de-tailing the potential risks to the patient and family (Fig. 2,A and B).

Corticosteroids are used in the perioperative periodand are instituted the night before surgery unless, ofcourse, the patient has required longer term use of ste-roids for control of cerebral edema; 125 mg of SoluMedrolis administered intravenously the night before surgery,the morning of surgery, and every 6 hours thereafter for 24to 48 hours depending on the patient’s condition. There-after, a rapid or slow steroid taper is appropriate, depend-ing on the patient’s postoperative status. Anticonvulsants(either phenobarbital or Dilantin) are administered preop-eratively even in those patients who have not had sei-zures, with appropriate blood levels being ascertainedprior to the initiation of surgery. It seems that Dilantin ispoorly tolerated by the elderly patient, and therefore wehave tended in more recent times toward the increaseduse of phenobarbital. If the patient has partial complex© 1991 The American Association of Neurological Surgeons

40


Figure 2. A, contrast-enhanced CT prior to embolization. B, contrast-enhanced CT after embolization showing tumor necrosis.

Figure 1. A, lateral selective external angiogram prior to embo-lization. B, lateral x-ray film of the skull after embolization show-

ing the cast of the tumor produced by Gelfoarn powder emboliimpregnated with Conray.

seizures, phenobarbital may be replaced with Tegretol inthe postoperative period.

SURGICAL TECHNIQUEThere are no special precautions in terms of anesthetictechnique in the management of these tumors. Mannitol(250 ml of a 20% solution) is administered intravenouslyover a period of 15 to 20 minutes as the skin incision isbeing made. An indwelling urinary catheter is essential.Pneumatic compression boots can be used in an effort toprevent postoperative pulmonary embolism. These mustbe applied in the operating room once a patient has beenanesthetized but should not be used if the patient has

been at bed rest for a significant period of time prior tosurgery. They are then continued until the patient is fullyambulatory.

Skin Incision and Bone FlapAfter induction, the patient’s head is shaved and a Faulkner-type incision is marked on the scalp (Fig. 4A). For right-sidedincisions a shoulder roll is placed beneath the right shoulderand the head turned toward the left side. The Mayfield three-pin head holder is applied in such a way that the sprocket forattachment to the operating table will be directed perpen-dicular to the floor when the patient’s head is turned 60 de-grees to the left of midline. A single pin of the



41

RANSOHOFF : LATERAL SPHENOID WING MENINGIOMA

head holder is placed ipsilateral to the skin incision andthis should penetrate the skull approximately 4 cm poste-rior to the external auditory meatus. The head is then se-cured to the operating table using the Mayfield table clampso that the head is elevated above the heart level, rotated60 degrees, and slightly extended, and the operating tableis slightly flexed (Fig. 3). After the skin is cleansed, theincision is marked on the scalp using a No. 10 blade start-ing at the level of the zygoma a few millimeters anterior tothe tragus and is extended superiorly behind the hairlinecurving anteriorly to end 3 cm lateral to the midline. Theposterior limb of the incision is modified to accommodatetumors which extend more posteriorly into the temporalfossa. Following draping and infiltration of the plannedincision line with 1% lidocaine containing epinephrine,the incision is started at the zygoma and carried downthrough the galea to the areolar connective tissue overly-ing the temporalis fascia. Care is taken in this region toidentify the superficial temporal artery and, if necessary,to dissect it using Metzenbaum scissors in order to retractit anteriorly with the reflected scalp flap. Occasionally, itbecomes necessary to sacrifice the posterior branch ofthis artery. Above the superior temporal line the incisionis carried down to the periosteum and, following the appli-cation of Raney clips to control bleeding, the skin is re-flected anteriorly using an elevator to dissect through thesubgaleal connective tissue. A sponge roll is used to sup-port the base of the scalp flap while it is retracted towardthe base and held using skin hooks (Fig. 4B). The perios-teum over the superior temporal line is then incised alongits superior margin using a No. 10 blade and the perios-

teum is stripped from the outer table of the bone and re-flected anteriorly. Using a Bovie electrocautery, thetemporalis muscle and fascia are incised beginning at thelevel of the zygoma, extending superiorly to end at thezygomatic process of the frontal bone. This frees a largeportion of the temporalis muscle to be reflected posteri-orly away from the area of craniotomy, thus avoiding in-jury to the frontalis branch of the facial nerve. In this waythis branch of the facial nerve is carried anteriorly andinferiorly with the remaining portion of the temporal muscleand fascia (Fig. 4C). A Bovie electrocautery is used toseparate the remaining temporal muscle from the temporalcrest and, most importantly, from the zygomatic processof the frontal bone anteriorly.

With the bisected temporalis muscle reflected anteri-orly and posteriorly away from the pterion, two burr holesare placed, one just superior to the zygoma near the floorof the middle fossa and the second just posterior andinferior to the frontal process of the zygoma which is thefloor of the frontal fossa. A craniotome is used to developa cranial opening beginning with the temporal burr holeand extending anteriorly until the sphenoid ridge is en-countered. The craniotome is then withdrawn and redi-rected through the frontal burr hole and again a cranialgroove developed posteriorly until the sphenoid wing isencountered. Following stripping of the dura from the in-ner table of the calvarium, the craniotome is then directedfrom the temporal burr hole superiorly to develop a cranialflap which extends a few centimeters superior to the supe-rior temporal line and anteriorly onto the supraorbital rimand floor of the frontal fossa. The resulting cranial

Figure 3. Patient position on the operating table.


42

Figure 4. A, skin incision. The course of the frontalis branch of nerve VII is shown. B, temporalis muscle incision outlined. C,temporalis flap elevated and bone flap outlined.



43

flap will be hinged on the sphenoid wing, and this is frac-tured as the cranial flap is lifted from the intact dura. Then,the dura. is dissected from the remaining sphenoid wingand small bony keel. Removal of this bone using a Leksellrongeur alternating with dural stripping results in a flat-tened greater wing of the sphenoid and allows unob-structed work in the region of the dura. of the pterionextending from the lateral portion of the middle and frontalfossae. The dura is then tacked to the surrounding perim-eter of the cranial opening.

Once the dura has been exposed, the ultrasound lo-calizer should be used to demonstrate the location of themeningioma in relationship to the bone flap and exposeddura so that additional removal of bone can be completedas necessary. The dural incision is usually initiated paral-lel to the floor of the frontal fossa allowing sufficient mar-gin of dura, i.e., 2 to 3 cm, in order to permit tenting of thedura to the adjacent muscle at the base, a technique thatprovides for good hemostasis (Fig. 5A). The dura is thenopened over the middle fossa and finally the incision iscarried across the area of the sphenoid wing using theclosed Metzenbaum scissors to dissect it from the adja-cent tumor. Here will be encountered a good deal of bloodsupply to the tumor from the middle meningeal artery whichwill be coagulated with the bipolar cautery. Thereafter,depending upon the height of the tumor, the dura is openedin a semicircular fashion, retracted toward the vertex, andheld in place with traction sutures. Again the ultrasoniclocalizer can be used to be certain that the entire tumor,which may be partially buried under a shelf of cortex, isaccessible through the operative exposure.

Tumor RemovalUtilizing the general principle of all meningioma surgery,that is, initially to devascularize the tumor from its duralattachment, the surgeon approaches along the base of theanterior fossa and coagulates the dural attachment alongthe sphenoid wing. At this point, the surgeon also ob-serves the relationship of the tumor to the adjacent brainand the extent of the arachnoidal plane which can be devel-oped between the tumor and the brain. This initial dissec-tion can be aided with the use of the operating microscope.If necessary, a resection of the lateral inferior portion of thefrontal lobe can be carried out with the use of suction andthe bipolar cautery. Elective resection of noncritical braintissue is far preferable to vigorous retraction as the latterwill inevitably lead to significant postoperative edema (Fig.5A). Once the total anterior aspect of the tumor has beendemonstrated to its most medial extent, a self-retaining re-tractor can be used to hold the exposure.

Then attention is focused on the middle fossa aspectof the tumor. There may be significant blood supply to the

tumor arising from the lateral aspect of the middle fossadura. Once again, elective resection of the anterior aspectof the temporal lobe, if necessary, is far preferable to vig-orous retraction. Having exposed both the temporal aswell as frontal aspects of the tumor, the surgeon thenproceeds along the “knife edge” of the sphenoid wing,devascularizing the tumor in this area. These tumors al-most always displace the knee of the middle cerebral ar-tery and adjacent branches medially (Fig. 5B) but not thevenous drainage to the dura at the lateral aspect of thesphenoid wing; these veins must be divided carefully inorder to provide good exposure of the tumor.

If the tumor is large, one can consider debulkingbefore attempting to dissect the superior aspect of thecapsule from the adjacent brain and middle cerebral ar-tery. If the tumor has been successfully embolized, it isoften sufficiently necrotic to be easily debulked withsimple suction. The ultrasonic aspirator is also of greatvalue in carrying out this maneuver (Fig. 5B). As onedebulks the tumor, small arterial branches within the tu-mor itself will come into the suction apparatus and peri-odically should be coagulated. In large tumors, the mostmedial aspect of the attachment of the tumor to the duralbase will not yet have been fully visualized and caremust be taken not to violate the medial aspect of thecapsule. Here, once again, the ultrasonic localizer can beof value in determining the amount of tumor which hasbeen removed and the amount which remains mediallyand superiorly. Proceeding in this fashion, a small amountof tumor may be left attached at the dural base which willbe attended to later as the important aspect of the opera-tive procedure now is to remove the major bulk of thetumor from the intracranial cavity.

As the superior and medial aspects of the tumorcapsule are dissected from the adjacent brain, the oper-ating microscope should be used. Depending on thevectors of growth, the middle cerebral artery may be dis-placed directly medially or somewhat posteriorly. Reviewof the angiography at this time can assist in the localiza-tion of this vessel. As the tumor capsule is retractedlaterally and inferiorly the adjacent brain and arachnoidplane are exposed and cottonoids should be placed inthis plane of dissection; it is important not to place onecottonoid on top of the next, but rather replace each asthe dissection proceeds in the depth of the wound. Oncethe main trunk of the middle cerebral artery has beenidentified at the most superior aspect of the tumor cap-sule, one proceeds with careful microsurgical techniqueto peel the vessel off the tumor capsule (Fig. 5C). Oftena major vessel will have indented the tumor, so to speak,and whereas initially one may have the impression thatthe vessel is encased in tumor, meticulous dissection may


44

Figure 5. A, dura reflected and tumor partial ly exposed.Dotted lines indicate areas of frontal and temporal lobes forelective resection, if needed. B, relationship of the middlecerebral artery to the tumor. Ultrasonic aspirator on the

tumor for debulking. C, tumor debulked. Dissection of thecapsule from the middle cerebral artery and adjacent brain. Abipolar cautery is used to coagulate small branches feedingthe tumor.



45

achieve a total removal of the tumor from this area withoutdamaging the vessel. If necessary, however, a small frag-ment of capsule may remain attached to the middle cere-bral artery to be reexamined once the major portion of thetumor has been removed. The appearance of cerebrospi-nal fluid welling up from the medial aspect of the capsulewill indicate proximity to the base of the tumor. With pro-gressive debulking and retraction of the capsule, alwaysidentifying the major arterial vessels, a complete grossremoval of the tumor can usually be achieved.

After removing the bulk of the tumor there will be agood view of the base of the skull and adjacent sphenoidwing. It is important to remove the entire dural attachmentas recurrence of meningiomas occurs almost always fromthe dural involvement at the base of the skull. Not only isvisual examination important, but also an intraoperativereview of the gadolinium-enhanced MRI scan will showthe operator the extent of dural involvement. The CO

2

laser can be very effective in vaporizing the dura. down tothe bare bone and probably is the preferred method ofremoving the dural base of the tumor. The CO

2 laser can

be either hand-held or manipulated through the operatingmicroscope. Adjacent brain structures should be protectedwith cottonoid patties. If the lateral aspect of the sphe-noid wing itself is invaded with tumor, this should be re-moved either with a high-speed air drill or with rongeurs.If a small amount of meningioma has been allowed to re-main adjacent to the main trunk of the middle cerebralartery, this area should be inspected before removal of thebrain refractors. Any tumor that cannot be dissected fi-nally from this major vessel should be gently coagulatedwith the bipolar cautery. The wound should be thoroughlyirrigated to evaluate any areas of venous oozing from thecerebral structures. It is our practice to then coat the en-tire exposed area with a thin layer of oxidized cellulose.The superior dural flap should then be placed over theexposed brain and wherever possible closed both fron-tally and temporally. The area that cannot be completelyclosed can be covered with a pledget of collagen film.Whereas this substance has not been generally utilized inneurosurgical procedures, we have found it to be veryeffective as a dural replacement. It does not require sutur-ing to the adjacent dural edges and fully protects the un-derlying brain and adheres nicely to the dural surface.The collagen film is placed over the area as it comes from

the sterile package and then is flooded with irrigating so-lution covered with a large cottonoid patty and suctionapplied. When the patty is removed the film will havefallen into place and the edges can be gently tucked in, inthis instance along the base of skull both in the frontaland middle fossae.

ClosureThe bone flap is inspected for any involvement by menin-gioma which should be removed with the high-speed drill.If a large area of bone flap has been removed, this can bereplaced with stainless steel or titanium screen. The boneflap is wired in place and the closure carried out in routinefashion. A subgaleal Jackson-Pratt drain is placed and ahead dressing applied, being careful to protect the ex-posed ear from compression.

COMPLICATIONSIn the postoperative period the two major complicationsto be avoided are seizures and hematoma formation. BothDilantin and phenobarbital should be used intraopera-tively; postoperatively, however, one drug or the othershould be discontinued. If the patient had been on anti-convulsant therapy for preoperative seizures, that regimeshould be continued at the time of discharge. Steroidsshould be tapered rapidly over a period of three to sevendays. If the patient’s postoperative course is satisfactory,a noncontrast CT scan should be obtained 48 hours post-operatively to rule out any intracranial hematoma and toassess the degree of brain edema. The latter can serve asa guide to the program of steroid withdrawal.

It is important, however, to consider a postoperativeCT scan prior to extubation if there has been any damageto the middle cerebral artery or any degree of brain swell-ing at the completion of the operative exposure. Long-term follow-up should include a repeat MRI scan withgadolinium enhancement three months postoperativelyand, if this is entirely negative, a further MRI scan in oneyear to confirm total removal. If residual tumor is known toremain at the base of the skull or small fragments are at-tached to the critical cranial vessels, similar studies shouldbe considered at somewhat more frequent intervals. Con-sideration of postoperative radiation therapy to the baseof the skull can be important if any evidence of residualtumor or regrowth can be demonstrated.


46

SELECTIVE MICROSURGICALVESTIBULAR NERVE SECTION FOR

INTRACTABLE MÉNIÈRE’S SYNDROMEEDWARD TARLOV, M.D.

INTRODUCTIONIntractable Ménière’s syndrome is found in a small pro-portion of patients with complaints of dizziness. Fluctuat-ing hearing loss, tinnitus, and bouts of violent vertigocharacterize the disorder. Vertigo is often the most dis-abling feature of the disorder. Most patients with Ménière’ssyndrome are managed by otolaryngologists. Diuretics,salt restriction, and vestibular sedatives are ordinarily theinitial treatments of choice. In the years since ProsperMénière first described the disorder, a variety of surgicaltreatments have been used when medical measures havefailed. Endolymphatic shunting operations carried out inan effort to relieve tinnitus and hearing loss have beendisappointing for relief of vertigo. Labyrinthine destruc-tive procedures that can relieve vertigo will invariablyabolish hearing, a disadvantage since the disorder maybecome bilateral in about 20% of instances. Selective mi-crosurgical vestibular nerve section can relieve the ver-tigo with preservation of hearing. This is an excellent op-eration for sufferers of Ménière’s disease who are nottotally deaf in the affected ear and who are refractory tomedical management.

The diagnosis of true Ménière’s syndrome is not com-mon among the whole spectrum of patients with com-plaints of dizziness. Strict clinical criteria for diagnosisshould be used. In most instances the onset is betweenages 30 and 60, with the young and very old rarely beingaffected. The recurring attacks usually increase in fre-quency. Ultimately, there is almost always a permanenthearing loss in untreated cases, and vertigo may continueafter deafness is complete. The incidence of the disorderhas been estimated at 1: 1 00, 000 population.

Vertigo is the first symptom in the majority of patientsand can be the most disabling feature of the disease. Tin-nitus is almost always present and maybe the most reli-

able indicator of which labyrinth is involved in unilateralcases. The stage of fluctuating hearing loss may last froma few weeks to several years before it becomes fixed. Fluc-tuations of deafness usually correspond with bouts ofvertigo. A premonitory aura with fullness in the head oraffected ear is common. Drop attacks (“the otolithic crisisof Tumarkin”) occasionally occur as a result of loss oftonic influences of the otoliths. In severe attacks the pa-tient may be thrown to the ground or, if seated as whendining, may suddenly become disoriented with respect togravity. Nystagmus may accompany the attacks, but itsdirection is not of localizing value.

Electronystagmography with caloric testing is usu-ally abnormal and may demonstrate canal paresis withdiminished responses to warm and cold water or air on theaffected side. Occasionally, directional preponderance orspontaneous nystagmus to the side opposite the diseasedlabyrinth occurs. On audiological testing the majority ofcases show low-frequency perceptive hearing loss withfairly good discrimination, loudness recruitment, and nega-tive tone decay found on impedance audiometry. Brainstemauditory-evoked response testing demonstrates normallatencies.

The most important differential diagnosis is that ofacoustic neuroma. The latter tends not to cause severevertigo as in Ménière’s disease. The loss of vestibularfunction is usually slow, producing mild ataxia rather thanparoxysmal disturbances. Magnetic resonance imagingwith gadolinium in suspicious cases is helpful in exclud-ing the possibility of an acoustic neuroma. Brainstem au-ditory-evoked response testing can identify an eighthnerve lesion but is, of course, not specific for its nature.

Pathological examination of the temporal bones inpatients who died following operation has shown dila-tion of the endolymph spaces at the expense of the peri-lymph spaces. This is thought to cause symptoms be-cause it is necessary to have equal perilymph andendolymph pressures for normal cochlear and vestibu-© 1991 The American Association of Neurological Surgeons

47

lar function. The situation is analogous to hydrocephalusor glaucoma in that there is an accumulation of an excessendolymph most probably due to deficient reabsorption.

Because of the relatively high incidence of bilateralinvolvement, efforts at preserving existing hearing or im-proving hearing have considerable potential value. Nev-ertheless, destruction of the labyrinth has continued as astandard otological operation when hearing preservationis not a consideration. This may be performed as atransmastoid or a permeatal procedure. It results in a totalloss of hearing and is not a desirable form of treatment inpatients with unilateral Ménière’s syndrome if there isuseful hearing in the affected ear or if there is loss ofhearing on the other side from other causes. It is also notdesirable in bilateral Ménière’s syndrome if both ears areaffected early in the disease, if one labyrinth has beendestroyed before disease appears in the other ear, or ifhearing is poor in one ear and is rapidly failing in the other.

A variety of operations to relieve the excess pressureof endolymph have been carried out, principally aimed atpreserving or improving hearing and relieving tinnitus.The idea of creating a fistula from the endolymphatic sacto the mastoid cavity or to the subarachnoid space hasproduced a standard series of shunting alternatives. Con-troversy about the efficacy of these procedures has ex-isted. Medical labyrinthectomy with streptomycin is indi-cated in the occasional instances in which there is nolabyrinthine function on the opposite side. It would bemost accurate to say that at this time the value of shunt-ing procedures alone for Ménière’s syndrome is unproven.Because the procedure can be combined with vestibularnerve section and because the latter relieves the mostdisabling symptom, however, the shunting proceduresseem worthy of further trials.

Because the endolymph pressure is usually lower thanthe cerebrospinal fluid (CSF) pressure, the precise physi-ological mechanisms involved in any benefit from shunt-ing the endolymph to the subarachnoid space are un-known.

The pioneer neurosurgeons had broad experiencewith vestibular nerve section in this disorder, and the op-eration proved to be quite reliable. Dandy refined the op-eration to the extent that only two deaths occurred amonghis 587 cases. Vertigo was relieved entirely in 90%. Fivepercent were unchanged, and 5% were worse. In the ma-jority of these 587 cases the entire eighth nerve bundlewas sectioned with total loss of auditory as well as vesti-bular function. Fifty-four of Dandy’s patients had facialparalysis, of whom 17 had permanent facial paralysis. Of95 patients in whom only the vestibular portion of thenerve was sectioned, 9 had improved hearing, 27 had un-changed hearing, 46 were worse, and 13 were totally deaf.

The experience of Olivecrona was similar. Falconer in alater era was able to preserve hearing on the operated sideto some degree in his eight patients. The microsurgicaltechnique described herein permits consistent postop-erative hearing preservation.

GENERAL INDICATIONS FOR SURGERYIN MÉNIÈRE’S SYNDROME ANDCHOICE OF OPERATIONSince Ménière’s syndrome is primarily treated byotologists, most of the operations in recent years havebeen carried out as otological services. When hearing onthe affected side is absent and the patient is mainlytroubled by severe whirling attacks, labyrinthectomy isindicated. Labyrinthectomy may be carried out via atransmastoid or permeatal approach. This operation is ef-fective in eliminating vertigo. The imbalance following laby-rinthectomy is usually short-lived, and good compensa-tion for this usually occurs without significant residualimbalance.

When hearing loss is the major symptom and themajor therapeutic effort is to be aimed at preserving hear-ing, with the vertigo as a minor component, an endolym-phatic shunting procedure is indicated.

The operation described here is indicated when bear-ing preservation is a consideration and when the vertigi-nous attacks are severe. The morbidity of the operation hasbeen quite low, and it has proved thus far to be very effec-tive for control of whirling vertigo, with this most disablingsymptom having been eliminated in almost all cases.

We have carried out this procedure as a combinedneurosurgical-otological operation. The vestibular nervecan safely be exposed via three routes: an extradural ap-proach along the floor of the middle fossa, in thecerebellopontine angle or medial internal auditory canal,and in the most lateral portion of the canal where the su-perior and inferior vestibular nerves are separate. Theo-retical considerations might make it desirable to sectiononly the superior vestibular nerve in order to reduce post-operative imbalance. Our experience, however, has beenthat postoperative imbalance from section of the entirevestibular nerve is minimal. The middle fossa approachrequires some degree of retraction on the temporal lobe.For these reasons we have not employed the middle fossaexposure of the vestibular nerve, and the exposure of thevestibular nerve in the internal auditory canal carries arisk to hearing. Accordingly, we have favored section ofthe vestibular nerve in the posterior fossa itself.

It may be worthwhile to review the relative advan-tages and disadvantages and our experiences with theapproach anterior to the sigmoid sinus compared withthe exposure that we now use routinely, posterior

TARLOV : SELECTIVE MICROSURGICAL VESTIBULAR NERVE SECTION

48


to the sigmoid sinus. In our initial cases, we used an oto-logical exposure through the mastoid anterior to the sig-moid sinus.

The mastoid is burred away, exposing the sigmoidsinus. The large mastoid emissary vein is skeletonized inthis approach, and a wide bony removal over the sigmoidsinus is carried out. As the operation is carried out in thesupine position, we have never had difficulties with airembolization or with handling of the sigmoid sinus in anyway. The wide bony removal over the sigmoid permitsgentle downward extradural retraction of the sigmoid andcerebellar hemisphere to facilitate a less tangential angleof visualization of the vestibular bundle as it passes intothe internal auditory meatus. Depending on the shape ofthe lateral wall of the posterior fossa, which is somewhatvariable, exposure of the dura anterior to the sigmoid si-nus is nearly tangential to the dura, a factor that makes atight dural closure somewhat more difficult. The duralopening is bounded anteriorly by the internal auditorymeatus and posteriorly by the sigmoid sinus. An openingis made in this small dural area. In an individual with ashort muscular neck or one with a thick mastoid, there islittle leeway to alter or vary the surgeon’s line of visualiza-tion of the intradural structures. The exposure of the eighthnerve bundle, although adequate to section the vestibu-lar nerve, is quite limited. For orientation, it is helpful togain a view of the trigeminal nerve superiorly and theninth, tenth, and eleventh nerves inferiorly. We have notencountered any difficulties requiring hemostatic controlof adjacent vessels, but the exposure anterior to the sig-moid sinus would be somewhat limited to accomplish thisif necessary. Positive identification of the eighth nervecomplex, by confirmation of its proximity to the flocculus,visualization of the fifth nerve rostrally and the ninth. tenth,and eleventh nerve complex below, and identification ofthe bony internal auditory meatus, is facilitated by thewider exposure through the posterior fossa. We have pre-ferred the posterior approach, since dural closure of thevery thin, tangentially exposed dura anterior to the sig-moid sinus is difficult. With the anterior approach we havehad several CSF leaks, a not infrequent problem intransmastoid or translabyrinthine exposure of the poste-rior fossa. Accordingly, we have since modified the op-eration in the manner described below and have found theexposure posterior to the sigmoid sinus to be more simpleand expeditious. In addition, the retrosigmoid approachgives a wide view of the posterior fossa. In the supineposition with the head turned contralaterally, it is quitestriking how little cerebellar retraction is necessary onceCSF has been aspirated. The structures behind the en-dolymphatic sac are clearly seen from within the dura when

the procedure is carried out in this manner. Morbidity hasbeen minimal, and dural closure over a patch graft hasnever been a problem.

POSITIONWe formerly considered the advantages of the sitting posi-tion virtually indispensable in posterior fossa surgery. Dur-ing development of our own microsurgical approaches totic douloureux, particularly in the elderly, it became clear tous that the cerebellopontine angle can be safely and widelyexposed in the supine position. The view obtained oncethe surgeon becomes familiar with the use of this positionis virtually identical to that obtained in the sitting positionexcept that the surgical field is rotated 90 degrees. Thesitting position has many advantages, especially for largecerebellopontine angle tumors. Nevertheless, we have beenusing the supine position increasingly for routine cranialnerve surgery in the posterior fossa, including surgery fortic douloureux, small acoustic neuromas, and this operationfor Ménière’s syndrome. The hazards of the sitting posi-tion, principally relating to hypotension and air embolism,are eliminated without losing most of the advantages of thesitting position. The head of the operating table is slightlyelevated, a step that markedly reduces venous pressure.The neck is rotated contralaterally and is moderately flexed.Two fingers can be inserted beneath the chin when theneck is in proper position. When CSF has been aspiratedfrom the cisterna magna, almost no cerebellar retraction isnecessary. No arterial or central venous line is necessary.Figure 1 demonstrates the patient’s head position from thesurgeon’s viewpoint. The surgeon is seated behind thepatient’s head. The operating microscope is brought in fromthe surgeon’s left to permit easy access by the scrub nurseto his right hand.

SURGICAL TECHNIQUEThe overall orientation in the supine position from thesurgeon’s viewpoint is demonstrated in Figure 1A. Weprefer a Mayfield three-point head fixation unit that al-lows the use of the Leyla Yasargil self-retaining retractorsystem. A paramedian incision about 1 cm posterior tomastoid prominence is used as indicated. Minimal hairshaving is necessary. A small laterally placed craniectomyis carried out. This extends over the junction between thetransverse and sigmoid sinuses. The dura is opened infe-riorly at first. Ordinarily, the cerebellar hemisphere is some-what full at this stage. With a cottonoid over the cerebel-lar hemisphere, CSF is aspirated from the convexitysubarachnoid space as shown in Figure 2A, to obtain aslack cerebellum as shown in Figure 2B. The force of grav-ity provides most of the necessary cerebellar re-

49

Figure 1. A, position of head turned to right with patient supine.In our early experience the opening as made at area A to bring usdown to dura anterior to the sigmoid sinus. We have since changedthe procedure to make our opening posterior to the sigmoid sinus(area B). B, the skin incision. C, dural exposure posterior to thesigmoid sinus. (A: Modified from Tarlov EC. Microsurgical vesti-

bular nerve section for intractable Ménière’s syndrome: Techniqueand results. Clin Neurosurg 1986;33:667-684. B and C: Modifiedfrom Tarlov EC, Oliver P. Selective vestibular nerve section com-bined with endolymphatic sac to subarachnoid shunt for intractableMénière’s syndrome: Surgical technique. Contemp Neurosurg1983;4(26):1-8.)



50

Figure 2. A, showing aspiration of CSF from surface subarachnoidspace over cerebellar hemisphere. B, slack exposure resulting fromCSF aspiration. Minimal retraction and changing the angle of visu-alization results in the view in C. C, lower cranial nerves in viewand vestibular portion of vestibulocochlear nerve sectioned. The

retractor’s function is largely to protect the cerebellar hemisphere.(B and C: Modified from Tarlov EC, Oliver P. Selective vestibularnerve section combined with endolymphatic sac to subarachnoidshunt for intractable Ménière’s syndrome: Surgical technique.Contemp Neurosurg 1983;4(26):1-8.)



51

Figure 3. A, B, C, and D, distraction of cut vestibular nerve toexpose facial nerve without manipulation of cochlear nerve. E,slack exposure prior to closure. F, dural closure over patch graft.

(Redrawn in color from Tarlov EC. Microsurgical vestibular nervesection for intractable Ménière’s syndrome: Technique and results.Clin Neurosurg 1986:33:667-684.)



52

Figure 4. A, prior technique no longer used, demonstrating inci-sion over mastoid. B, dural opening anterior to sigmoid sinus. C,limited exposure of vestibulocochlear bundle. D, section of vesti-

bular bundle. (Modified from Tarlov EC. Microsurgical vestibularnerve section for intractable Ménière’s syndrome: Technique andresults. Clin Neurosurg 1986;33:667-684.)



53

traction from here on. Petrosal veins are identified at thispoint, the arachnoid is swept medially off their junctionwith the petrosal sinus, and the veins are coagulated. Theyare then divided with curved microscissors, leaving theusual cuff adjacent to the petrosal sinus for safety. Theretractor is then used to place the arachnoid over the in-ternal auditory meatus on a slight stretch. The retractor’sfunction here is largely to protect the cerebellar hemisphere;as the redundant arachnoid forms a cistern at this point,the arachnoid can be swept medially and left intact overmuch of the cerebellopontine angle cistern system.

On a number of occasions, tortuous vessels in theregion of the eighth nerve complex have been observed.We have not infrequently seen these during the course ofoperations for tic douloureux, in patients without eighthnerve signs or symptoms. Although it has been suggestedthat vascular cross-compression may be a causative fac-tor in Ménière’s syndrome, no causative relationship be-tween such vessels and the symptoms of Ménière’s syn-drome has, in our opinion, ever been proven, and we wouldnot consider it presently advisable to carry out vasculardecompression in severely symptomatic patients withMénière’s syndrome, although the microvascular decom-pression operation has, we believe, proven its value in thetreatment of tic douloureux. Once the retractor has beencorrectly positioned a wide view of the eighth nerve com-plex from the internal auditory meatus to the brainstem isobtained. The vestibular portion of the eighth nerve oc-cupies the superior 50% of the combined bundle. It isusually slightly more gray than the more inferiorly lyingcochlear division. At the level of the internal auditorymeatus a plane of cleavage between the vestibular andcochlear divisions may be visible. Gently depressing thevestibular portion of the nerve inferiorly allows visualiza-tion of the facial nerve lying anterosuperiorly. Vesselscoursing along the eighth nerve bundle must be preservedif hearing preservation is to be accomplished. Frequentlythere are few or no vessels to contend with, but it is occa-sionally necessary to carefully dissect around a small ves-sel, separating it from the nerve sharply if it courses acrossthe vestibular nerve itself. A clear plane of cleavage be-tween the vestibular and cochlear portions cannot alwaysbe visualized. When this is the case, we section the supe-rior 50% of the combined cochlear and vestibular bundle.No manipulation of the cochlear portion is carried out.Following such sections, no postoperative vestibular func-tion on this side has been demonstrable postoperatively.As the cochlear nerve and its blood supply are delicate,no manipulation of the cochlear nerve is carried out. Theview obtained once the vestibular nerve has been sec-tioned is indicated in Figures 2C and 3D.

In an effort to determine its value in guiding the sur-geon to preserve cochlear nerve function, we have usedintraoperative brainstem auditory evoked response test-ing. With the availability of facilities for this testing, theprocedure itself is simple enough. A special microphoneis placed on the ipsilateral ear. Once the appropriate scalpleads have been placed, satisfactory intraoperative trac-ings can be obtained. Even with rapid averaging tech-niques the time delay necessary for the interpretation doesnot permit immediate feedback to the surgeon carryingout his manipulations. Accordingly, the test may demon-strate that some surgical manipulation previously carriedout has caused a change in the evoked response latency.This information would only be helpful to the surgeon if itwere available during the course of each movement, whichit is not. Thus, although the procedure is technically fea-sible, we have not found it helpful in preserving cochlearfunction.

A pericranial graft taken in the initial phase of theprocedure is used in the dural closure. The muscles, sub-cutaneous tissues, and skin are then closed in layers, anda small dressing is applied. Steroid preparation and post-operative treatment for several days have reduced theside effects of the procedure. Postoperative vertigo seemsto be proportional to preoperative vestibular function asdetermined by electronystagmography, as would be ex-pected. Patients with the most severe vestibular impair-ment preoperatively have been the least disturbed in theearly postoperative period. Patients have all been dis-charged within one week and have minimal residual ataxiain one month.

With this approach, CSF leakage has not been a prob-lem. Our earlier transmastoid exposures did lead to CSFleakage, and reclosure of the wound was necessary onseveral occasions in view of the thinness of the dura ante-rior to the sigmoid sinus and its oblique angle in relationto the surgeon’s view (Fig. 4). No facial weakness or anyother unwanted effect outside the eighth nerve systemhas occurred.

The procedure can be carried out in the mannerdescribed without producing loss of hearing. Among50 patients operated upon so far we have operatedseveral times on the only hearing ear and on severalpatients with normal hearing preoperatively and post-operatively.

All of the patients have had relief from their whirlingattacks of dizziness thus far, and, as this is the most dis-abling feature of the disorder, this fact is worthy of note.The procedure appears to be promising in the surgicaltreatment of intractable Ménière’s syndrome when reliefof vertigo with hearing preservation is desired.


54

CHIARI MALFORMATIONS ANDSYRINGOHYDROMYELIA IN CHILDREN

W. JERRY OAKES, M.D.

INTRODUCTIONChiari malformations, or hindbrain hernias, are being diag-nosed and operated upon with increasing frequency. Forthe purposes of this chapter, two separate entities will bediscussed. The Chiari I malformation is characterized bycaudal descent of the cerebellar tonsils. The brain stem andneocortex are typically not involved and the patient doesnot suffer from a myelomeningocele. Syringomyelia is com-monly but not invariably present. The Chiari II malforma-tion is almost always seen in conjunction with spina bifidaand is a more severe form of hindbrain herniation. The neo-cortex and brain stem are dysmorphic and the cerebellarvermis (not the tonsils) is displaced into the cervical spine.Accompanying the vermis are dysmorphic and elongatedaspects of the medulla and lower pons as well as the loweraspect of the fourth ventricle. Again, syringomyelia is com-monly associated with this lesion. Not discussed in thischapter is the rare Chiari III malformation.

CHIARI I MALFORMATIONWith the advent of magnetic resonance imaging (MRI) thedetection of caudal displacement of the cerebellar tonsilsand the presence of an associated syrinx has become safeand accurate. Typically the tonsils are at least 3 mm belowthe plane of the foramen magnum. They lose the roundedappearance of their caudal pole and become pointed or“peg-like.” This is associated with obliteration of the sub-arachnoid space at the craniocervical junction with the im-paction of tissues into this confined region. When all of theabove criteria are not met, the situation should be judged inconnection with the clinical symptomatology of the pa-tient. The presence of syringomyelia or other developmen-tal anomalies will assist in the interpretation of the intradu-ral findings at the craniocervical junction.

Patients with a symptomatic Chiari I malformation aregenerally offered operative intervention. The more severethe neurological deficit the stronger the case for interven-

tion. When occipital pain is the only symptom and no neu-rological signs are present, the degree of disability from thediscomfort should be carefully weighed against the risks ofthe procedure, prior to the implementation of surgical inter-vention. When syringomyelia is present, I generally favorintervention even with minimal symptoms. Intracranial pres-sure should be normalized prior to consideration ofcraniocervical decompression. Approximately 10% of pa-tients with Chiari I malformation will have hypertensivehydrocephalus and ventriculoperitoneal shunt insertionshould precede other considerations. Flexion and exten-sion views of the cervical spine are also important to re-solve questions of spinal stability and other bony anoma-lies. If significant basilar invagination is present, this issueshould be addressed prior to a posterior procedure whichmay add to spinal instability. It should be emphasized thatcomputed tomography of the brain without the use of sub-arachnoid contrast is notoriously poor in its ability to de-tect the presence of a significant Chiari malformation.

Once a candidate for surgery has been appropriatelychosen, the patient is prepared with preoperative corti-costeroids and a broad spectrum antibiotic. The patient ispositioned prone (Fig. 1) in a pin-type head holder withthe neck flexed. The head of the table is elevated some-what, but no central venous access is mandatory sincelowering the head will eliminate the gradient for air embo-lization. A chest Doppler monitor is used for the detectionof air embolization and to monitor slight changes in thepatient’s pulse. Patients are not allowed to breathe spon-taneously. This significantly lowers the likelihood of seri-ous pulmonary complications postoperatively. Musclerelaxants are allowed to become fully effective during theinduction of anesthesia to avoid the Valsalva maneuverduring placement of the endotracheal tube. A severe Val-salva maneuver has been associated with progression ofsymptoms in some patients.

The skin incision is made from a point 2 cm belowthe external occipital protuberance to the midportionof the spinous process of C-2 (Fig. 2A). It is quite© 1991 The American Association of Neurological Surgeons

55

Figure 1. Optimal positioning of a patient for exposing a Chiari malformation.

OAKES : CHIARI MALFORMATIONS AND SYRINGOHYDROMYELIA


56


Figure 2. A, operative exposure of a Chiari I malformation. Astent has been placed through the fourth ventricular outlet. B,Chiari I malformation associated with syringomyelia; midlinesagittal section.© 1991 The American Association of Neurological Surgeons

57

unusual for the tonsilar tissue to descend below the levelof the upper portion of C-2 (Fig. 2B), and by not removingthe important muscular attachments at C-2 the postopera-tive pain is significantly decreased and the likelihood ofpostoperative spinal deformity seen in conjunction withsyringomyelia is substantially lessened. The avascularmidline plane of the occipital musculature is divided withmonopolar current. No incision transecting muscle is nec-essary in this procedure since the plane dividing the leftand right muscular bundles completely separates thesetwo groups. A small amount of fat will mark this naturalcleavage plane. Again using the monopolar current, themuscle insertion immediately above the foramen magnumis separated from the occipital bone and the posterior archof C-1. The arch of C-1 itself is removed as well as theoccipital bone immediately above this (Fig. 2A). There isno need for lateral exposure and bone laterally situated isleft intact. This minimizes risk to the vertebral veins andarteries. The bone edges are waxed and the dura is openedin the midline. Initially, the dura over C-1 is opened. Care istaken as the incision is extended across the circular sinusnear the foramen magnum. This sinus can be formidableand should not be approached nonchalantly. With thedura retracted laterally, the arachnoid is opened in themidline. The subarachnoid adhesions are lysed with a sharpinstrument and are not simply torn. The arachnoid edge isthen clipped to the dural edge with metal clips.

The tonsils, which can be recognized by their verticalfolia. are separated in the midline to expose the floor of thefourth ventricle. Care is taken to free the caudal loop ofthe posterior inferior cerebellar artery and avoid damageto this vessel or the branches. Again adhesions are cutrather than torn. On separating the cerebellar tonsils, aveil of arachnoid is sometimes encountered. This veilshould be opened widely. Obstruction to cerebrospinalfluid flow can also occur from the posterior inferior cer-ebellar arteries. These vessels may approximate in the mid-line and be adherent to one another. They should be sepa-rated and mobilized laterally with great care. Options atthis point differ but I prefer to place a stent (Fig. 2A). Thisis composed of a short length of ventricular catheter (3 to5 cm), and on the outer surface of the catheter a section ofsoft drainage tube is sewn in place. The tip of the ven-tricular catheter is removed leaving an open hollow con-duit. The stent has two lumens, one within the ventricularcatheter and a second between the catheter and the drain.This stent is then carefully placed with one end of thetube lying in the midportion of the fourth ventricle and theother in the cervical subarachnoid space distal to the lowermargin of the cerebellar tonsils. Care is taken to positionthis stent in the subarachnoid space, not the subdural

space. The stent is held in position with a fine suturesewn to an avascular portion of the medial aspect of thepia over the tonsil. Closure is then accomplished with agenerous dural graft, either of cadaveric origin or har-vested from the nuchal ligament. A central dural tackupsuture is used to further expand the subarachnoid spacein this area. Closure is with absorbable suture which isknown to react minimally in the subarachnoid space.

Following the operation, patients may experience somenausea and vomiting as well as hiccups. These are almostalways self-limited. Neurological deficits which are well-established prior to the operation are unlikely to reversefollowing operative manipulation. Long-standing pain andtemperature loss is very unlikely to return. Hand and armweakness with fasciculations and loss of muscle bulk mayimprove functionally but will not normalize. A particularproblem exists when pain is a major component of thepresentation. Children and adolescents infrequently havea major problem with pain. Adults, however, may be quitediscouraged by the persistence of discomfort in the neck,shoulders, and/or arms. Pain may very well persist despitea physiologically successful operation with obliterationof the syrinx cavity. This limitation of surgical interven-tion should be carefully explained to the patient prior tosurgery. Mild scoliosis (less than 35 degrees) may im-prove or simply stabilize, whereas more severe spinal de-formity may well progress despite adequate treatment.With the advent of MRI scanning the status ofsyringohydromyelia can be assessed easily. If a sizablesyrinx persists months to years after craniocervical de-compression and symptoms attributable to this lesion areserious or progressive, consideration can be given to alaminectomy over the lower aspect of the syrinx and theplacement of a syrinx to subarachnoid shunt or to a syrinxto peritoneal shunt. If a syrinx to subarachnoid shunt ischosen, placement of the distal catheter in the free sub-arachnoid space is an important technical maneuver. Cath-eters can easily be mistakenly placed in the subdural spacewithout benefit to the patient.

CHIARI II MALFORMATIONChildren with a myelomeningocele may develop symp-toms referable to their hindbrain hernia. Symptoms andsigns are generally age-specific, with infants developinglower cranial nerve disturbances (difficulty with swallow-ing, weak cry, inspiratory wheeze, aspiration pneumonia,absent gag, and opisthotonos) and older children morecommonly developing progressive upper extremity spas-ticity. Ataxia of the trunk or appendages is recognizedmuch less often. Since some degree of hindbrain hernia-tion is present in the vast


58

majority of spina bifida patients, MRI evidence of hind-brain herniation must be accompanied by progressive orsevere symptomatology to warrant operative intervention.Many patients will remain clinically stable for long peri-ods despite significant deformity. As many as one-thirdof patients will develop difficulty with phonation, swal-lowing, or apnea by age three years. If the “asymptom-atic” remainder were followed for a longer period or if lessserious symptoms were considered significant this one inthree figure would undoubtedly be higher. Because thesymptoms of the Chiari II malformation are frequently lifethreatening, symptomatic Chiari II malformation is the lead-ing cause of death in the treated myelomeningocele popu-lation today. When treated conservatively, as many as 10to 15% of all patients will die from the malformation by theage of three years.

The decision for surgical intervention is controver-sial. Because there is a significant likelihood of stabiliza-tion or actual improvement with conservative care, somewould argue against operative intervention. This is sup-ported to some degree by autopsy material that demon-strates hypoplasia or aplasia of vital lower cranial nervenuclei. Against this, however, is the experience of numer-ous surgeons who have seen dramatic improvement inmany patients following decompression. In addition, ob-jective evidence of physiologic functioning has been re-ported to improve with both brain stem evoked responsesand CO

2 curve following operation. With these conflict-

ing pieces of evidence one can quickly appreciate thesurgeon’s dilemma.

With increasing experience, my willingness to opera-tively intervene is increasing. This is due to the relativelylow incidence of operative complications and the clearimprovement demonstrated by some patients. Poor resultsare more commonly due to a delay in offering operativeintervention. Once serious difficulties are clinically evi-dent with breathing, swallowing, or phonation, the situa-tion may very well be irreversible. In that case, the bestthat operation can be expected to do is to maintain thepoor level of lower cranial nerve function seen immedi-ately prior to operation. Problems with aspiration pneu-monia, apnea, and other life-threatening difficulties mayvery well persist.

The solution to this problem does not seem to be acontinuation of a conservative approach, accepting a 10 to15% mortality. Rather, an earlier identification of patients athigh risk for serious problems, and offering this group in-tervention, seems to be a more logical option. Being able todetect this high-risk group prior to the development of irre-versible life-threatening problems is a key provision.

If serious problems with phonation, swallowing, or

breathing are detected and normal intracranial pressure ispresent, urgent intervention is appropriate when full sup-port of the child is proposed. It is also important to empha-size that normalization of intracranial pressure is a prerequi-site to consideration of craniocervical decompression.Patients with questionable shunt function are well servedto first have their shunt revised. If progressive or serioussymptoms persist after adequate shunt revision, decom-pression of the craniocervical junction can be contemplated.Again, the MRI has made the diagnostic evaluation of thisgroup of patients almost risk-free and quite precise.

As with the Chiari I patients, corticosteroids and abroad spectrum antibiotic are given preoperatively. Ste-roids are not a substitute for a needed craniocervical de-compression. The anesthetic management and position-ing of the patient are similar to those for the Chiari I patient.Of some difference, however, is the fact that decompres-sion should extend to the level of the caudally displacedposterior fossa tissue. This is frequently below the levelof C-4. By removing this additional bone and displacingthe musculature, the risks of cervical deformity are sub-stantially increased even if the laminectomy is kept quitemedial away from the facets. Since the lower portion of thefourth ventricle is usually not within the posterior fossa,the occiput may need to be removed minimally if at all (Fig.3B). If it is elected to open the dura over the posteriorfossa, great care is necessary. The transverse sinus in thepatient with spina bifida is frequently placed very near ifnot at the level of the foramen magnum (Fig. 3B). An un-knowing opening of the dura and sinus in this area maywell lead to an operative disaster. The elasticity of thetissues of the cervical spine is pronounced. In removingthe laminal arch of small infants, each bite with the rongeurneeds to be crisp and clean. Undue distortion of the spi-nal cord may occur if this principle is not followed. It isimportant to study the preoperative MRI for the positionof the fourth ventricle, the cerebellar vermis, and the pos-sibility of a medullary kink. The position of all these struc-tures is critical to the intradural exploration.

Once the dura is opened, finding the caudal extentof the fourth ventricle can be difficult (Fig. 3A). Intraop-erative ultrasound may be of help in localizing this struc-ture. The choroid plexus usually maintains its embryonicextraventricular position, marking the caudal end of thefourth ventricle. When present, this is a reliable intraop-erative marker. Unfortunately, dense adhesions andneovascularity at points of compression or traction maybe found, especially near C-1, and this may make dissec-tion treacherous. The fourth ventricle may be coveredby vermis with its horizontal folia or the choroid plexusmay simply lie within the displaced ventricle.


59

Figure 3. A, operative exposure of a Chiari II malformation. B, Chiari II malformation: midline sagittal section.



60

If the purpose of the intradural manipulation is toopen the foramen of Magendie and provide free egress ofcerebrospinal fluid from the fourth ventricle, this can beaccomplished by the placement of a stent as with the ChiariI malformation patients. It is necessary to find and openthe tissue widely over the caudal aspect of the fourthventricle. It may happen that several planes of dissectionare developed before the floor of the fourth can be ad-equately appreciated. It is important during the explora-tion of each of these avenues that vascular and neuraltissues be preserved and that natural planes be devel-oped so that no irreparable damage to the delicate tissuesof the lower brainstem occurs. The caudal aspect of amedullary kink can easily be mistaken for the appropriatetarget. This dissection is one of the most difficult in pedi-atric neurosurgery. Errors or simple tissue manipulationmay convert a tenuous portion of the medulla or lower

pons to permanently damaged tissue. The surgeon shouldalways bear in mind the risk-benefit ratio for each of hisactions, and this particular area is unforgiving of evensmall excesses of manipulation. Grafting of the dura andclosure are similar to the previous description.

In addition to the avoidance of problems with infec-tion, hemorrhage, and increased neurological deficit, pa-tient selection and the timing of intervention are critical tothe successful outcome of decompressing a patient with aChiari II malformation. Despite what was thought to beappropriate and timely intervention, an alarmingly highpercentage of patients with lower cranial nerve abnormali-ties treated surgically eventually progress. This raises thequestion of whether the current strict selection criteria aretoo restrictive and whether less symptomatic infantsshould be considered for decompression. This area ofspeculation remains in dispute.


61

CAROTID BODY TUMORSFREDRIC B. MEYER, M.D.

THORALF M. SUNDT, JR., M.D.


CLINICAL PRESENTATIONThe most common presenting symptom in a patient with acarotid body tumor is a palpable neck mass. This may bedetected either by the patient or by the physician during aroutine examination. Other symptoms often include hoarse-ness, dysphagia, stridor, or, possibly, tongue weakness.

On examination, the neck mass is located below theangle of the jaw and is often laterally mobile but verti-cally fixed because of its attachment to the adventitia ofthe carotid artery. Although most of these masses have atransmittable pulse, an audible bruit is infrequent. Theincidence of cranial nerve involvement has been esti-mated at 20% and usually includes the vagus and hypo-glossal nerves. Rarely, a patient may also present with aHomer’s syndrome. The stridor and dysphagia may besecondary to either vagus involvement or compressionof the pharynx by the adjacent tumor. In either situation,these specific findings are most suggestive of a largetumor extending to the base of the skull.

In any patient with a carotid body tumor and hyper-tension, it is important to consider catecholamine secre-tion by the tumor or the possibility of multicentricityincluding a pheochromocytoma elsewhere. Therefore, onpreoperative evaluation, it is important to consider thepossibility of excess catecholamine production whichwould have significant anesthetic implications.

The differential diagnosis includes a bronchial cleftcyst, carotid artery aneurysm, primary or metastatic car-cinoma, adenopathy, nerve sheath tumor, and glomusjugulare tumor. All suspicious neck masses should beevaluated by radiological imaging studies. Injudiciouslocal biopsy may result in uncontrollable hemorrhage.

DIAGNOSISThe best diagnostic test at present for evaluating the ana-tomic pathology is a transfemoral cerebral angiogram.The presence of an enhancing oval mass widening theangle of the carotid bifurcation with displacement of both

the internal and external carotid arteries is essentiallypathognomonic of a carotid body tumor. Although theblood supply is primarily from the bifurcation and exter-nal carotid artery, contribution from the internal carotidartery, vertebral artery, and thyrocervical trunk can occur.On angiography it is important to note the cephalic ex-tent of the tumor blush to help plan surgical exposure.Bilateral carotid angiography is important to evaluatethe potential for collateral cerebral blood flow and toidentify possible multicentric paragangliomas of the con-tralateral carotid body and glomus jugulare.

Contrasted enhanced computed tomography (CT)and magnetic resonance imaging are useful to demon-strate the lateral and medial extent of the tumor. This isespecially important in noting the displacement of thepharynx. However, it may be difficult to determine whetherthe mass is an aneurysm as opposed to a carotid bodytumor on these diagnostic studies. Ultrasonography hasalso been advocated as a noninvasive diagnostic test inpatients with a strong family history or a history of acontralateral carotid body tumor.

NATURAL HISTORYAND SURGICAL SELECTIONThe malignancy potential of carotid body tumors hasbeen reported to range from 2.6 to 50%. The pathologicalcriteria for malignancy are based on the standard criteriaof cellular atypia and mitoses, local invasion, and dis-semination. However, unique for these tumors is that thehistological appearance does not correlate strongly withthe potential for malignancy. Currently, the metastaticrate for carotid body tumors is approximately 5%. Al-though metastatic spread occurs most commonly to theregional lymph nodes, metastases to the brachial plexus,cerebellum, lungs, bone, abdomen, pancreas, thyroid, kid-ney, and breast have been reported. Metastases shouldnot be confused with multicentricity of paragangliomasat other sites in the body.

Carotid body tumors will grow relentlessly if notresected. Several retrospective studies have reported amortality rate of approximately 8% in untreated cases.

62


Although some authors have reported palliation with ra-diation therapy alone, it is generally agreed that radia-tion is not an acceptable primary treatment. In addition,some authors have recommended radiation therapy forincomplete resections, although there is no convincingevidence to demonstrate that this is effective. The mor-bidity associated with unresected tumors is significantand includes progressive lower cranial nerve palsies, dys-phagia, airway obstruction, and extension to the skullbase with infiltration of the central nervous system. Ac-cordingly, in an otherwise medically healthy patient,complete surgical removal is the recommended treatmentof choice.

Epidemiological studies have demonstrated thatthere are two forms of this disease. First, there is the morecommon sporadic form in which there is a 5% incidenceof bilateral carotid body tumors. The second, less com-mon, form is a familial disease with an autosomal domi-nant transmission. Within this second group, there is a32% incidence of bilateral tumors. Therefore, if a posi-tive family history is obtained in the initial evaluation ofa patient, early examination of other family members isstrongly recommended because the ease of resection isbased on the size of the tumor. When patients who haveundergone resection of a carotid body tumor on one sidedevelop a contralateral tumor, surgical resection is rec-ommended only if 10th nerve function is intact on thepreviously operated side. Otherwise the risk of bilateral10th nerve palsies with a high potential for repeated as-piration would be unacceptably great. In that specificinstance it would be most prudent to observe the tumorthrough noninvasive means such as ultrasound or CTscanning.

SURGICAL TECHNIQUEThe surgical approach advocated here emphasizes sixfundamental concepts:

1) The preservation of cerebral blood flow during andafter the operation is critical. Therefore, all patients aremonitored with intraoperative electroencephalography.Furthermore, patients with large tumors in whom tempo-rary carotid artery occlusion may be required have intra-operative baseline occlusion xenon-133 cerebral bloodflow studies. Trial balloon occlusion of the internal carotidartery during the preoperative angiographic assessmentmay also be of benefit.

2) Distal exposure of large tumors at the base of theskull is obtained by mobilization of the parotid gland. Thisapproach facilitates identification of the lower cranial nervescephalad to the tumor and aids in their preservation.

3) These tumors are dissected in the capsular-adven-titial plane as opposed to the subadventitial plane whichhas been advocated by some surgeons. This plane is de-

veloped by using bipolar coagulation and magnifica-tion. This minimizes the risk of arterial wall injury anduntimely hemorrhage.

4) Great effort is taken to maintain the integrity ofthe external carotid artery since it is a potential source ofcollateral flow.

5) Although some authors recommend the routine useof a shunt in large tumors, shunts are only used when theelectroencephalogram and cerebral blood flow studies dem-onstrate insufficient perfusion during carotid artery occlu-sion. This minimizes the risks associated with a shunt andavoids an unnecessary arteriotomy. In most cases, meticu-lous dissection eliminates the need for either a temporarycarotid artery occlusion or shunt placement.

6) Since exposure is critical in successful removal ofthese tumors, a longitudinal incision is used extendingfrom the ear to the suprasternal notch along the anteriorsternocleidomastoid muscle. Although cosmetically lessappealing than the horizontal incision advocated bysome surgeons, it permits excellent exposure of both thedistal and proximal carotid arteries. In addition, the ipsi-lateral leg is prepared and draped for surgery in case asaphenous vein graft is required.

The patient is intubated with a wire endotrachealtube. It is important to have not only an arterial line formonitoring of blood pressure but also good venous ac-cess in case a rapid transfusion is required. As indicatedin Figure 1, the patient is positioned in a similar mannerto that used for a carotid endarterectomy. The ipsilateralcervical region is draped extending from the clavicleup to the face both anterior and posterior to the ear. Thelower part of the incision is similar to that used in aroutine carotid endarterectomy and parallels the ante-rior border of the sternocleidomastoid muscle, endingseveral centimeters superior and lateral to the supraster-nal notch. The cervical segment of the incision runsalong the anterior border of the sternocleidomastoidmuscle and ascends to a point just behind the lobe ofthe ear. In the lower quarter of the postauricular sulcus,it drops to the bottom of the ear, skirts the earlobe, andthen ascends in a pretragal skin crease to the superiorborder of the zygoma. After the proximal common ca-rotid artery is identified and vascular loops are placedaround it, the parotid gland is then mobilized prior tofurther dissection of the tumor.

The superficial cervical fascia is incised and the poste-rior border of the parotid gland is exposed and elevated. Theanterior-inferior surface of the auricular cartilage is followeddeep to its “pointer,” the triangular projection of cartilage atits medial limit. The temporoparotid fascia is incised be-tween the mastoid process and the posterior margin of theparotid gland, and the facial nerve is found adjacent to thefascia. A finger placed on the mastoid tip and directed ante-

63

MEYER AND SUNDT : CAROTID BODY TUMORS

Figure 1. The skin incision is located along the anterior borderof the sternocleidomastoid muscle but then extends to in frontof the ear to facilitate parotid mobilization. The probe adjacentto the parietal boss is for intraoperative cerebral blood flow

monitoring. The scalp has been wired with electrodes for con-tinuous intraoperative electroencephalographic monitoring.(Copyrighted by Mayo Foundation.)

riorly, the cartilaginous “pointer,” and the palpable junc-tion of the external auditory meatus (the tympanomas-toid suture) all point to the main trunk of the 7th nerve.Once the main trunk is identified, the lower division andmarginal mandibular nerve which form the upper limit ofthe deep dissection can be traced forward by sharp dis-section and elevated safely using mobilized parotid tis-sue as a “bundle.” The posterior belly of the digastricmuscle is followed to its point of insertion in the mastoidgroove and divided there. The stylohyoid muscle liessuperior and parallel to the digastric muscle and shouldalso be divided, exposing the deeper stylomandibularligament which must also be resected for adequate distalexposure of the internal carotid artery.

Although this relationship between nerves andmuscles is referred to by anatomists as a “retroparotid fossa,”the tissue is in fact densely bound by deep, thick cervicalfascia; until this fascia is divided it is not possible to el-evate or mobilize the parotid gland superiorly and thus itis not possible to expose and isolate the distal internalcarotid artery. For high exposure it is usually necessary toidentify the origin of the 7th nerve, but seldom is it neces-sary to trace this nerve distally into the parotid gland it-self. The use of fishhooks as refractors rather than heavyself-retaining refractors often avoids damage to the mar-ginal mandibular branch of the facial nerve.

After mobilization of the parotid gland and expo-sure of the distal internal carotid artery, the proximal com-

mon carotid artery is followed distally by dissection ofthe deep fascia anterior to the sternocleidomastoid muscle.This dissection is carefully extended cephalad to the bi-furcation where the caudal limits of the tumor are usuallyencountered. At this point vascular tapes are then placedaround the internal and external carotid arteries prior tofurther dissection of the tumor. Baseline and occlusionxenon-133 cerebral blood flow studies andelectroencephalography are performed. It is important toknow the potential for collateral blood flow ahead of timein case the carotid artery must be quickly occluded forhemostasis.

As depicted in Figure 2, the common facial vein isoften incorporated into the tumor capsule and must beligated along with veins draining into it from the sur-rounding tissue. The tumor is then isolated along its me-dial and lateral borders. The proper plane of dissection isidentified between the lower pole of the tumor and thecommon carotid artery using bipolar forceps under mag-nification. Since the tumor’s main blood supply is fromthe carotid bifurcation and external carotid artery, thedissection delineates this attachment first (Fig. 3). Usu-ally there is an areolar plane between the tumor and ar-tery except for its subadventitial attachment at the bifur-cation. With the use of bipolar cautery, the multipleperforating arteries arising from the vasa vasorum arecoagulated and divided (Fig. 3). The tumor is usually fedby large proximal branches of the external carotid artery,


64


Figure 2. The parotid gland has already been mobilized and therostral extent of the tumor delineated. The common facial veinis often incorporated into the tumor and must be ligated. Note

that fishhooks are used to retract both the skin and the parotidgland to prevent cranial nerve injury. (Copyrighted by MayoFoundation.)


65


Figure 3. After dissection of the medial and lateral borders of thetumor from the surrounding tissue, its attachment to the commoncarotid and external carotid arteries is attacked with bipolar cautery.

Not depicted in these diagrams are the vascular loops which have beenplaced around the common carotid artery and its major divisions forquick hemostasis if necessary. (Copyrighted by Mayo Foundation.)


66


which must be ligated individually (Fig. 4). The same istrue for feeding arteries from the vertebral artery and thy-rocervical trunk which develop in very large lesions. Thetumor is grasped with forceps and rotated superolaterallyto expose the tumor-carotid artery interface. By dissect-ing in the periadventitial layer close to the arteries, therisks of injuring the superior and recurrent laryngeal nervescan be minimized.

It should be noted that we have encountered, on sev-eral occasions during routine carotid endarterectomies, ananteriorly located aberrant vagus nerve. Furthermore, intwo cases of carotid body tumors, the vagus nerve wasactually incorporated within the tumor bed and had to becarefully dissected out. Identification of the vagus nervein reference to the tumor is greatly facilitated by mobiliza-

tion of the parotid gland. The other cranial nerve whichcan be injured at this point in the dissection is the hypo-glossal nerve. Usually the tumor displaces the hypoglos-sal nerve superiorly. Again, identification of the nerve inthe submandibular region is important for its preservation.The mandibular branch of the facial nerve can also beinjured by excessive retraction under the angle of the jaw,but the prior parotid mobilization usually provides suffi-cient room. In tumors with a large lateral extension, thespinal accessory nerve should be identified and protected.

After dissection and ligation of the feeding arter-ies from the common and external carotid arteries, thelateral and superior poles of the tumor are further mo-bilized (Fig. 5). Laterally and somewhat posteriorly itis common for the tumor to derive a large share of its

Figure 4. The tumor is dissected from the arteries in an areolarplane between the capsule and the adventitia. Only the tumor’sorigin from the adventitia is removed in a subadventitial plane.

Bipolar cautery is essential to coagulate the multiple perforatingvessels of the vasa vasorum without injury to the parent artery.(Copyrighted by Mayo Foundation.)


67


Figure 5. The superior and lateral poles of the tumor are further mobilized after the main blood supply tothe tumor has been ligated. (Copyrighted by Mayo Foundation.)


68


Figure 6. As a last step, the posterior wall of the tumor is dis-sected from the carotid artery complex and the cranial nerves.Rotation of the tumor improves exposure of this interface andfacilitates protection of the superior laryngeal nerve. After tu-

mor removal, adequate hemostasis is obtained and the wound isclosed in multiple layers over one drain placed between the ca-rotid artery and sternocleidomastoid muscle. (Copyrighted byMayo Foundation.)


69


blood supply from the carotid sheath. These vessels arequite large but can be usually controlled with bipolarcoagulation. This approach leaves the medial posteriorattachment between the internal and external carotid ar-teries for last. In this manner, better visualization of thevagus nerve and its very important branch, the superiorlaryngeal nerve, is achieved. As the tumor is elevatedfrom its bed, the superior laryngeal nerve is dissectedaway from the tumor capsule (Fig. 6).

After excision of the tumor, the arteries are carefullyinspected for any arterial wall injury, and cerebral bloodflow is again measured. If blood flow is low, or if a seg-ment of the artery appears suspicious, the artery is oc-cluded and a local arteriotomy is made. The appropriatearterial repair and, if necessary, endarterectomy are per-formed with a saphenous vein patch graft repair. The oc-casional massive carotid body tumor may extend up toand erode the foramen lacerum and petrous bone. Tumorsthat extend into the posterior fossa need to be approachedthrough a combined neck dissection and suboccipitalcraniectomy similar to that utilized in removal of the glo-mus jugulare tumor. However, remnants of tumor leftwithin the foramen lacerum are probably unresectable.

POSTOPERATIVE COMPLICATIONSThe major morbidity associated with carotid body tumorresection is lower cranial nerve palsies, specifically in-jury to branches of the vagus nerve including the supe-rior and recurrent laryngeal nerves. Rarely, the hypoglo-ssal nerve may be injured along with the sympatheticchain. In our experience, cranial nerve palsies occurredonly in tumors which were greater than 5 cm in length.Documentation of damage to the superior laryngeal nerveis difficult unless electromyography of the cricothyroidmuscle is performed. Clinically, these patients presentwith difficulty in swallowing and possibly aspiration.The proximity of the superior laryngeal nerve to the re-current laryngeal nerve means that, if there is injury toone, both are usually involved. However, hoarseness aloneimplies recurrent laryngeal nerve injury which can bedocumented on pharyngeal examination. Evidence forsuperior laryngeal nerve injury includes difficulty withswallowing or aspiration. If there is injury of the recurrentlaryngeal nerve, Teflon injection of the paralyzed vocalcord achieves good palliative relief. If injury to the supe-rior laryngeal nerve leads to persistent aspiration, then apercutaneous gastrostomy may ultimately be required.

70

OLFACTORY GROOVE MENINGIOMASJOSHUA B. BEDERSON, M.D.

CHARLES B. WILSON

INTRODUCTIONMeningiomas arising near the midline of the anterior cra-nial fossa may be divided according to their specific ori-gin from the anterior faIx, cribriform plate, medial orbitalroof, ethmoid region, planum sphenoidale, tuberculumsellae, diaphragma sellae, or anterior clinoid processes.Based on similarities between the symptoms, signs, andoperative approaches, traditionally these tumors havebeen separated into two groups, the olfactory groovemeningiomas and the suprasellar meningiomas. Olfac-tory groove meningiomas characteristically attain a largesize prior to diagnosis, unlike the suprasellar meningio-mas, which arise closer to the optic chiasm and causemuch earlier compression of the visual pathways.

CLINICAL PRESENTATIONAs with other meningiomas, women are affected morefrequently than men. The incidence is very low prior toage 30, and peaks in the fifth and sixth decades. Theclinical course spans a range of months to years, lastinglonger than 3 years in one-fourth of patients. The clinicalpresentation involves visual loss, frontal lobe dysfunc-tion, headache, anosmia, or seizures, either alone or incombination. Focal motor deficits are rarely noticed bythe patient, although on careful neurological examina-tion abnormalities may be detected in nearly one-third ofpatients. Similarly, anosmia is an uncommon presentingcomplaint, but it is almost universally detected on ex-amination. Unilateral sparing of smell is rare, but whenpresent it is important to document so that an attempt canbe made during surgery to preserve this function.

RADIOLOGICAL STUDIESHigh quality preoperative radiographic examination isessential. This should include magnetic resonance imag-ing (MRI) with gadolinium enhancement. Sagittal MRIreveals the relationships between the tumor and the ante-rior visual pathways, sellar contents, ventricular system,

and the anterior cerebral arteries (Fig. 1A). In tumors withbilateral extension, which is usual, sagittal or coronalMRI indicates the volume of tumor on each side, contrib-uting information essential for preoperative planning (Fig.1B). The extent of peritumoral edema and mass effect, thedegree of bony involvement, and the configuration ofthe frontonasal sinuses are all carefully noted. When MRIis unavailable, late generation computed tomography(CT) is obtained without and with intravenous contrastadministration, and should include direct coronal imag-ing. Although bilateral cerebral angiograms, demonstratethe relationships between the tumor and the anterior ce-rebral artery branches, comparable information can beobtained on a high-field strength MRI scanner, and forthis reason cerebral angiography is redundant. Depend-ing on the precise origin along the subfrontal dura, theprimary blood supply comes from branches of the ante-rior or posterior ethmoidal arteries. Capsular arteries arerecruited from the pial circulation. Preoperative embo-lization is precluded because only very rarely does thetumor receive any contribution from the external carotidcirculation.

PREOPERATIVE PREPARATIONRare contraindications to surgery can be found in elderlypatients whose asymptomatic tumors are discovered in-cidentally during workup of unrelated medical problems.However, there are currently no effective alternatives tosurgery. Unless medical problems, primarily cardiopul-monary, pose an unacceptable risk, removal is advised.The potential risks of surgery should be explained topatient and family and include, but are not limited to,stroke due to damage to anterior cerebral arterial branches,visual loss, frontal lobe retraction injury, postoperativehemorrhage, cerebrospinal fluid fistula, frontonasal si-nus infection, meningitis, and tumor recurrence.

Careful preoperative medical evaluation of cardiac,pulmonary, and renal status is performed. All patientsare treated preoperatively with dexamethasone and ananticonvulsant. Prior to surgery, 1 autologous unit offibrin adhesive and 2 units of packed red blood cells© 1991 The American Association of Neurological Surgeons

71

Figure 1. A, sagittal MRI showing the relationship betweenan olfactory groove meningioma and the anterior cerebralarteries, anterior visual pathways, sellar contents, and the ven-tr icular system. B, a T2 weighted axial MRI showing

peritumoral edema and dorsal displacement of the anteriorcerebral arteries. In addition, the volume of the tumor on eachside of the midline is indicated, which helps in preoperativeplanning.

are reserved in the blood bank. Arterial pressure is moni-tored by a radial artery catheter. Large-bore intravenousaccess is obtained, and a urinary bladder catheter is in-serted. All patients are fitted with sequential compressionstockings. Preoperative prophylactic intravenous antibi-otics are administered prior to the skin incision, and shouldinclude both Gram-positive and Gram-negative coverage.Dexamethasone, 20 mg, is administered intravenously.

Because intracranial pressure (ICP) is often elevatedin patients with larger tumors, induction of general endot-racheal anesthesia includes a period of preintubation maskhyperventilation after administration of a narcotic or thio-pental. Hypotension is rigorously avoided due to the riskof decreasing cerebral perfusion in the presence of increasedICP. Nasotracheal intubation is inadvisable due to the riskscaused by tumor involvement of the cribriform and eth-moidal regions. After intubation, controlled hyperventila-tion is used to keep the PCO

2 initially between 28 and 30

mm Hg. If further reduction of ICP is needed during expo-sure, the PCO

2 can be dropped below 25 mm Hg. Mannitol,

500 ml of a 20% solution, is administered intravenously atthe time of the skin incision.

SURGICAL TECHNIQUEPositioningThe patient is placed in the supine position, with theoperating table gently flexed and all pressure points care-fully protected. The head is elevated above the heart,facing straight up, slightly extended, and placed in three-point skeletal fixation (Fig. 2).

Operative TechniqueIn planning the operation, several features of olfactorygroove meningiomas are important: their usual bilateral-ity, frequently large size, broad dural attachment throughwhich most of the blood supply is derived, and intimaterelationships to the anterior cerebral arteries and, in thelargest tumors, the anterior visual pathways.

A coronal skin incision is used (Fig. 3A) and scalphemostasis is obtained by the application of compres-sive plastic clips. Dissection is carried out in the subgalealareolar plane, leaving the underlying pericranium intact.The scalp is reflected and retracted anteriorly to the su-praorbital ridge over rolled sponges (Fig. 3B). The poste-rior scalp flap is gently reflected

BEDERSON AND WILSON : OLFACTORY GROOVE MENINGIOMAS


72

Figure 2. Operative positioning of a patient for removal of an olfactory groove meningioma.

Figure 3. A, coronal skin incision. B, retraction of the anteriorand posterior scalp flaps; the dashed line represents the line of

periosteal incision. C, location of burr holes and outline of thebone flap.




73

back to expose the coronal suture. The pericranium isthen incised posteriorly along the coronal suture and bi-laterally along the superior temporal lines to the supraor-bital ridge. It is elevated from the bone, leaving its ante-rior attachment intact. Although a periosteal elevator canbe used, we have found it helpful to elevate the pericra-nium by firmly sweeping it forward with a dry sponge.Preservation of the intact pericranium is important forlater closure of the frontonasal sinus and for repair ofdural defects over the ethmoid sinus. Exposed tissues arecovered with a moist sponge.

Although a unilateral subfrontal approach has beenused by others, we prefer a bifrontal craniotomy. A freebone flap or osteoplastic bone flap can be used. The goalof this exposure is to open directly onto the orbital roofs,so that there is no overhanging frontal bone and brainretraction can be minimized. Bilateral burr holes areplaced in the “key holes” just behind the zygomatic pro-cess of the frontal bone, inferior to the temporal line.When using an osteoplastic bone flap, and for a right-handed surgeon using a free bone flap, a third burr hole ismade over the right anterior temporal fossa. This providesthe base for an osteoplastic flap and the access for laterexposure of the optic chiasm. The fourth burr hole is madeover the left posterior frontal lobe. A single midline ante-rior burr hole (the fifth) is made just above the nasion, sothat its inferior aspect is at the midline floor of the ante-rior fossa. With few exceptions, the frontal sinus will beentered. The sixth and seventh burr holes are made oneither side of the superior sagittal sinus just anterior tothe coronal suture (Fig. 3C). The bone is cut using the aircraniotome posteriorly and the hand-held Gigli saw forthe supraorbital cuts. This allows the latter cuts to beflush with the orbital roof. The osteoplastic flap is re-flected laterally and covered with a moist sponge. Thefree flap is stored in saline-soaked sponges. The com-monly used antibiotic solutions may be cytotoxic andshould not be used for storage of the bone flap. Bleedingfrom the sagittal sinus is controlled with thrombin-soakedGelfoam, and epidural hemostasis is obtained by place-ment of epidural tackup sutures. At this time, wire holescan be made in the bone edge and in corresponding pointsalong the bone flap.

The frontal sinuses are nearly always entered. Theyare denuded of mucosa, including that portion of thesinuses within the reflected bone flap. The frontal sinuseshave a variable anatomy and may extend posteriorly alongthe orbital roof to the planum sphenoidale, or laterally tothe edge of the orbit. All the mucosa must be removed inorder to avoid subsequent formation of a mucocele. Thenasofrontal ostia are plugged with muscle, and the si-nuses are filled with Gelfoam soaked in antibiotic solu-

tion and packed with bone wax (Fig. 4). Later, the sinuseswill be covered with a pericranial flap, but this should bedelayed until after the tumor has been removed. It is some-times necessary to resect much of the subfrontal dura (towhich pericranium would normally be sewn) when it isinvolved by tumor, and it may be necessary to use part ofthe pericranial flap to repair the dural defects. In somecases the frontal sinuses are small and despite openinginto them the mucosa remains intact. In this setting themucosa can be left in situ, carefully displaced toward theostia, and covered with Gelfoam and pericranium.

Transverse dural incisions are made over each fron-tal lobe just above the anterior edge of the craniotomyand are carried to the edge of the sagittal sinus. Usingself-retaining refractors and protecting the brain withmoist cottonoid strips, the frontal lobes are retracted lat-erally and the sagittal sinus is doubly ligated with 3-0silk sutures placed under direct vision through the con-vexity dura on each side and through the falx just be-neath the sinus (Fig. 5). These are left uncut for laterreapproximation of the dura. The sagittal sinus and falxare then incised. Larger tumors may incorporate the freeedge of the falx, but it can be cut above the tumor tomobilize the convexity dura.

The frontal lobes are gently retracted posterolaterally,exposing the rostral pole of the tumor near the midline.With the exception of cortical veins, vascular attachmentsbetween the frontal lobes and the tumor capsule are bipo-lar coagulated and divided, and an incision is made inthe base of the tumor at its junction with the anterior dura(Fig. 6). Every effort is made to preserve the cortical ves-sels by “sweeping” or “brushing” them off the tumor cap-sule. The anterior aspect of the tumor capsule is thenentered, and piecemeal removal of the tumor can begin atits most rostral dural attachment. Because bilateral fron-tal lobe retraction constitutes a major insult to the brain,any further brain retraction is focused on the right frontallobe. Most of the tumor decompression can be safelyaccomplished from this side, reserving further left frontallobe retraction for dissection of the plane between tumorand brain (see below). All tumor removal is performedunder the operating microscope.

The Cavitron ultrasonic aspirator or cutting cauteryloop is passed along the base of the tumor in the plane ofthe orbital roof and is then swept upward through the over-hanging tumor. Simultaneously, this separates the base ofthe tumor from the floor of the frontal fossa and interruptsthe tumor’s major blood supply (Fig. 7). All attachments oftumor to dura and bone are removed along the floor of theanterior fossa. The exposed feeding arteries entering thesessile base of the tumor are coagulated or their foraminapacked with bone wax. Subsequent internal tumor decom-


74


Figure 5. Technique of ligation of the superior sagittal sinus.


75

Figure 6. Gentle retraction of frontal lobes exposing the olfactory groove meningioma; piecemeal removal of the tumor is begun.

Figure 7. Sagittal section depicting the dissection of the tumor from the floor of the frontal fossa.




76


pression can proceed with relatively little bleeding ifundercutting precedes removal of the overhanging tu-mor. Piecemeal tumor removal is continued in this man-ner with little or no additional frontal lobe retractionuntil only a shell of tumor remains. As the posterior por-tion of the capsule is approached, great care is taken towatch for embedded branches of the anterior cerebral ar-teries.

As the tumor’s periphery is approached from within,the capsule becomes pliable and tends to collapse in-ward and forward. To minimize brain retraction the tumoris gently mobilized forward (rather than retracting thefrontal lobes posteriorly) by use of retention sutures orheavy forceps gently applied to the tumor capsule. Thesurface of the frontal lobe is gently brushed away fromthe tumor with moist cottonoid strips as the capsule ismobilized (Fig. 8). With the exception of invasive men-ingiomas, this plane is well-defined. At this stage it isnecessary to advance the brain refractors into the planebetween tumor and frontal lobe. If internal decompres-

sion and anterior mobilization of the capsule have beenachieved, this can be done with minimal additional re-traction pressure on the brain.

As the posterior part of the tumor capsule is dissected,major branches of the anterior cerebral arteries are identi-fied and spared. Some knowledge of their orientationwill have been obtained from the MRI scans, and theseshould be available in the operating room. Small branchesmay be embedded within the tumor and if they cannot bedissected free they should be divided to avoid avulsionfrom the pericallosal artery. At this stage it is sometimespossible to identify uninvolved dura of the medial sphe-noid wing behind the tumor and, working medially, toexpose the right anterior clinoid process, internal carotidartery, and optic nerve. With large tumors that overhangthese structures it may be necessary to use the right-sided exposure obtained during the opening to workdown the sphenoid wing from the right side. This can befacilitated by temporarily rotating the operating tableto the left. Large tumors compress the optic nerves

Figure 8. Technique of separation of the tumor capsule from the subfrontal cortex.


77

and chiasm, and the capsule must be reflected anteriorlyusing gentle dissection to separate it from the depressedchiasmal complex (Fig. 9A).

With the optic nerves and chiasm and carotid arter-ies protected by cottonoid strips, the remaining tumoralong with its entire dural attachment is removed (Fig.9B). This may require following the tumor into the eth-moid or sphenoid sinuses. Hyperostotic bone should beremoved with an air drill. The extent of bone removal isdetermined by the age of the patient, the vascularity ofthe bone, and the involvement of the paranasal sinuses.

ClosureExposed bone is packed with bone wax and any duraldefect over the sphenoid and ethmoid sinuses is coveredwith a pericranial graft. The pericranial graft can be freeor, when the anterior dura has been removed, it can be leftattached. This has the advantage of simultaneously cov-ering the previously packed frontonasal sinuses. If anextensive bone defect has been created, it is packed witha graft of adipose tissue or muscle prior to closure of theoverlying dural defect. The covered floor of the frontalfossa is then sealed with autologous fibrin adhesive.

Hemostasis in the friable tumor bed may be the mostdemanding part of the procedure. The retractor blades areremoved one at a time, and the underlying protectivecottonoids are floated off with direct irrigation. Bipolarcoagulation is used to occlude the thinwalled veins onthe surface of the compressed brain. The irrigating solu-tion should remain clear after Valsalva’s maneuver. Thebrain cavity is then lined with a sheet of Surgicel.

The dura is first reapproximated using the sagittalsinus sutures and then closed in a watertight fashion ifpossible. If the dura has been torn extensively duringreflection of the bone flap, it may be necessary to repairthe defect with a pericranial patch graft. If it has not beendone already, the pericranium is reflected over the packedfrontal sinuses and sewn to the dura. Epidural hemostasisis obtained using tack-up sutures, bone wax, bipolar co-agulation, and either Gelfoam or Surgicel. The dura istacked up to the surrounding bone if not already done,and the intervals between tack-up sutures are gentlypacked with strips of Surgicel. All dural bleeding pointsare coagulated and the sagittal sinus is covered with astrip of Gelfoam if any oozing persists. In the case of anosteoplastic flap, the bone flap is waxed. The undersurfaceof the detached temporalis muscle is inspected and anybleeding points coagulated. The bone flap is replacedand wired to the surrounding skull with 28-gauge stain-less steel wire. The anterior wires are tightened first tominimize the cosmetic defect caused by the gap

between bone flap and skull. The anterior and midlineburr holes are covered with titanium wire mesh or filledwith methylmethacrylate cement. The incisions into thetemporalis muscle are closed. The scalp flap is inspectedand any bleeding points coagulated. The galea is closed,beginning with a single suture joining the flaps in theexact midline. Raney clips are removed and large bleed-ing vessels coagulated as they are encountered. The skinis closed with vertical mattress nylon sutures.

POSTOPERATIVE CAREOf primary concern in the postoperative period is brainedema. Although no important cortical draining veinsare interrupted by this procedure as a rule, frontal loberetraction may cause significant brain edema. Swellingmay begin soon after surgery and is maximal 2-3 dayslater. Dexamethasone, 16-24 mg/day, is administered andcan be increased temporarily if edema worsens. Serumosmolarity is measured every 6 hours and 50 g of manni-tol is administered intravenously to maintain an osmo-larity between 285 and 300. Modest restriction of freewater intake is maintained for 3 days.

Unlike masses in other locations, postoperative fron-tal fossa hematomas may not cause focal neurologicaldeficits until brain stem herniation is imminent. There-fore a high index of suspicion should be maintained, andany deterioration in mental status must be investigated.New postoperative visual or motor deficits, depressedmental status, or any deterioration of neurologic status inthe postoperative period should prompt a CT scan.

Compression of the hypothalamus or pituitary stalk bytumor and surrounding edema may lead to transient postop-erative diabetes insipidus. Strict recording of intake, output,daily weight, and serum electrolytes is routine. Urine outputgreater than 200 ml/hour in combination with a urine spe-cific gravity less than 1.005 may be sufficient to institutetreatment with aqueous vasopressin, administered subcuta-neously, and titrated to reduce urine output and maintainnormal serum electrolytes and osmolarity.

If cerebrospinal fluid (CSF) rhinorrhea occurs it istreated initially by lumbar subarachnoid CSF drain-age via a closed system. In the immediate postopera-tive period while brain swelling is maximal, great caremust be taken not to precipitate rostrocaudal brain stemherniation by this treatment. The leak will stop duringdrainage, which is continued for 4 days. After this pe-riod, the drain is clamped and evidence of furtherleakage is sought. If there is any question as to theexistence or location of a CSF leak, labeled cottonpledgets are placed bilaterally in the superior nasalcavity under direct vision. Radionuclide is injected


78

Figure 9. A, separation of the tumor capsule from the opticnerves and chiasm. B, view of the floor of the anterior cranial

fossa after complete removal of the tumor along with its duralattachment.

into the subarachnoid space and the cotton pledgets ex-amined 24 hours later for evidence of radioactivity. Dif-ferential activity in the pledgets can direct subsequenttransnasal or, rarely, subfrontal reparative procedures tothe proper location.

Late postoperative bone flap infections are rare,but transgression of the frontonasal sinuses may createa predisposition for this complication. Removal of the

bone flap may be required, followed by a course of antibi-otic therapy and delayed cranioplasty.

Tumor recurrence is rare following complete ex-cision of a true olfactory groove meningioma. Whenit does occur, reoperation is generally indicated. Ifcomplete excision is not possible, or in rare cases ofinvasive meningioma, postoperative radiotherapy isindicated.



79

CEREBRAL ANEURYSMSAT THE BIFURCATION OF

THE INTERNAL CAROTID ARTERYEUGENE S. FLAMM, M.D.

INTRODUCTIONThis chapter will focus on the operative management ofpatients with aneurysms arising in the region of the bifur-cation of the supraclinoid internal carotid artery. Thesecomprise approximately 5% of all intracranial aneurysms.Of our 40 patients with aneurysms of the bifurcation ofthe internal carotid artery, 70% were Grades I-III, 10%were Grades IV and V, and 20% had unruptured aneu-rysms. There is no specific clinical presentation that in-creases the probability of finding an aneurysm in thislocation. The special surgical considerations for this an-eurysm concern the local anatomy, particularly the ar-rangement of the perforating arteries near the bifurcationof the carotid artery and the path of the anterior choroidalartery in relation to the aneurysm. Although there aremany similarities in the management of all patients withsubarachnoid hemorrhage (SAH) from aneurysms, spe-cific details of this location will be stressed.

SELECTION OF PATIENTSAND TIMING OF SURGERYAll patients with SAH from aneurysms at the carotid bi-furcation are considered for surgery regardless of age.Some consideration must be made for the neurologic gradeof the patient. In those patients who are Grades I-III andalert, surgery is carried out on the earliest regular operat-ing day. In the past two years we have abandoned the ideathat surgery must be performed within the first three daysafter SAH or else delayed until two weeks. At the presenttime, we will operate on any day after SAH if the patientis alert and does not have significant vasospasm on an-giography or increasing velocities as measured bytranscranial Doppler ultrasonography.

OPERATIVE INDICATIONSIn patients who have had a SAH, there is little to con-traindicate surgery of a ruptured aneurysm of the carotidbifurcation provided that the medical and neurologicalcondition of the patient will permit surgery to be per-formed safely. The indications for surgery of an unrupturedaneurysm in this location are similar to those for otherlocations. Because of the position of this aneurysm deepin the medial sylvian fissure, there is an increased ten-dency for intracerebral hemorrhage, and, for this reason,surgery prior to SAH may be even more appropriate withthis aneurysm location.

PREOPERATIVE PREPARATIONThe preoperative care of patients with SAH is as impor-tant to their overall outcome as the actual surgical proce-dure. The major steps, once the diagnosis of a SAH ismade, include determining the neurologic status of thepatient, localizing the site of the aneurysm, and embark-ing on a course that will prepare the patient for surgicalobliteration of the aneurysm.

Timing of AngiographyWe prefer to carry out angiography soon after the patientis admitted. This is done not on an emergency basis butwhen the full neuroradiologic team is available. Althougha second study prior to delayed surgery is necessary, thefirst angiogram provides an early diagnosis which is ofobvious importance for managing the patient.

Preoperative ObservationAll patients are cared for in a specialized neurosurgi-cal intensive care unit. General issues of managementdeal with control of the patient’s blood pressure, seda-tion and relief of headache, and fluid balance. A goalfor blood pressure is at the patient’s norm prior to SAH.Care is taken to avoid hypotensive levels, partic-© 1991 The American Association of Neurological Surgeons

80


ularly in patients with a previous history of hyper-tension.

FluidsNo attempt is made to restrict fluid intake in patientswith SAH. The usual daily intake is 2500-3000 ml/ dayeither orally or intravenously. This is regulated by use ofa central venous. pressure maintained between 8 and 10cm H

2O. Colloid is liberally used to maintain blood pres-

sure and central venous pressure in the desired ranges.The is particularly important in patients who developcerebral vasospasm either with or without clinical signsof ischemia.

AnticonvulsantsAll patients with SAH are placed on phenytoin as soon asthe diagnosis has been made. This is continued postop-eratively for three to six months even in patients whohave never had a seizure.

CorticosteroidsCorticosteroids are not used routinely during the preop-erative period unless there is evidence of increased in-tracranial pressure. In preparation for surgery patients arestarted on methylprednisolone 12 hours prior to surgery.An intravenous dose of 250 mg every six hours is given;this is continued at this level during the first three post-operative days and then tapered over the next four to fivedays. While on steroids, patients receive antacids or his-tamine blockers.

SURGICAL TECHNIQUEAnesthetic Techniques and Aids to ExposureTo maximize the exposure of the circle of Willis and re-duce the amount of retraction required, several adjunctsare utilized. An intravenous infusion of 20% mannitol(0.5-1.0 g/kg) is begun at the time of the skin incision.Patients also receive furosemide (40 mg) on call to theoperating room. Another important adjunct is spinal drain-age. A catheter introduced through a Touhey needle isinserted into the lumbar subarachnoid space after induc-tion of anesthesia. The drainage is not opened at thistime. When the dura has been exposed and tented thespinal drainage is begun. This delay prevents strippingaway of the dura which may cause epidural bleeding thatis difficult to control. Furthermore it is easier to open theleaves of the arachnoid in the sylvian fissure if there issome cerebrospinal fluid (CSF) present. In addition tothe relaxation of the brain and reduction of the need forretraction that this method provides, it facilitates themicrodissection since the surgeon can work in a drierfield and does not constantly have to remove CSF whileworking on the aneurysm. Spinal drainage is

avoided in patients with moderate amounts of atrophysince intracranial access to CSF is easy. The risk of exces-sive removal with subsequent collapse of the brain isthus avoided.

The final step to achieve a slack brain is to main-tain the PaCO

2 in the range of 25-30 mg Hg before the

dura is opened; thereafter PCO2 is kept between 30 and

35 mg Hg.

Operative PositioningAll intracranial microsurgery should be performed withthe head secured by skull fixation. This is necessary toreduce any movement of the head which will interferewith the microdissection that is carried out at × 16-20magnification. A system that permits this as well as theattachment of self-retaining refractors and other adjunctsis particularly helpful.

The head is turned to a full lateral position and thevertex is dropped slightly toward the floor. The zygomais almost parallel with the floor. This permits better directvisualization of the region of the optic nerve and carotidartery and reduces any obstruction to vision by thetemporalis muscle or floor of the anterior cranial fossa. Afolded sheet or shoulder roll should be placed beneaththe ipsilateral shoulder to reduce the amount of stretchplaced on the carotid artery, jugular vein, and brachialplexus (Fig. 1A).

DrapingSkin towels are sutured or stapled to all areas in front ofthe hairline. Towel clips are never used. This reduces thechance of damage to the skin and facilitates the reflec-tion of the scalp flap.

Skin IncisionAs shown in Figure 1B, a curvilinear incision at or justbehind the hairline is used. It extends from the level ofthe zygoma, 1 cm in front of the tragus, to a point be-tween the midline and pupillary line. The incision linemay be infiltrated with local anesthetic but epinephrineis not used. Care is taken to preserve the superficial tem-poral artery although this is not always possible. Thepterional craniotomy is the same for virtually all aneu-rysms of the carotid circulation and upper basilar artery.

Operative ProcedureAlthough aneurysms at the distal end of the internalcarotid artery are among the less frequent aneurysmsof the anterior circulation, they can be a most chal-lenging problem because of the size that they may at-tain, the increased likelihood of intraoperative rup-ture, and the involvement of several major vessels,namely the anterior and middle cerebral arteries, the

81

FLAMM : CEREBRAL ANEURYSMS

Figure 1. A, overall view of the patient positioned for a rightpterional craniotomy. The head is turned fully to the left and the

right shoulder is elevated. B, detail of the incision line which isbehind the hairline and is only slightly curved.

anterior choroidal, as well as the internal carotid arteryitself.

The temporalis fascia is incised with a scalpel; thispermits closure as a separate layer which reduces postop-erative swelling in the area. The scalp flap is reflectedinferiorly in a single layer using the cutting cautery toseparate the temporalis muscle from the skull. The bonylandmarks that are utilized are the zygomatic process ofthe frontal bone, the lateral margin of the supraorbitalridge, and the zygoma. The muscle is reflected downwarduntil these structures

are visualized or, in the case of the zygoma, easily pal-pated (Fig. 2A).

A single burr hole is placed in the temporal re-gion. From this hole a free bone flap measuring 5 × 4cm is created with the craniotome. An initial cut ismade from the burr hole to the sphenoid wing as far asis possible. The craniotome is then returned to the burrhole and the remainder of the flap created; it is usuallynecessary to crack the bone at the sphenoid wing byelevating the flap. It is essential that the bony openingbe flush with the floor of the frontal fossa. (Fig. 2B). A


82


Figure 2. A, burr hole and bone flap superimposed within the scalp flap. B, bone flap removed. C, bonyopening after reduction of the sphenoid wing. Dural incision is indicated.


83


minor variation is made when a left- or right-sided expo-sure is used. For a right-handed surgeon, a leftsided cran-iotomy should have 1-2 cm more frontal exposure to per-mit the free introduction of instruments without cominginto contact with the margin of the flap. This is less of aproblem when the exposure is on the right side becauseinstruments in the surgeon’s right hand will cross the tem-poral lobe.

The lateral aspect of the sphenoid wing is thenrongeured away so that it will offer no obstruction to theline of vision. It is not necessary to drill this away be-cause 1.5 cm of the wing can easily be removed withrongeurs. After dural tenting sutures have been placedthe spinal drainage is opened.

A linear dural incision is made along the base of theexposure, 2 cm from the craniotomy margin. Care is takennot to coagulate the dura so that a good dural closure canbe performed. If necessary, the middle meningeal arteryis secured with small titanium clips (Fig. 2C).

It is difficult to describe all the nuances of the dis-section techniques used in aneurysm surgery. The firststep is the identification of the optic nerve. Once this hasbeen seen the exposure is maintained with a self-retain-ing retractor. The landmarks for identifying the optic nerveare the olfactory tract and the sphenoid wing. The nerveis found at the point of intersection of these two struc-tures (Fig. 3A). A second self-retaining retractor is placedon the medial edge of the temporal lobe. At times it isadvisable to coagulate and divide some of the temporalbridging veins before any retraction is carried out.

The operating microscope is now brought into useand the dissection proceeds at increasing magnifica-tions as the carotid artery and aneurysm are exposed.The first step of the microdissection is to divide variousarachnoid connections. This frees the aneurysm fromany undue traction and increases the room in which tooperate within the subarachnoid space. The arachnoidbetween the frontal lobe and optic nerve is first divided.This plane is extended laterally into the medial portionof the sylvian fissure. As the arachnoid is divided addi-tional retraction increases the exposure of the carotidartery. Once this arachnoid is opened the dissection canproceed along the carotid toward the point of origin ofthe aneurysm (Fig. 3B).

Even before these planes are well-established thesurgeon should develop a mental picture of the locationof the neck. Should rupture occur before the dissectionhas been completed, it is helpful to have a good idea ofwhere the neck is so that a rapid and accurate dissectionand application of the clip can be carried out while bleed-ing is controlled by the suction.

A specif ic detail of importance when operatingupon carotid bifurcation aneurysms is to identify

both the anterior cerebral and middle cerebral arteriesbefore beginning the dissection of the neck of the aneu-rysm. The dissection of the neck should begin by work-ing along the respective Al and MI segments back towardthe aneurysm rather than starting the dissection directlyat the neck (Fig. 3C). In my experience this reduces therisk of intraoperative rupture of the aneurysm by reduc-ing the manipulation of the dome itself.

A watertight closure of the dura should also be at-tempted. The bone flap must be securely fastened to pro-vide for a good cosmetic result. It is advisable to tie thesuperior frontal bone flap suture first so that this area getsthe best approximation of the bone edges.

A subgaleal drain with a closed vacuum system isplaced beneath the scalp flap for 24 hours. This reducesthe periorbital swelling that may develop.

Vascular Considerations

Posterior Communicatingand Anterior Choroidal ArteriesThe relationship of several branches of the carotid, middlecerebral, and anterior cerebral arteries to an aneurysm ofthe carotid bifurcation must be considered during the dis-section and when the clip is applied. One must be certainnot to overlook a relationship between the aneurysm andbranches of the posterior communicating and anterior cho-roidal arteries. Although these arteries arise proximal tothe bifurcation of the carotid artery, their course posteri-orly may carry them very close to the neck or fundus of ananeurysm at the top of the carotid artery (Fig. 3D). Thismust be remembered when placing a clip across the neckof the aneurysm, the blades that pass deep to the carotidartery must not occlude these other vessels as they passthrough the subarachnoid space behind the carotid artery.The artery should be located before the clip is applied sothat it is not included in the clip as it passes deep to theaneurysm. This is a small but important detail of this aneu-rysm location that should be stressed.

Recurrent Artery of HeubnerBy virtue of its recurrent nature, the artery of Heubnercourses laterally past the region of the carotid bifurca-tion. It is essential that any relationship between the an-eurysm and this vessel be delineated to avoid compro-mising this very important artery.

Lenticulostriate VesselsAlthough the lenticulostriate arteries that arise fromthe A1 and M1 segments, respectively, are usuallynot in the immediate vicinity of the carotid bifurca-tion, their position must be confirmed in each case tobe certain that they are not compromised by the dissec-

84



85


Figure 3. A, initial view through the microscope to indicatethe relationship of the olfactory tract, optic nerve, carotidartery, and sphenoid wing. B, exposure of the proximalsupraclinoid carotid artery. Note the placement of the re-fractors to facilitate the opening of the sylvian fissure. C,

identification and dissection of both the anterior and middlecerebral arteries as well as the aneurysm neck. D, final dissec-t ion in preparat ion for appl icat ion of the cl ip. Note therelationship of the branches of the carotid artery to the neckof the aneurysm.

86


tion or the clip. Occasionally, a very distinct vessel of100-200 µm may be found in close association with theneck of the aneurysm.

Clip ApplicationThe selection of the appropriate clip should begin earlyin the course of the dissection so that the surgeon has agood idea which clips and clip appliers will be best suitedfor the particular aneurysm. It is important to rememberthat the neck of an aneurysm becomes wider than theinitial diameter when it is compressed. It is important tohave a clip long enough to account for this increase.

Although smaller carotid aneurysms can usually besafely clipped by placing the clip at right angles with theparent vessel, it is often better to apply the clip so that theblades are parallel with the parent vessels (Fig. 4, A andB). This is particularly important when dealing with largerthick-walled aneurysms. Failure to do this increases thechances of compromising the lumen of the vessel or pro-ducing a kink in the parent vessel.

An additional problem encountered with carotidaneurysms, especially larger ones, is the tension withinthe aneurysm. This often prevents the clip from closingcompletely; there is also an increased chance of ruptureif the clip does not completely obliterate the aneurysmwhen it is applied. Several techniques are available toreduce the tension within the aneurysm. Temporary oc-clusion of the internal carotid artery in the neck dramati-cally reduces the pressure in the supraclinoid carotid ar-tery and within the aneurysm. This is utilized for largeaneurysms arising from the proximal supraclinoid carotidartery. For aneurysms at the bifurcation of the carotidartery, temporary clips can be applied directly to thesupraclinoid carotid artery. They must be positioned sothat the working space necessary to clip the aneurysm isnot compromised.

Another technique that has been helpful with largethick-walled aneurysms is suction decompression. A 21-gauge scalp vein needle with the flanges removed isconnected to the operating room suction (Fig. 4C). Bypuncturing the dome of the aneurysm where it is thick,blood can be auctioned through the aneurysm and theintraluminal tension reduced. Although this may notcause a thick walled aneurysm to collapse, the aneu-rysm becomes softer and more pliable. The clip can thenbe closed down easily and more safely. Blood loss hasnot been more than 100 ml when this technique hasbeen employed.

In almost all cases the aneurysm should be punc-tured and opened after it has been clipped. Only inthis way can the surgeon be certain that the goal ofobliterating the aneurysm has been achieved. It is

surprising how often an aneurysm may bleed when this isdone after a seemingly perfect clip application. The onlyexception to this procedure is when no further adjust-ment of the clip is safe or possible. While this occurs withsome of the larger ophthalmic region aneurysms, it shouldnot pose a problem for the distal carotid aneurysms. Incases where the aneurysm is not punctured, a postopera-tive angiogram is used to ensure complete obliteration ofthe aneurysm.

Intraoperative Neurophysiologic MonitoringWe have come to rely on intraoperative monitoring formost aneurysms. When dealing with aneurysms of thecarotid bifurcation, particularly when temporary occlu-sion of the carotid artery is anticipated, we utilize bothquantitative electroencephalography (EEG) and soma-tosensory evoked potentials (SEPs). Electrodes for stimu-lation and recording these parameters are applied beforethe patient is draped. Oftentimes the nature of the aneu-rysm is such that no departure can be made from the pro-cedure even when changes in EEG and SEP are noted.They should, however, be used to alert the surgeon of theneed for possible readjustment of the retractor positionor pressure. Changes in these neurophysiologic guide-lines are also helpful to determine the advisability ofcontinuing temporary occlusion of the carotid artery.

Specialized InstrumentationOperating MicroscopeThe usual arrangement for the operating microscope isthe Contraves mount for the Zeiss microscope. The pre-ferred objective lens is 300 mm; this allows ample roomto work between the lens and point of focus. It is essentialto be able to focus the microscope without using one’shands; this allows the surgeon to keep both hands in thefield at all times and make minor adjustments of the fo-cus as the dissection proceeds. This is particularly impor-tant when the arachnoid is being dissected from the neckof the aneurysm. The surgeon should be seated, and thesurgeon should be free to move and change positions, aswell as to shift the microscope, with minimal effort.

InstrumentationFairly rigid, moderately heavy bipolar forceps with dif-ferent tip sizes and angulations are used. By using heavierstainless steel, rather than titanium instruments, it is pos-sible to use them both for dissection and coagulation.The lighter instruments do not have sufficient spring tobe used in a spreading fashion to divide the arachnoidand do not provide enough proprioceptive feedback tothe surgeon.

A variety of sharp arachnoid knives is an important

87


Figure 4. A, incorrect clip placement may result in distortion ofthe carotid bifurcation. B, placement of the clip parallel to thebifurcation of the carotid artery preserves the normal anatomical

relationships and normal flow patterns. C, final view of occludedaneurysm with suction decompression to facilitate clipping andensure complete obliteration.


88


part of the equipment needed. These can be used effec-tively in places where scissors cannot; small disposableknives such as the Beaver blades can be used to goodadvantage for this purpose. Less traction is transmitted tothe aneurysm if the arachnoid is divided sharply ratherthan tearing it with forceps.

An adequate array of microsurgical instruments isnecessary. In addition to the standard instruments, a groupof dissectors with varying curved and straight tips, bothon bayonet and straight handles, are helpful for retract-ing vessels and the aneurysm itself. A No. 7 vented Frazier-type suction tip is used on the left side with a vacuum of100 mm Hg. A larger tip is kept on the right side for thefirst assistant in the event of rupture of the aneurysm.

Although it is not appropriate to discuss all of thedetails of aneurysm clips here, certain points should bestressed. Several suitable clips should be selected andloaded by the surgeon before the dura is opened. My ownpreference is for larger clips that will fit the aneurysm. Inthe past few years I have relied on a full complement ofSugita-type clips.

The use of a self-retaining system for brain retrac-tors is essential for all microsurgery and particularly forsurgery of aneurysms. Since the operative exposure issmall, there is usually insufficient room for the hands ofan assistant on a brain retractor. Furthermore, the retrac-tion must be gentle and precise; this can only beachieved with mechanically secured retractors. For an-eurysms of the carotid bifurcation, a minimum of twoself-retaining refractors are needed to part the lips of thesylvian fissure.

POSTOPERATIVE MANAGEMENTAND COMPLICATIONSThe goals of postoperative care are to maintain ad-equate cerebral perfusion, to reduce any postoperativecerebral swelling and increase in intracranial pressure,and to prevent the occurrence of seizures. These aimscan be accomplished by continuing the preoperativemedical regimen of corticosteroids, anticonvulsants,and fluids.

Upon transfer to the recovery room, blood pressure ismaintained at normal to slightly elevated levels. Centralvenous pressure is maintained at 6-8 cm H2O by admin-istering colloid and/or whole blood. Corticosteroids, inour practice 250 mg of methylprednisolone every sixhours, are maintained at this level for three days and thentapered over the next five days. Levels of anticonvulsantsare determined initially after surgery and during the post-operative period.

Since most aneurysms have been opened intraopera-tively we do not routinely perform postoperative angiog-raphy. If there is any change in the patient’s neurologiccondition or if the aneurysm was not opened at surgery,angiography is carried out.

At present all patients are being managed with adihydropyridine calcium channel blocker to reduce theincidence of cerebral vasospasm and delayed ischernia.The efficacy of this regimen is monitored with transcranialDoppler measurements and follow up angiography whenindicated by neurologic deterioration. We prefer to docu-ment the occurrence of vasospasm rather than presumethat it is the cause of any neurologic deterioration.

89

TREATMENT OF UNILATERAL ORBILATERAL CORONAL SYNOSTOSIS

JOHN A. PERSING, M.D.JOHN A. JANE, M.D.

INTRODUCTIONSynostosis of the coronal suture may involve all or just apart of the suture. Different degrees of involvement of thesuture result in distinctly different skull shapes.

The patient with partial or unilateral coronal synos-tosis is characterized by ridging of the prematurely fusedhalf of the coronal suture, flattening of the ipsilateralfrontal and parietal bones, bulging of the ipsilateral squa-mous portion of the temporal bone, and bulging of thecontralateral frontal and parietal bones. Radiographically,in addition to sutural sclerosis, the harlequin abnormal-ity (relative elevation of the greater wing of the sphenoidbone ipsilateral to the fused suture) is present. Basal com-puted tomographic (CT) scanning demonstrates narrow-ing of the sphenopetrosal angle, ipsilateral to the fusedcoronal suture, and deviation of the anterior cranial basefrom the midline toward the side of the fused coronalsuture.

The patient with bilateral coronal synostosis, how-ever, characteristically has ridging of the coronal suturebilaterally, flattening of the caudal portion of the frontalbones and supraorbital ridges, and bulging of the ceph-alad portions of the frontal bones. The occiput is flat-tened, and the squamous portion of the temporal bone isexcessively prominent. The vertex of the skull is moreanteriorly situated than normal. The skull frequently takeson a turribrachycephalic, or “tower shaped,” appearance.

The diagnosis of bilateral coronal synostosis is usu-ally made clinically but is supported radiographicallyshowing “sclerosis” of the coronal suture, associated withthe bilateral harlequin deformity of the greater wing ofthe sphenoid in the orbits. CT scanning is supportive indemonstrating bony fusion across the sutural marginsand a narrowed sphenopetrosal angle bilaterally.

Because the shapes with differing suture involve-ment are so disparate, so are the operative treatments. Thetreatments for unilateral and bilateral coronal synostosiswill be addressed separately.

UNILATERAL CORONAL SYNOSTOSISPreoperative preparation includes autologous or desig-nated blood donation for patient use perioperatively, andprophylactic antibiotics to be used intraoperatively, ex-tending 24 hours postoperatively. No steroids,anticonvulsants, or lumbar cistern drains are used.

Anesthetic monitoring technique includes arte-rial and central venous lines, Foley catheter, and O

2

monitor for blood loss (and replacement); and Dop-pler, endtidal CO

2, and nitrogen monitors for venous

air embolism. The child’s trunk and extremities arewrapped circumferentially around with soft cotton,and all irrigation fluids are warmed intraoperativelyto preserve body heat.

The timing of surgery in patients with coronal synos-tosis, optimally, is within the first few weeks of life. Thisis done in order to reduce the potential effects of increasedintracranial pressure on brain growth and to take advan-tage of the ameliorative effect on skull shape by the re-maining normal growth of the brain. Bone remodelingtechniques are also easier in the younger child.

Distinction is made between the child who is lessthan 1 year of age with respect to bone remodelingtechniques versus the child who is older than 1 yearof age, as the bone becomes much more brittle withincreasing age.

The patient is placed supine on the operating tableon a padded headrest. Only minimal hair (maximum width,1 cm) is clipper-prepared at the time of surgery. Remain-ing hair anterior to the intended incision line, if longenough, is covered with masking tape or aluminum foilfollowing braiding. The face is prepared into the field tobe able to judge orbital and frontal symmetry intraopera-tively. The bifrontal region is exposed by a bicoronalincision extending to the tragal region bilaterally. Dis-section of the anterior scalp flap is in the supraperiostealplane to reduce blood loss.

Burr holes are placed bilaterally at the pterion, andparasagittally posterior to the coronal suture (Fig. 1).Alternatively, if the anterior fontanel is patent, entry to© 1991 The American Association of Neurological Surgeons

90



91

PERSING AND JANE : UNILATERAL OR BILATERAL CORONAL SYNOSTOSIS

Figure 1. A, unilateral coronal synostosis. Bifrontal craniotomy isperformed removing both the hypoplastic and protuberant frontalbone. Further removal of the basal extension of the involved coronalsuture (blue dashes) into the cranial base is performed by rongeur. B, 1,the shortened mediolateral axis of the supraorbital rim is augmentedby an interposition bone graft (yellow), supported posteriorly by anadditional bone graft along the supraorbital rim. The whole rim is thenreplaced inferiorly to equal the supraorbital rim height on the oppositeside. 2, the zygomatic process of the frontal bone is cut obliquely andthe supraorbital rim and lateral wall of the orbit moved anteriorly to

increase the superolateral projection of the orbital rim (3). C, thefrontal bone undergoes radial osteotomy and is reshaped to achievesymmetrical frontal contour. The squamous portion of the temporalbone ipsilateral to the fused suture is removed, and radial osteotomiesare performed to allow bone remodeling. D, the squamous portion ofthe temporal bone is returned in place and secured posteriorly andinferiorly. The frontal bone is secured to the advanced orbital rim. Asegment of anterior parietal bone is removed to create a neocoronalsuture. E, the temporalis muscle is advanced forward and attached tothe lateral portion of the orbital rim and frontal bone.

92


the epidural space may be gained at the fontanel’s mar-gin, in place of a separate burr hole. A bifrontal cran-iotomy, incorporating the coronal suture, is performed,avoiding placement of a burr hole in the region of theglabella. In young patients, a bifrontal craniotomy yield-ing a solid single bone segment may not be possible asthe metopic suture is patent. This patency allows thetwo frontal bone segments to be independently mobile.However, the periosteum, left intact by supraperiostealdissection, aids to tether the bones and keep them inrelative alignment.

The greater wing of the sphenoid bone which isthickened and displaced superiorly, ipsilateral to the fusedcoronal suture, is rongeured away to the level of the lat-eral supraorbital fissure. This removes the lateral aspectof the abnormal basal portion of the “coronal ring” (e.g.,the frontosphenoid suture). The orbit ipsilateral to thefused suture is shortened mediolaterally. The superiororbital rim then is augmented by interposition and sup-porting bone grafts placed following osteotomies to theroof and lateral orbital wall (Fig. 1B). As the superior rimalso is often displaced cephalad compared to the con-tralateral side, it is positioned caudad to equal the heightof the opposite orbital rim.

The orbital rim is advanced to a slightly overcor-rected position and not tethered to the cranial base or theposterior body of the zygoma. Instead, bone grafts, har-vested from the parietal region, are inserted posterior tothe advanced orbital rim to wedge it forward. They aresecured to the rim only by suture or wire. The dura isplicated in the frontal region contralateral to the fusedcoronal suture to achieve frontal symmetry. Radially ori-ented osteotomies are placed into the center of the con-vex “contralateral” (to the fused coronal suture) frontalbone and the flattened “ipsilateral” frontal and parietalbones (Fig. 1C). This bone is remodeled by selective frac-tures using the Tessier rib bender to achieve the desired,symmetrically convex, form. The squamous portion ofthe temporal bone ipsilateral to the fused suture is re-moved by osteotomy and cut radially to allow contour-ing with the mallet and Tessier rib bender. The temporalbone is attached by wire suture to the surrounding boneanteriorly, inferiorly, and posteriorly to prevent recur-rence of the temporal bulging. The frontal bone plate isattached to the orbital rims superiorly and laterally but isnot attached posteriorly to the parietal bone. A rim ofpartetal bone approximately 5-10 mm wide is excised tocreate, in effect, a “neo”-coronal suture (Fig. 1D). Thetemporalis muscle is reflected anteriorly to fill in the gapbetween the orbital rim, which has been advanced, andthe squamous portion of the temporal bone (Fig. 1E). Thescalp flap is then closed in two layers (galea and skin).

“LATE” UNILATERAL CORONALSYNOSTOSISIn the patient with unilateral coronal synostosis treatedafter 1 year of age, the correction of skull irregularities isinitially much the same as in the child less than 1 year ofage. Burr holes are placed in the pterion region bilater-ally and the parasagittal region medially behind the hair-line. No burr holes are placed in the glabella region; thefrontal bone is fractured forward here following weaken-ing osteotomy of the outer table of the skull with a side-cutting air-driven drill bit. A bifrontal craniotomy is per-formed, leaving 5 mm of frontal bone height cephalad tothe apex of the orbital rim. The dura is plicated in thefrontal region contralateral to the fused suture in the areaof excess frontal prominence. An osteotomy of the orbitalrim ipsilateral to the fused suture is performed as describedearlier. This orbital rim is advanced to a slightly overcor-rected position compared to the contralateral side. Thehollow, immediately posterior to the advanced orbitalrim, is filled in with bone chips harvested from the pari-etal bone region, either as split or full segment cranialbone grafts.

The bulging in the squamous portion of the temporalbone is addressed by elevating this bone by craniotomyand radially cutting it. Kerfs or endocranial grooves throughthe inner table of the skull are oriented transverse to theradial osteotomies, on the concave surface of the bone.The dura is plicated locally and the bone is flattened bymallet and returned into place. A major distinction is nowmade from the younger patient (less than 1 year of age) inthe frontal bone remodeling (Fig. 2). The frontal bone iscut into vertically oriented “slats” of bone. The perios-teum is allowed to remain on the external surface of thebone to aid in bone alignment if fracture of the more brittlebone occurs during reshaping. Kerfs, or bone-weakeningchannels, are placed transverse to the long axis of the bone“slats” endocranially, to allow more accurate remodelingof the bone and achievement of frontal symmetry. The newlyshaped “frontal bone” is attached to the supraorbital mar-gin and to the more posteriorly located temporal and pari-etal bones if the child is older than 3 years of age. A gap isallowed to remain posteriorly (creating a neocoronal su-ture) if the child is between 1 and 3 years of age. The scalpis then closed. The nasal radix deviation is ordinarily notsurgically corrected in infancy.

BILATERAL CORONAL SYNOSTOSISThe patient with bilateral coronal synostosis is preparedfor surgery similar to the patient with unilateral coronalsynostosis, including arterial pressure and air embolusmonitoring. One major difference, however, relates tooperative positioning.

93


Figure 2. Late unilateral coronal synostosis. A bifrontal cran-iotomy is performed, but contrary to the younger patient, only aminimal amount of the lateral wing of the sphenoid is removed.The frontal bone is cut into vertically oriented slats. Theundersurface of each slat receives kerfs (green arrows) oriented

transversely and is remodeled to achieve a symmetrical form. Theorbital rims are advanced by orbital osteotomy with bone graftsinserted to lengthen the mediolateral axis of the superior orbit.Additional bone grafts are placed in the region of the pterionanteriorly to prevent postoperative hollowing (inset).

The patient is placed in a modified prone position inorder to correct both the frontal and occipital abnormali-ties associated with bilateral coronal synostosis. Beforeplacing the patient in this position, however, it is impor-tant to assess the stability of the cervical spine and thecraniovertebral junction by preoperative lateral cervicalspine roentgenograms in flexion and extension. Position-ing the patient on the operating table is greatly aided bya vacuum-stiffened bean bag to mold to the upper bodyand neck. The face and arms must be padded with largeamounts of cushioning foam to prevent pressure soresand compression nerve palsies.

The approach for the child in infancy under theage of 1 begins with a supraperiosteal dissection ofanterior and posterior scalp flaps to expose the frontal,parietal, and occipital regions. A transversely orientedperiosteal incision is placed 1 cm above the orbitalrims, and an incontinuity, subperiosteal, andsubperiorbital dissection is completed. The temporalismuscle is elevated out of the temporal fossa, and theoccipital musculature, from the nuchal line in the oc-cipital bone cephalad to the level of the superior rimof the foramen magnum caudad, is likewise elevatedsubperiosteally. The skull abnormalities are then reas-

sessed. If there is an absolute elevation of the height ofthe skull associated with shortening of the skullanteroposteriorly, accompanied by periorbital hypopla-sia, as is ordinarily the case, the procedure proceeds asfollows: Burr holes are placed in the pterion regions bi-laterally, and parasagittally in the anterior parietalbonejust posterior to the coronal suture (Fig. 3, A and B).Similarly, a biparieto-occipital bone graft is outlined withmultiple burr holes adjacent to the sagittal and trans-verse sinuses. If the metopic or sagittal sutures are widelypatent and preclude elevation of the frontal or parieto-occipital bones as a single piece, separate frontal orparieto-occipital bone grafts should be elevated on eachhalf of the skull. Once the bone is elevated both frontallyand parieto-occipitally (Fig. 3C), further dissectionepidurally maybe carried out below the level of the trans-verse sinus to allow the surgeon to fracture outward (pos-teriorly) the occipital bone. These outfractures or “barrelstaves” increase the “bony capacity” by enlarging theperimeter of the skull locally, allowing later brain anddural displacement in this region as the height of theskull is reduced. Barrel stave osteotomies in the occipitalbone in the midline and paramedian regions are longerthan those placed further laterally, to achieve elonga-


94



95


Figure 3. A, bilateral coronal synostosis. The skull in bilateralcoronal synostosis demonstrates flattening occipitally and bulgingfrontally with an anteriorly displaced skull vertex. B, burr holes areplaced as indicated on the lateral view of the skull. C, a bifrontalgraft is elevated. A biparieto-occipital bone graft is developed poste-riorly leaving a bony bridge between the two hemicrania across theocciput. The remaining occipital bone undergoes barrel stave os-teotomy. An osteotomy in the orbital roof, lateral wall, and floormobilizes the orbital rim in the form of the letter “C.” The orbital

rims are advanced and secured with parietal bone grafts inserted intoan oblique osteotomy in the zygoma. D, the vertex of the skull isshifted posteriorly by severing the parietal bone struts, removing asegment of bone, and relocating the strut posteriorly (blue arrow).This reduces the prominence of the frontal “bossing.” The basal andparietal bone segments are cinched together while the intracranialpressure is monitored. E, the frontal and temporal bone grafts arereattached to the basal skull, but the parieto-occipital bone graft isallowed to “float” with attachments only to the underlying dura.

96


tion of the anteroposterior axis of the skull without fur-ther widening of the parieto-occiput. The occipital bonebarrel staves are fractured posteriorly, at the base of theskull, and then inwardly at their distal margin so they donot create a pressure point on the overlying scalp.

In the lateral sphenoid region, the thickened andabnormally elevated superior portion of the greater wingof the sphenoid bone is removed by rongeur to the levelof the lateral supraorbital fissure. With this bone removal,the basal extension of the coronal suture, the lateralfrontosphenoid suture, is removed. “C”-shaped orbitalosteotomies (superior, lateral, and inferior orbital wallosteotomy to the level of the inferior orbital foramen) areperformed in both orbits to increase orbital rim projec-tion bilaterally. The advanced orbital rims are held for-ward by parietal bone grafts which are wedged in theosteotomy site in the body of the zygoma and secured tothe rim anteriorly. A craniotomy is performed to removethe abnormally convex squamous portion of the tempo-ral bone. The bone is reshaped by a combination of radialosteotomies into the center of the convexity and con-trolled fractures of the bone segments with the Tessier ribbender. The dura is plicated locally, and the temporalbone returned, to be secured to the surrounding base ofthe skull posteriorly and inferiorly.

At this point, two parietal bone struts remain, ex-tending from the vertex of the skull to the basal portionof the temporal and parietal bones. These twocephalocaudally oriented struts of bone are severed attheir caudal interface with the more basal skull and areshifted posteriorly approximately 1-2 cm (Fig. 3D). Theshift is performed in order to reduce the bulging contourof the dura in the superior frontal region and replace pos-teriorly the elevated vertex point which has become dis-placed anteriorly. If further contouring of the dura isneeded, dural placation sutures are placed at this time.

in order to safely reduce the height of the skull, atwist drill hole is placed in the right paramedian parietalbone and a pressure transducer is inserted to measureintracranial pressure (ICP) (Fig. 3E). Wire and nylon su-tures are passed through drill holes in the lateral bonestruts and the basal skull. The wire is slowly cinched downover the course of 30-60 minutes, and the intracranialpressure is continuously monitored under conditions ofnormocapnia and normotension. Presently, we recom-mend that while the vertex of the skull is being cincheddown, cerebral perfusion pressure of approximately 60mm Hg be maintained. During the initial stages of theheight reduction, however, the intracranial pressuremay be elevated (>20 mm Hg) for short periods. If thesurgeon does not observe a rapid reduction in intrac-ranial pressure over the course of 1-2 minutes, then a

smaller increment in reduction of the cranial vault heightis necessary over a longer period of time. Ultimately, thesame degree of height reduction may be achieved, but itmay take longer to do so. Caution must be exercisedduring this maneuver to prevent brain injury.

While the incremental cinching-down process is be-ing achieved, the frontal and parietal bone are remodeledto give the contour desired. Radially oriented osteoto-mies in the frontal and parieto-occipital bones allow forreshaping by a series of controlled fractures with the mal-let and the Tessier rib bender. After completing the re-shaping of the frontal bone, it is attached with wire su-tures to the orbital rim. The posterior aspect of the frontalbone is not secured with wire and is allowed to “floatfree.” The parieto-occipital bone segment adjacent to thebarrel staves in the occipital bone is shortened to allow agap of approximately 510 mm between the bone edges.This is done in order to encourage further displacementof the neurocranial capsule posteriorly, postoperatively.The parieto-occipital bone graft is attached to the sur-rounding dura, but is not fixed to bone posteriorly tofurther aid this posterior displacement process.

The temporalis muscle is advanced forward to attachto the lateral portion of the supraorbital rim and the scalpincision closed. No drains are used. A postoperative skullmolding cap is routinely used as an additional guide toaid skull shape normalization for an additional 2-3months.

“LATE” BILATERAL CORONAL SYNOSTOSISThe patient who is older than 1 year of age is treatedsimilarly in terms of positioning and craniotomy lines tothe child less than 1 year of age. The difference is that thebone at 1 year of age is more difficult to reshape, fractur-ing more readily, by the methods employed at less than 1year of age. Therefore, the operative technique is modi-fied as follows.

Following bifrontal and biparieto-occipital cran-iotomy, the orbital rims are advanced forward in the shapeof the letter “C” bilaterally. Bone grafts are wedged in atthe osteotomy site in the zygoma and secured bytransosseous wires. Barrel stave osteotomies are performedin the parieto-occipital region; the staves are fracturedposteriorly to elongate the anteroposterior axis of theskull. The squamous portion of the temporal bone is re-moved by craniotomy. The remaining cephalocaudallyoriented parietal bone struts extending from the vertex ofthe skull are severed, and an intracranial pressure moni-tor is inserted in the right parietal bone. The parietal bonestruts are displaced posteriorly approximately 1 cm,and the height of the skull is slowly reduced by cinch-ing down on the wire loops in the basal parietal and

97


Figure 4. Late bilateral coronal synostosis. Bifrontal and biparieto-occipital bone grafts are elevated. The orbits undergo “C”-shapedorbital rim osteotomy and advancement. Occipital barrel stavesare developed. The parietal bone struts are shifted posteriorly and

shortened to reduce the frontal contour projection. The frontalbone and parieto-occipital bone are split into vertically orientedslats and undergo remodeling aided by kerfs (green arrows) placedon the endocranial surface of the bone.

temporal bones. We have observed that it requires a longertime for intracranial pressure reductions to occur in olderchildren when compared to similar incremental reduc-tion in vault height in younger children, perhaps due tothickened dura with increasing age or scarring of the durafrom previous surgery. Therefore, smaller increments ofadjustment should be anticipated over a given time pe-riod. The cerebral perfusion pressure should be maintainedat approximately 60 mm Hg, under conditions ofnormocapnia and normotension. The frontal and parieto-occipital bones are then cut into vertically oriented “slats,”approximately 2 cm wide, leaving the periosteum at-tached to the external surface of the skull bone (Fig. 4).Transversely oriented kerfs are placed through the in-ner table of the skull (concave surface of the bone), andcontrolled fractures are performed on the “slats” for re-shaping. The individual bone segments are wired or su-tured with long-acting absorbable suture to give thedesired form. The temporal bone, because of its thin-ness, ordinarily can be molded by radially oriented os-teotomies and kerfs on the inner table of the skull, with-out cutting the bone into “slats.” This allows easierfixation to the basal skull, posteriorly and inferiorly.Bone chips from the parietal region are used to fill

in the temporal hollow behind the advanced orbital rim,and the temporalis muscle is transferred anteriorly andattached to the supralateral orbital rim.

Patients who are older than 3 years of age may havethe bone fixed in a slightly overcorrected position toallow for further brain growth, yet still provide the stabil-ity necessary for reconstruction of the skull. In the childbetween 1 and 3 years of age, the brain is still growingrapidly, and provision must be made for this by allowinggreater room for growth by not fixing the remodeled bone,particularly in the anteroposterior axis, while still em-ploying the bone remodeling techniques for the olderchild. Skull molding caps are generally not used postop-eratively in patients undergoing this procedure who areolder than 1 year of age.

The complications associated with this procedure arefew, but the possibility of cerebral infarction or mortality isstill present. Blood loss and hypothermia are the immediateintraoperative concerns, and greater blood loss should beexpected with osteotomies of the basal vault bone. Postop-eratively as well, diffuse oozing from all the osteotomy linesneeds to be anticipated and frequent checks of the bloodcount should be taken. If intravenous fluid loading is notgiven at the initiation of the operative procedure, venous air


98


embolism may occur. Lastly, the major concern with thereduction in skull height is the possibility of cerebralherniation. If the skull height is reduced with the provi-sion for ICP monitoring, and the guidelines for

reduction are met as described earlier, the risk is smalland no negative neurologic sequelae have yet been notedin our cases.

99

CONVEXITY MENINGIOMASARAH J. GASKILL, M.D.

ROBERT H. WILKINS, M.D.

INTRODUCTIONThe convexity meningioma arises from the meninges overthe cerebral convexity, and as it grows it indents the un-derlying brain. Its main blood supply ordinarily comesfrom regional meningeal arteries such as the anteriorbranch of the middle meningeal artery shown in Figure 1.At times the convexity meningioma may grow throughthe dura mater into the overlying calvarium; hyperosto-sis of the overlying bone is a typical response. When thatgrowth pattern occurs, the tumor may also receive someof its blood supply through scalp vessels such as thesuperficial temporal artery. In rare instances, the menin-gioma grows completely through the calvarium and pro-duces a second continuous mass beneath the scalp.

PATIENT SELECTIONThe mere presence of a convexity meningioma is notjustification for its removal. For example, a small asymp-tomatic tumor that has been discovered by accident in anelderly woman can be followed by serial computed to-mography (CT) or magnetic resonance imaging (MRI)examinations, and if it remains asymptomatic and doesnot increase significantly in size, it does not require sur-gical excision. On the other hand, surgical treatment isappropriate in an otherwise healthy patient with a rea-sonably long life expectancy, especially if the tumor is ofmoderate or large size, if it has been shown by serial CTor MRI studies to be increasing in size, or if it is produc-ing symptoms.

SURGICAL TECHNIQUEIn designing an operation to remove a convexity menin-gioma, the surgeon should try to interrupt as much of theblood supply as possible early in the operation. Ordi-narily this is done as the scalp flap is reflected and thedura mater is opened. In the unusual circumstance of avery vascular lesion, preoperative embolization of thetumor should be considered.

Because a convexity meningioma is ordinarily a be-nign tumor that is easily accessible, every effort shouldbe made to obtain a cure by a complete tumor removal.This dictates the removal not only of the main tumor, butalso of any portion of the dura mater and bone thought tobe invaded by the tumor. Therefore, the surgeon shouldplan for dural grafting, which is usually necessary, andcranioplasty, which is also required at times.

The operation may be performed with the patient ina supine, lateral, prone, or semi-sitting position, depend-ing on the specific location of the lesion. In general, thehead should be positioned so the region of the tumor ismost prominent. The proposed cranial opening shouldbe centered on the tumor and should be larger than thetumor to permit its removal.

For the ordinary convexity meningioma, we preferto use a scalp flap that is based anteriorly, laterally, orposteriorly and is turned down away from the superiorsagittal sinus (Fig. 2A). A free bone flap (i.e., nothinged on the temporalis muscle or other soft tissue)


Figure 1. Selective external carotid arteriogram, lateral view. Thetumor blush of a convexity meningioma is demonstrated. Theprincipal source of blood supply to the meningioma is the anteriorbranch of the middle meningeal artery.


100


is then created in one of several ways. Several burr holescan be made and connected with a high-speed drill, cran-iotome, Gigli saw, or rongeurs (Fig. 2A). or a single perfo-ration can be made and the bone flap formed with a high-speed drill or craniotome. The latter technique is fasterand there are fewer irregularities in the skull when thebone flap is replaced at the end of the operation. How-ever, the dura mater is more likely to be cut flush with thebone edge (especially in an elderly individual in whomthe dura mater is thin and is very adherent to the calva-rium), making it more difficult to close in watertight fash-ion at the end of the operation.

Depending on the vascularity of the tumor, there maybe excessive bleeding from the bone edges along the lineof separation around the craniotomy flap. This can be mini-mized by rapidly separating the bone flap from the under-lying dura mater. This may be done by using two periostealelevators inserted into the bony opening at some distanceapart, to pry the bone flap away from the dura. A thirdinstrument such as a No. 2 Penfield dissector is then usedsimultaneously or subsequently to separate the bone flapfrom the underlying dura and tumor (Fig. 2, B and C). Thesurface of the dura is covered temporarily with absorbablegelatin sponge and cottonoid strips to control bleeding

Figure 2. Scalp and cranial flap for removal of a convexitymeningioma. A, outlines of typical scalp and bone flaps for re-moval of a convexity meningioma. B and C, elevation of the

bone flap and separation of the underlying dura mater and tumorfrom the inner table of the skull.


101

GASKILL AND WILKINS : CONVEXITY MENINGIOMA

from this source, and bone wax is spread into the exposeddiploic spaces of the calvarium around the periphery of thebony opening to stop bleeding from that source (Fig. 3).

Periodic drill holes are made through the bone edgearound the cranial defect and heavy nonabsorbablesutures are inserted for use during the closure (Fig. 3).Cottonoid strips are placed between these sutures andthe skin edge to prevent possible bacterial contamina-tion. The dura. is then opened circumferentially aroundthe tumor, with a margin of uninvolved dura. at least 1cm in width (Fig. 4). As the dura. is opened, meningealvessels are sealed shut with bipolar coagulation.Hemostatic clips may be used, but will cause somedegree of artifact on postoperative CT and MRI

examinations. The dura mater is then tacked up aroundthe periphery of the bony opening using interrupted 3-0nonabsorbable sutures (Fig. 4, inset). A brain spoon orsimilar shield is inserted during this suturing to preventneedle injury to the underlying brain. At the previouslymentioned drill holes, the dura is tacked to the bone.Between these it is tacked to adjacent soft tissue such asthe periosteum or temporalis fascia.

If it is small, the meningioma. and its attached duraare gradually separated from the adjacent and underly-ing cerebral cortex and are removed en bloc (Fig. 5). If itis larger, and especially if a portion of the cortex overliesthe equator of the tumor, the center of the meningiomacan be removed by various instruments

Figure 3. Application of wax to the bone edges for hemostasis and a technique of creation of drill holes foranchoring the bone flap.


102


Figure 4. Technique of dural opening; insertion of dural tack-up sutures (inset).


103


Figure 5. Technique of removal of a meningioma en bloc. Bipolar coagulation of vascular feeders to the tumor (inset).


104

Figure 6. The dura mater is closed in watertight fashion using a graft of periosteum or temporalis fascia.As the bone flap is replaced, the dura is tacked up to it.



105


(depending on its consistency) such as bipolar cautery,suction, cutting loop cautery, ultrasonic aspiration, andlaser vaporization. This reduces the bulk of the tumorand permits the separation of its external surface from thebrain without excessive brain retraction.

In either case, the separation of the tumor from thebrain is done gradually around its circumference and thenbeneath it as the tumor is retracted away from that portionof the cortex receiving the surgeon’s attention (Fig. 5).The surgeon coagulates and divides all vessels that bridgebetween the cortex and the tumor but tries to leave theintrinsic cortical vessels intact. At all times the surgeonshould attempt to minimize the damage done to the cor-tex by the process of separation; however, some damagealmost always occurs, and the surgeon should warn thepatient preoperatively that a temporary or permanentneurologic deficit, appropriate to the location and sizeof the tumor, might result. We find it helpful to insert acottonoid strip between the tumor and cortex at eachlocation after the separation has been made (Fig. 5).Thus, when the surgeon has completed the separation

around the periphery of the meningioma, a circumferen-tial series of cottonoid strips is in place. The tumor is thenlifted out of its bed and any remaining attachments arecoagulated and divided (Fig. 5, inset).

After hemostasis has been achieved, the defect leftby the tumor removal is filled with warm irrigation solu-tion and the wound is closed in anatomical layers. A fas-cia] graft ordinarily can be obtained from the exposedperiosteum or temporalis fascia to be sewncircumferentially in watertight fashion to fill the duraldefect (Fig. 6). If not, fascia lata or a dural substitute canbe used. If a small portion of the bone flap appears to beinvolved by tumor, it should be removed before the flapis replaced; if the majority is involved, the entire flapmay need to be sent to the pathology laboratory. As thebone flap is sutured back into position, the underlyingdura and dural graft are tacked up to its center (Fig. 6).Any residual bony defects can be f illed withmethylmethacrylate or some other cranioplasty material.The scalp is closed in two layers and a compressive headdressing is applied.

106

OCCIPITAL LOBECTOMYMILAM E. LEAVENS, M.D.

INTRODUCTIONOccipital lobectomy is a useful procedure that may bepartial or complete, depending on the patient’s patho-logic condition and the extent of involvement of theoccipital lobe and adjacent brain. Gliomas, arterio-venous malformations, metastatic tumors, and, less of-ten, meningiomas are the usual indications for perform-ing a lobectomy.

The boundaries and important anatomical featuresof the occipital lobe and adjacent brain are shown inFigure 1. The occipital lobe has roughly a pyramid shape,with four triangular sides. The relationship of these exter-nal features of the occipital lobe to the cranial sutures,dural sinuses, ear, mastoid bone, and vein of Labbé isshown in Figure 2. Placement of the bone flap is based onthese anatomical features so that the exposure is adequatefor performing the lobectomy while avoiding injury tothe dural sinuses and avoiding opening of the mastoidair cells.

SURGICAL TECHNIQUEThe anesthesia used is continuous Sufenta (sufentanil cit-rate) opioid infusion with Norcuron (vecuronium) musclerelaxant, and a small amount of Forane (isoflurane, USP)inhalation agent. The patient is placed in the lateral posi-tion, supported by a bean bag and axillary roll, with thehead in a three-quarters position on a doughnut (Fig. 1C)or in a three-pin fixation head holder. This is followed byroutine preparation of the skin and draping. The patient isinfused intravenously with an antibiotic, usually 1 g ofAncef (cefazolin sodium) during the operation and thenpostoperatively every 6 hours for 24 hours. The patientalso receives intravenously a single maintenance dose of300 mg of phenytoin and 4-6 mg of dexamethasone every4 hours. The PaCO

2 is kept between 25 and 30 mg Hg, and,

if needed to further control intracranial pressure, 0.5 to 1 g/kg of mannitol is given intravenously.

The scalp incision and flap, position of the boneflap, and dural incision for a right-sided lobectomy areshown in Figures 1 and 2. The occipital pole is adjacentto the torcular herophili and is near the inion.

The parieto-occipital fissure is near the lambdoidsuture plane at the midline. The lambdoid suture is 6.5cm from the inion.

Incision and ExposurePart of the skin incision is in the midline extending fromjust above the inion to beyond the parieto-occipital fis-sure. The length of the midline scalp incision will vary,depending on whether the lobectomy is to include part orall of the occipital lobe in either hemisphere, or the oc-cipital lobe and posterior temporal and parietal lobes inthe nondominant hemisphere. The upper end of the mid-line scalp incision continues laterally, extending overthe upper mastoid region, but it may end more anteriorlyabove the ear if a larger bone flap is needed. The com-bined scalp, occipital muscle, and periosteum are dis-sected off the calvarium in one layer.

Identifying the middle of the inion, visualizing thesagittal suture, and knowing the position of the uppermastoid bone, are vital to locating the superior sagittaland transverse sinuses and placing the burr holes prop-erly. The venous sinuses are about 0.75 cm wide. Burrholes 1 and 2 (Fig. 2) are placed 1.5 cm lateral to themidline. Burr hole 2 is placed 2 cm from the inion. Burrhole 3 is on a plane through the middle of the mastoidbone and 5.5 cm from the tip of the mastoid. The bone cutbetween burr holes 2 and 3 is above the transverse sinusand mastoid bone. The position of the remaining burrholes depends on the size of the tumor and the amount ofexposure required. A rim of temporalis fascia may be leftattached to the free bone flap and resutured during clo-sure. During the formation of the bone flap a suction trapis used to collect bone dust, which is used during theclosure.

Dural tack-up and wire holes are placed in the calva-rium and bone flap. Dural tack-up sutures are used but leftuntied until the time of closure. Intraoperative ultrasound isinvaluable in locating the gross boundaries of the intracere-bral lesion. The dura is opened with a T incision. The base ofthe dural flaps is along the dural sinuses. A view of theexposed brain after opening the dura is shown in Figure 2.The point marking the 75% distance from the nasion to theinion and the point on the upper lateral margin of the© 1991 The American Association of Neurological Surgeons

107

LEAVENS : OCCIPITAL LOBECTOMY

Figure 1. A and B, lateral and medial surfaces of the occipitallobe. C, head position. D, scalp incision and flap, position ofthe bone flap, dural incision. RF, rolandic fissure; AG, angulargyrus; POF, parieto-occipital fissure; OP, occipital pole; PON,

preoccipital notch; VL, vein of Labbé; CF, calcarine fissure;BV, bridging veins to superior sagittal sinus; BV, bridging veinsfrom occipital pole to torcular herophili; PCA, posterior cere-bral artery.

orbit are useful. When they are connected, the resultingline marks the plane of the sylvian fissure. The suprama-rginal gyrus, which is anterior to the angular gyrus, willbe on the turned-up distal end of the sylvian fissure.

Shown in Figure 2 is the angular gyrus with its surpris-ingly close position to the occipital pole and the midline,which is a position of considerable significance to plan-ning an occipital lobectomy on the dominant hemisphere.The location of the angular gyrus has been measured to be35 ± 4 mm from the occipital pole and 24 ± 2 mm from themidline. On a fresh cadaver brain the location of the angu-lar gyrus in the two hemispheres measured 35 and 43 mmfrom the occipital pole and 27 and 25 mm from the mid-line. The intact, functioning angular gyrus may be resectedin the nondominant hemisphere but should not be re-

sected in the dominant hemisphere because of the morbidityof persistent postoperative dysphasia. Localization of speechin the temporal and parietal lobes in the language-dominanthemisphere is very variable. Rarely does speech representa-tion extend into the occipital lobe. It is reasonable, however,to perform an occipital lobectomy in the dominant hemi-sphere if that procedure is needed to excise a neoplasm or anarteriovenous malformation, although it is not advisable toextend the cortical resection beyond the occipital lobe with-out first localizing the posterior extent of speech localizationby stimulation mapping techniques. Well-demarcated occipi-tal tumors extending anteriorly in the dominant hemispheremay be removed without use of such techniques. After theoccipital lobectomy, the tumor that remains anteriorly iscarefully removed, preserving the brain adjacent to it,


108


Figure 2. Exposed occipital lobe shown with adjacent brain andvascular anatomy. 1, 2, and 3 (circled), burr holes 1, 2, and 3; MB,mastoid bone; SF, sylvian fissure: LS, lambdoid suture; TS, trans-

verse sinus; PTL, posterior temporal lobe; SSS, superior sagittalsinus; 75%, point on skull at 75% distance from nasion to inion; I,inion. For additional abbreviations, see the legend to Figure 1.

using a bipolar cautery, suction, laser, and the Cavitronultrasonic aspirator (CUSA) with little or no brainretraction.

The preoccipital notch, the most lateral border of theoccipital lobe, is 6 cm from the occipital pole adjacent tothe lateral aspect of the transverse sinus and near the vitalvein of Labbé at its termination into the transverse sinus.Because occlusion of this vein may result in a devastat-ing hemorrhagic infarction of the brain, it should be iden-tified and preserved during the lobectomy. An exceptionwould be its sacrifice in order to remove tumors that haveinfiltrated the adjacent temporal and parietal lobes in thenondominant hemisphere.

LobectomyThe initial steps in performing the lobectomy after

exposure of the cortex is retraction to identify, coagulate,and divide medial bridging veins (Fig. 3) and then, withsimilar retraction, to locate the occipital pole. The oc-cipital pole bridging veins are preserved until the resec-tion is completed. Measurements are made from the oc-cipital pole to mark the I

1 (first incision posterior to the

angular gyrus) cortical incision for a resection limited tothe occipital lobe. The approximate distances from theoccipital pole to the preoccipital notch of 6 cm, to theangular gyrus of 3.5 cm, and to the parieto-occipital fis-sure of 4.5 cm are measurements used to avoid injuringthe angular gyrus while making the I

1 incision.

The I2 (second incision anterior to the angular

gyrus) cortical incision in the nondominant hemi-sphere includes part of the posterior parietal and tem-poral lobes and will be determined by the size and


109


Figure 3. Retracted occipital lobe shows bridging veins. I1 (first

incision posterior to the angular gyrus) marks the plane of theoccipital lobe resection. I

2 (second incision anterior to the angular

gyrus (nondominant hemisphere)) marks the incision for a largerresection in the nondominant hemisphere. For additional abbre-viations, see the legends to Figures 1 and 2.

position of the neoplasm or arteriovenous malformation(Fig. 3). The vein of Labbé is identified so it can be pro-tected and preserved.

The cortical incision is made using bipolar cautery,suction, and scissors to coagulate and divide the piamater and cortical vessels. Retraction is confined to theoccipital lobe being resected. With the aid of a head-light, loupes, and retraction, the incision through thewhite matter is slowly advanced, as are the incisionsthrough the pia mater of the occipital lobe and corticaltissue adjacent to the falx and the tentorium cerebelli.The branches of the posterior cerebral artery in theparieto-occipital fissure and the calcarine fissure areidentified, coagulated, and divided. If there is an oc-cipital ventricular horn it will also be divided. Hemo-stasis should be maintained and blood should not beallowed to pool in the ventricle or subdural space. After

the occipital lobe has been disconnected from the brain,the bridging veins connecting it to the torcular herophiliare coagulated and divided. The specimen is removed,leaving the cavity shown in Figure 4. Hemostasis mustbe absolute before closure and is obtained by bipolarcautery, with Surgicel (oxidized cellulose) placed, whereneeded, on the cut surface of the brain, and by fillingthe cavity temporarily with saline-soaked cotton balls.

Closure and Technical PointsThe dura is closed with continuous 4-0 Neurolon. Beforeclosure of the dura is complete, air is removed from thelobectomy site by filling the cavity with saline solution.

Illustrated in Figure 5 are advantageous techniquesfor making the scalp incision, securing the bone flap,and bandaging the patient. Figure 5A illustrates themethod of using the unipolar cutting and


110


Figure 4. View of the intracranial cavity after I2 incision and resec-

tion of the occipital lobe and part of the temporal and parietal lobesin the nondominant hemisphere. TC, tentorium cerebelli; V, ven-tricle. For additional abbreviations, see the legends to Figures 1-3.

coagulating machine (Neomed or Bard) with a straight,small electrode blade with a 7/8-inch insulated shank(ME 2280, American V. Mueller).

The Neomed is set at 20 cutting and 20 coagulation,with fine adjustments made as needed. The initial milli-meter incision is done with a No. 15 steel knife blade; therest of the incision of the scalp and pericranium is donewith the electrosurgical unit. Steady, moderately fast 2-3-mm deep cuts (Fig. 5A-2) are used, retracting the woundedges as the cuts are made and keeping the blood suckedaway. Small bleeding sites from the skin edges and sub-cutaneous fat are controlled with the unipolar cauterycoagulation current (Fig. 5A-3). This is done lightly andbriefly. In some patients, bipolar cautery may be neededto control bleeding from larger vessels. Infiltration of thescalp with 1% lidocaine with 1/100,000 epinephrine isnot mandatory but may be used before making the scalp

incision to add to hemostasis. With proper application ofthis technique, the patient will lose only a few millilitersof blood. Hemostats and scalp clips are not needed forhemostasis. The wounds heal well and look no differentfrom those made without the electrosurgical cutting blade.Some surgeons produce a satisfactory scalp wound usingan electrosurgical machine alone without using a steelknife initially.

Craniotomy bone flaps cut with power tools are rela-tively small considering the bone opening they have to fill.If not sutured back into place with care, they may eventu-ally become depressed. To avoid this, the bone flap is ro-tated so that it touches the calvarium at three or four places(Fig. 5B). In those places it is firmly secured to the calva-rium with 26-gauge stainless steel wire. These wire suturesare twisted tightly and cut, leaving 4-5-mm lengths that areforced into the outside wire holes. Each wire is tapped


111


Figure 5. A, optional scalp incision technique using a steel knifeblade in combination with the unipolar electrosurgical cutting andcoagulation current. B, method of rotating and fixing the bone flapin place with 26-gauge stainless steel wire. C, craniotomy bone

defects shown filled with bone dust; the use of a subgaleal drain isillustrated. D, dressing is held in place with No. 7 Surg-O-Flextubular elasticized net bandage.


112


flat on the skull using the handle of a periosteal elevatoror a small mallet. To foster healing of the wound so thatthe bone flap becomes firmly fixed in place and withoutscalp depression over defects in the skull, the burr holesand empty spaces around the flap are filled with bonedust (Fig. 5C), which is washed free of blood before it isapplied. Reopening the wound months or years later re-veals the bone flap held firmly in place by the wire su-tures, the bone dust permeated by fibrous tissue, and, insome patients, new bone formation.

After hemostasis of the scalp is achieved, and to pre-vent the development of a significant postoperativesubgaleal hematoma, the galea is sutured firmly to each ofthe wire sutures and to each dural tack-up suture hole in thecalvarium using 3-0 Vicryl suture. The dura, galea, and boneflap are sutured together using 3-0 Vicryl sutures throughone or two pairs of holes placed in the bone flap. In largecraniotomy wounds or wounds that are not perfectly dry, aDavol medium 1/8-inch drain with enclosed suction is usedin the subgaleal space for 12 to 24 hours (Fig. 5C).

The galea is closed with interrupted 3-0 Vicryl, andthe scalp is closed with continuous 4-0 nylon or withstaples. A transfusion is rarely required for this operativeprocedure.

The wound is covered with bacitracin ointment,Adaptic dressing, 4 × 4-inch sponges, and abdominalpads. In small and moderate-sized craniotomy wounds,the dressing is held in place with a “ski cap” made fromNo. 7 tubular elasticized net bandage (Fig. 5D). For pa-tients with large craniotomies and those who require asubgaleal drain, a standard pressureroll bandage is used.

COMPLICATIONSPerioperative mortality is 1% or less. A postoperativecomplete homonymous hemianopsia is expected. Injuryto the angular gyrus in the dominant hemisphere willresult in dysphasia manifested by dyslexia, dysgraphia,and acalculia. Failure to close the dura properly and sealopened mastoid air cells with muscle or pericranium su-tured firmly in place may result in a cerebrospinal fluidfistula, meningitis, or brain abscess.

Intracranial pressure may remain elevated postop-eratively if the neoplasm has not been completely ex-cised or if blood has been allowed to collect and remainin the ventricle and subdural space. Concavity of thescalp over burr holes and over a depressed bone flap oc-curs if the bone flap has not been properly secured andburr holes filled in or covered.

113

SPINAL MENINGIOMASMICHAEL N. BUCCI, M.D.

JULIAN T. HOFF, M.D.

INTRODUCTIONMeningiomas are the second most common tumor of thespinal canal, accounting for 20% of intraspinal neoplasms.Most of the tumors are intradural and extramedullary inlocation. Two-thirds occur in the thoracic spine; 80% arein women. The duration of symptoms is variable. Rootpain, segmental motor changes, and spasticity are com-mon early findings.

PREOPERATIVE CONSIDERATIONSMyelography and postmyelography computed tomogra-phy usually demonstrate the typical “crescent-shaped”appearance of an intradural extramedullary lesion. Mag-netic resonance imaging is now the diagnostic procedureof choice, which demonstrates these lesions in multipleplanes (Fig. 1).

Preoperative preparation usually consists of dexam-ethasone (10 mg, by mouth or intravenously) at mid-night and on call to the operating room. Concurrent ad-ministration of either antacids or a H

2 receptor blocker is

also advisable. Preoperative antibiotic prophylaxis isoptional, usually consisting of either 1 g of nafcillin orcephalothin intravenously.

SURGICAL TECHNIQUEThe operation is usually performed with general anesthe-sia; however, spinal anesthesia may be considered in high-risk patients with low thoracic or lumbar lesions. Centralvenous access and arterial pressure monitoring are help-ful, but not essential.

The operation is performed with the patient in theprone position. All pressure points are well-padded. Acomfortable head rest is essential. For patients with upperthoracic or cervical tumors, three-point skull fixation withthe cervical spine in neutral position is best.

Prior to preparation and draping, anterior-posteriorroentgenograms are obtained in patients with thoracicmeningiomas in order to best localize the lesion. Alterna-tively, the patient may be taken to the radiology depart-ment prior to surgery where fluoroscopy can be employedto localize these lesions relative to surface anatomy. Al-though intraoperative evoked potential monitoring may

be beneficial in some cases, intradural extramedullarytumors can be resected safely without this additionalmonitoring.

After preparation and draping, a midline incision isemployed and the fascia is incised down to the spinousprocesses. The muscular attachments are stripped bilater-ally in the subperiosteal plane, exposing the laminae outto the facet joints (Fig. 2A). The spinous processes andlaminae are carefully removed using either bone rongeurs,a high-speed air drill, or both. Enough bone is removedboth above and below the tumor to permit an adequatedural opening (Fig. 2B).


Figure 1. Magnetic resonance image of a thoracic meningioma,T2 weighting, with a paramagnetic contrast agent administered.Note the clear definition of this intradural extramedullary lesion.


114


Figure 2. A, thoracic spine, posterior elements, with muscularattachments stripped in the subperiosteal plane. B, decompressive

thoracic laminectomy exposing the dorsal dura. Enough bone isremoved both above and below the tumor to permit safe removal.

At this point, the operating microscope is broughtinto the field. The dura. is opened in the midline, takingcare to preserve the arachnoid intact. A microdissector isthen used to separate the dura. and arachnoid. Dural tack-up sutures are placed and cottonoid pledgets are laidover the exposed dura. Tension on the sutures providesbetter exposure and relative hemostasis from the epidu-ral space (Fig. 3A). The tumor is usually visible beneaththe arachnoid.

If the tumor is located dorsally, the arachnoid is care-

fully opened by sharp dissection (Figs. 3B and 4). Thearachnoid edges are secured to the dura with clips. If thetumor is ventrally located, an additional arachnoid layermay be present lateral to the exposed spinal cord.

The tumor is best debulked prior to dissection of thetumor-cord interface. Bipolar coagulation and piecemeal re-moval using an ultrasonic surgical aspirator or cutting bipo-lar coagulator minimize spinal cord manipulation, providinga safe and gentle method for removal (Fig. 3C). The planebetween tumor and spinal cord is usually well-developed and


115

BUCCI AND HOFF : SPINAL MENINGIOMAS

Figure 3. A, illustration of the dural opening, exposing an intactarachnoid layer overlying the tumor. B, arachnoid opening revealsthe extramedullary neoplasm causing spinal cord compression. A

well-developed plane exists between the lesion and the spinal cord.C, illustration of piecemeal tumor removal using bipolar electro-cautery.


116


Figure 4. Intraoperative photograph of a cervical (C1-2) menin-gioma prior to resection. Note the extramedullary location.

Figure 5. Intraoperative photograph of the cervical spinal cordfollowing resection of a meningioma. Note the vertebral arterylocated ventral to the spinal and spinal accessory nerve rootlets.

easy to delineate after central debulking. The tumor edgesfold in, allowing circumferential arachnoid dissection andeventual delivery of the tumor capsule (Fig. 5).

For ventrally located tumors, division of the dentateligaments and possibly of selected dorsal rootlets facili-tates both tumor exposure and removal.

Meticulous hemostasis is essential prior to dural clo-sure. A watertight closure follows. Muscle, fascia, and

subcutaneous tissue are closed in layers. The skin is gen-erally closed with a subcuticular suture, and then Steri-strips are placed. Drains are not used.

The major postoperative complications include in-fection, cerebrospinal fluid leakage, and increased neu-rologic deficit. Perioperative antibiotic prophylaxis, care-ful wound closure, and dexamethasone administrationlessen these risks.



117

PERCUTANEOUS TRIGEMINALGLYCEROL RHIZOTOMY

RONALD F. YOUNG, M.D.

PATIENT SELECTIONPatients are generally selected for any form of surgicaltherapy of trigeminal neuralgia based on previous failureof pharmacological treatment. The diagnostic criteria fortrigeminal neuralgia are: 1) sharp, knife-like pain of du-ration from a fraction of a second up to several seconds toone minute; 2) pain confined within one or more of themajor peripheral divisions of the trigeminal nerve; and3) the presence of trigger zones from which innocuousstimuli trigger the patient’s characteristic pain.

Ablative procedures such as percutaneous glycerolrhizotomy (PGR) should not be utilized for patients withatypical facial pain. However, some patients present withhistories somewhat atypical for trigeminal neuralgia withgenerally longer-lasting pain and sometimes a constantache between pains. The latter is particularly true of pa-tients treated with carbamazepine. Patients who lack trig-ger phenomena and whose pain extends beyond thetrigeminal distribution should not be considered for per-cutaneous trigeminal rhizolysis.

Patients with recurrent trigeminal neuralgia after pre-vious surgical procedures such as microvascular decom-pression or radiofrequency rhizolysis may be treated suc-cessfully with PGR as may patients who developtrigeminal neuralgia as a consequence of multiple scle-rosis. The procedure is most appropriate for patients withpain in the second division alone, the second divisionand the first division, the second and the third division,or all three divisions. Patients with exclusively third di-vision trigeminal neuralgia respond poorly to percutane-ous glycerol rhizolysis. The exact reason for the failuresin such patients is unclear, but PGR produces maximalsensory changes in the second and first divisions andminimal or no change in the third division.

There are few, if any, absolute contraindications tosurgery. Patients with hypertension require appropriatemedical therapy preoperatively and intraoperatively.

PREOPERATIVE PREPARATIONA preoperative cerebral imaging study, preferably a mag-netic resonance scan or computed tomographic scan withintravenous contrast, is recommended to exclude struc-tural lesions as the cause of the patient’s trigeminal neu-ralgia. Recommended preoperative laboratory studiesinclude a complete blood count, screening of electrolytelevels, and assessment of clotting ability utilizing theprothrombin time and the partial thromboplastin time.X-ray films of the chest are obtained for patients over theage of 40 or with a previous history of pulmonary dis-ease, and an electrocardiogram is obtained for patientsover the age of 40 or with a previous history of cardiovas-cular disease.

Preparation of the patient for the procedure includesinsertion of an intravenous line for administration of intra-venous sedation. Blood pressure is monitored every fiveminutes with an automatic sphygmomanometer. I havenot monitored arterial blood pressure by an indwellingcatheter except in a rare patient such as an individual withan unruptured intracranial aneurysm. The patient is pre-pared for the procedure with intravenous midazolam, ashort-acting benzodiazepine, and fentanyl, an opiate anal-gesic. Dosages are adjusted according to the patient’s bodysize, age, and previous drug utilization. The parasym-patholytic agent glycopyrrolate is also administered toprevent bradycardia which may accompany penetrationof the foramen ovale. A broad spectrum antibiotic such ascefazolin is administered intravenously about 30 minutesprior to beginning the procedure in an attempt to preventa rare case of meningitis following PGR.

OPERATIVE PROCEDUREThe procedure is carried out with the patient in thesupine position. The head is rotated approximately 15°away from the side of the patient’s pain (Fig. 1). Theskin of the side of the face around the angle of themouth is prepared with suitable topical antiseptic. Ster-ile drapes are then applied across the upper chest. Be-ginning at a point about 2.5 cm lateral to the angle ofthe mouth on the side of the patient’s pain, the skin© 1991 The American Association of Neurological Surgeons

118


Figure 1. Angled fluoroscopic view used to identify the foramenovale and place the needle for PGR. Plane A represents the mid-sagittal plane which is rotated 15° (plane B) away from the side of

pain. The fluoroscopic beam is then angled 40° from vertical andthe central ray is directed about 2-3 cm lateral to the angle of themouth. The foramen ovale is then seen as in Figure 2.


119

YOUNG : PERCUTANEOUS TRIGEMINAL GLYCEROL RHIZOTOMY

and deep tissues are infiltrated with approximately 2-3ml of 1% Xylocaine solution. The anesthetic agent isintroduced first by raising a small skin wheal with a 25-gauge needle and then with use of a 22-gauge, 1.5inchneedle. Care should be taken that the oral mucosa is notpenetrated with the needle used to introduce the localanesthetic.

A C-arm fluoroscopy unit is employed and the beamis angled about 40° from vertical and the central ray ofthe beam is directed at a point about 2.5 cm lateral to theangle of the mouth on the side of the patient’s pain (Fig.1). The foramen ovale is then identified by a brief fluoro-scopic view. Hyperextension of the patient’s neck is un-necessary, but slight changes in the angle of the fluoro-scopic beam may be required in order to place the foramenovale in an “en face” position. In addition, slight rota-tions of the head greater than or less than the original 15°maybe required to place the foramen ovale in the centralportion of the fluoroscopic image. Using this technique,the foramen ovale appears on the edge of the shadow ofthe petrous bone midway between the ramus of the man-dible laterally and the maxilla medially (Fig. 2).

When the foramen ovale has been identified, theneedle which will be used for glycerol injection is intro-duced at a point about 2.5 cm lateral to the angle of themouth on the side of the patient’s pain. When the pain ison the left side, the needle is grasped in the right

hand and the index finger of the left hand is placed in-side the patient’s mouth. When the patient’s pain is on theright side, the needle is grasped in the left hand and theright index finger is placed within the oral cavity. Thetechnique of insertion of the needle is the so-called“Härtel technique.” As the needle is introduced, the glovedfinger inside of the mouth prevents penetration of theoral mucosa. As soon as the tip of the needle passes thepterygoid plate, the position is viewed briefly fluoro-scopically. The needle direction is then changed as re-quired and advanced incrementally under intermittentfluoroscopic viewing.

The key to successful instillation of glycerol intothe retrogasserian cistern is accurate puncture of the fora-men ovale. The ideal point for puncture of the foramenovale is in the center of the medial half of the foramen(Figs. 3 and 4). Punctures in the lateral half of the fora-men may result in the needle tip being placed beneaththe temporal lobe or in the temporal lobe (Fig. 5). Whenthe needle is placed along the medial edge of the fora-men ovale, the needle may enter the cavernous sinus. Ifthe needle is placed too close to the anterior or posterioredges of the foramen ovale, extradural or subdural place-ment of the needle tip rather than subarachnoid place-ment is likely. As the needle tip enters the foramen ovale,slight contraction of the masseter muscle frequently oc-curs and the patient may experience a brief twinge ofpain. In addition, firm resistance is felt as the foramen ispenetrated. As soon as the foramen is penetrated, no at-tempt is made to advance the needle further at this point.

The patient’s head is then rotated until it is in theneutral position and the C-arm fluoroscope is changed

Figure 2. The right foramen ovale as viewed fluoroscopically, asdemonstrated in Figure 1. The ramus of the mandible is lateral andthe maxilla medial. The foramen appears on the edge of the shadowof the petrous bone. (From Young RF. J Neurosurg 1988; 69:39-45.With permission.)

Figure 3. Correct position for needle puncture of the right fora-men ovale just medial to the midpoint of the foramen. Satisfactorypunctures may be made more medial but not more lateral to thispoint. (From Young RF. J Neurosurg 1988; 69:39-45. With per-mission.)


120


Figure 4. Cutaway anatomical view of needle puncture of the leftforamen ovale. The needle is in the medial half of the foramen andmidway between the anterior and posterior borders. The mandibu-

lar division is seen exiting from the skull via the foramen. Since theneedle passes through the nerve, contraction of the masseter musclewith jaw closure often accompanies puncture of the foramen.


121


Figure . Incorrect puncture of the right foramen ovale. The needleis too lateral. Punctures at this point may yield cerebrospinal fluidfrom the subarachnoid space beneath the temporal lobe. Glycerolshould not be injected at this point. The needle should be with-drawn and the foramen repunctured. (From Young RF. J Neurosurg1988; 69:39-45. With permission.)

to the lateral view (Fig. 6). The depth of the needle is thenascertained (Figs. 7 and 8). The needle is slowly advancedand the stylet is removed intermittently to ascertain flowof spinal fluid. The needle is advanced only 1-2 mm in-crementally. As soon as cerebrospinal fluid flow is ob-tained, further advancement of the needle is stopped. Ifthe needle tip passes posterior to the shadow of the clivuswithout obtaining spinal fluid, then the needle is againslowly withdrawn and rotated slightly during the with-drawal, again observing for cerebrospinal fluid flow. Ifcerebrospinal fluid (CSF) flow is not obtained, the needleis withdrawn from the foramen ovale, the patient’s head isagain rotated 15°, and the angled fluoroscopic view isutilized for repuncture of the foramen ovale. A second orperhaps at most a third attempt to obtain cerebrospinalfluid flow is made utilizing the same technique with slightdifferences in the point of repuncture of the foramen ovale.Surprisingly, a second puncture in a relatively similarposition to the first may give cerebrospinal fluid flowwhen the first puncture did not.

If cerebrospinal fluid flow is not obtained, a deci-sion must be made as to how to proceed. Preoperativediscussion with the patient concerning this decision isessential. Two possibilities present themselves. The firstincludes an injection of glycerol even though cere-

Figure 6. Lateral fluoroscopic view used to assess needle depth. The head is placed in the neutral position and thefluoroscopic beam is horizontal.



122


Figure 7. Metrizamide trigeminal cisternogram viewed with alateral fluoroscopic beam as in Figure 6. The needle is positionedlow in the cistern in good position for glycerol injection for painin the mandibular division, mandibular and maxillary division, orall three divisions. A slight needle advancement would be ideal formaxillary division pain alone or a combination of maxillary andophthalmic division pain. (From Young RF. J Neurosurg 1988;69:39-45. With permission.)

Figure 8. Metrizamide trigeminal cisternogram, demonstrating anexcessively posterior position of the needle tip not recommendedfor glycerol injection. The needle enters the cistern very posteri-orly suggesting that the foramen ovale was punctured near theposterior rim. Repeat puncture of the cistern is recommendedfrom this position before glycerol is injected. Compare the needleposition to that shown in Figure 7. (From Young RF. J Neurosurg1988; 69:39-45. With permission.)

brospinal fluid is not obtained. In this instance, the like-lihood for successful amelioration of the patient’s pain isprobably at most 50%. The second alternative, which ismy preference, is to proceed with radiofrequency trigemi-nal rhizotomy at this point. I use a needle which allowsboth glycerol injection through a Luer locking hub aswell as radiofrequency rhizolysis using a thermistorradiofrequency heating probe.

Other authors have recommended a retrogasseriancisternogram using water-soluble contrast material in-jected through the needle to confirm that the needle tipis within the cistern as well as to estimate the volume ofthe cistern (Figs. 7 and 8). My experience suggests thatcisternography is not helpful and may indeed be coun-terproductive. In order to carry out cisternography, thepatient must be placed in the sitting position to fill thecistern and avoid the contrast material running into theposterior fossa. The cistern must be emptied by placingthe patient supine and then PGR must be carried out withthe patient sitting again. These manipulations risk dis-lodging the needle tip from the cistern. Furthermore, ifCSF is obtained and the needle is correctly placed in theforamen ovale as previously described, the needle tip willbe in the cistern. If CSF is not obtained, there is littlevalue in injecting contrast material. The technique de-scribed for penetrating the foramen ovale is the only onewhich will reliably place the needle tip in the cistern.

Radiographic techniques which employ submentovertexor anteroposterior and lateral views only cannot ensurethat the foramen ovale is correctly punctured.

If clear cerebrospinal fluid is obtained after punctureof the foramen ovale, then the patient is brought into thesitting position and the head is flexed so that theorbitomeatal line is angled slightly below the horizontal.The lateral fluoroscopic view is used to ascertain that theneedle position has not changed. In elderly patients withslack facial tissues, it is possible for significant acciden-tal withdrawal of the needle to occur as the patient isbrought into the sitting position. Occasionally, it may benecessary to hold the needle in place while the patient ismoved and at all times during glycerol injection, butgenerally this is not required. When the depth of theneedle has been confirmed to be the same as in the su-pine position and when cerebrospinal fluid flow is con-firmed, the patient is ready for glycerol instillation.

Sterile anhydrous glycerol is used. The glycerol isprepared by the hospital pharmacy in 2-ml ampules,and I ml of glycerol is drawn into a 1 -ml tuberculinsyringe using an 18-gauge needle. The 18-gauge needleis removed and the tuberculin syringe containing theglycerol is attached to the needle in the foramen ovaleafter removal of the stylet. The glycerol is injectedslowly, in increments of 0.05 ml or approximately 1drop. The hub and bar rel of the needle gener-



123


ally contain a volume of about 0. ml of cerebrospinalfluid, and therefore I make an initial total injection ofapproximately 0.15 ml of which only 0.05 ml is actuallyinjected because of the dead space just mentioned. Within30 to 60 seconds, the patient will usually describe sensa-tions of tingling, pins and needles, or burning in the dis-tribution of the branch of the trigeminal nerve closest tothe needle tip. Most commonly, the patient describes thefeeling of paresthesias around the upper lip and angle ofthe mouth. Further glycerol is injected to a minimumvolume of about 0.15 ml actual injected volume. Aboutfive minutes after the initial injection, testing of thepatient’s facial sensation is begun with a pin and the cor-neal reflex is monitored intermittently. If little or no anal-gesia is obtained, further glycerol is injected in 0.05-mlaliquots and facial sensation is tested further. When thepatient notices a significant decrease in sharpness of thepin, usually at least a 50% decrease compared with thenormal side, the injection is terminated. If the cornealreflex is decreased or the patient complains of paresthesiasand burning in the eye and the trigeminal neuralgia painis not within the first trigeminal division, no further glyc-erol is injected. The final total volume of glycerol variesfrom approximately 0.15 ml to about 0.55 ml, with anaverage of about 0.35 to 0.4 ml. At the conclusion of theprocedure, the needle is withdrawn and pressure is ap-plied over the injection site. The patient is maintained inthe sitting position with the head flexed forward as previ-ously described for approximately one hour. The patientmay be transferred from the radiology suite to a recoveryarea as soon as the needle is withdrawn.

SPECIALIZED INSTRUMENTATIONA portable C-arm fluoroscopy unit with television repro-duction of the enhanced image is essential. A simple 20-or 22-gauge, 3.5-inch spinal needle may be used for thisprocedure, but, as previously described, I use a needlewhich may be utilized either for glycerol injection or forradiofrequency rhizotomy. If cerebrospinal fluid flow isnot obtained from the needle or if pain is located exclu-sively within the third trigeminal division, I prefer theradiofrequency technique.

COMPLICATIONSIn my experience, virtually all patients who will experi-ence long-term relief of trigeminal neuralgia experienceat least some alteration in facial sensation. About two-thirds of patients experience analgesia of a more signifi-cant degree such that pain sensation is reduced by 50%or more compared with the normal side. Reduction orabsence of the corneal reflex occurs in 5-10% of patients

in whom the pain does not involve the ophthalmic divi-sion. Unlike the corneal reflex changes which occur withradiofrequency rhizolysis, there is rarely a complete ab-sence of the corneal reflex. There may be regions of thecornea from which the reflex cannot be obtained, but acareful examination will usually reveal some areas of thecornea from which the reflex can still be obtained. I havenot experienced in my own patients nor have I seen anypatient suffering from neuroparalytic keratitis or blind-ness following PGR of the trigeminal nerve.

I have likewise not seen any patient of my own whohas experienced anesthesia dolorosa after PGR but I haveseen one patient in whom anesthesia dolorosa resultedafter three different attempts at glycerol rhizotomy. Weak-ness of the muscles of mastication is likewise an unusualcomplication with glycerol rhizotomy. I have never ob-served paresis of cranial nerves other than the trigeminal,nor inadvertent puncture of foramina other than the fora-men ovale with this technique. Rarely a large hematomamay form rapidly in the tissues of the cheek which pre-vents the needle tip from reaching the cistern. In suchpatients, pressure is applied manually, an ice pack is used,and the procedure is delayed until the hematoma resolves.

In my opinion, the key to the avoidance of complica-tions is the incremental injection of glycerol with inter-mittent monitoring of facial sensation. In rare cases, de-spite apparent perfect placement of the needle within theretrogasserian cistern and despite the flow of cerebrospinalfluid, the patient experiences little or no immediate alter-ation in sensation upon glycerol instillation. In such pa-tients, I usually instill a total volume of about 0.35 ml.Surprisingly, in a few such patients, a slow onset of facialanalgesia has developed over a period of 30-90 minutes.

Using the needle placement technique describedhere, accidental puncture of any foramen other than theforamen ovale should not occur. Likewise, accidentalplacement of the needle beneath the temporal lobe or inthe temporal lobe should not occur,

Occasionally the patient may experience a burst ofrepetitive typical trigeminal neuralgic pains which maylast for as long as several hours. For this reason, patientsare monitored in the postoperative recovery room for atleast two hours following withdrawal of the needle, andintravenous narcotics are used in the instance of a showerof repetitive pains. The vast majority of patients experi-ence complete relief of trigeminal neuralgia within oneto two hours. An occasional patient will experience reliefof pain over 24 hours or more. I have not seen patientswho have persisted with trigeminal neuralgia for 24 to 48hours after glycerol rhizotomy in whom the pain has even-tually resolved, unlike the experience reported by others.

124

LUMBAR HEMILAMINECTOMY FOREXCISION OF A HERNIATED DISC

PATRICK W. HITCHON, M.D.VINCENT C. TRAYNELIS, M.D.

PATIENT SELECTIONTraditionally, a patient who presents with low back painradiating into one lower extremity is thought to harbor aunilateral lumbar herniated intervertebral disc providedother lesions have been ruled out. Other conditions thatcan mimic a herniated disc include primary or metastatictumors of the spine, pelvis, or retroperitoneum, spondy-lolisthesis, and spinal infections, to mention a few. It is tobe emphasized that the majority of patients with a herni-ated disc, with or without radiculopathy, improve withbedrest.

Plain roentgenograms of the lumbar spine ruleout fractures, lytic or blastic lesions, or spondylolis-thesis. Magnetic resonance imaging (MRI) has beengaining favor in the past 2-3 years as a noninvasivepotentially diagnostic study. The value of MRI liesin its noninvasive feature, although its potential inidentifying herniated discs is restricted to relativelylarge lesions. Where MRI is unavailable, a plain com-puted tomographic (CT) scan can be helpful. Shouldthe above studies be nonconfirmatory, and the pa-tient is considered a surgical candidate, the most re-liable study is that of lumbar myelography with awater-soluble contrast agent, followed by CT. Myel-ography need only be performed when the patient isconsidered for surgical treatment after failure of con-servative therapy.

All nonsteroidal anti-inflammatory medications,phenothiazines, and antidepressants are discontinued 1week prior to surgery. Routine blood work, includingplatelet count, prothrombin time, and partial thrombo-plastin time are mandatory. All patients undergo x-rayfilms of the chest and an electrocardiogram, and thosewith a history of angina pectoris or those over 65 years ofage have a consultation with a cardiologist. Typing andscreening of the patient’s blood without cross-matchingis sufficient.

SURGICAL TECHNIQUEPositioningA number of devices are available for positioning of thepatient undergoing lumbar disc surgery. They include foamlaminectomy rolls, the Wilson frame, and the Cloward frame.The choice of a device is based on its performance in re-ducing intra-abdominal pressure and flexing the lumbarspine, its ease of application, and the distribution of bodyweight to obviate pressure sores. We have adopted theCloward frame, which satisfies most of the above criteria.The lateral decubitus position is reserved for obese pa-tients or those for whom the prone position is contraindi-cated, such as patients with a prominent transplanted kid-ney. All patients are transported to the operating roomwearing thigh-high antithromboembolic hose. Generalanesthesia is induced and the patient is then positioned onthe Cloward frame with straps across the thighs and withthe head turned to one side (Fig. 1).

Once intravenous fluids have been started, the firstdose of a 2-day course of prophylactic cefazolin sodiumis administered. The skin is shaved and an adhesive plas-tic drape is applied to both buttocks to exclude theperineum from the surgical field. The lumbar region isscrubbed with Betadine soap for minutes and then paintedwith a 10% solution of povidone-iodine followed by iso-propyl alcohol. With a sterile marking pen the proposedsurgical incision is outlined on the patient’s spine, usingthe iliac crests and spinous processes as landmarks. Plainx-ray films and/or myelographic studies should be postedon a viewbox in the operating room. Once the proposedincision is marked, the surgical field is draped using firsta plain or iodinated adhesive plastic sheet, then towels,and, finally, cloth or disposable sheets. Sutures or staplesare applied on the corners to fix the drapes. Traditionally,the authors have preferred to infiltrate a proposed inci-sion with 1% Xylocaine with 1/200.000 epinephrine forhemostasis.

125

HITCHON AND TRAYNELIS : LUMBAR HEMILAMINECTOMY

Figure 1. The patient is positioned prone on the Clowardframe with the hips flexed and the arms extended. The abdo-

men is suspended within the frame, reducing intra-abdominalpressure.

Incision and ExposureThe skin incision for a single level hemilaminectomymay be 2-3 inches long, depending on the patient’s weight.The greater the retraction necessary, the longer the inci-sion. For simplicity, this chapter will address removal ofan L4-5 herniated disc on the right side, with the surgeonstanding on the ipsilateral side and the assistant on theopposite side.

The skin incision is carried down to the fascia with ascalpel. Bleeding points are controlled with bipolar co-agulation. At this point, with an Allis clamp applied to aspinous process as a marker, a lateral x-ray study is ob-tained to confirm the level. A self-retaining Adson retrac-tor is utilized to expose the underlying fascia. With theelectrocautery, the fascia is incised and the paraspinalmuscles dissected away from the spinous processes andlaminae in the subperiosteal plane. Manual retraction ofthe muscle is performed either with a Cushing or Cobbperiosteal elevator. To prevent potential interlaminar slip-page of the instrument, the largest elevator is preferred. Toachieve nontraumatic dissection with minimal blood loss,the cutting current has been found to be most useful inexposing the leading and trailing edges of the laminae allthe way laterally to the facet joint. Inadvertent entranceinto the facet joint is to be avoided in order to reduce thechances of developing postsurgical arthritis. The laminaeare polished using an elevator and a 4 × 8-inch spongefolded upon itself to measure 2 × 8 inches. One sponge isused per level. This process is repeated for improved expo-sure. All sponges should be marked with a radiopaque

thread. It is important for both the surgeon and scrub nurseto keep count of the sponges to avoid unnecessary delaysof wound closure in case of a miscount.

Once the sponges have been withdrawn, a 1 ¼-inchTaylor retractor is inserted to maintain exposure. Theshortest possible retractor should be used so as not tointerfere with the surgical procedure. The pointed tip ofthe retractor is engaged lateral to the facet joint and 2-3pounds of weight are suspended from the proximal endto maintain its position (Fig. 2). Excessive weight is un-necessary and may predispose to the dislodgement of theretractor and injury to the facet joint.

Hemilaminectomy and DiscectomyThe caudal edge of the L4 lamina is now removed a dis-tance of 1-1. 5 cm rostrally using a Leksell rongeur. Bleed-ing from the underlying cancellous bone is controlled withbone wax. An angled curette is utilized to free the ligamen-tum flavum from the undersurface of the L4 lamina. Thisdissection is necessary to facilitate insertion of the 40°Kerrison footplate for the completion of the partial hemil-aminectomy of the caudal edge of L4. An oval window iscreated that extends laterally to the facet joint and mea-sures 2 × 1.5 cm. From here on the operation requires magni-fication with loupes, or a microscope if two surgeons areinvolved. Using a No. 15 blade, the ligamentum flavum isincised as far medially as is possible, parallel to its fibers.This opening should be sufficient to advance a half-inchcottonoid into the epidural space. The cottonoiddisplaces the thecal sac away from the ligamen-


126


Figure 2. Following subperlosteal dissection of the laminae involved, the Taylor retractor is engaged lateral to the facet joint,providing exposure.


127


Figure 3. Following laminotomy and dissection of the epiduralspace, the affected root is retracted medially, exposing the her-

niated disc beneath. The laminotomy extends laterally to thefacet joint.

turn flavum, thereby reducing the chance of injury to thethecal sac. The ligamentum flavum is then grasped by aCushing forceps or Allis clamp and is pulled laterally.With the cottonoid in place, the ligamentum flavum isfurther incised to the facet joint. A Scoville curette com-pletes the excision of the ligamentum flavum by cuttingit between the sharp edge of the curette and the facetjoint. Following removal of the ligamentum flavum, asmall laminotomy of the rostral edge of L5 will furtherimprove exposure. Inspection of the epidural space re-veals fat and epidural veins. Gently the epidural vesselsmay be coagulated and the space lateral to the L5 nerveroot dissected in search of the herniated disc. If accom-plished hastily, this dissection may result in an excessiveamount of blood loss. Bipolar coagulation is performedwith low current to prevent sparking and potential injuryto the L5 nerve root.

Once the nerve root and thecal sac are identified, theD’Errico or Love root retractor is applied against the L5nerve root which is then retracted medially by the assis-tant (Fig. 3). Retraction of the nerve root may be quitedifficult with a chronic herniated disc because of adhe-sions and scarring between the fragment and the affectedroot. On the other hand, if the herniation is only weeksold, this retraction may be quite simple. In the presenceof scarring, a No. 4 Penfield or Church scissors may beneeded to free up the nerve root from the adhesions. Incase of persistent bleeding from the epidural veins, a 1/4-inch cottonoid may be inserted rostrally and another cau-dally for tamponade.

With the nerve root retracted, the herniated disc comesinto view. If a free fragment exists within the canal, thisextruded portion may be removed using a straight pitu-itary forceps. Occasionally, the herniation may besubligamentous; if so, a small incision into the posteriorlongitudinal ligament is sufficient for the disc hernia-tion to deliver itself. Once the herniated fragment hasbeen excised, attention is then directed to the annulus.This annulus harbors a degenerated disc and, furthermore,is defective, having allowed a disc fragment to herniate.It is our belief that such a disc is pathologic and is likelyto result in recurrent herniations. A vigorous disc exentera-tion, performed unilaterally, reduces the incidence of sucha postoperative complication.

With the root still retracted, the annulus is incisedwith a No. 15 blade. The medial vertical incision is per-formed first and then the annulus is incised laterally bothrostrally and caudally adjacent to the endplates of L4 andL5, respectively. The sharp edge of the blade should at notime be directed medially toward the nerve root or thecal.sac. Finally, the lateral edge of this window is incised andthe window is removed, allowing access to the disc space.The disc exenteration that is performed henceforth depends,to a great extent, on the size of the window created. A No. 1or 2 straight curette is initially utilized to enlarge thewindow. The angled Cone curettes, starting with the me-dium and proceeding to the large, are advanced into thedisc space to the level of the elbow on the curette (Fig.4). This prevents penetration of the annulus anteriorlyand possible vascular injury. The end-plates of


128


Figure 4. Following removal of the free fragment or thesubligamentous component, a window is incised in the annulus.The window should be large enough to allow the introduction of

the Cone curettes for further disc exenteration. The curette isto be manipulated so that the cutting edge remains within thedisc space.

L4 and L5 are scraped, fragmenting the disc material andcartilage which is subsequently removed with pituitaryforceps. Disc removal cannot be accomplished with theforceps alone. Undue vigor with the Cone curettes mayresult in penetration of the end-plates and entry into thecancellous vertebral body. This results in bleeding thatcan be controlled only with bone wax applied with adissector, and/or Avitene. The disc is removed with a sweep-ing motion of the curette from medial to lateral, and fromventral to dorsal. Forceful irrigation into the disc spacecan often yield retained fragments that have not yet de-livered themselves. Once the removal is thought to becomplete, careful inspection of the L5 nerve root is per-formed medially and along its course toward the neuralforamen. A right-angled dural separator may be advancedventral and medial to the nerve root for the identificationof retained fragments. The separator should be rotatedsuch that these fragments are swept out lateral to the nerveroot. A malleable bent probe is now advanced along thenerve root toward the neural foramen, both dorsally andventrally, in search of residual disc fragments. Followinginspection, the wound is irrigated and all cottonoids areremoved. Large wads of Gelfoam or Surgicel should notbe left in the epidural space, and are removed. If bleedingpersists, Avitene may be applied and the excess removedby irrigation.

In the presence of sufficient subcutaneous adiposetissue, a fat graft measuring approximately 1.5 × 2 cmmay be easily harvested unilaterally and applied over thenerve root. By preventing adhesions between the spinalmuscles and the nerve root, this fat graft may facilitate

future dissection, if necessary. The Taylor retractor is nowremoved and bleeding points from the muscle coagu-lated. Drains are not used.

ClosureThe fascia is approximated using 2-0 Vicryl, which is anabsorbable suture material. Sutures in the muscle are notnecessary. The fascial closure usually provides sufficientobliteration of dead space and good apposition of the muscleagainst the laminae and spinous processes. The subcutane-ous tissue may be approximated with 3-0 Vicryl; staples or3-0 nylon in a vertical mattress suture are used for skin clo-sure. A dressing is applied consisting of two sponges se-cured into position with an elastic-type adhesive tape. Anexcessively thick dressing is unnecessary and may be un-comfortable when the patient lies on his back.

POSTOPERATIVE CAREPostoperatively the patient is allowed activity as tol-erated. On the first postoperative day the patient isexpected to start sitting up for meals and to begin walk-ing. Postoperative analgesia is provided with meperi-dine hydrochloride, 100 mg, and Vistaril, 25 mg, intra-muscularly every 3-4 hours as necessary. Thesemedications may be switched to the oral route at thepatient’s request. Should the patient develop lumbarmuscle spasms, a muscle relaxant, such as diazepam in5-10-mg doses intramuscularly, maybe prescribed asnecessary. Patient-controlled analgesia pumps or tho-racolumbar immobilizers are not warranted for thistype of surgery. By 5-7 days postoperatively, the


129


patient is able to ambulate independently and may bedischarged home. Prior to release from the hospital, physi-cal therapy consultation is sought to guide the patient innonstrenuous exercises that are to be continued at home.Skin sutures may be removed 7-10 days postoperatively.The patient is generally unable to return to employmentuntil after a 6-week postoperative examination. At thattime, x-ray films may be obtained if deemed necessaryby the physician. Lifting in excess of 30 pounds is notrecommended for up to 3 months from surgery. A com-promise can usually be achieved by the patient and hisemployer. If satisfied with the outcome of surgery, thesurgeon may discharge the patient with instruction toreturn should the need arise.

COMPLICATIONSAn early but rare postoperative complication within thefirst 2 days of operation is cauda equina compression byan epidural hematoma. This demands immediate evacua-tion following diagnosis if reversal of neurological defi-cit is to be achieved. A resurgence of back pain, with or

without radiculopathy, and an elevated erythrocyte sedi-mentation rate, days or weeks postoperatively, is pathog-nomonic of discitis. In the case of aseptic discitis, radio-logic studies will show sclerosis of the adjacent end-platesrather than loss of cortical margins. Treatment usuallyconsists of bedrest, bracing, and nonsteroidal anti-inflam-matory drugs. The presence of fever, chills, and leukocy-tosis suggest a bacterial wound infection. If the CT scanor MRI shows an abscess in the subcutaneous tissues over-lying the laminae, open surgical drainage, packing, andadministration of intravenous antibiotics are indicated.With purulent discitis and osteomyelitis, plain roentgeno-grams may show destructive changes of the end-plates ifthe condition has been present for at least 3 weeks. Aneedle aspiration of the disc space and identification ofthe organism is necessary to guide the subsequent antibi-otic treatment regimen. A generally cited overall compli-cation rate with hemilaminectomy and discectomy is lessthan 5%. In the absence of a complication, the surgicaltreatment for lumbar herniated disc is one of the mostrewarding in neurosurgery.

130

TRANSORAL SURGERYFOR CRANIOVERTEBRALJUNCTION ANOMALIES

ARNOLD H. MENEZES, M.D.

INTRODUCTION AND PATIENT SELECTIONAbnormalities at the craniovertebral junction have beenrecorded for several centuries. But, it is only in the pastfour decades that antemortem recognition of these le-sions has stimulated surgical therapy. Several approachesto the anterior craniovertebral junction have been devel-oped. The transoral-transpalatalpharyngeal route is mostfrequently used for decompression of this area. Advancesin neurodiagnostic imaging and microsurgical instrumen-tation have expanded the use of this approach.

The bony abnormalities of the craniovertebral junc-tion may be divided into reducible and irreducible cat-egories. Primary treatment for reducible craniovertebraljunction (CVJ) lesions is stabilization. Surgical decom-pression of the cervicomedullary junction is performedwhen irreducible pathology is encountered. Should thelesion be ventrally situated, a transoral-pharyngeal de-compression is required. In dorsal compression, a poste-rior approach is used. If instability exists following eithersituation, a posterior fixation is necessary. Thus, the fac-tors that influence the surgical approach to lesions of thecraniovertebral junction are: 1) reducibility, 2) the direc-tion of encroachment, and 3) the type of lesion.

These factors are assessed with plain x-ray films,polytomography, and myelography with computed to-mography. Dynamic studies determine anteroposteriorstability in the flexed and extended positions and verti-cal stability under traction. Magnetic resonance imag-ing, including a dynamic study, is currently the proce-dure of choice (Fig. 1, A and B).

Mere identification of a ventrally placed lesion at thecraniovertebral junction is not an indication for an ante-rior transoral approach. The main indication for the transoraloperation is irreducible ventral compression of thecervicomedullary junction, whether it is from bone, granu-

lation tissue, or abscess. The exposure obtained is from theclivus to the C2-C3 interspace (Fig. 2). Bony tumors andchordomas have been approached in a similar manner. Pri-mary intradural tumors such as schwannomas and menin-giomas. are best approached via the dorsal route becausethey reside within the subarachnoid space. However, shouldthis not be possible, a ventral transoral operative approachmay be necessary. On occasion, midline vertebral basilarartery aneurysms have been clipped via this route.

PREOPERATIVE PREPARATIONA magnetic resonance imaging (MRI)-compatible halois used for cervical traction and stabilization prior to thetransoral procedure. A high caloric intake is advised.Oropharyngeal cultures are obtained three days prior tothe operation. No antibiotics are used unless pathologicflora are present.

The entrance to the oral cavity must provide a work-ing distance of 2.5 to 3 cm between the upper and lowerteeth. This span may not be available in patients afflictedwith rheumatoid arthritis with involvement of the tem-poromandibular joints. In such a circumstance, a medianmandibular split with midline glossotomy may be neces-sary to gain access into the oral cavity.

Evaluation of lower cranial nerve and brain stem func-tion is extremely important. Difficulty in swallowing,tracheal aspiration, or repeated upper respiratory infec-tions would mandate a tracheostomy at the start of theoperative procedure. A tracheostomy maintains a safeopen airway should obstruction from lingual swellingoccur early in the postoperative convalescence.

Sensory evoked responses and brain stem laten-cies from median nerve stimulation are assessed be-fore surgery to allow for comparison during the opera-tion. Custom-built teeth guards for the upper and lowerdentition are obtained via impressions taken a few daysprior to surgery. They protect the teeth during© 1991 The American Association of Neurological Surgeons

131

Figure 1. A, T1-weighted MRI of the craniovertebral junction.There is assimilation of the atlas and severe basilar invagination.The odontoid process invades the medulla oblongata (open arrow).

B, axial T1-weighted MRI made 15 mm above the plane of theforamen magnum. The odontoid process grossly indents the me-dulla (open arrow).

Figure 2. Illustration of the rostrocaudal exposure (between ar-rows) obtained via the transoral operation.

surgery when the self-retaining Dingman mouth retractoris used.

SURGICAL TECHNIQUE

AnesthesiaThe patient is brought sedated to the operating roomwith halo traction set at 5-7 pounds. Topical oropharyn-geal and nasopharyngeal anesthesia is used and supple-mented with bilateral superior laryngeal nerve blocks tofacilitate an awake fiberoptic oropharyngeal intubationwith the patient in the best position judged from preop-erative dynamic studies. Following this, the patient ispositioned appropriately for the operative procedure andexamined awake to check the neurologic status. Generalanesthesia is then administered. Nasopharyngeal intuba-tion is to be avoided since it disrupts the integrity of thehigh naso-oropharyngeal mucosa. The patient is posi-tioned supine with the head resting on a Mayfieldheadholder and traction is maintained throughout theentire operative procedure. In circumstances where thebony anomaly or tumor has been recognized to violatethe dura and the subarachnoid space, a lumbar spinalsubarachnoid drain is installed.

Operative ProcedureA tracheostomy is performed when the operation in-volves the high nasopharynx: and the craniocervicaljunction. Operative procedures at the level of atlas and

MENEZES : TRANSORAL SURGERY FOR CRANIOVERTEBRAL JUNCTION ANOMALIES



132


axis may not require a tracheostomy. In the latter circum-stance, an armored endotracheal tube is secured to an inci-sor tooth in the midline and held in place by the Dingmanself-retaining tongue retractor during the operation. A tra-cheostomy is essential when the operative procedure in-volves the clivus and there is brain stem dysfunction. Whenthe operation is limited to the upper cervical spine, the softpalate is drawn up into the high nasopharynx via suturesattached to soft rubber tubing passed through the nostrilsand then withdrawn. The abdomen is prepared for possibleharvesting of fascia and fat grafts, should this be necessaryafter intradural exploration.

A gauze pack occludes the laryngopharynx. The oralcavity, nasopharynx, and oropharynx are cleansed firstwith 10% povidone-iodine solution, then with hydrogenperoxide, and finally rinsed with saline. The circumoralarea is prepared in the usual fashion and isolated.

A modified Dingman self-retaining mouth retractorallows depression of the tongue and lateral retraction ofthe checks (Fig. 3). One-half percent lidocaine solution

with 1/200,000 epinephrine is injected into the medianraphe of the soft palate. A midline incision is made ex-tending from the hard palate to the base of the uvula,deviating from the midline at its base. Stay sutures pro-vide for lateral retraction and are held in place via thesprings of the Dingman mouth retractor.

The operating microscope now provides magnifica-tion and illumination. Topical 5% cocaine is applied tothe posterior pharyngeal mucosa and its midline is infil-trated with 0.5% lidocaine solution with 1/200,000 epi-nephrine. A midline incision is made into the posteriorpharyngeal median raphe (Fig. 4A). This extends fromthe rosal clivus to the C2-C3 interspace. The posteriorpharyngeal wall is now retracted laterally using stay su-tures for self-retaining exposure. The prevertebral fasciaand longus colli muscles are reflected free of their osseousligamentous attachments to expose the caudal clivus andthe atlas and axis vertebrae. The anterior longitudinal liga-ment and the occipital ligaments are dissected free of theirbony attachments to expose further the caudal clivus, ante-

Figure 3. This drawing illustrates the exposure of the oral cavityand pharynx with the Dingman mouth refractors in place. The softpalate has been incised, exposing a portion of the hard palate at the

apex of the incision. The view through the operative microscopeis within the circle.


133


Figure 4. Drawing of views through the microscope. A, the softpalate is retracted laterally with stay sutures; the posterior pharyn-geal wall is incised. B, the longus colli muscles as well as the posteriorpharyngeal wall are retracted to expose the clivus, the anterior archof C1, and the body of C2. The dens is revealed when the anteriorarch of the atlas is resected. There is granulation tissue around the

dens and behind the clivus. C, the caudal clivus is resected to exposethe invaginated odontoid process which is now cored out. D, theshell of the odontoid process is removed along with a part of thebody of the axis. E, the cruciate ligament is visible after removal ofthe odontoid process. The rectorial membrane is incised vertically.F, the dura bulges ventrally when decompression is complete.


134


rior arch of the atlas, and the anterior surface of the body ofthe axis (Fig. 4B). This exposure is between 3 and 3.5 cm inwidth (Fig. 5A). Further lateral exposure is inadvisablebecause of the risk of destruction of the eustachian tubeorifices, entrance into the vertebral canal and damage tothe vertebral arteries, and injury to the hypoglossal nerves.

The anterior arch of the atlas is now removed with ahigh-speed drill for a width of 3 cm. Resection of thecaudal clivus is carried out in a similar fashion depend-ing on the pathology (Fig. 4C). The soft tissue ventral tothe odontoid process is cleared with rongeurs. The odon-toid process is cored out in a rostrocaudal direction using

a high-speed cutting burr initially and then a diamondburr. It is then separated from all of its ligamentous at-tachments using microsurgical technique (Fig. 4D). Ifthere is chronic instability, parmus will be encountered.This should be carefully resected by cauterization andpiecemeal removal. Division of the odontoid process atits base and downward traction is to be avoided becauseincomplete resection as well as a dural tear may result.The removal of the odontoid process is completed withup-biting Lee-Smith Kerrison rongeurs with a 1-mmfootplate. Angled curettes and microbiopsy forceps alsofacilitate the removal (Fig. 5B).

Figure 5. Microphotographs depicting the operative exposureand transoropharyngeal decompression of the craniovertebral junc-tion of the patient depicted in Figure 1. A, exposure of the clivus,the arch of the atlas, and the body of the axis. B, the shell of the

cored odontoid (od) being removed. C, the rectorial membrane(te) is seen above the cructate ligament (cr) after the odontoidprocess is removed. D, the dura (d) bulges into the wound once therectorial membrane (te) is incised.


135


Removal of pannus from behind the odontoid processand from the posterior fossa should be done piecemealwith angled and ring curettes.

Preoperative lateral pleuridirectional tomography,as well as axial magnetic resonance imaging, will guidethe surgeon as to the extent of removal of the odontoidprocess or abnormal bony lesion (Fig. 4E). Extraduraltumors, likewise, are now resected. The tectorial mem-brane must be incised carefully to allow adequate duraldecompression (Fig. 4F). A blunt nerve hook insertedbetween the tectorial membrane and the dura preventsdural penetration (Fig. 5, C and D). Changes in brainstem latencies require gentle handling of tissues and mini-mal compression of the neural structures.

An intradural lesion at the upper cervical spine,craniocervical border, or behind the clivus necessitatesa cruciate incision of the dura. The vertical componentof this starts inferiorly proceeding up in a rostral direc-tion with careful cauterization of the circular sinus atthe foramen magnum. The dural turgidity may be re-duced by draining spinal fluid through a previouslyplaced lumbar subarachnoid drain. The dural leaves arereflected to the four quadrants to allow exposure of thelower pons, medulla, and cervicomedullary junction.Once the intradural operation is complete, it is essentialto bring the dural leaves together with 4-0 polyglycolicacid sutures. Fascia harvested from the external obliqueaponeurosis is now placed against the dural closure.This may be held in position by plasma glue if avail-able. A fat pad reinforces the fascia.

Aerobic and anaerobic cultures are obtained fromthe depths of the wound prior to closure. Bacitracin pow-der and microfibrillar collagen are placed over the clo-sure. The longus colli muscles are approximated in themidline with 3-0 polyglactin sutures. Anatomical approxi-mation of the posterior pharyngeal musculature is ob-tained with similar sutures using an interrupted figure-of-eight technique. Then, horizontal mattress suturesapproximate the posterior pharyngeal mucosa. There isno muscular layer over the clivus, and, hence, mucosalapproximation must be done carefully at the rostral endof the incision. A blanket of Gelfoam is then placed overthe mucosal closure of the nasopharynx.

The closure of the soft palate is done by bringing thenasal mucosa together with interrupted 3-0 polyglactinsutures. The muscularis as well as the oral mucosa of thesoft palate are approximated with interrupted verticalmattress sutures of similar strength.

In situations where surgery is necessary through theclivus, or in patients with platybasia and a foreshortenedclivus, the hard palate must be exposed and its posterior8-10 mm resected using Kerrison rongeurs. During this

step, it is essential to preserve the mucosa in the nasal aswell as the oral aspects of the hard palate. This techniqueallows high nasopharyngeal exposure without splittingthe mandible or doing a median glossotomy.

Postoperative CareFollowing the operation, the patient’s head is elevated to10-15° and is maintained in 5-7 pounds of skeletal trac-tion. This permits drainage of nasal and oral secretionswithout pooling at the operative site. Intravenoushyperalimentation is the rule for the first five to six daysduring which no oral intake is permitted. At the end ofthis time it is important to assess craniocervical stabilitywith flexion-extension dynamic pleuridirectional tomog-raphy, visualizing the facets. Vertical instability is checkedwith and without cervical traction. Should instability bepresent, a dorsal fixation procedure is necessary. In theevent that the craniocervical junction is stable, the tra-cheostomy is gradually allowed to close. The oral intakeis advanced from a clear liquid diet at the end of the firstweek to a full liquid diet in a fortnight. Solid food isgiven at the end of the third week.

If the dura was opened and a fascial graft was usedfor repair, intravenous antibiotics and spinal drainage arecontinued for 10 days after surgery.

COMPLICATIONSThe most dreaded postoperative complication is cere-brospinal fluid leakage and meningitis. This may be im-mediate or delayed. An immediate cerebrospinal fluidleak is best handled via immediate reapproximation ofthe wound and closure in the manner described. A de-layed cerebrospinal fluid leak implies retropharyngealinfection. This must be treated with intravenous antibiot-ics, spinal drainage, and elevation of the head. Intrave-nous hyperalimentation promotes wound healing by sec-ondary intention. Should this not occur, a surgical closurebecomes necessary.

Wound dehiscence can be avoided by delaying oralintake. Retropharyngeal infection may present itself withdehiscence within 7-10 days. Control of infection andpossible lateral extrapharyngeal drainage of the site maybe necessary. Mucosal coverage will occur if an adequatecaloric intake is maintained.

Nasal speech and regurgitation of food are manifes-tations of dehiscence of the palate; closure of the sepa-rated edges may be done early or late. Inadequate andimproper closure of the soft palate will lead tovelopalatine incompetence.

Failure to recognize craniocervical instability canlead to disastrous vascular complications as well ascervicomedullary compression.

136

ANTEROLATERAL CERVICALAPPROACH TO THE

CRANIOVERTEBRAL JUNCTIONDENNIS E. McDONNELL, M.D.

INTRODUCTIONSynonyms for this operation are the transcervical, ante-rolateral, parapharyngeal, or submandibular approach tothe ventral craniovertebral junction (atlas, axis, and cli-vus). Surgical exposure of the ventral aspect of thecraniovertebral junction has always been considered dif-ficult and dangerous, and rightly so. Since 1956, whenSmith and Robinson described gaining access to the an-terior cervical spine by dissecting through fascial planes,surgeons have steadily become bolder in directly attack-ing pathologic processes involving the vertebral bodiesor discs, and lesions in the ventral spinal canal. This ap-proach is part of every spinal surgeon’s armamentariumfor dealing with the segments of C3 to T1. It has notgained popularity with treatment of conditions rostral toC2. The transoral route is presently the preferred ventralroute to the atlas, axis, and clivus.

Surgeons have been hesitant to adopt this approachbecause of the complexity of anatomy, concentration ofvital structures, and difficulty in exposure that is entailedin the approach to this region. However, there are someinherent advantages to this approach that might warrantconsideration. This approach gives a sterile surgical fieldas opposed to the contaminated transoral field. It offerswider exposure of the lateral masses of C1 and C2 than ispossible via the transoral route. Also, the lower cervicalsegments are available, if necessary, whereas they are notfor the transoral route.

PATIENT SELECTIONCharacteristically, a symptomatic lesion affecting thecervicomedullary junction of the central nervous system(CNS) is insidious, subtle, and relentless in its progres-sion. The ultimate cause is mechanical compression ofthe CNS resulting in pain, myelopathy, and, ultimately,apnea and death, if the compression is not reversed in

some way. The lesion maybe congenital, developmental,or acquired and may involve the osseous elements, sup-portive ligaments, vascular structures, and/or the CNS.Craniovertebral instability may be superimposed, furthercompromising the patient.

It is essential that the anatomy of the lesion be im-aged, misalignment be discerned, and instability be de-termined before a treatment plan can be decided for thepatient. Plain x-ray films of the craniocervical junctionshould include anteroposterior, open-mouth, and lateralviews. Dynamic flexion and extension in the lateral pro-jection will help to assess instability. The magnetic reso-nance (MR) image will determine the position of the com-pressing lesion as well as the type and degree of CNSdistortion present. The computed axial tomogram (CT)will give important information regarding the integrityof the bone of the body of C2, odontoid process, lateralmass articulations, arches of C1, and occipital structures.Positive contrast cisternography combined with CT and/or polytomography gives additional information regard-ing bone and soft tissue distortions.

A lesion within or distorting the ventral subarach-noid space extending from the pharyngeal tubercle ofthe basiocciput caudally to even the lower cervical seg-ments is amenable to this approach. A lesion producingsymptoms of pain, weakness, incoordination, dysphagia,nystagmus, deafness, other cranial nerve dysfunction, orurinary incontinence, and having an inherent tendencyof progression, demands direct removal if the process isto be arrested. If the process is chronic, the patient maybedebilitated and nutritionally compromised. Thepatient’s respiratory reserve may be impaired whichfurther increases the risk for any surgical procedure. Abedridden, debilitated, catheterized patient harboringa CNS compressive lesion presents a perplexing anddifficult challenge to the surgeon. Anatomical, physi-ological, nutritional, psychological, and neurologicalfactors must all be assessed before performing this pro-© 1991 The American Association of Neurological Surgeons

137

McDONNELL : ANTEROLATERAL CERVICAL APPROACH TO THE CRANIOVERTEBRAL JUNCTION

cedure. Therefore, this cannot be rushed into; the risksmust be assessed and an organized plan of action must beformed.

Retraction of the superior pharynx for exposure ofthis region results in swelling with relative compromiseof the upper airway. Postoperative respiratory embarrass-ment should be expected and managed. Respiratory im-pairment will be aggravated by any preoperative myel-opathy. Postoperative pharyngeal swelling will alsoimpede oral alimentation. If there is generalized debility,then nutritional depletion will be compounded furtherby the stress of the surgery and impaired deglutition.Indwelling urinary catheter, vascular access lines, alimen-tation tube, endotracheal tube, and external subarach-noid drain may be in place for an extended time, raisingthe threat of systemic infection. A catabolic state willpotentiate the risk of such infection. Therefore, nutritionalsupport must be carefully planned preoperatively.

It is strongly recommended that a treatment planalgorithm, as described by Menezes, be followed in es-tablishing a management plan for the patient with anupper cervical or craniovertebral lesion. By doing so,problems and complications can be reduced or avoidedand chances for an optimal outcome improved. Ifcraniovertebral instability is expected either pre- or post-operatively, then this must be specifically addressedpreoperatively. Likewise, if the intradural compartmentis purposefully entered (one of the advantages of thisapproach) then the cerebrospinal fluid (CSF) must bediverted, at least temporarily, to avoid a CSF collectionunder tension in the dissected cervical tissues or a CSF-cutaneous fistula.

Severe neurologic impairment is not a contraindica-tion to this procedure. The statement, “the patient is tooill to tolerate surgery,” just does not hold, if there is anyneurologic function to preserve caudal to the lesion.Obviously, if there is coagulopathy, sepsis, systemic ma-lignancy, or cardiac compromise, then such a patient isnot a candidate for this surgery until these conditions arereversed.

The alternative procedure is the transoral ap-proach. The best procedure most often is the procedurewith which the surgeon has the most experience and ismost comfortable. The transoral approach is most com-monly used by surgeons approaching this region. There-fore, when dealing with extradural lesions localized tothe C1 and C2 segments it may be preferred. However,such lesions are just as amenable via the transcervicalroute. If the lesion extends rostral to the pharyngealtubercle of the basiocciput, then an alternative to thetranscervical route should be considered. If the duramust be entered or the lesion extends caudal to C2,and not rostral to the pharyngeal tubercle, then thetranscervical route is preferred.

PREOPERATIVE PREPARATIONThe patient with a craniovertebral chronic compressivelesion is in serious jeopardy from compression and/orinstability. The rationale for treatment is surgical correc-tion of the pathological tension, distortion, and com-pression of the neuraxis in order to reestablish neuralconductivity and to renew regional cervicomedullaryblood flow. The treatment plan for these patients is oftenmultistage and is outlined as follows:

1. Attempt reduction by cervical traction to establishrealignment or to demonstrate irreducibility of anymalalignment.

2. Tracheostomy/feedinggastrostomy or jejunostomy.3. Anterior resection of the encroaching structures

(bone, ligaments, parmus, exudate, or tumor).4. Immobilization in skeletal traction between stages.5. Occipito-C1-C2-C3 fusion by the posterior approach.6. Halo brace immobilization, usually for six months.

These steps must be understood and accepted by thepatient and family before commencing the anterior spi-nal operation.

Respiratory reserve will be a critical factor in theimmediate postoperative management. Therefore, formalassessment of pulmonary function is crucial, particularlyif there is a history of chronic lung disease. Tracheostomyand feeding gastrostomy should be done electively; thegastrostomy established one week preoperatively allowsfor it to mature so that it is immediately available for use.The tracheostomy is placed either at the time of gastros-tomy or just prior to the transcervical procedure. Adequatenutrition is essential to avoid bacterial infections and toassist early mobilization. This is best ensured by preemp-tive and elective direct transabdominal enteric access.

Refrain from using preoperative antibiotics to avoid al-tering the patient’s normal flora and establishing resistantnosocomial bacteria. A broad spectrum antibiotic (i.e.,cefazolin) is given just prior to starting the procedure. A topi-cal antibiotic is also placed in the irrigation (i.e., bacitracin).

If entry through the dura is planned, a lumbar sub-arachnoid drain should be placed before starting the pro-cedure. It need not be draining during the procedure, butit is established for continuous drainage postoperatively.This reduces CSF hydrostatic pressure at the surgical site,thus avoiding a CSF fistula.

SURGICAL TECHNIQUE

AnesthesiaThe patient is positioned supine while awake. If it is de-cided that a tracheostomy is not needed, then an awakeintubation is done. The hypopharynx, larynx, and tra-chea are anesthetized locally with a topical anesthetic.The trachea is then intubated via the

138


nasotracheal route. The fiberoptic bronchoscope is help-ful and facilitates this maneuver. This avoids hyperex-tension, which may aggravate compression of theneuraxis. General anesthesia is maintained by an intrave-nous narcotic agent, and a combination of inhalationagents consisting of isoflurane, nitrous oxide, and oxy-gen. A low dose of sodium thiopental is given as well. Ifmotor evoked potentials or the electromyogram. is to bemonitored, long-acting muscle relaxants should beavoided. Infiltration of the skin incision with 0.5%lidocaine with 1/200,000 epinephrine will aid hemosta-sis in the superficial tissues.

A lumbar subarachnoid drain is placed with the pa-tient in the lateral decubitus position either under localanesthesia or after induction of general anesthesia. Greatcare must be used in positioning the patient for this, par-ticularly if there is instability.

It is advantageous to monitor blood pressurethrough a direct arterial line, central venous pressure,end expiratory carbon dioxide tension, and neuro-physiologic responses throughout the procedure.Monitoring of somatosensory cortical evoked re-sponses from the stimulation of the median and com-mon peroneal nerves is helpful. Motor evoked re-sponses will probably also be measured as a routine inthe future. An indwelling Foley catheter is often al-ready in place for patients with advanced myelopathy.Hourly measurement of urine output is obtained as anadditional guide to fluid balance. The intraoperativeadministration of diuretics is not ordinarily necessaryfor this procedure. Sequential compression pneumatichose are placed on the lower extremities for prophy-laxis against venous stasis and thromboembolism.

Operative PositioningThe operating table should be set up beforehand to acceptthe “C-arm” fluoroscope. Usually the C-arm is oriented inthe transverse plane for lateral projection beneath the headof the table, so that it is out of the way of the anesthesiolo-gist, surgeon, and assistant. Fluoroscopic control of theoperative field is very helpful for maintaining orientationin the sagittal plane. For some tables it is necessary toplace a padded radiolucent board on the table and to ori-ent the head of the patient to overhang the foot of the table,to accommodate the C-arm coming in beneath the tableand the patient’s head, neck, and shoulders.

The patient’s head is rotated 30° to the contralateralside and extended in order to raise the mandible up andaway from the surgeon’s line of sight to the field (Fig. 1).It is supported on a sponge doughnut or horseshoe head-rest. Of course, if the patient’s head is being maintainedin a halo traction system. the sponge doughnut is notneeded.

Skeletal traction is often necessary to maintain posi-tion in the face of spinal instability. If a halo ring is inplace, an extra thick pad or short mattress can be used onthe operating table to allow the patient’s head to go intoextension, by having the head overhang the end of thepad, supported by the halo ring and traction. Rope at-tached to the tongs or halo and 10 to 15 lbs of weight fortraction can be hung over an ether screen support.

The anesthesiologist, anesthesia equipment, andmonitoring devices are all at the head of the patient. Thesurgeon and assistant are opposite each other at thepatient’s side, and the scrub nurse is next to the surgeontoward the patient’s feet.

The microscope is mounted with the assistant’s ob-servation port and laser. It will straddle the C-arm at thehead of the table and be brought from the side oppositethe surgeon. The laser console is positioned on thesurgeon’s side toward the patient’s head. This allows thearticulated arm to attach to the microscope opposite tothe observer port and out of the way of the operatingteam. The high-speed drill can be kept sterile in a basinon a ring stand out of the way and brought into the fieldonly when needed.

The neck and chin are shaved and scrubbed withdetergent as is either the lateral thigh or lower abdomen,if a fascial graft is needed for subsequent dural closure.The neck incision is marked and injected with the localanesthetic/epinephrine combination.

Plastic adhesive towel drapes positioned around themarked incision sites will help isolate the operative areaswhich are then covered by an adhesive incise drape. Thisisolates the tracheostomy as well as other contaminatedsites from the operative fields.

The operating microscope and laser articulated armare draped with commercially available plastic drapes.The C-arm is also draped with transparent plastic drapes.

Skin IncisionThe choice of incision is dependent on the surgeon’s expe-rience and the specific craniovertebral exposure requiredfor the task at hand. Four choices are available: transverse,oblique-vertical, hockey-stick, and “T.” The first descrip-tion of this approach by Stevenson recommended the Tincision. A vertical extension or direction of the incision isnecessary only if exposure caudal to C4 is necessary. Oth-erwise, the transverse incision is quite adequate and cos-metically more appealing; it also serves to relatively pro-tect the superior laryngeal nerve (SLN). Regardless of theincision, the secret to adequate exposure with this approachis wide dissection of the cervical fascial planes. The con-cept that the vertically oriented incision facilitates theexposure and obviates the necessity for wide fascial dis-section is erroneous and misleading.

139


Figure 1. The patient’s position is oriented with the head ex-tended and rotated contralateral to the side of approach. The

curvilinear incision (dotted line) is 2 cm below and parallel to theedge of the mandible.


140


The transverse incision is my preference, and this will bethe perspective presented here. The incision is 2 cm infe-rior and parallel to the lower edge of the mandible, andextends from the angle of the mandible posteriorly to thebase of the mental protuberance beyond the midline an-teriorly (Fig. 1). Care is taken to avoid the marginal man-dibular branch of the facial nerve supplying the mentalmuscles of the lower lip.

The side of approach, right or left, depends on thelesion. If there is unilateral lower cranial nerve impair-ment, the approach should be made from the side of im-pairment to avoid possible injury to the intact contralat-eral cranial nerves. However, with this procedure thesurgeon’s perspective will be at approximately a 20° anglefrom the midsagittal plane and a 30° angle from the trans-verse/horizontal plane; this gives an easier view of deepstructures contralateral to the side of approach. Thereforethe side of approach is an important factor in preopera-tive planning, particularly with lesions eccentric fromthe midline. In dealing with midline lesions, the side ofapproach is usually that of the surgeon’s preference.

Dissection and ExposureThe key to adequate exposure is wide sharp dissection ofeach layer or plane of the cervical fascia, beginning withdevelopment of a wide subcutaneous flap on each side of

the incision superficial to the platysma muscle (Fig. 2). Therewill be anatomical landmarks that identify each plane andguide the surgeon’s way. Each landmark is dissected free ofits fascial investment and is preserved both anatomicallyand functionally. Fine-toothed forceps and delicatesemisharp dissecting scissors are the instruments to use here.It is helpful, if not essential, for the surgeon’s assistant topick up the cervical fascia with fine-toothed forceps justopposite to the surgeon’s forceps and follow along thesurgeon’s course, giving countertraction on the fascia as thedissection with scissors progresses. This maneuver helps todefine the fascial plane being dissected. Likewise, the sur-geon must move the forceps and lift the fascia close to thescissors, keeping the fascia taut with countertraction fromthe assistant’s forceps at all times. The fascial planes areavascular with no major intervening structures. When liftedand opened, the areolar fibrous texture of the fascia is re-vealed. Its transparency allows a view of the structures itcontains. Such perspective is gained only with countertrac-tion on the fascia and sharp dissection. Wide opening ofeach fascial layer in a sequential, methodical manner willensure adequate exposure of the deeper structures, whilepreserving the intervening structures. The dissection can becompared to that required on a cadaver in the anatomy labo-ratory for anatomical presentation. Therefore, this proce-dure requires a “cadaveric” dissection of the cervical

Figure 2. Wide dissection of the subcutaneous layer opens themost superficial layer of the cervical fascia, exposing the outer

surface of the platysma muscle.


141


fascia and its enclosed structures. Anatomical landmarkswill lead the way.

The route will be through the platysma muscle to thesubmandibular trigone, beneath the submandibular gland,inferior to the digastric muscle, inferior to the hypoglos-sal nerve, superior to the greater cornu of the hyoid bone,past the lateral aspect of the superior pharyngeal con-strictor muscle, to the retropharyngeal space and theprecervical fascia. The previously dissected skin edgesand subcutaneous flaps are retracted with a Weitlaner-type self-retaining retractor, thus exposing the superfi-cial surface of the platysma muscle.

Platysma MuscleThe medial edge of the platysma is grasped in the mid-line. A hole is cut in the median fascial raphe to gainentrance to the next fascial layer. The dissecting scissorsis used to undermine and develop the fascial layer in themidline vertically in a cephalocaudal direction. The fas-cia] sheet thus formed is then cut vertically in the mid-line for a length of 6 cm from the mandibular symphysisto the median notch of the superior thyroid cartilage (Fig.3). This defines the medial edge of the platysma muscleand initiates vertical access and will allow freer retrac-tion. The medial edge of the platysma is grasped andelevated. The undersurface of the platysma is then dis-sected and freed. The platysma muscle can then be

transected across its fibers parallel to the direction of theprimary incision for the full length of the exposure (Fig.4). The edges of the platysma are further undermined toform muscle flaps along both the superior and inferioredges of the incision, which are then incorporated andspread apart by the Weitlaner retractor.

Submandibular GlandThe next fascial layer is identified by the submandibulargland which bulges forth beneath its transparent invest-ment (Fig. 5). The inferior edge of the gland is graspedand elevated with the assistant giving countertraction.The fascia is then opened, undermined, and dissectedparallel to the line of the incision. The facial artery andvein will be encountered crossing the field of dissectionposterolateral to the submandibular gland. These vascu-lar structures should be gently grasped and elevated bytheir adventitia. They are then dissected axially alongtheir course. This maneuver further opens the subman-dibular fascial plane. The facial vein is clamped,transected, and ligated. The facial artery is preserved (Fig.6). Dissection of the facial artery proximally in the lateraldirection will lead to the carotid sheath, which is thelateral limit of the exposure. When the facial artery isdissected, it can be fully retracted and need not be sacri-ficed, as it serves as an orientation landmark. Elevatingthe inferior edge of

Figure 3. The medial border of the platysma muscle is found at themidline and is split vertically from the mental symphysis to the

superior notch of the thyroid cartilage; this initiates access to thenext layer of cervical fascia.


142


Figure 4. The platysma muscle is transected; lifting the tissueswith grasping forceps and giving countertraction assists with thedissection. This exposes the next fascial layer.

Figure 5. Retraction of the platysma exposes the submandibulargland and the facial artery and vein. Dissection of these structuresopens the next layer of cervical fascia.



143


the submandibular gland with the superior edge of theincision using the Weitlaner retractor exposes the nextlandmark, the tendon of the digastric muscle.

Digastric Muscle and TendonThe next fascial layer is identified by the tendon of thedigastric muscle, which is a substantial, glistening yellow-white cord running parallel to the course of the incisionbeneath the inferior edge of the submandibular gland (Fig.7). The fascial sling around the digastric tendon attaching itto the greater wing of the hyoid bone is transected along thecourse of the tendon. This frees the tendon so that it can beretracted rostrally toward the mandible. This retraction isfacilitated by dissecting the lower edge of the anterior andposterior bellies of the digastric muscle and freeing theirundersurface. When the digastric tendon is retracted towardthe mandible, the next fascial layer becomes evident. Thehypoglossal nerve comes into view coursing just deep,slightly inferior, and parallel to the digastric tendon.

Hypoglossal NerveThe hypoglossal nerve is gently dissected along its courseand carefully preserved (Fig. 8). Posterolaterally the dis-section is carried along the nerve trunk toward the de-scending hypoglossal ramus, which is another guide tothe region of the carotid artery. Again, it is not necessary

to dissect along the medial border of the carotid sheathunless segments caudal to C4 are to be exposed. Thusfreed, the hypoglossal nerve is retracted superiorly, ex-posing the hyoglossus muscle. The greater cornu of thehyoid bone now comes into view.

Hyoid BoneThe greater cornu of the hyoid bone can now be seen andpalpated. The fascia overlying it is grasped and openedalong the course of the hyoid bone to the carotid sheath(Fig. 9). The carotid artery is easily palpated and is thelateral-most limit of the dissection. It is retracted later-ally by a right-angled, Army/Navy, or Sauerbruch retrac-tor. This maneuver opens the retropharyngeal space. It isnot necessary to cut any muscles, nerves, or vessels. Atthis point there is concern for compromise of the superiorlaryngeal nerve (SLN). The SLN courses deep to the in-ternal carotid artery along the middle pharyngeal con-strictor muscle toward the superior cornu of the thyroidcartilage. The SLN is therefore caudal and inferolateralto the route of the exposure described here. Entrance tothe retropharyngeal space by this procedure is alongthe greater cornu of the hyoid bone and adjacent to thesuperior pharyngeal constrictor muscle. The SLN is notseen in the dissection of this approach. However, theSLN is vulnerable to stretch injury from retraction.

Figure 6. The anterior belly of the digastric muscle is exposedwhen the facial vein is transected and the submandibular and iselevated and retracted superiorly.© 1991 The American Association of Neurological Surgeons

144


Figure 7. Transecting the fascial investment of the digastric muscleexposes and frees the next landmark and opens the next layer ofcervical fascia.

Figure 8. The digastric tendon is separated from its fascial sling at thehyoid bone and is retracted superiorly. This exposes the hypoglossalnerve, the next landmark, and the next layer of cervical fascia.



145


Figure 9. Dissection of the hypoglossal nerve opens this layer ofcervical fascia and allows retraction of the nerve to expose thenext landmark, the greater cornu of the hyoid bone. Opening the

fascia along the hyoid bone exposes the lateral wall of the superiorpharyngeal constrictor muscle.


146


Wide dissection of the fascial planes as described herewould tend to protect the SLN from retraction injury, asless force is required to separate the tissues that are freedby the fascial dissection. The SLN is not involved in thesoft tissues retracted superiorly for this exposure; it wouldbe vulnerable to transection if the deep cervical fasciawas opened vertically in the lateral exposure here to gainaccess to C4 or lower cervical segments through this route.The SLN must be specifically identified and preserved ifC4 or caudal segments are to be exposed by this route.

The superior constrictor muscle of the pharynx isretracted medially by a deep right-angled retractor. Theretropharyngeal areolar tissue is opened with scissors.The fat pad in the retropharyngeal space confirms thelocation. The anterior surface of the cervical spine is eas-ily palpated. The prominence of the anterior tubercle ofC1 is noted and orients the surgeon by directing the pal-pating finger rostrally. The precervical fascia and musclesare now in view. The midline of the cervical spine orientsthe midsagittal plane and is identified between the lon-gus colli and longus capitis muscles on either side, withthe C1 anterior tubercle in the center (Fig. 10).

Longus Colli MusclesThe medial borders of the longus colli muscles are cau-terized and elevated by sharp dissection from the antero-lateral surfaces of C2 and C3. The muscles converge atthe midline and cramp the surgeon’s view of the midline.A tooth-bladed self-retaining retractor is now insertedwith the blades engaged in the dissected medial walls ofthe longus colli muscles (the deep blades of the Clowardor Caspar retractor will do, but the Apfelbaum modifiedretractor system is more suited for this region).

This preliminary soft tissue retraction initiates accessto deep structures. The microscope with CO2 laser attachedis now adjusted to the field. A quick glimpse with thefluoroscope in the lateral projection will assist the orienta-tion. The laser is very advantageous in separating the lon-gus colli and longus capitis muscles from their medialattachments (Fig. 11). Laser dissection facilitates the ex-posure of the anterior arch of C1 and the lateral mass ar-ticulation of the atlas and axis. It allows muscle separationup to the pharyngeal tubercle of the basiocciput. Visualiz-ing these most rostral structures requires verticalretraction using a deep, narrow, right-angled retractor

Figure 10. Retraction of the superior pharyngeal constrictor opens the retropharyngeal space andexposes the precervical fascia and the longus colli muscles,


147


Figure 11. The anterior tubercle of C1 identifies the midline. Thelongus colli muscles are dissected from the anterior arch of C1 and

the anterior surface of C2 and C3. The laser controlled by themicroscope attachment facilitates this.


148


blade. This retractor can be self-retaining or handheld byan assistant. Here, again, the Apfelbaum retraction sys-tem offers an advantage in angulation of view. The me-dial one-half of the C1 and C2 lateral masses as well asthe anterior rim of the foramen magnum and adjacentbasiocciput (structures rostral to the anterior arch of C1)should be in view before proceeding further

Median Tubercle of the C1 Anterior ArchAligning the surgeon’s sight trajectory with the midsag-ittal plane and the conscious discernment of the angle ofapproach using the C1 anterior tubercle as a guide willhelp maintain proper orientation (Fig. 11). The perspec-tive of view to the structures of interest is upward andangled at 45°. The inferior and anterior aspect of the archof C1 and the base of the dens as well as the intervalbetween the dens and lateral mass articulations are seen.Laser removal of overlying soft tissue clears the view. Thedens, body of C2, and atlantotransverse ligament can beremoved without removing the anterior arch of C1, if sodesired. The transverse ligament is a tough, thick, well-defined, pale yellow ligamentous belt behind the dens.This ligament comes into view after the dens has beenresected. It is a guide after the arch of C1 and the odon-toid process are removed, but it may be obscured by somepathologic factors such as occur in rheumatoid arthritis.

Anterior Rim of the Foramen MagnumThe anterior rim of the foramen magnum is the next land-mark just above the arch of C1. The basiocciput can bepalpated and seen between the attachments of the longus

colli and longus capitis muscles. This can be drilled awayif entrance into the ventral aspect of the posterior fossa isrequired. The pharyngeal tubercle is the rostral-most land-mark and limit to this approach.

ResectionA high-speed drill with a cutting burr is used to resect thebone (straight or angled handpiece). The drill should notencumber the surgeon’s view. Usually the anterior arch ofC1 can be removed to give a full view of the dens andadjacent articulations. The dens should be resected fromthe apex caudally, keeping its base intact (Fig. 12). Itshould be thinned out like an eggshell. The diamondburr is used at this point to avoid inadvertent tearing ofthe adjacent soft tissues. These lateral and posterior os-seous cortical remnants are dissected with a Rosen-typemicrodissector from the periosteum and adjacent liga-ments. They can be lifted away with either a 3-0 curetteor a micro Kerrison rongeur. The medial wall of the lat-eral masses can be removed to widen the ventral expo-sure; this will be necessary if the lesion is intradural.The atlantotransverse ligament must now be identifiedto verify proper orientation (Fig. 13). This ligamentalong with the apical bursa and cruciform and tectorialligaments must be removed. Separation from the under-lying dura is accomplished by microdissection. The CO

2

laser slightly defocused at 5 to 10 watts allows removalof these structures without mechanical pulling. Theseligaments and pannus can be thick, resilient, and diffi-cult to remove when there has been long-standing insta-bility at C1-C2. If the anterior rim of the foramen

FIgure 12. The anterior arch of Cl has been removed, exposing the dens. The dens is best drilled awayfrom its apex to its base.


149


Figure 13. The transverse ligament is the next landmark. It, alongwith the posterior longitudinal ligament, must be removed to ad-equately decompress the neuraxis.

magnum is to be removed, it should be drilled away atthis point using a diamond bit.

The ventral aspect of the dural sac must bulge andpulsate into the resection site (Fig. 14). Only then is ad-equate decompression of the neuraxis assured. The durais recognized by its glistening pale gray color and longi-tudinal fibers. It may be transparent enough to reveal thepial vessels of the underlying spinal cord.

If the lesion is intradural, the dura. can now be openedlongitudinally. Traction sutures on the dural edge help towiden the exposure. The P-2 cutting needle on a 5-0polydioxanone suture or the PR-1 needle on a 5-0polyglycolic acid suture can be manipulated here for thispurpose. If the lesion is a neoplasm, the laser is indis-pensable for atraumatic removal and compensates for the“long reach” required via this approach. Vascular lesionsare dealt with as unique circumstances.

Control of the bleeding from venous sinuses andepidural veins can be challenging at times in this region.Microbipolar coagulation is the mainstay for hemosta-sis. Waxing bone edges will control bleeding from thissource. I prefer microfibrillar collagen (Avitene) andgentle suction compression with a cottonoid pledget forcontrol of bleeding around the dura and neuraxis.

Closure Techniques

Dural ClosureThe dura cannot be closed primarily here. A dural graftwill be needed, using either autograft or allograft fascia.

The graft is tacked on all four sides at several placesusing 5-0 suture on the P-2 needle; a running stitch canalso be used. The milliwatt CO

2 laser can be used to “weld”

the edges. The most effective dural closure supplement isfibrin glue. The commercial form of fibrin glue is not yetapproved. I use cryoprecipitate instilled topically overthe graft followed by spraying it with thrombin(Thrombostat). This results in a thick layer of coagulumwhich helps to seal the dural closure.

CSF DiversionAn external lumbar subarachnoid drain should be placedbefore commencing the procedure if an intradural proce-dure is planned. If the dura is inadvertently opened, thelumbar subarachnoid drain should be placed before thepatient is transported from the operating room. The lum-bar catheter should be tunneled subcutaneously for sev-eral centimeters before being brought out through theskin to the external reservoir. This will allow CSF drain-age for several days and minimize infection at the CSFdrain site. Drainage of about 300 ml of CSF per each 24-hour period will reduce hydrostatic pressure at the surgi-cal site and avoid a CSF fistula.

Spinal StabilizationSpinal stabilization is of critical importance and pre-operative plans are implemented. Skull tongs are usu-ally in place before starting, and light traction (5 to10 lbs) is maintained throughout the procedure. Trac-tion can be maintained as a temporizing measure

150


Figure 14. Bulging of the dura into the resection site ensuresadequate decompression.

postoperatively until final stabilization is accomplished.Osseous fusion in situ is an option. The C1-C2 lateralmasses are available for interarticular arthrodesis via thisapproach. An intact C1 anterior arch can be used as agraft purchase. These alternatives imply a stableatlantooccipital articulation, which must be discerned.

An external orthosis (halo brace) is rarely adequateas the sole source for either temporary or long-term sta-bility following this procedure. It is often essential as asupplement to surgical arthrodesis of the craniovertebraljunction and is continued for several months.

Occipitocervical arthrodesis is required as a subsequentprocedure in the face of craniovertebral instability. The pa-tient is kept in skull traction until this is accomplished.

DrainsSubsequent buildup of serum or CSF in the dissectionsite is best evacuated via a closed suction silicone rubbercatheter brought out from a separate stab wound. Usuallythis can be removed in 48 to 72 hours. This helps to re-lieve tissue pressure on the upper airway and reduce bac-teria] culture media in the wound.

Suture ClosureAbsorbable rather than nonabsorbable suture material ispreferred for wound closure. The platysma muscle is closedas a separate layer. Interrupted horizontal mattress stitchesplaced separately in the subcutaneous layer give strengthto the closure. Skin edge approximation with a continu-

ous subcuticular stitch will provide a cosmetically pleas-ing scar, particularly if a transverse incision has been used.

MonitoringNeurophysiologic MonitoringSomatosensory and motor evoked responses are beingused with increased frequency for these procedures. Theirtrue efficacy and role are controversial and yet to be de-termined. Theoretically, they have merit in warning thesurgeon of a potentially injurious maneuver and may beof benefit. Anesthesia methods must be modified to sup-port such monitoring if it is used.

Radiographic MonitoringThe C-arm fluoroscope is a helpful, if not essential, ad-junct for this procedure. It should be positioned for lateralprojection at the outset and kept in place during the proce-dure. It offers the surgeon an ongoing means of orienta-tion, monitors progress of the resection, and warns of anyshift in alignment. In Figure 15, the intraoperative lateralfluoroscopic view shows the trajectory of view and theretractor position using this approach; the dissector is be-hind the dens and the anterior arch of C1 can be preservedif desired. The standard x-ray unit can be used as an alter-native but is more inconvenient and time consuming.

SPECIALIZED INSTRUMENTSRetractorsAdequate retraction is essential here because of the


151


Figure 15. This lateral fluoroscopic view demonstrates the angleof access and the retractor position. The dissector is along theposterior wall of the dens.

deep access and angle of approach. Hand-held retractorsare used during the cervical fascia and soft tissue dissec-tion. A narrow pediatric Deaver retractor (modified byfirst straightening the bow and then bending the blade toa 90° angle so that blade tip measures 8 cm) can be usedto lift soft tissues rostrally for initial precervical dissec-tion of the superior corner at C1 and the clivus.

The deep blades of the Cloward or Caspar self-re-taining retraction systems are adequate for ongoing re-traction after dissection of the longus colli muscles. How-ever, they are not designed for retraction in this region;therefore, their purchase on the tissues is not secure andthey can migrate out of position. The surgeon must beaware of this as the procedure progresses. The Apfelbaummodification of the Caspar retractor is designed specifi-cally for this region and may eliminate the retractionproblem here. Additionally, some table-mounted retrac-tion devices are being designed that may improve visual-ization. Regardless of the retraction system used, atten-tion is directed to avoiding injury to the structures beingretracted. There is no substitute for adequate soft tissuedissection to ease retraction strain and facilitate safe ex-posure using this approach.

DrillsA high-speed drill system controlled under microsurgi-cal technique is required. The drill shaft must be long

enough to avoid line of sight obstruction. The A attach-ment for the Midas Rex drill or comparable handpieceworks well; an angled handpiece for this drill is now avail-able, which will facilitate its use. The long-angled hand-piece of lower speed drills are applicable as well. Bothcutting and diamond burrs are needed for safe bone re-section.

LaserThe microsurgical CO

2, laser facilitates dissection of

paraspinal muscles, resection of ligaments, and excisionof pannus or tumor. Complete removal of offending le-sions is possible even in hard-to-reach areas. Mechanicaltrauma is also avoided, thus improving the safety of theprocedure. Use of safe microsurgical laser technique andappropriate energy densities is mandatory.

Ultrasonic AspiratorThe ultrasonic aspirator can be an advantage in the re-moval of selected soft tissue lesions, particularly if theyare vascular neoplasms. The long-angled handpiece willbe needed, because the short or straight pieces will not fitinto the field and will obscure the line of sight.

MicroinstrumentsMicrodissectors of various designs will be useful inthe separation of soft tissue from bone, particularlyaround the dens. Micro Kerrison rongeurs, both 45°and 90°, and micro pituitary forceps will be needed forsoft tissue removal. An angled or straight 2-0 or 3-0curette will assist in delicate bone removal. Bayonet-ted instruments are needed to keep the line of sightfree. Long, insulated bipolar forceps are indispensablefor hemostasis.

SuturesA small, strong, cutting, and semicircular needle is neededfor suturing in this cramped space. The P-2 Ethicon needleworks well here. The preferred suture is 5-0 polydioxanonewhich can be used as a dural traction suture, for duralpatch repair, or for dural graft closure. An alternativeneedle and suture for this purpose is 5-0 Dexon Plus on aDavis & Geck PR-1 needle.

COMPLICATIONS

Neurological ComplicationsIncreased motor paresis, respiratory paralysis, sensoryloss, and systemic sympathectomy with loss of auto-nomic control are the devastating effects of neuraxisinjury at the cervicomedullary junction. Aggravationof a preexisting myelopathy is always an inherentrisk of operating in this region. This risk is even greaterwhen instabil i ty is present. Often, instabil i ty


152


is produced or increased with surgical resections in thisregion.

Tight compression of the neuraxis by misalignedbone, inflammatory granulation tissue, or neoplasm mustbe relieved to effect adequate decompression. Such ma-nipulation always adds increased risk of aggravated in-sult to neural structures by direct trauma or damage totheir vasculature. Microsurgical technique is critical tolimiting these deleterious occurrences.

Peripheral and cranial nerves are at risk during dis-section of the cervical fascia and retraction of the softtissues of the submandibular triangle. The hypoglossalnerve is particularly vulnerable as it is directly dissectedand retracted. Often there is transient tongue weaknesslasting a few weeks. The marginal mandibular branch ofthe facial nerve can be severed or stretched as it coursesaround the angle of the mandible, thus causing a saggingof the lower lip. The vagus nerve can be stretched, usuallyby retraction; this is evidenced by hoarseness of the voiceand is usually self-limited. These changes are most oftendue to dysfunction of the superior laryngeal nerve.

Vascular ComplicationsInjury to the carotid artery is a potential which is notlikely to occur because the dissection is medial to it andthe retraction is deep to it. Dissection and freeing of themedial rostral carotid sheath will reduce transmitted strainof retraction to the carotid artery. The vertebral artery is atrisk of laceration particularly while drilling or curettingthe lateral recesses of the C2 and C3 vertebral bodies andthe intervening disk. The surgeon must keep these struc-tures continuously in mind throughout the procedure.

Venous access may be a problem if the patient isdebilitated. Peripheral veins may be thrombosed, and sep-sis from indwelling peripheral and central venous lines isan ever-present burden. Central venous access is neededduring the operative procedure to guide fluid balanceand may be needed postoperatively for parenteral nutri-tion. Fastidious attention and sterile technique are re-quired to reduce problems with venous access.

Peripheral venous thrombosis and pulmonary embo-lism is an ever-present threat to the patient who is bedrid-den or under prolonged anesthesia. Sequential compres-sion pneumatic hose, mini-dose heparin, and a rotatingbed may help reduce this risk.

Respiratory ComplicationsUpper airway obstruction from soft tissue swelling is amajor postoperative problem with this procedure. This isparticularly the case with myelopathic and debilitatedpatients. Even the patients who are neurologically intactmay require a prophylactic tracheostomy to avoid this

problem. At the very least, patients will require an endot-racheal tube for several days to ensure an adequate air-way. Pneumonitis either from aspiration or atelectasismust be combated continually in the immediate postop-erative period. Positive end-expiratory pressure of 5 to10 cm H

2O on the ventilator is prophylactic for atelecta-

sis. A rotokinetic bed assists pulmonary drainage. A feed-ing jejunostomy may prevent aspiration.

Visceral ComplicationsPharyngeal perforation will negate the advantage of thetranscervical procedure over the transoral route. If unde-tected and not repaired it will lead to catastrophic septicfasciitis and/or a retropharyngeal abscess. Anatomic dis-section of the fascial planes and cautious retraction arerequired to avoid this; with attention to detail, this prob-lem is uncommon and can usually be avoided. It becomesa more imminent concern when scarring from previousoperations in or radiation to this region obscures land-marks and tethers retraction. The use of smooth-tippedrather than tooth-tipped retractor blades may help avoidinjury to the pharynx. I prefer the toothed retractor bladesbecause they anchor more securely into the longus collimuscle edge. However, proper seating of the blades be-neath the dissected medial edge of the longus colli muscleis mandatory.

A CSF fistula will occur if the subarachnoid space isopened and not specifically closed. A lumbar subarach-noid catheter will help prevent this from occurring andhelp close it when it does occur.

Structural ComplicationsInstability cannot be overemphasized. An aggravation ofmyelopathy occurs if this is not treated. Skull traction isrequired from the outset. A brace and/or surgical arthrod-esis may be needed.

Nutritional ComplicationsAdequate nutrition is critical for recovery of any stressedpatient. Nutritional insufficiency is compounded by apreoperative debilitated state, which is often the casewith patients harboring lesions in this region. The pha-ryngeal and upper airway edema that occurs after thisprocedure will impede deglutition for several days. Theresulting aggravation of a nutritional catabolic state canbe averted by a preoperative gastrostomy orjejunostomy.Optimal nutritional support will also help to avoid sepsisand to facilitate early ambulation and rehabilitation ofthese patients.

Hemodynamic ComplicationsHypovolemia and impaired sympathetic tone can com-

153


bine, dangerously reducing cardiac output. Postural hy-potension is regularly noted. This can further impair neu-rologic function. Attention to central venous pressureand fluid maintenance, and the administration of col-loid/crystalloid parenteral fluid combinations for fluidbalance support, will help avert this. Supplemental sym-pathomimetic agents maybe used, either intermittentlyor continuously for an adequate response.

Avoiding ComplicationsComplications will occur even with the best laid plans,particularly with the involved surgical procedure de-scribed here. Patient selection and preparation will helpreduce their incidence. Seven actions that will help ob-tain an optimal ultimate result, in my opinion, are:

1. Careful selection of a treatment plan based on analgorithm that considers reduction of malalignmentand specific imaged anatomy of the neural compres-sion;

2. Preemptive and electivea) Tracheostomyb) Feeding gastrostomy or jejunostomy;

3. Hydrodynamic control of CSFa) External lumbar subarachnoid drain (catheter

tunneled under the skin)b) Permanent CSF-peritoneal shunt if required;

4. Wide anatomic dissection of fascial planes;5. Microsurgical technique under fluoroscopic control;6. Nursing on a rotokinetic bed with skull traction;7. Intensive postoperative support.

154

CORRECTION OF MALPOSITIONOF THE ORBITS

JOHN A. PERSING, M.D.

INTRODUCTIONMalposition of the orbits is a congenital deformity that isfrequently seen in craniofacial anomaly centers. Thereare two commonly encountered forms of this abnormal-ity involving the mediolateral axis of the orbits.

The first is telorbitism, which refers to a wideneddistance between the two orbits. This is to be distinguishedfrom the second type, hypercanthorum, in which there isa widened distance between the medial canthi alone (thedistance between the lateral canthi is normal). The wid-ened distance between the canthi may be due to bone,soft tissue, or both being abnormally displaced laterally.In this chapter, only the condition of the bone displace-ment will be considered.

PATIENT SELECTIONThe appropriate operative procedure for correction forthe patient with hypercanthorum entails translocation ofthe medial portion of the orbits only, whereas in the pa-tient with telorbitism both the medial and lateral orbitalwalls are translocated medially. Patients are chosen forthe two different operative approaches based on the clini-cal measurement of a widened intercanthal distance (me-dial and lateral) and plain radiographic and computedtomography scan demonstration of widened bonyintercanthal distance. Surgery is elected usually at ap-proximately four years of age, unless a concurrent abnor-mality such as an encephalocele is present, providing theopportunity for onestage treatment. Age four is chosenbecause this allows for correction of the abnormality whenthe bone is sufficiently strong to avoid inadvertent frac-ture, the orbit has reached relative maturity, and the childhas not reached school age, so that major deformity canbe corrected, or at least ameliorated, prior to critical peerinteraction at school age.

PREOPERATIVE CONSIDERATIONSThe risks of the operative procedure include potential

injury to the brain and visual system. including enoph-thalmos, extraocular muscle entrapment, optic nerve dam-age, cerebrospinal fluid leakage, and recurrence of theintercanthal deformity postoperatively. It is unclear as towhether translocation of the orbits medially in early child-hood negatively affects midfacial growth.

Preoperative preparation includes prophylactic useof intravenous antibiotics at the time of surgery to coveranaerobic flora in the frontal and paranasal sinuses, andprophylactic, short-term, anticonvulsants. No steroidsare used.

A comprehensive anesthetic technique is advocatedwhich includes orotracheal intubation, monitoring forblood loss with central venous catheters, arterial line,and Foley catheter, and monitoring for air embolus byDoppler, end-tidal CO

2, and nitrogen monitors. Hypoten-

sion is induced by increasing concentrations of the inha-lation anesthetic agent at the time of craniotomy to mini-mize blood loss. Autologous or designated donor bloodtransfusion is preferred if blood transfusion is necessary.Spinal drains are placed in the lumbar cistern followingthe induction of anesthesia. The drains are not opened,however, until burr holes have been made and the cran-iotomy is about to be performed. The child’s trunk andextremities are wrapped with a soft gauze to maintainbody heat, and all irrigation fluids are warmed before useto prevent hypothermia.

SURGICAL TECHNIQUEThe initial preparation for the treatment of patientswith hypercanthorum or telorbitism is the same. Thepatient is placed supine with the head on a well-pad-ded headrest and the neck slightly extended. Drapingof the patient includes full exposure of the scalp to theregion of the midportion of the vertex of the skull andto the level of the mouth caudally. Hair removal in thescalp, if performed at all, is minimal, to a maximum of1 cm wide, corresponding to the course of the bicoronalincision. Hair anterior to the skin incision lineis braided to remain out of the way during© 1991 The American Association of Neurological Surgeons

155

PERSING : CORRECTION OF MALPOSITION OF THE ORBITS

the operative procedure and is covered with masking tapeor heavy aluminum foil. A bicoronal incision is performed,extending down to the level of the tragus anteriorly, toallow for easy dissection periorbitally. The dissection ofthe anterior scalp flap is performed in a supraperiostealplane to the level of the orbital rims, followed by eleva-tion of the periosteal flap from the same bicoronal inci-sion site to be used later as a covering flap for the anteriorcranial fossa floor defect created by the orbit transloca-tion procedure.

Bilateral pterion burr holes and one parasagittal burrhole posterior to the coronal suture are placed. A bifrontalcraniotomy line is drawn on the skull leaving approxi-mately 1 cm height of frontal bone superior to the apex ofthe superior orbital rim. A bifrontal craniotomy is per-formed. The outer table of the midline frontal bone in theglabellar region is removed with a side cutting burr, andthe frontal bone is fractured for-ward. Alternatively, asfrequently is the case in the region of excess bone in theglabellar region, a burr hole may be placed in the area ofintended bone removal (Fig. 1A).

HypercanthorumIn patients with hypercanthorum, the remaining bifrontalbone segment is bisected, leaving a supraorbital bar, ap-proximately 5-6 mm in width, cephalad to the medialportion of the orbital rim. If the nasal profile is accept-able, a segment of midline bone maybe left approximately3 mm wide to simulate a new nasal bridge. Two additionalapproaches exist, however, if the nasal profile is unac-ceptable. The midline bone may be removed entirely, leav-ing a 5-mm bone segment laterally on each medial or-

bital rim (Fig. 1B). When the orbital rims are translocatedmedially, the medial border of the orbital rim defines anew, more acceptable nasal profile. The second and mostoften used alternative is to leave a 3-mm segment of na-sal bone in the midline to serve as a base scaffolding foron-lay bone graft augmentation of the dorsum of the nose.With all these techniques, the medial orbital osteotomyis usually performed with a sagittal or oscillating saw toavoid unwanted fracture of the nasal and lacrimal bones.

The frontal lobes are allowed to reposition posteri-orly by cerebrospinal fluid drainage for an osteotomy inthe orbital roof extending posterior to the midpoint ofthe globe’s anteroposterior axis (Fig. 1C). The mediallimit of the osteotomy is the lateral cribriform plate,avoiding injury at this time to the olfactory nerve fi-bers. Characteristically, the cribriform. area is exces-sively widened and will obstruct medial translocationof the orbital rim. Therefore, the anterior-most olfactoryfibers are divided, and the proximal segments of thesenerve fibers and surrounding dura are oversewn to pre-vent cerebrospinal fluid leakage postoperatively. Theanterolateral portion of the ethmoid air cells are removedby rongeur, to allow subsequent unobstructed move-ment of the orbits medially.

If the medial canthi position relative to one anotherand to the anteroposterior axis of the nasal bone is ac-ceptable, effort is made to preserve their attachment tothe lacrimal bone. To avoid displacement of the canthiduring dissection, a subcilliary, conjunctival or Caldwell-Luc incision is made infraorbitally to complete the cau-dal dissection and osteotomy paralleling the superior or-bit osteotomy (Fig. 1, D-F).

Figure 1. A, a burr hole is placed in the glabellar region at the siteof intended bone removal (green) to allow for safe dissection of themidline dura and sagittal sinus. Orbital osteotomies are located, asshown, to include provision for removal of the medial inferiorportion of the nasal process of the maxilla so as not to impinge onthe nasal airway following medial orbital translocation. A support-ing frontal bone bar, approximately 5 mm tall, is left above themedial superior orbital rim. B, 1, the midline nasal bone has satis-factory projection but the breadth is too great. The midline nasalbone is left in situ, and resection of excess bone occurs in theparamedian location (green). 2, the midline projection is unac-ceptable. The midline bone is removed and the medial orbital walls,when translocated medially, form the new nasal profile. 3, theexisting nasal profile is deficient but, rather than excising the

midline nasal bone, it is allowed to remain in situ to serve as basescaffolding for dorsal augmentation by placement of a cantileveredbone graft. C, 1, view of the anterior skull base from above. Theorbital roof osteotomy extends from the midportion of the orbitlaterally to the cribriform plate medially. 2, the osteotomy ex-tends posteriorly well behind the midpoint of the axis of the globe.Bone is removed medially (green) adjacent to the cribriform plateto allow for medial translocation of the orbital roof. D, in order toplace osteotomies in the inferior orbital region, either atransconjunctival, Caldwell-Luc, or subciliary incision is made. E,in transconjunctival and subciliary approaches, a preseptal dissec-tion is preferred to expose the inferior orbital rim. F, the inferiororbital osteotomy is placed at the level of the infraorbital foramenin children to avoid damage to developing tooth buds.

156



157


A transversely oriented osteotomy is placed in the para-median frontal bone, leaving 5-6 mm width of frontal boneas the medial extension of the supraorbital bar.

If the midline nasal bone is to remain in situ, the os-teotomy line does not extend into the plane of the medialnasal bones. The segment complex of frontal, ethmoid,and more laterally situated nasal bones is removed byosteotome bilaterally (Fig. 2A).

The medial rim with superior and inferior rim extensions,in the form of the letter “C.” are mobilized. Ordinarily, thegreatest resistance to movement is encountered in thedeep nasomaxillary region. Prying of the rim with an os-teotome usually frees the bony and soft tissue attach-ments. Care must be taken, however, to avoid injury to thenaso-lacrimal duct during this maneuver. As the orbits aretranslocated medially, infolding of nasal cartilage (septaland upper lateral) and mucosa usually results, requiringtrimming of the excess tissue. The openings in the mu-cosa are oversewn to prevent gross air and bacterial con-tamination into the epidural space postoperatively. Themedial orbital walls and the nasal bones are trimmed withthe air drill to approximately 3-4 mm wide so that the over-all distance between dacryon and dacryon is approximately10 mm or less (Fig. 2B). Trimming is done at this time ratherthan earlier, to avoid fracture of the medial orbit during theprying maneuvers. The nasal process of the maxilla is ag-gressively trimmed to avoid occlusion of the airway as theorbital rim is moved medially. The orbital rims are placed in

position but are not yet secured.If the medial canthi need to be repositioned, a transnasal

medial canthapexy may be performed. It is easier to per-form the initial stages of the canthapexy prior to stabiliza-tion of the orbital bones because of the greater visibilityafforded by the mobile and widely separated bone seg-ments. A drill is used to open the entry point into theposterior superior lacrimal bone (Figs. 2, C and D, and 3A)to avoid fracture of this fragile bone. If the medial orbitalwall is exceedingly thin and does fracture, a split calvarialbone graft (Fig. 3B) from the parietal region may serve asa substitute for the medial orbital wall. It will provide sta-bilization of the medial canthus and avoid anterior as wellas lateral migration of the canthus postoperatively. Themedial orbital rims are then translocated medially and se-cured to each other and the frontal supraorbital bar.

If the midline nasal profile remains unacceptable aftertranslocation of the medial orbits, a cantilevered costo-chondral bone graft may be secured to the existing nasalprofile (Fig. 3, C and D). Costochondral rib grafts are theprimary choice graft material because the cartilaginoustip, unlike even calvarial membranous bone, resists re-sorption extremely well. To lessen the likelihood of epi-dural infection postoperatively, the pericranial flap pre-viously elevated at the time of anterior scalp flap dissec-tion is tacked over the ethmoid air cells on the anteriorcranial base. The scalp flap is then closed in two layers,galea and skin.

Figure 2. A, dissection is carried out posterior to the lacrimal crestmedially. An oscillating saw is used to complete the superior andinferior orbital osteotomy. The posterior medial orbital wall os-teotomy is completed by an osteotome. Particular attention is nec-essary to remove enough bone in the ethmoid air cell region to allowfor unimpeded translocation of the medial portion of the orbitalrim. B, following mobilization of the orbital rims, the midline bone

and remaining medial orbital rims are trimmed (green) with a shap-ing burr to achieve the thinnest possible final nasal midline (ordi-narily 8 to 10 mm encompassing the medial orbital walls and themidline bony strut). C, the medial canthi are elevated with an ac-companying periosteal pennant from the dorsum of the nose. D, adrill hole (green) is placed in the posterior superior lacrimal bone forthe periosteal pennant and the canthus to be passed transnasally.

158



159



160


Figure 3. A, an air-driven burr is used to penetrate the lacrimalbone perpendicular plate of the ethmoid to allow threading of thecanthal tendon. Any significant opening in the nasal mucosa isoversewn. B, if the lacrimal bone is too fragile or otherwiseunusable, a split calvarial bone graft can be used to serve as abuttress substitute for the lacrimal bone. The canthal periosteal

pennants are tightened following medial translocation and fixa-tion of the orbital rims. C, augmentation of the nasal bridge isaccomplished following translocation of the orbits by placing awedge of bone beneath the cantilevered costochondral cartilagegraft. D, the graft is secured to the underlying medial orbital rimsby transosseous wiring.

161


Hypertelorism (Telorbitism)For patients with hypertelorism (telorbitism), the opera-tive approach is much the same as just described forhypercanthorum, but, in addition, the lateral orbit ismoved with the medial, as a single unit.

The patient is positioned supine, following place-ment of a lumbar cerebrospinal fluid drain, and a bifrontalcraniotomy including temporalis muscle elevation is per-formed. The supraorbital bar is left 1 cm wide at the levelof the orbital rim apex (Fig. 4, A-C). The supraorbital barwill be transversely bisected when translocation of theorbit is performed, leaving a 5-mm-wide supraorbital barfor fixation, and a 5-mm-wide orbital rim at the rim’sapex. The orbital roof and lateral orbital wall are cut pos-terior to the midpoint of the globe following careful pos-terior repositioning of the anterior tip of the temporallobe. If correction of an accompanying orbital malrota-tion is not necessary, horizontally oriented osteotomiesare placed on the anterior surface of the maxilla throughor below the level of the infraorbital foramen (Fig. 4D). Inyoung children, the more cephalad osteotomy is desir-able to avoid injury to developing tooth roots. Olfactoryfiber section and frontonasoethmoid resections are per-formed as described previously for the treatment of

hypercanthorum.In patients with hypertelorism, the medial canthi may

require repositioning. This is performed by transnasalcanthopexy as described earlier (hypercanthorum). Like-wise, the need for repositioning of the lateral canthi ap-proximately 2 mm above the medial canthi on the hori-zontal axis also may be evident. After the orbits havebeen translocated and secured, the attachment point forthe lateral canthi in most cases, is placed just inside theorbital rim (Fig. 4, E-H). When severe exorbitism (globeprotrusion beyond the eyelids secondary to a constrictedorbit volume) coexists, the canthi are attached to orbitalrim bone on the external surface of the zygomatic pro-cess of the frontal bone. This reduces the projection ofthe globe beyond the eyelid. The temporal fossa is filledwith calvarial bone chip grafts and the temporalis muscleis advanced forward to be attached to the orbital rim, inorder to prevent an “hourglass” deformity or hollowingpostoperatively in the temporal region.

The incisions are closed in two layers, galeal andskin. The nose is packed with a petroleum-basedgauze. No drain is inserted in the galeal region toavoid aspiration of nasopharyngeal bacteria into thesubgaleal space.

Figure 4. A, a bifrontal craniotomy is performed with a 1-cm-tallsupraorbital bar left above the apex of the orbital rim. It is bisectedleaving a 5-mm-thick supraorbital rim which may be tr anslocatedmedially, following removal of paramedian frontal and nasal bone(green), and a 5-mm supraorbital bar to which the orbital rim boneis affixed. B, note resection of the nasal process of the maxilla toavoid impingement on the nasal airway. C, the orbits are thentranslocated medially. D, bone grafts are inserted posterior to thelateral rim of the orbit and in the region of the zygoma to preventa postoperative hourglass deformity. A drill is used to trim the

lateral portion of the supraorbital bar. E. the normal position ofthe lateral canthus is approximately at or 2 mm above the level ofthe medial canthus. F, drill holes are placed in the frontal process ofthe zygoma. The lateral canthus is attached transosseously to theinternal surface of the zygomatic process of the frontal bone, ifthe patient does not demonstrate globe proptosis. G, if the patienthas accompanying significant proptosis, the canthus is placed onthe external surface of the frontal process of the zygoma in aneffort to reduce the malrelationship between the globe and theeyelids and restore normal lid/globe anatomic relationships (H).

162



163


POSTOPERATIVE CAREThe patients are monitored postoperatively by clinicalmeasures, and intracranial pressure (ICP) monitoring isnot used routinely. This is because placement of a lumbarcistern drain allows for false low recordings of intracra-nial pressure despite clinical evidence of cerebral edema.This may lead to a tardy diagnosis of elevated ICP orintracranial hemorrhage. Because of this, pain controlpostoperatively should still allow for clinical assessmentof neurologic status.

COMPLICATIONSComplications from the operative procedure are rela-tively few. The major immediate concerns relate to cere-bral edema and/or intracranial hemorrhage, and injuryto the visual system, either to the globe or optic nerveby trauma or hematoma, or to the extraocular muscle

system. Also, it is important to note that if sufficientbone is not removed from the medial portion of the crib-riform plate as the orbit is translocated medially, thereis the possibility of impingement of the medial rectusmuscle on the corner of the remaining bone, which mayrequire reoperation. Later concerns include cerebrospi-nal fluid leakage, subdural or epidural infection, andosteomyelitis. The possibility of cerebrospinal fluid leak-age and meningitis should be significantly reduced bywatertight dural closure supported by the use of fibringlue at the suture line, with further support by the peri-cranial flap overlying the dural closure. Unresolved prob-lems are soft tissue relaxation at the medial canthal re-gion resulting in an apparent redevelopment ofhypercanthorum, and the possibility of growth distur-bance on the nasomaxillary and midface regions, withsurgery performed in early childhood.

164

REMOVAL OF CERVICAL OSSIFIEDPOSTERIOR LONGITUDINAL LIGAMENT

AT SINGLE AND MULTIPLE LEVELSRALPH B. CLOWARD, M.D.

INTRODUCTIONOssif ication of the posterior longitudinal ligament(OPLL) has been recognized as a cause for cervical spinalstenosis resulting in severe compressive myelopathy. Thislesion was first described by Japanese authors in 1960and numerous articles have since delineated its pathol-ogy and recommended techniques for its surgical treat-ment. Although it is quite prevalent in the Far East, OPLLis encountered less often in Caucasians. It is consideredimportant in the differential diagnosis of cervical myel-opathy. My experience with this lesion antedates the ini-tial 1960 published report by almost 10 years. My firstcase was diagnosed in 1951 and has been followed for 38years. The first operated patient, by laminectomy in 1956,did not improve.

Since 1958, all of my OPLL operations have been bythe anterior surgical approach. Direct access to the lesionregardless of its size and extent was accomplished byremoving the greater part of the vertebral bodies and theintervening discs. The resulting spinal defect was filledwith a large, well-fitting cadaver bone graft obtained fromthe bone bank and sterilized with ethylene oxide gas.

TWO OPERATIVE TECHNIQUES

Multiple Level OPLLIf the ossified lesion extends longitudinally over two ormore levels of the spine and occupies a minimum of 60%of the anteroposterior diameter (Fig. 1, A-E) a multiplelevel surgical technique is used. The surgical opening inthe spinal canal must expose the length of the symptom-atic lesion and be wide enough to expose the extremes ofthe lesion to facilitate its total removal. There must beextra room to insert a dural patch if necessary.

Operative TechniqueThe standard anterior surgical approach for treatment ofcervical disc lesions described elsewhere in this Atlas isused. A generous stripping of the longus colli muscles isessential to obtain the wide transverse exposure. Two setsof self-retaining retractor blades with teeth will exposethe number of disc spaces required for the vertical expo-sure. All discs are excised and the disc spaces are com-pletely cleaned to the maximum depth of the disc spacewhere the hard bony lesion is encountered.

The Cloward anterior cervical instruments are em-ployed, using the guide, the guard, and the drill attachedto a Hudson drill handle (Fig. 2A). The largest drill, 16mm in diameter, is used. The depth to be drilled is deter-mined by measuring the disc space with the depth gaugeand then projecting the drill 1 or 2 mm longer. The drill,therefore, will encounter and remove a millimeter or so ofthe anterior surface of the OPLL.

Successive drill holes are made in the adjacent,cleaned out disc spaces at two, three, or four levels (Fig.2B). If the patient has small vertebral bodies, the bonebetween the drill holes may be totally removed by thelarge drill. With large vertebral bodies, a narrow isthmusof bone will remain. This is nibbled away with a largebone rongeur (Fig. 2C). The remainder of the large open-ing in the spine is shaped with a high-speed drill using a“pineapple” burr (Fig. 3A). The upper and lower roundedends of the exposure are squared off, and then the lateralwalls are joined and finally bevelled slightly, wider atthe top than the bottom. The lateral margins of the OPLLare carefully separated from the loose areolar attachmentsto the dura. Then, the lesion is shaved down with a high-speed diamond drill. Starting at its thickest area, it isgradually reduced to a thin shell (Fig. 3B). A small hole ismade in the thinnest area to expose the dura. The thinangled osteophyte elevator is inserted and the undersideof the lesion separated from the dura. The upbiting 3© 1991 The American Association of Neurological Surgeons

165

CLOWARD : REMOVAL OF CERVICAL OSSIFIED POSTERIOR LONGITUDINAL LIGAMENT

Figure 1. Multilevel OPLL. A-C, a computed tomogram, axial view,showing the thick ossified ligament occupying greater than 60% of

the sagittal diameter of the spinal canal. D and E, sagittal and coronalreconstructed views showing the longitudinal extent of the OPLL.


166


Figure 2. A, use of the Cloward drill assembly to make multipletrephine holes in the cervical spine. B, appearance of the spineafter three successive drill holes have been placed. C, the interven-ing parts of the vertebral bodies between the drill holes are re-moved with a large bone rongeur. © 1991 The American Association of Neurological Surgeons

167


Figure 3. A, use of a high-speed drill to complete the bone removaland convert the bone defect into a smooth, bevelled, rectangular

channel. B, drilling the ossified posterior longitudinal ligament. C,insertion of the bone graft. D, technique of securing the bone graft.


168


mm 40° cervical rongeur is inserted and the remainingshell of the bony lesion is nibbled off bit by tiny bit, untilit is completely removed. Removal of the lesion may bedone under the microscope, although this is not essential.

If the dura is accidentally torn, the tear is suturedwatertight with fine sutures. If the lesion has invaded anddestroyed an area of dura, the defect should be patchedwatertight with a dural substitute. This is only possiblewith a wide dural exposure.

In one case in Japan, the dura had been invaded andincorporated into the OPLL over the entire length of thelesion (three levels). There was insufficient dura lateralto the lesion to attach any kind of dural substitute. There-fore, a large piece of Gelfoam was placed over the spinalcord and the bone graft inserted. The complications offree-flowing spinal fluid and meningitis almost cost thepatient her life. The cerebrospinal fluid drainage stoppedafter nine months and the patient finally made a com-plete recovery.

A large bone graft is shaped with an air drill to fittightly into the spinal defect so that the cancellous boneof the iliac crest is in close proximity to the sides of thespinal defect as well as to the ends (Fig. 3C) These longgrafts are secured to the spine at the end or the side withsmall loops of wire (Fig. 3D) In Figure 4, postoperative x-ray films (Fig. 4, D and E) and computed tomographyscans (Fig. 4, A-C) show long bone grafts and a widespinal canal.

Postoperative CareThe postoperative care is the same as that of the anteriorcervical disc operation. The patient requires only a softcollar postoperatively, with a tight-fitting secured bonegraft.

Single Level OPLLOPLL responsible for a severe neurologic def icit(radiculomyelopathy or quadriparesis) can occur from a

Figure 4. A-C, postoperative computed tomogram in axial view,showing a properly seated bone graft and evidence of completelyexcised OPLL. D and E, lateral and anteroposterior views of the

cervical spine showing the bone graft secured to the cervical spinewith wire loops (arrows).


169


lesion localized and confined to one level (Fig. 5) or resultfrom trauma to the end of a longer OPLL. A conventionalsingle level anterior discectomy is made with a large drillto make an opening sufficient to remove the intraspinallesion. The hole in the disc space is made with a 14- or 16-

mm drill and then enlarged with an air drill to a round or arectangular contour. The opening into the spinal canal issufficient to gain access to and totally remove the intraspi-nal lesion. The bone graft is fashioned to fit the spinaldefect, which may be either round or rectangular (Fig. 6).

Figure 5. A-D, a computed tomogram and cervical spine filmshowing OPLL at a single level (C3-4).


170


Figure 6. A postoperative film showing optimal placement of thebone graft and total removal of the OPLL.

CONCLUSIONSDirect and total removal of ossif ied and hypertro-phied posterior longitudinal ligaments of the cervi-cal spine is extremely effective in relieving thesymptoms and neurologic deficits of cervical myel-

opathy. Personal experience with this operation hasdemonstrated good to excellent results in 93-95%of patients. Over 75% of the patients operated uponhave reversed their functional def icits and becomeneurologically normal.


171

TECHNIQUE OF VENTRICULOSTOMYJOSEPH H. PIATT, JR., M.D.

KIM J. BURCHIEL, M.D.

INTRODUCTIONVentriculostomy is a rudimentary neurosurgical skill.Because it requires no special manual dexterity, ventricu-lostomy is one of the first procedures mastered by theneurosurgical trainee, but even the experienced operatorcan encounter difficulties if basic aspects of techniqueare neglected. This chapter first describes cannulation ofa normal ventricular system in general terms. The ana-tomic details of the frontal and the posterior approachesto ventriculostomy are then presented. Common errorsare highlighted.

GENERAL CONSIDERATIONSVentriculostomy is a daily exercise on a busy neurosurgi-cal service. The most frequent indication is the need fortemporary decompression of the ventricular system inthe setting of hydrocephalus. In the past, ventriculos-tomy has seen heavy use in the measurement of intracra-nial pressure (ICP), but several other techniques that en-tail less risk of ventriculitis are now available for use as amonitoring tool per se. If elevated ICP requires cerebrospi-nal fluid (CSF) drainage as a therapeutic measure, how-ever, ventriculostomy is still indispensable. Ventriculos-tomy may also be a preliminary step in a variety of otherneurosurgical operations, such as CSF shunt insertion,Ommaya reservoir insertion, excision of posterior fossalesions that have caused hydrocephalus, and functionalstereotactic procedures.

In theory, the ventricular system can be cannulatedfrom any site on the surface of the skull, but in practiceonly two sites are in common use: one near the midlinejust anterior to the coronal suture and another just ante-rior to the lambdoid suture. Beginning at these sites andutilizing simple external landmarks, the surgeon can can-nulate the ventricular system safely and easily. The pre-cise location of the standard sites is discussed below. It isprobably no more difficult to hit the ventricle from theone site than from the other, but it is our impression thanthe posterior site is more difficult. Selection of a site is

usually determined by extraneous considerations suchas the underlying disease process, the location of preex-isting incisions and burr holes, and the surgeon’s trainingand experience. In the absence of compelling reasons tothe contrary, it is customary to perform ventriculostomyon the right side in order to minimize the risk of injury tothe dominant hemisphere.

Whichever approach is selected, frontal or posterior,external landmarks guide ventricular catheter insertion,and these landmarks must not be obscured by surgicaldrapes. It is usually possible to feel the bridge of the nosethrough the drapes, but it is often impossible to locatethe ear. One solution to this problem is to mark sites nearthe nasion and the tragus with plastic syringe containercaps. These caps can be secured at the correct locationswith the adhesive tapes used to affix precordial stetho-scopes to the chest, and they are easy to feel throughconventional draping. Alternately, the scalp, face, andear can be draped all together in a single field with alarge, transparent, adhesive plastic drape. The scalp isshaved widely in order to permit subgaleal tunneling ofthe Silastic ventricular catheter. The incision, the mid-line, and the course of the nearby sutures are markedbefore draping, not after.

Although a twist drill is adequate for most situa-tions, a burr hole is advantageous because it allows he-mostasis under direct visual control as well as flexibilityin aiming the ventricular catheter. The dura is coagulatedand opened widely, and the pia-arachnoid is opened aswell. If an attempt is made to push a Silastic catheter witha wire stylet through a hole too small, the catheter willdrag on the dural aperture, and the stylet will puncturethe catheter tip and protrude into the brain parenchyma.

Once the site for the burr hole has been selected,the f inal position of the catheter is determined bythree geometric variables: the alignment of the cath-eter in the sagittal plane, alignment in one other per-pendicular plane, either the coronal or the axial, andthe depth of catheter insertion. For infants and smallchildren, the depth of catheter insertion can be esti-mated from the computed tomography (CT) scan, butin an adult with a nondisplaced ventricular system it© 1991 The American Association of Neurological Surgeons

172


is never necessary initially to insert the catheter morethan 6 cm deep to the outer table of the skull. If it isproperly directed, the catheter will encounter CSF at thisdepth. Strict control of the depth of insertion of the cath-eter and stylet minimizes the risk of neurological com-plications. Catheters are available with impregnated spotsat the 5, 6, and 7 cm marks. Alternatively, the cathetermay be marked with a loose ligature.

When the return of CSF is not immediate, the naturaltendency to insert the catheter further must be resisted. IfICP is not greatly elevated, it is helpful for the anesthesi-ologist to normalize arterial pCO

2 so that the cranial cav-

ity is sufficiently pressurized to force immediate ventingof CSF out the catheter as soon as the ventricular system ispunctured. An air lock in the lumen of the catheter may beeliminated by gentle irrigation with a small quantity ofsaline, and the distal end of the catheter may then bedropped in order to siphon ventricular CSF. If the catheterhas been positioned in the frontal horn, body, or trigone ofthe lateral ventricle, the flow of CSF should be free. If it isnot, the possibility that the catheter has been misplacedmust be considered. Other less capacious CSF spaces thatcan be cannulated inadvertently include the temporal horn,the interhemispheric fissure, the third ventricle, the sylvianfissure, and even the basilar cisterns. An intraoperativeplain skull radiograph or pneumoventriculogram will dis-tinguish among these possibilities. If there is an open fon-tanel or a craniectomy defect, intraoperative ultrasonogra-phy may be useful as well.

When free flow of CSF has been achieved, the ven-tricular catheter can be advanced without the stylet to itsfinal depth and tunneled under the galea to an exit site atleast 5 cm distant.

The principal complications of ventriculostomy arecentral nervous system infection and intracranial hemor-rhage, and they occur at low but not negligible rates. Inthe setting of ICP monitoring, positive CSF or cathetertip cultures are encountered at a rate of slightly less than10%. In only about half of these instances is there CSFpleocytosis or other signs of established ventriculitis.The risk of infection for an individual patient rises withthe duration of monitoring, the number of ventriculosto-mies performed, the requirement for other neurosurgicalprocedures, and the presence of intraventricular hemor-rhage. To minimize the risk of infection, it is commonpractice to replace the ventriculostomy catheter after 5days. Tunneling the catheter under the galea to an exitsite at least 5 cm away from the site of insertion probablyreduces the rate of infection as well. There is no consen-sus on the efficacy of prophylactic antibiotics. The riskof hemorrhage visible on CT along the course of the cath-eter maybe as high as 2%, but, prior to the CT era, symp-

tomatic hemorrhages requiring treatment were recognizedat much lower rates. Failure to cannulate the ventricularsystem may be considered a complication, as well; inexperienced hands the failure rate is as low as 1%.

Ventriculostomy is performed frequently for a widevariety of indications on the assumption that the risks ofinfection and injury to the brain are very low, but lapsesin technique quickly vitiate the assumption of low risk.Adequate lighting, draping, positioning and immobili-zation of the patient, instrumentation, and anesthesia areall critical. Patient restlessness disrupts sterile techniqueand makes landmarks difficult to maintain. Insufficientanesthesia for incision of the scalp and for tunneling ofthe catheter allows surges in systemic arterial pressureand central venous pressure that can exacerbate intracra-nial hypertension. Hemorrhage from the cortical surfaceoccasionally requires illumination, hemostatic agents,and the bipolar cautery for control. Although we recog-nize that accepted standards of practice vary from onemedical center to another, we believe that requirementsfor anatomic control and surgical asepsis favor perfor-mance of ventriculostomy in the operating room with theassistance of an anesthesiologist whenever the patient’scondition permits.

FRONTAL VENTRICULOSTOMYThe patient is positioned supine with the neck in a neu-tral position and the brow up.

Proper localization of the burr hole site is critical. Asite 3 cm from the midline, at the midpupillary line, and1 cm anterior to the coronal suture is recommended (Fig.1). Sites closer to the midline risk encounters with thelarge bridging veins draining the frontal lobes into thesagittal sinus or, catastrophically, with the sagittal sinusitself; a distance of 3 cm makes adequate allowance forthe usual uncertainty in determination of the midline.Sites further lateral invalidate the usual landmarks fordirecting ventricular catheter insertion. Sites further pos-terior begin to encroach on motor cortex, and anteriorsites require incisions on the forehead. The coronal su-ture can almost always be palpated in the midline.

Once the site for a burr hole has been selected, it isnecessary to align the catheter in the coronal plane andin the sagittal plane. In the coronal plane the cathetermust be pointed at the glabella (Fig. 2A). In the sagittalplane the target is a point about 2 cm anterior to thetragus of the ipsilateral ear (Fig. 2B). If the landmarkshave been successfully identified through the drapes,and if the catheter is properly aligned, CSF will be en-countered at a depth of 6 cm. The terminal portion of thecatheter will lie in the frontal horn of the lateral ventriclewith its tip near the foramen of Monro (Fig. 2C).

173

PIATT AND BURCHIEL : TECHNIQUE OF VENTRICULOSTOMY

Figure 1. For frontal ventriculostomy a burr hole site is selectedabout 3 cm from the midline, at the midpupillary line, and 1 cmanterior to the coronal suture. In the coronal plane the catheter isdirected toward the glabella.

Figure 2. In the coronal plane the ventricular catheter is passedfrom an entry site 3 cm from the midline toward the glabella (A).In the sagittal plane the catheter is directed from an entry site just

anterior to the coronal suture toward a point about 2 cm anteriorto the external auditory meatus (B). The tip of the catheter comesto rest in the frontal horn close to the foramen of Monro (C).



174


POSTERIOR VENTRICULOSTOMYThe patient is positioned three-quarters supine with aroll under the ipsilateral shoulder and with the head turnedfully toward the contralateral. shoulder (Fig. 3). It is ex-pedient for the head to rest with the brow at the horizon-tal plane or even slightly below.

Proper selection of the burr hole site is critical forposterior ventriculostomy as well. A burr hole near themidline allows insertion of a catheter down the length ofthe body of the lateral ventricle and also allows utiliza-tion of the medial canthus for orientation. A point 8 cmsuperior to the inion and 3 to 4 cm lateral to the midlineis suitable (Fig. 4). These coordinates place the burr holejust above the lambdoid suture. For patients in whom theinion cannot be palpated, such as young infants, the lamb-doid suture itself can be used for orientation: a site in theparietal bone just anterior to the suture 3 cm from themidline is satisfactory.

For posterior ventriculostomy a palpable marker atthe glabella permits orientation through the drapes in boththe sagittal and axial planes. From the posterior approachthere is a natural tendency to wander across the midline,and the surgeon should take care to aim at the ipsilateralmedial canthus in the axial plane. In the sagittal plane, themarker at the glabella is itself the target. If the catheter isproperly directed, it will encounter CSF at a depth of 6 cmas it punctures the dorsolateral wall of the trigone of thelateral ventricle. From this point the catheter can be ad-vanced into the body of the lateral ventricle without thestylet. For CSF shunts the depth of ventricular catheterinsertion can be measured from the CT scan. It is custom-

ary to place the tip of the catheter at the most extreme endof the frontal horn in order prevent obstruction of the cath-eter perforations by choroid plexus (Fig. 5). For temporaryventricular drainage or for ICP monitoring, the depth ofcatheter insertion is less critical. As for frontal ventriculos-tomies. externalized catheters should be tunneled to anexit site several centimeters from the burr hole in order tominimize the risk of infection.

The most problematic aspect of posterior ventricu-lostomy is locating the site for the burr hole. The surgeonmay be tempted to pick a more inferior site correspond-ing to the tip of the occipital horn, where the ventricularsystem is closest to the cortical surface. If, as in CSF shuntinsertion, the goal is to place the tip of the catheter in thefrontal horn, an inferior site is a mistake. From this ap-proach the surgeon faces the difficult task of slipping thecatheter up over the hump of the thalamus (Fig. 5). In-stead, the catheter may fall into the temporal horn and,because the thalamus itself is the obstacle, it can be in-jured by the stylet. A more lateral site above and behindthe pinna has been popular as well, perhaps because po-sitioning the patient is simpler for this approach. Thetrigone of the lateral ventricle, viewed en face from thisperspective, presents a large cross-sectional target area. Itis, however, a shallow target, and from this lateral ap-proach the surgeon must somehow persuade the catheterto turn frontally as soon as it encounters CSF. Further-more, because the surgeon has no good landmark for ori-entation of the catheter in the axial plane, the thalamusand internal capsule are at risk for injury by a catheterdirected slightly too far anteriorly.

Figure 3. For posterior ventriculostomy it is necessary to place aroll under the ipsilateral shoulder in order to turn the head fullyhorizontal.© 1991 The American Association of Neurological Surgeons

175

PIATT AND BURCHIEL : TECHNIQUE OF VENTRICULOSTOMY

Figure 4. For posterior ventriculostomy a burr hole is placed 8 cmabove the inion and 3 to 4 cm off the midline, just anterior to the

lambdoid suture. The catheter is directed toward the medial canthusin the axial plane and toward the glabella in the sagittal plane.

Figure 5. For posterior ventriculostomy the optimal length of theventricular catheter can be estimated from the CT scan. In thispatient the left ventricular catheter is slightly too long, but such anerror has the advantage of placing the catheter tip perforationsanterior to the choroid plexus and reducing the risk of catheterobstruction by ingrowth of tissue. The right ventricular catheterwas inserted through a burr hole situated inferior to the recom-mended site; this catheter had to be guided up over the hump of thethalamus, as its passage in and out of the plane of the CT scanindicates.



176

CEREBELLAR MEDULLOBLASTOMAARTHUR E. MARLIN, M.D.SARAH J. GASKILL, M.D.

INTRODUCTIONMedulloblastoma is a midline tumor occurring in theposterior fossa. While these tumors occur throughout life,they are far more common in children than in adults.They represent 15-20% of brain tumors in children with apeak incidence at 8 years of age and a 2:1 male/femaleratio. The tumor was first described by Bailey and Cushingin 1925. They believed it originated from thepluripotential medulloblast-a primitive cell which hasnever been identified. More recently, the tumor has beenthought to originate from cells of the subpendymal re-gion of the fetus and has thus been termed a primitiveneuroectodermal tumor of the posterior fossa.

The classical presentation is usually that of increasedintracranial pressure related to hydrocephalus. The aver-age symptom duration is two months. The most commonpresentation is, therefore, morning headaches and vomit-ing, papilledema, and ataxia.

The diagnosis of a midline posterior fossa tumor ismade by computed tomography (CT) or magnetic reso-nance imaging (MRI) (Fig. 1). Typically, on CT, the tu-mor has a slightly increased density and enhances withcontrast. There tends to be a low signal intensity on T1weighted MR images and enhancement with gadolinium.On T2 weighted images, the tumor has a high signal in-tensity. There are atypical features in about half the cases.In general, the tumor fills the fourth ventricle from thecisterna magna to the aqueduct of Sylvius, causing hy-drocephalus.

With the diagnosis of a midline posterior fossa massin a child, surgery is mandated because gross total removalprovides the best outcome regardless of tumor pathology.Any alternative therapy produces a poorer outcome.

PREOPERATIVE CONSIDERATIONSTiming of surgery will depend on the degree of hydro-cephalus and state of the patient. Unless the child is inpoor nutritional status and medical condition, the soonerthe surgery, the better. If the patient is not medically stable,

then increased intracranial pressure can be controlled withexternal ventricular drainage (EVD) until operative in-tervention is appropriate. With EVD, there is a definitebut small risk of upward herniation. This risk can be fur-ther decreased by careful control of intracranial pressurewith the avoidance of sudden cerebrospinal fluid (CSF)drainage and decompression. It is the senior author’s pref-erence to avoid EVD if possible and operate soon afterthe diagnosis is made. Pretreatment with steroids isthought to be beneficial and is always done, especiallywhen there is some delay in surgery. If possible, steroidsare not given until after the enhanced CT scan, as thismay alter the enhancement. Anticonvulsants are not rou-tinely used in posterior fossa surgery. Prophylactic anti-biotics are given for placement of the external ventricu-lar drain, if used, but are not routinely used for the surgeryitself. There is no definitive study to suggest that they areeffective in this circumstance.

We prefer to use the sitting position for medulloblas-toma surgery (Fig. 2). Anesthesia technique therefore re-quires a central venous catheter and Doppler monitoringfor venous air embolism. Arterial blood pressure, oxygensaturation, and end-tidal C02 are also monitored. Legwraps are used to prevent hypotension. The patient ishyperventilated to decrease intracranial pressure. Man-nitol or Lasix, however, are not routinely given. If, onexposure, the posterior fossa is tight, the lateral ventricleis cannulated and EVD is established.

POSITIONING AND DRAPINGThe major debate of operative positioning centers onthe sitting versus the prone or lateral decubitus(parkbench) type of position. The argument againstthe sitting position is the increased incidence of airembolism. The advantage of the sitting position is theexcellent exposure and decreased intracranial pressure.Although air embolism is best known in the sittingposition, it has been reported in the prone positionand, in fact, occurs in any position where the head ishigher than the heart. Because of the incidence of airembolism and the possible disastrous consequences,© 1991 The American Association of Neurological Surgeons

177

MARLIN AND GASKILL : CEREBELLAR MEDULLOBLASTOMA

Figure 1. Sagittal (A) and axial (B) T1-weighted MRI scans withgadolinium enhancement of a midline medulloblastoma. Note the

relationship of the tumor to the floor of the fourth ventricle,aqueduct of Sylvius, and cisterna magna.

special precautions must be taken in the sittingposition.

Pin sites should be wrapped with Vaseline gauze toeliminate these as a source of air embolism. Hemostasismust be meticulous. Air may enter the diploic spaces ofthe bone, so these must be carefully waxed. The Dopplermonitor should always be loud enough so the surgeoncan hear it to identify air at the same time as the anesthe-siologist. The Doppler monitor should be tested severaltimes with a rapid saline infusion during the operation.With changes in the Doppler sounds, the anesthesiolo-gist should attempt to aspirate the air from the right atriumand the surgeon should attempt to eliminate the source.A wet lap sponge should always be available to cover thewound when air is identified. The wound is then gradu-ally exposed to find the source. Jugular compression orValsalva’s maneuver may help.

Before surgery, all the members of the operative teamshould familiarize themselves with the head holder andstabilizing apparatus. If necessary, a drill should be doneso that all involved know how to quickly and effectivelychange the patient’s position to supine if needed. This ismore readily accomplished in the sitting position thanthe prone, and the anesthesiologist has the best access tothe patient in this position.

The patient’s head is placed in the three-pin headholder. The foot board is used as a seat for a small child sothe shoulders are at least three inches off the table whenthe head of the table is removed. The table is graduallyplaced in the sitting position with careful arterial pres-

sure monitoring, first by flexing the table and then byelevating the back. The head is held straight by the sur-geon while the fixation device is first secured proximalto the patient, and then tightening proceeds distally. Thehead should be flexed, taking care not to compromisevenous drainage or the airway. A spiral, flexible endotra-cheal tube is used. The chin should be about a finger’sbreadth off the chest. Once secured in this fashion, theentire table and patient, now as a single unit, can be tiltedslightly forward. During the procedure, upward and down-ward movement of the table will be required. Drapingshould allow for this movement. The arms are folded andsupported on the patient’s lap to prevent a traction bra-chial plexus palsy.

Once positioned and shaved, the incision is marked(Fig. 2). The inion is the major landmark. The occipitalsquama, C1, and C2 will need to be exposed. In general,an incision 4 cm above the inion and 8 cm below to aboutthe spine of C5 will be adequate. A burr hole incision isalso marked in case EVD is necessary. This is 4 cm lateralto the inion and 6 cm above it. The position of the innercanthus of the eye from this incision should be notedbefore draping so a ready pass can be made into the ven-tricle, if necessary. The incision is then infiltrated withsaline in small infants, or Xylocaine with epinephrine, toallow for dissection of the skin edges and hemostasis.The initial drapes are held in place with skin staples,which most adequately and expeditiously secure them.Sheets and a craniotomy drape with an incorporated win-dow and irrigation collection bag are used.


178


Figure 2. Top, the sitting position is demonstrated. Bottom, the incision is made with respect to theinion, 4 cm above and 8 cm below.


179


SURGICAL TECHNIQUEThe skin incision is made and Raney clips placed on theskin edges. To do this, each skin edge needs to be dissectedwith blunt dissection in the scalp and sharp dissection in theneck, taking care to leave the fascia intact. A gauze is usedover the skin in infants and small children to prevent necro-sis from the pressure of the Raney clips. The skin is thendissected laterally, using retraction with Weitlaner or cer-ebellar retractors, exposing the fascia. The fascia and musclesare opened in a Y fashion (Fig. 3) to decrease the likelihoodof a postoperative pseudomeningocele. The intersection ofthe limbs of the Y should be low on the skull for easiestclosure but still over the skull. The upper limbs of the Y areincised with the cutting current down to bone. Using peri-osteal elevators, the muscle is then stripped to the nuchalline. Care is taken not to go beyond this line as that wouldeliminate the advantage of the Y. As the muscle is stripped,emissary veins will be encountered on both sides. Theseneed to be waxed. The V flap is retracted with a heavy suture.The surgeon then continues the midline incision, strippingthe muscles from the arch of C1 and the spine and laminaeof C2. The remaining muscles on the occipital squama mustbe stripped laterally on each side to the region of the mas-toid process, which may be poorly developed in a child.

Two burr holes are now placed in the occipital squamaon each side of the midline. After stripping the dura fromthe bone, a craniectomy with rongeurs or a craniotomywith an air drill, such as the Midas Rex, can be performed.The bone here need not be replaced because the angle ofthe skull and the heavy musculature provide protection.The bone is removed superiorly until the inferior portionof the transverse sinus is visualized, and laterally to theregion of the mastoid. The sigmoid sinuses need not beexposed. Care should be taken in the midline. Here, thebone can have a “keel-like” projection intracranially, andthe occipital sinus can be quite large. Also, care shouldbe taken with the arch of C1, which may be cartilaginousin the young child. The craniectomy should be performedbefore the removal of the arch of C1 and the spine of C2;these provide protection in case an instrument slips. Theinferior pole of the tumor should be exposed; the preop-erative MRI will indicate the need to remove C1 and C2.Bone edges are carefully waxed to prevent air embolism.

Assessment of the intracranial pressure in the poste-rior fossa can now be made. It is surprising that even withlarge lesions, with the sitting position and hyperventila-tion, the dura will not be tight. If this is the case, it can beopened as described below.

If the dura is tight, then the burr hole incision ismade. The calvarium is perforated. The dura is coagu-lated and incised. An external ventricular drain is placed.

This is tunneled to exit at a distance from the incision.CSF is slowly allowed to escape until the dura can beopened without cerebellar herniation. Care must be takennot to decompress the ventricles too rapidly, as anextraaxial hematoma, subdural or epidural, may result.Intracerebral hemorrhage may even occur.

The dura is also opened in a Y fashion with the infe-rior limb slightly off the midline (Fig. 4). First, the dura isopened over each hemisphere to the region of the occipi-tal sinus. At this point, either small hemostats or hemoclipsare used to isolate the sinus before it is transected. Thesinus is then tied off with a 4-0 silk suture and the clips orhemostats removed. The V dural flap is elevated and re-tracted superiorly with a suture. The dural opening isthen extended inferiorly off the midline to the caudalportion of the foramen magnum or the inferior pole of thetumor. These dural flaps are retracted by suturing them tothe paravertebral muscles. The table-patient unit is tiltedfurther forward to provide better visualization.

The foramen magnum and usually the tumor are vi-sualized. Using a 22-gauge needle, the arachnoid of theforamen magnum is penetrated. CSF can be taken forcytology; the significance of this cytology, however, isunclear. The arachnoid is then widely opened. This canbe done with the needle or forceps.

If the inferior pole of the tumor is seen, it is gentlyelevated with a small retractor or Penfield dissector. Thefloor of the fourth ventricle is visualized and a cottonoidis placed on it (Fig. 5). If the tumor is not visualized, thecerebellar tonsils are elevated, the floor of the fourth ven-tricle is visualized, and a cottonoid is placed.

At this point, the Yasargil self-retaining retractor isbrought into the operative field. The cerebellar hemi-spheres are slightly retracted, if necessary, to clearly ex-pose the vermis. The arachnoid and pia over the midlinevermis are coagulated with bipolar cautery, and a linearincision is made into the vermis. Using blunt dissection,this is extended until the tumor is encountered. Planeslaterally are then developed using bipolar coagulation,gentle retraction, and cottonoids. At this point, biopsiesare taken. Medulloblastomas are usually friable and quitereadily auctioned.

As one extends deeper and superiorly, the planebetween tumor and cerebellum may become indis-tinct, but the demarcation between normal and ab-normal tissue is readily visible. The tumor should bemanipulated very gently with careful concern regard-ing the anatomy involved. If the tumor is not readilysuctioned, the Cavitron ultrasonic aspirator (CUSA)should be used. The table can be further tilted andelevated. Working at the exposed surface. the tumor

180


Figure 3. The Y-shaped fascial incision is shown.


181


Figure 4. A Y-shaped incision is made in the dura, taking care to tie off the occipital sinussuperiorly, and making the inferior limb slightly off the midline.


182


Figure 5. Top, tumor is seen after the dura is opened and thevermis is partially split. The vermis will be further split, and thetumor plane developed. A cottonoid is placed on the floor of the

fourth ventricle. Bottom, after tumor resection, the enlarged aque-duct of Sylvius can be visualized.


183


will come inferiorly into the field. Working with gentlesuction, the rostral portion of the fourth ventricle andaqueduct will come into view. Once seen, attention shouldbe turned caudally to the cottonoid on the floor of thefourth ventricle. The tumor should now be gently liftedfrom the fourth ventricle and the cottonoid advancedrostrally. This portion of the tumor can now be removedwith forceps and suction. The first goal of surgery, tounblock the aqueduct. has now been accomplished.

Using the CUSA, the remaining tumor is now re-moved. Up to this point, there is usually oozing from thetumor which is not easily controlled until the tumor isremoved. Large bleeding vessels are controlled with bi-polar coagulation. If the tumor does not readily come offthe floor of the fourth ventricle or is invading the floor, itis “shaved” off with gentle suction here, and the brainstem is not violated. Most commonly, the tumor does notinvade the floor of the fourth ventricle and is readilylifted from it. In this fashion, the second goal of surgery,gross total tumor removal, is accomplished.

Once hemostasis has been achieved, the tumor cav-ity is inspected for possible residual tumor (Fig. 5). He-mostasis can be assessed by Valsalva’s maneuver or jugu-lar venous compression. Questionable areas of tissueare sent for frozen section. No hemostatic agents are leftbehind.

Once satisfaction has been achieved, the dura isclosed with 4-0 silk. If watertight dural closure cannot be

obtained, a fascial patch is used. Gelfoam is used to coverthe dural closure. The paravertebral muscles and fasciaare now closed with 2-0 Vicryl sutures. A good fascialclosure is very important to prevent a pseudomeningocele.The subcutaneous tissue is closed with 3-0 Vicryl. Theskin is closed with 3-0 Vicryl in a subcuticular fashionfollowed by the application of Steristrips.

POSTOPERATIVE CONSIDERATIONSThe table is now tilted back to the original position, andthe three-pin head holder is removed. A soft cervical col-lar may be used for comfort. The patient is extubated inthe sitting position and is kept sitting for the first 48hours.

Within 48 hours after surgery, a CT scan, withoutand with contrast, is done to assess the extent of tumorremoval and the degree of hydrocephalus. This must becompared to the preoperative study. Signif icantpneumocephalus will be apparent, but some decrease inventricular size should be obvious. A significant numberof patients, perhaps 40%, will subsequently need aventriculoperitoneal shunt.

One week postoperatively, CT myelography is doneto rule out metastases and stage the patient for adjunc-tive therapy. With this, CSF is sent for cytology, and bonemarrow aspiration is performed. Currently, CT myelogra-phy appears superior to MRI for evaluation of spinalmetastases. This will probably change in the near future.

184

INTRODUCTIONCavitation of the injured portion of the spinal cord fol-lowing trauma is a common phenomenon. As a conse-quence of this posttraumatic cavitation, a syrinx mayform within the substance of the cord from several monthsto many years following the injury. The presenting symp-toms and signs of a syrinx are varied and may be as subtleas a slight change in sensory level or as dramatic as in-tractable pain. A constant awareness of the potential forsyrinx formation in the patient with a spinal cord injuryis the most effective way to ensure that these lesions arefound.

DIAGNOSISPrior to the advent of magnetic resonance image (MRI)scanning, the diagnosis of posttraumatic syringomyeliawas difficult. One technique used to search for a syrinxwas postmyelographic computed tomography (CT) scan-ning, looking for an increase in the radiodensity of cystswithin the spinal cord (Fig. 1). Another commonly usedtechnique was air myelography. The “collapsing cordsign” was indicative of a positive test. Both of these testshad faults that produced both false negatives and falsepositives. MRI scanning has significantly improved theaccuracy of syringomyelia diagnosis and follow-up (Fig.2, A and B). The entire spinal cord may be visualized in asagittal plane, and the extent of a posttraumatic syrinxmay be seen. Unfortunately, a large percentage of pa-tients with spinal trauma also have some form of internalfixation device in place to facilitate spinal stabilization.Such ferrous materials may disrupt the image of the spineobtained with the MRI and prevent visualization of somecritical areas of the spinal cord. Sampling of informationfrom the spinal cord above or below the metal artifact isstill possible and treatment planning may be carried outusing this information.

PREOPERATIVE PREPARATIONSAs with all types of surgical procedures, preoperativediscussions with the patient should include discussion

of risks versus potential benefits of therapy. In additionto the standard risks associated with any surgical proce-dure, some authors have reported dramatic loss of neuro-logic function following decompression of the syrinx.

The position for placement of the syrinx to sub-arachnoid shunt is the major preoperative decision tobe made. In general, the shunt catheter should go intothe most dependent portion of the syrinx cavity butabove the level of injury in a complete lesion. MRIimages should be able to show the extent of the cyst,position of the cyst within the substance of the cord,and some of the intracystic anatomy. The best surgical

SHUNTING OF APOSTTRAUMATIC SYRINX

DAVID J. GOWER, M.D.

Figure 1. An example of a delayed contrast CT scan of a cervicalposttraumatic syrinx. Note the area of contrast enhancement withinthe parenchyma of the spinal cord.© 1991 The American Association of Neurological Surgeons


185

GOWER : SHUNTING OF A POSTTRAUMATIC SYRINX

Figure 2. A, a 35-year-old patient four years after a lower tho-racic compression fracture with an incomplete spinal cord injury.The patient presented with an ascending sensory level andmidthoracic pain intractable to medical therapy. Note the large

syrinx extending from C2 down into the thoracic spinal cord. B,one month following a thoracic syrinx to subarachnoid shuntingprocedure. The syrinx is now decompressed in the cervical regionand the patient’s pain is improved.

method for treatment of a syrinx is to shunt the collectedspinal fluid back into the subarachnoid space or into theperitoneal cavity. The shunt tubing in the spinal cordshould be placed in the most dependent area of the cystbut with consideration of the concentration of function(i.e., thoracic rather than cervical). The size of the cyst ateach level is also important, and it is preferable to placethe tubing in an area with the greatest amount of dilationto reduce the risk of injury to remaining function of thecord. In those patients with thoracic or conus lesions,placement of the shunt in the thoracic region is prefer-able to the cervical area. The drainage end of the shuntshould be placed in the subarachnoid space but in somecases of trauma this may not be possible and the patientshould be prepared for possible diversion of cerebrospi-nal fluid into the peritoneal cavity.

Anesthetic ConsiderationsGeneral anesthesia is the anesthetic of choice. While somesurgeons are capable of performing complex operationsupon the spine under local anesthesia, the risk to thepatient with small amounts of movement while placingthe tubing may be great. Preoperative medications mayinclude a course of dexamethasone and/or a cepha-

losporin prophylactic antibiotic. The patient should bemildly hyperventilated to reduce intracranial/intraspi-nal pressure and the potential for rapid dilation of thesyrinx following the opening of the dural envelope. Pneu-matic sequential compression stockings are usually usedon the patient’s lower extremities to reduce the risk ofdeep venous thrombosis and potential pulmonary embo-lism. Monitoring of somatosensory evoked potentials inconjunction with possible rectal sphincter electromyo-graphy may also be useful.

Operative PositioningThe patient should be positioned prone on the table ina manner appropriate for the area of the spinal cordthat is to be approached. Chest rolls are helpful in de-creasing venous congestion and reducing bleeding.Patients with a cervical lesion should be placed in tongson a circular headrest with care being taken to protectthe eyes or in a three-point fixation system. The armsshould be placed at the side with appropriate padding.With a thoracolumbar approach, the patient should bepositioned in a prone position with the arms up onarmboards. Before the skin of the patient is prepared,an x-ray f ilm of the area should be taken to


186


locate the correct position for the shunt; a radiopaquemarker may be placed upon the skin to identify the loca-tion. A mark should be made on the skin lateral to the siteof the incision with a skin marker or with a needle, butthe scratch should not cross the midline.

DrapingThe wound should be draped out in the standard fashion sothat the mark for proper location is easily seen in the opera-tive field. The site of the incision should be marked with anappropriate skin marker and should be covered with an ad-hesive drape that may or may not be impregnated with io-dine. The field should be wide enough to allow access to theabdomen should a syrinx to peritoneal shunt be required.Frequently during this period of time the circulating nursein the room win pass the shunt tube onto the back table. Thetubing should be in contact with the room air for the mini-mum amount of time to decrease the risk of dust attaching tothe surface by electrostatic attraction. The tubing should beplaced directly into a solution of Bacitracin (50,000 units in250 ml of saline) and kept in the solution until ready toinsert into the spinal cord.

SURGICAL TECHNIQUE

Skin IncisionThe skin incision should be in the midline. Since theappropriate vertebral level was previously identified ra-diographically, the incision should be centered over thislevel. The length of the incision usually can be limited to7-10 cm.

Operative ProcedureThe paraspinous musculature should be retracted usingan appropriate retractor, and the wound should be mademeticulously free from bleeding sites. The lamina overthe site that had been previously decided upon should beremoved using a high-speed drill. In most cases only asingle level needs to be removed with some of the bottomof the lamina above and the top of the lamina below alsodrilled away. Wide exposure on the side of the proposedshunt should be carried out. The bleeding from the boneedges may be controlled with bone wax or Gelfoampressed into the diploë; the bleeding from epidural veinsis controlled using bipolar electrocautery. The guttersbetween the dura and the bony edge are lined withGelfoam moistened with thrombin in order to controlbleeding, and the wound is lined with cotton sheets.

The dura is opened by a small nick with a sharpinstrument, leaving the arachnoid intact (Fig. 3). Onemethod of further opening the dura is to insert the tip of ablunt nerve hook and gently pull the instrument superi-orly to split the dura along the natural lines of division.One should be careful to leave at least 2-3 mm of un-opened dura at each end of the wound in order to facili-tate later closure in a water-tight fashion.

The actual site for the myelotomy should be con-sidered once the dura is open. Midline opening of thespinal cord will potentially damage the dorsal columnsand decrease the patient’s proprioceptive function fromthe lower extremities and should be avoided. The pref-erable area to enter the cord is in an area of obviousthinning or in the dorsal root entry zone. The

Figure 3. A sharp hook is used to elevate the dura so it maybe opened with a scalpel.


187


side of the entry should be determined prior to the operationby careful examination of the MRI scan. Damage to thethoracic dorsal root entry zone will give at worst a dermatomalpatch of numbness around the patient’s thorax.

The arachnoid should be opened paramedian andspinal fluid released. The edges of the arachnoid can begently picked up using a forceps and tacked to the edgeof the dura using hemaclips. Usually two to threehemaclips are necessary to ensure that the arachnoid istacked up out of the way. This portion of the procedurewill allow an easy identification of the subarachnoid spaceto ensure that the placement of the subarachnoid portionof the syrinx to subarachnoid shunt is not in anextraarachnoid position. The subarachnoid space shouldbe observed at this time. In most cases there should beminimal scarring and access to the subarachnoid space,but, where access is poor, a shunt of cerebrospinal fluidinto the peritoneal cavity may be necessary.

Attention should be directed to the spinal cord wherethe pia mater is cauterized using the bipolar cautery on alow power setting. Care should be taken not to damagelarge vessels on the surface of the cord. In those patientswhose spinal cord appears reasonably normal withoutdilation or thinning, a 26-gauge needle should be passedinto the cord to puncture the cyst and ensure that thesyrinx extends into the operative site. The fluid withinthe posttraumatic syrinx should be clear and colorless.

Yellow fluid or bloody fluid would imply a different patho-logic process. A myelotomy approximately 1 cm in lengthis made by first incising the pia with a No. 11 blade andthen using bipolar forceps and a microdissector to openthe cord into the cyst longitudinally.

In patients with syringomyelia secondary to trauma,often the syrinx is asymmetrical (Fig. 4) with several blindpouches that do not allow the easy passage of the shunt-ing catheter. The proximal arms of the shunt (Fig. 5) shouldbe passed in each direction within the syrinx cavity (Fig.6). The entire length of the shunt tube should be used, ifpossible.

In those cavities that are not large enough to accom-modate the entire shunt tube, the tube should be trimmedso that it does not stretch or distort the spinal cord. Greatcare should be taken while inserting the tube into thespinal cord to ensure that the tubing is within the syrinxcavity and is not dissecting a separate plane outside thecavity through the parenchyma of the cord. After place-ment of one end of the small “K” or “T” tube, it is neces-sary to kink the tubing at its midpoint and have an assis-tant hold the middle connector while the other end of thetubing is passed into the syrinx cavity. The spinal cordshould receive the minimal amount of manipulation pos-sible while at the same time maintaining the smallestpossible myelotomy so that the tubing does not slip out.Once the tube is in place, the distal end of the tube in-tended for the subarachnoid space needs to be placed.

The dentate ligaments can be identified on the lat-eral aspect of the spinal cord. The distal tube should beplaced under the leaf of arachnoid that has been pre-

Figure 4. Cervical MRI scan demonstrating posttraumatic syrinx.Note that the cavity is irregular and is paramedian. This Informa-tion should be considered when planning placement of a syrinx tosubarachnoid shunt tube.

Figure 5. Photograph of the small silicone plastic tubing used toform the syrinx shunt. The tubing is very flexible and may becustom fit to the syrinx cavity.

© 1991 The American Association of Neurological Surgeons© 1991 The American Association of Neurological Surgeons

188


Figure 6. The proximal arms of the shunt are passed into the syrinx.


189


served along the dura, and passed ventral to the dentateligament (Fig. 7). This will ensure that the tubing is sub-arachnoid. There should only be a minimum of bleedingduring this procedure, and all bleeding must be stoppedand the wound completely dry before closure is started. Ifultrasonography is available, it can be used to examinethe spinal cord at this time to look at the tubing withinthe syrinx cavity. Comparison of the spinal cord beforeand after placement of the shunt tube should ensure thatthe syrinx is adequately drained,

In those patients with poor access to the subarach-noid space, the distal aspect of the shunt should be con-nected to a standard peritonea] shunt catheter which ispassed through a subcutaneous tunnel to a separate inci-sion on the flank. The tubing is passed into the peritonealcavity and the separate wound closed. The tubing shouldbe carefully tacked to the paraspinous tissue in a mannerthat will not kink the tubing. This is sometimes best ac-complished by bringing the tubing down in a loop aroundthe next spinous process before bringing it out throughthe fascia. If at all possible, one would hope to see cere-brospinal fluid leaking from the distal end of the catheterbefore its placement into the peritoneal cavity.

ClosureClosure is started by removing from the wound the metalclips that were used to hold the arachnoid back. These aretaken off the field and not left within the wound. The

arachnoid is allowed to fall back into position. The durais then closed using either an interrupted or running stitch.Stitches should be placed very close together to ensurethe best chance of a watertight closure.

After closure of the dura, Valsalva’s maneuver per-formed by the anesthetist increases the subarachnoidpressure; any leaks that are identified should be rein-forced using sutures. In those patients in which duralclosure is difficult, fibrin glue consisting of cryopre-cipitate mixed in equal proportions with topical throm-bin is helpful in producing a watertight closure. Severalstitches can be placed to reapproximate the muscle bel-lies and then sutures placed for the tight closure of thefascia, again to reduce the chances of cerebrospinal fluidleakage. The skin is usually closed with a running ny-lon suture in a locking stitch to provide a firm barrieragainst cerebrospinal fluid leakage. Percutaneous drain-age of the wound should be avoided if at all possiblesince this will leave a track to the surface. Skin suturesshould be left in these wounds for longer than the usualperiod of time, sometimes up to two weeks, to ensurethat the wound is well-healed.

SPECIALIZED INSTRUMENTATIONRemoval of the lamina in these patients should be donewith a high-speed drill. The Midas Rex drill is particu-larly well-suited because of the large number of bitsavailable and the speed with which the drill turns.

Figure 7. The distal end of the shunt is placed within the subarachnoidspace ventral to the dentate ligament.© 1991 The American Association of Neurological Surgeons

190


The M8 or AM8 bits are particularly good for this type oflaminectomy. Microinstrumentation and an operating mi-croscope are also of value while placing the shunt tubinginto the syrinx cavity. Intraoperative ultrasonography isan interesting but not a necessary component of the pro-cedure to ensure that the tube is in place and the syrinxadequately drained.

POSTOPERATIVE COMPLICATIONSNeurological deterioration postoperatively should beevaluated in a method that is appropriate to the extentand rate of deterioration. Those patients who awaken withcatastrophic neurologic deficit should be studied on anurgent basis with CT or MRI scanning or wound re-ex-ploration if appropriate. In those patients who awakenwith a mildly worsened neurologic deficit, conservativetherapy and observation may be indicated since the ma-nipulation of the spinal cord may have made the patienttemporarily worse.

Cerebrospinal fluid leaks should be treated in anaggressive manner. Small leaks from the wound can betreated by reinforcing stitches of the wound edges. Majorleaks of cerebrospinal fluid should probably be re-ex-plored and the dura again closed at the site of leakage. Inthose patients in whom it is necessary to re-explore forcerebrospinal fluid leakage, use of fibrin glue or autolo-gous blood patching is helpful in reducing the problemsof continued leakage postoperatively.

A small superficial wound infection or stitch abscesscan be treated by local wound care. Superficial woundinfections should be treated by gentle cleansing of thewound with removal of debris. Deep infections should betreated in the operating room by opening the fascia andpacking the incision. Like most shunting devices, infec-tion of the tubing would necessitate removal of the pros-thesis from the spinal cord. Most likely, this form of in-fection would present as meningitis, and it would beunlikely that this could be cured with antibiotics alone.

191

DIRECT SURGICAL TREATMENT OFVEIN OF GALEN MALFORMATIONS

HAROLD J. HOFFMAN, M.D.

INTRODUCTIONThe clinical presentation of a patient with a vein of Galenmalformation is entirely dependent on the severity of shunt-ing through the fistula. Neonates with galenic malforma-tions characteristically have multiple fistulae (Fig. 1) whichshunt more than 25% of their cardiac output through themalformation, resulting in high output congestive heartfailure. In addition, low diastolic blood pressure reducescoronary blood flow and results in myocardial ischemiawhich aggravates the heart failure. Preferential flow throughthe malformation bypasses the cerebral cortex, leading tocerebral ischemia (Fig. 2). These infants are gravely illbecause of their associated cerebral and myocardial is-chemia and cannot be managed medically. Direct surgicalattack on these neonatal malformations of the vein of Galencan rarely be done with safety. Modern interventionalneuroradiologic techniques can lead to obliteration of manyof the fistulae feeding the galenic malformation. In doingso, the patient can be rescued from intractable heart failureand, if any residual malformation is left, surgery can beused to deal with this in a well patient who is no longer inheart failure and whose brain has been salvaged from is-chemia. In using such interventional techniques one mustbe cognizant of the characteristic constriction in the drain-ing straight sinus, which is probably a protective mecha-nism (Fig. 3).

The older infant and young child with a galenicmalformation typically has very few fistulous connec-tions to the vein of Galen (Figs. 4 and 5). Although thesechildren may show evidence of heart failure, they charac-teristically present with a greatly dilated vein of Galenwhich occludes the aqueduct and posterior third ven-tricle and produces hydrocephalus. These patients arecandidates for surgical obliteration of the fistula.

Older children and adults with galenic malforma-tions frequently have a relatively low flow angiomatousnetwork supplying the vein of Galen. These patients arerarely candidates for surgical attack on the lesion.

PREOPERATIVE PREPARATIONThe patient who has heart failure should be treated medi-cally and by interventional neuroradiologic techniques.Because these patients can suffer cerebral ischemia, theyare prone to seizures and anticonvulsant medicationshould be used. If significant hydrocephalus is present,this should be treated with a cerebrospinal fluid diver-sionary shunt.

OPERATIVE PROCEDURE

AnesthesiaNeonates and infants have a small blood volume which canbe severely compromised during surgery for a galenic mal-formation. Central venous and arterial lines are mandatoryfor monitoring purposes. A warming blanket must be usedto reduce heat loss during the operation. Mannitol, steroids,and hyperventilation can aid in producing a relaxed opera-tive field. The neonates and infants cannot be placed in apin fixation headrest because of fear of penetrating the skull.An infant horseshoe headrest is sufficient.


Figure 1. Lateral carotid arteriogram of a neonate with an aneu-rysm of the vein of Galen showing multiple fistulous feeders enter-ing the aneurysm.


192


Figure 2. Computed tomography scan of a neonate with an aneu-rysm of the vein of Galen and ischemic damage to the right cere-bral hemisphere.

Figure 3. Magnetic resonance imaging scan of an infant with ananeurysm of the vein of Galen showing characteristic constrictionin the straight sinus.

Figure 4. Lateral (A) and anteroposterior (B) views from a verte-bral arteriogram of an infant with an aneurysm of the vein of Galen

showing a large fistulous branch of the posterior cerebral arterywhich enters the aneurysm at its anterosuperior border.




193

HOFFMAN : DIRECT SURGICAL TREATMENT OF VEIN OF GALEN MALFORMATIONS

Figure 5. Lateral (A) and anteroposterior (B) views from acarotid angiogram showing a large fistulous branch of the pos-

terior cerebral artery which enters the aneurysm at i tsinferolateral border.

Operative RoutesA variety of routes have been used to deal with a galenicmalformation. These include transcallosal, subtemporal,and transtentorial routes.

The transcallosal route is the approach of choice forthose galenic malformations in which the feeding ves-sels enter the lesion over the dome of the sac, particularlyat its anterosuperior aspect. In those patients with mul-tiple feeders entering from both sides it may be necessaryto stage the transcallosal route-going in from one sideinitially and then approaching from the other side at asubsequent occasion.

The subtemporal approach is the best route to thosemalformations that are fed by one or occasionally by aunion of both posterior cerebral arteries. The fistula canbe clearly visualized by this route and successfullyclipped or ligated.

The transtentorial route should be used only for thosemalformations in which there is a single feeding vesselentering at the level of the tentorial notch. By openingthe tentorium the surgeon can obtain a clear view of thisvessel which is usually a branch of the posterior cerebralartery that can directly enter the vein of Galen. However,in neonates with a massive dilated venous drainage sys-tem the tentorium is extremely vascular; opening into itrisks massive hemorrhage and so the transtentorial routeis not used in such patients.

Transcallosal RouteThe patient is positioned supine with the head elevated

30° above the body and placed on a horseshoe headrestin anatomical position so that the nose faces directlyforward (Fig. 6A). The vertex is then prepared and draped.The skin incision is square, with the anterior limb of theincision just behind the coronal suture and the medialportion of the incision just across the midline on the farside, extending back to just in front of the lambda fromwhich the posterior transverse limb of the incision runsparallel to the anterior transverse limb. A free bone flap isturned and the dura opened. Care should be taken topreserve parasagittal draining veins. The dural openingis made in such a fashion as to preserve as many of thedraining veins as possible.

Utilizing self-retaining brain refractors, the hemi-sphere is retracted away from the falx and the corpuscallosum exposed (Fig. 6B). The corpus callosum is fre-quently thinned out by the underlying aneurysmal veinof Galen. In neonates there may be very large anteriorcerebral arteries lying on the corpus callosum and pen-etrating through it into the aneurysm. Using suction orthe ultrasonic aspirator, an opening is made through thecorpus callosum which brings one on to the dome of theaneurysm. The aneurysmal sac is not adherent to brainand its wall is sufficiently thick that it can be mobilizedso as to visualize the feeding arteries. As many feedingarteries as one can safely occlude should be clippedutilizing titanium clips (Fig. 6C). If one is successful,the sac will collapse or become less tense and the colorof blood will change from red arterial blood to bluevenous blood.


194


Figure 6. A, a line drawing showing the skin incision (dashed line)and bone flap for the transcallosal approach. B, refractors are on thefalx and the right cerebral hemisphere. One draining vein has beendivided in the course of elevating the dural flap. A distended feedinganterior cerebral artery branch can be seen beneath the arachnoid on

the corpus callosum. C, the anterior cerebral feeder has been clipped.Retractors are separating the divided edges of the corpus callosumincision. The anterior cerebral artery feeder is entering the aneu-rysm at its anterosuperior border. A posterior cerebral artery feedercan be seen entering posteriorly and laterally on the aneurysm.


195


Figure 7. (Continued)


196


Subtemporal RouteIn patients with a single branch of the posterior cerebralartery creating a fistulous tract into the vein of Galen oroccasionally into the vein of Rosenthal and from thereinto the vein of Galen, a subtemporal route can be used tovisualize the feeding vessel and directly ligate or clip it.This route is particularly useful in infants with a singleposterior cerebral artery branch feeding directly into agreatly dilated vein of Galen.

The patient is positioned prone with the head turnedto expose the side of the subtemporal approach (Fig. 7A).The skin incision extends from in front of the ear to abovethe squamosal temporal suture and goes back to the mas-toid. An osteoplastic flap is turned and the dura opened.If the brain has been properly relaxed by treatment of thehydrocephalus and utilization of mannitol and hyper-ventilation, the temporal lobe can be easily retracted (Fig.7B), the tentorial edge visualized, and the feeding poste-

rior cerebral artery branch going into the aneurysm li-gated or clipped (Figs. 7C and 8).

Transtentorial RouteIf the feeding vessel comes in more posteriorly, it is use-ful to open the tentorium and retract upward on the tem-poral and occipital lobes and downward on the cerebel-lum in order to expose the aneurysm and find the feedingvessel which can enter the vein of Rosenthal, thus dis-tending it as well as the vein of Galen.

The patient is positioned in much the same fashionas for the subtemporal route (Fig. 9A). However, the flapis more posteriorly placed so that one elevates the brainbehind the vein of Labbé (Fig. 9B). The tentorium can becut using a blunt hook and utilizing the monopolar cau-tery to coagulate and cut this structure. Once the fistu-lous vessel is exposed as it enters the vein of Rosenthalor vein of Galen, it can be clipped (Figs. 9C and 10).

Figure 7. A, a line drawing showing the skin incision (dashed line)and bone flap for the subtemporal approach. B, a retractor haselevated the temporal lobe. The vein of Labbé is preserved. C, the

temporal lobe has been further elevated, fully exposing the tento-rial edge, the feeding posterior cerebral artery branch, and theaneurysm. A clip has been placed on this feeder.


197




198


Figure 8. A, a lateral vertebral arteriogram showing a fistulousposterior cerebral branch entering the base of the aneurysm. B, a

postoperative lateral vertebral arteriogram of the same patientshowing the ligated fistulous branch and no filling of the aneurysm.

Figure 9. A, a line drawing showing the skin incision (dashed line)and bone flap for the transtentorial approach. B, a retractor iselevating the occipital lobe, exposing the tentorium. C, with further

elevation of the occipital lobe and with the tentorium divided, a largefeeding posterior cerebral artery branch is seen entering a distendedvein of Rosenthal which is continuous with the vein of Galen.



199




200


Figure 10. Operative photograph showing the split in thetentoriurn and the appearance of an aneurysmally dilated

vein of Rosenthal after clipping of a posterior cerebral ar-tery feeder.

Closure TechniquesThe dura is tacked up to the bone flap and an epiduraldrain is left in place for 12 hours. The bone flap is reat-tached using nonabsorbable sutures. The scalp is closedin two layers using absorbable sutures in the galea andstaples in the skin.

MonitoringNormal perfusion pressure breakthrough has been reportedin patients following direct surgery on an aneurysm of thevein of Galen. In most of these patients, the surgery hasinvolved interfering with the venous drainage from theaneurysm. I believe that if the surgery is restricted to cut-ting off the arterial supply to the aneurysm and leaving thevenous drainage intact, no problems should occur withnormal perfusion pressure breakthrough. However, corti-cal blood flow can be measured using a thermal diffusionflow probe which will allow any postoperative hyperemic

edema to be anticipated. It is important during the opera-tion to not let the patient become hypovolernic and tomaintain an adequate central venous pressure and a nor-mal systolic blood pressure.

COMPLICATIONSThe wall of an aneurysm of a vein of Galen can on occa-sion be relatively thin and so the aneurysmal sac must betreated with care to avoid disastrous hemorrhage. As men-tioned, normal perfusion pressure breakthrough has beenreported, complicating surgery for vein of Galen aneu-rysms. If the surgery is restricted to occluding the feedingarteries entering the vein of Galen, normal perfusion pres-sure breakthrough should not occur. Such a maneuver iseffective with these lesions in that, unlike the usual arte-riovenous malformation, these feeding arteries are largefistulous tracts going directly into the vein of Galen,which has a normal but enlarged venous drainage.


201

SPINAL NERVE SCHWANNOMAPHYO KIM, M.D.

BURTON M. ONOFRIO, M.D.

SELECTION OF PATIENTSIntraspinal schwannomas maybe subdivided as thosepurely intraspinal, abutting on, but not extending be-yond, the lateral limits of the mid-pedicle, or those dumb-bell-shaped, extending lateral to the pedicle. Dumbbelltumors may be primarily within the spinal canal with aminimal extent lateral to the pedicle, or primarilyextraspinal into the posterior triangle of the neck, or intothe thoracic, abdominal, or pelvic cavities with a mini-mal intraspinal component.

Most patients present with pain of long duration inthe involved dermatomes; however, they may present witha mass in the posterior cervical triangle or may have an“incidental tumor” found on routine chest x-ray (Fig. 1).Various forms of long tract deficits may cause hand and/or leg numbness and weakness if spinal cord or caudaequina compression is the early sign of the presence oftumor. When a sacral tumor is present, sacral pain, saddlehypalgesia, or a urinary or anal sphincter disorder may bethe first noted abnormality.

In a patient with von Recklinghausen’s disease, mul-tiple tumors are present. Determining the one or ones thatare symptomatic may pose a difficult problem, as well aschoosing which ones may be safely and appropriatelyremoved while not upsetting the homeostasis the othernonsymptomatic lesions have at that time with other brainand/or spinal cord segments.

The def initive diagnosis of spinal nerveschwannoma is made on the basis of radiographic stud-ies. Plain x-ray films of the spine (especially the ap-propriate oblique view) may show enlargement of theintervertebral foramen (Fig. 2) or erosion of the adja-cent vertebral bodies, pedicles, or transverse processes(Fig. 3). The cortical margins of the bone are usuallywell-maintained, denoting the slow growth pattern ofthese benign lesions. The tumor mass may be visual-ized by plain and enhanced computed tomography (CT)scans. Magnetic resonance imaging scans using T1weighted imaging with gadolinium enhancement and

T2 weighted imaging, either of the entire spine or tai-lored to the appropriate segments, best defines the massand the soft tissue interphase with the spinal cord orcauda equina. Water-soluble contrast myelography andpostmyelogram CT scanning show the relationship ofthe mass with the subarachnoid space. Myelography isparticularly useful to rule out dural and arachnoiddiverticulae which may be confused with solid or cys-tic tumors, especially in the posterior mediastinum.These spinal fluid diverticulae may not require sur-gery since they are usually innocuous incidental le-sions. Postmyelogram CT scanning may indicate, es-pecially in dumbbell tumors, those which are


Figure 1. Posterior mediastinal tumor found incidentally on aroutine chest x-ray film (arrows). Note left T9-T10 rib erosionwith intact bony cortical margins indicative of an indolent tumorgrowth pattern.


202


Figure 2. An oblique thoracic spine film shows marked forami-nal enlargement with intact pedicle and facet cortical margins(arrows).

Figure 3. Anteroposterior tomogram shows left vertebral body,pedicle (single arrow), and proximal rib (double arrows) erosion.

amenable to total removal solely by the anterior approach(Fig. 4).

If the segment is functionally critical, namely, spinalroots at the levels of C5-T1 and L3-S1, many surgeonshave attempted partial root preservation, fearing signifi-cant functional loss to a limb. Recent studies at the MayoClinic showed that detectable neurologic deficit occursonly infrequently after division of the root (C5-T1, L3-S1) together with the tumor. The trend was consistentthroughout the cases irrespective of tumor size and ex-tension (intradural, extradural, and intra-extradural mass).The postoperative deficits, which appeared in a smallgroup, were partial, and good functional recovery wasalways achieved after physical therapy. Therefore, if thetumor is inf iltrating diffusely into the host nerveroot(lets), our current practice is to remove the entire in-volved root to achieve complete tumor removal.

To plan an appropriate skin incision, especially incases of a thoracic region tumor, a skin marker is placedat the level of the lesion either during myelography orusing fluoroscopy the night before the operation. Whenthe sitting position is to be used for laminectomy fortumors in the cervical region, patients should undergoa thorough cardiac examination to rule out atrial sep-tal defects. Intraoperatively, the anesthesiologist takesmeasures to prevent air embolism, such as placinga central venous catheter in the atrium andattaching a Doppler monitor probe on the

Figure 4. Although vertebral body erosion has occurred, as shownon this postmyelogram CT scan, this multidensity posterior medi-astinal tumor causes no significant deformation of the subarach-noid space (arrows) and was removed entirely from the anterior,transthoracic approach.




203

KIM AND ONOFRIO : SPINAL NERVE SCHWANNOMA

anterior wall of the chest. Intraesophagealechocardiography may be useful in detecting and visual-izing the presence of air in the atrium during the surgery.Maintenance of a normal blood pressure range intraop-eratively is critical to prevent neurologic complicationssecondary to defective spinal cord autoregulation. Pro-phylactic antibiotics (we use vancomycin, 1 g, intrave-nously) are given during the induction of anesthesia. So-matosensory evoked potential monitoring is used tominimize the risks of injury in tumors causing signifi-cant compression of the spinal cord or cauda equina.

SURGICAL PROCEDUREConcerning solely extradural tumors, when the extradu-ral segment of the parent nerve root proximal to the tu-mor is long enough, an anterior approach with piecemealremoval and intraforaminal debulking will deliver theroot safely into view and permit a silver clip applicationproximally, thus precluding a spinal fluid leak. If thetumor is fibrous, not lending itself to debulking into theforamen or if the origin of the tumor from the root is tooclose to the common dural sac, a combined anterior andposterior approach will be needed (Fig. 5).

If the dumbbell tumor is in the cervical segmentsand has a significant intraspinal segment, we prefer aposterior approach with a total facetectomy at the appro-priate level. In the vast majority of cases, even when large,the extraspinal tumor may be delivered piecemeal. Thesurgeon, staying within the tumor capsule posteriorly,gently pulls on the tumor capsule while placing counter-

traction on the structures abutting the capsule. Using bi-polar coagulation, meticulous hemostasis may beachieved and clear visibility reduces the chance of in-jury to the carotid sheath and vertebral artery (Fig. 6).

For thoracic dumbbell tumors extending I cm be-yond the lateral aspect of the facet on the anteropos-terior imaging studies, the patient is placed in thelateral position and the chest, abdomen, and back are

Figure 6. This tumor has not extended beyond the confines of aneroded and expanded cervical spine segment (arrows) and may beremoved totally using the posterior laminectomy approach. Water-soluble myelography with postmyelogram CT scanning should be usedto define the possibility of an intradural component prior to surgery.

Figure 5. A and B, the parent root foramen, harboring a tumor,has been enlarged in both examples. The tumor in A demands a

posterior and anterior approach. The tumor in B may be removedtotally with an anterior approach.



204


Figure 7. The lateral position used for the thoracotomy whenemploying the anterior approach may be modified intraopera-tively to allow a posterior laminectomy approach, in the non-

obese patient, by rotating the patient forward, exposing the tho-racic spines posteriorly.

prepared and draped. A thoracic surgeon performs an in-tercostal incision between the ribs corresponding to thevertebral segment involved (Fig. 7). The team approachmaintains safety in reflecting the aorta on the left or thevena cava on the right. The head of the rib caudal to theforamen is removed. If the tumor is on the T8 nerve root,the head of the T9 rib is removed to visualize the forameninvolved. The intercostal artery and veins may be safelycontrolled both proximal and distal to the tumor andclipped. The tumor is debulked, maintaining the tumorcapsule integrity if possible. As the debulking is carriedmedially through the foramen, visualization of the par-ent nerve proximal to the tumor maybe possible if thetumor is soft or partially cystic. If a clip can safely beapplied to the root proximal to the tumor, a posteriorapproach is not needed. If the tumor is fibrous or there isnot a suitable proximal uninvolved segment of nerve longenough to clip, the procedure is terminated, silver clipsapplied to the tumor margins, the chest incision closed, adrain inserted into the pleural cavity and attached to un-der-water suction, and the patient turned to the proneposition. The back is prepared and draped and the spinousprocesses of the adjacent vertebral segments along withlaminae and facet joint are exposed widely. A total hemi-laminectomy of the cephalad and caudal segments is doneon the appropriate side. The total facetectomy is performedat the involved foraminal level, the ligmentum flavum isremoved, and a plane is achieved medial to the tumor.The parent root is found and doubly clipped or ligated toprevent a spinal fluid leak, and the remaining tumor isremoved. An orthopedist evaluates the need for a fusionat this juncture.

In aesthenic patients, the anterior intercostal ap-proach in the lateral position may be extended to includethe posterior vertebral elements for facet and lamina re-moval without closing the chest. By rotating the patient45° anteriorly, the posterior vertebral elements may beexplored with the chest still open.

Extending the safety limits of either the anterior orposterior approach is to be avoided. If needed, the two

approaches using the same anesthetic are well-toleratedin the thoracic, lumbar, or sacral segments. Working inthe depth of one approach to avoid a second incisionrisks spinal cord, major vessel, or visceral damage andthat philosophy is to be avoided (Fig. 8). For dumbbellsacral tumors, an anterior transdural approach plus a sec-ond phase, same anesthetic, prone posterior approach, isusually needed (Fig. 9).

Solely intradural tumors demand adequate visual-ization to delineate the proximal and distal parent nerveroot, to prevent avulsion from the ventral spinal cordwith disastrous subpial hemorrhage. Removing the laminaand performing a facetectomy, flush with the pedicle, onthe side of the tumor ensure an adequate field for safetumor removal.

Figure 8. The foramen is massively eroded by tumor. However,because of a tight irregular tumor common dural interface (ar-rows), intracapsular subtotal tumor removal anteriorly will likelyneed to be followed by a posterior partial hemilaminectomy andtotal foraminotomy to allow safe and total tumor removal.



205

KIM AND ONOFRIO : SPINAL NERVE SCHWANNOMA

Figure 9. A, a sacral tumor totally contained within the sacralcanal needs only a posterior approach. B, any extension anteriorto the sacral foramen will need an anterior and posterior approach.

When an intradural tumor extends through the foramen anteri-orly in the cervical segments, the tumor and nerve root are re-moved piecemeal. A fascia lata or homologous dural graft is usedto close the dura laterally after the tumor removal has beenaccomplished (Fig. 10, A and B). Rarely is there a need for asecond stage procedure anteriorly to remove “unreachable” tu-mor extending anteriorly. If there is residual tumor, the secondstage anteriorly is done several weeks later after the dural grafthas been well-incorporated, to prevent a cerebrospinal fluid col-lection in the anterior neck after the second stage procedure.

Dumbbell intradural thoracic tumors with significant anteriorextension should also be staged to avoid a spinal fluid leak com-municating with a negative pressure intrapleural space, possiblyprecipitating low pressure headaches and postoperative subdu-ral intracranial hematomata caused by the persistent cerebrospi-nal fluid pleural fistula. For dumbbell intradural tumors affectingthe spine below the diaphragm, the posterior and anterior ap-proach may be done the same day, with the surgeon attemptinga watertight dural closure. The lack of a significant negativepressure system intraabdominally makes a persistent extraspinalcerebrospinal fluid collection unlikely.

Figure 10. A, totally intradural tumor, B, an intra-extradural extraspinal dumbbell tumor.



206

Our preference in case of a laminectomy for a cervi-cal region tumor is the sitting position, since this allowsexcellent exposure with gravity drainage of blood andcerebrospinal fluid, maintaining a relatively bloodlessfield. A Gardner headholder with three pins is placed onthe head, with the pins on the frontal or parietal areaavoiding the temporal squama. The upper half of thetable is raised about 60°, and the lower half is raised toelevate the legs. The headholder is secured to a frameconnected to the siderail of the table, and the neck is

positioned in moderate flexion to make the alignmentof the vertebrae approximately vertical. The flexion ofthe neck should not be excessive to prevent occlusionof the jugular vein or anterior cord compression inpatients harboring moderate spondylosis. To minimizethe risk of air embolization pertinent to the position,meticulous precautions should be made, such as wax-ing all exposed bone edges and remaining subperiostealas much as possible when reflecting the muscles priorto performing the laminectomy.


207

COMBINED CRANIOFACIAL RESECTIONFOR ANTERIOR SKULL BASE TUMORS

EHUD ARBIT, M.D.JATIN SHAH, M.D.

INTRODUCTIONThe rationale for combined craniofacial resection of askull base tumor is that this procedure can achieve a com-plete monoblock excision of the tumor with adequatemargins. Our experience suggests that this is the mosteffective approach to local control of disease, and, incombination with radiation therapy and chemotherapy,it may also improve long-term survival.

A wide variety of tumors may involve the anteriorskull base. They may be grouped into three main catego-ries according to their site of origin: 1) tumors that origi-nate in the nasal cavity or paranasal sinuses and extendto the skull base; 2) tumors arising from the orbital con-tents or lacrimal gland, which may ultimately erode theorbit and skull base; and 3) tumors of intracranial originthat transcend the skull base to involve the orbit, nasalcavity, paranasal sinuses, or infratemporal fossa. The lastgroup are usually either meningiomas or metastases.

The most common tumors in these areas are malig-nant and arise in the paranasal sinuses. Most are of epi-thelial origin and include, in order of frequency, epider-moid carcinoma, undifferentiated carcinoma, andadenocarcinoma. Fewer than one-third of these tumorsare of salivary gland origin. Approximately 10% areesthesioneuroblastomas arising from the olfactory epi-thelium. There are also rare cases of various sarcomas,lymphomas, metastases, and ectopic meningiomas. Tu-mors of paranasal sinus origin most often occur in thefifth and sixth decades of life, with the exception of theesthesioneuroblastoma, which is more common in thethird and fourth decades of life.

PRESENTING SYMPTOMSPresenting symptoms and signs are seldom characteristicand are often confused with those of sinusitis or allergic

rhinitis. The most common complaints are nasal stuffi-ness, obstruction, nonspecific sinus discharge, and occa-sionally local pain. Less frequent symptoms include pe-riorbital edema, pain over the cheek and forehead, andexcessive lacrimation. Rare symptoms include epistaxisand cerebrospinal fluid (CSF) rhinorrhea.

Neurologic symptoms indicate that the disease ismore advanced, with extension of the tumor beyond thesinuses or into the cranial cavity. Anosmia indicates in-volvement of the olfactory nerves and is commonly asso-ciated with esthesioneuroblastoma and olfactory groovemeningioma. However, essentially all tumors that erodethe cribriform plate or infiltrate the nasal cavity can causeanosmia. Hypesthesia or pain in the cheek may indicateextracranial involvement of the maxillary nerve or ero-sion of the skull base in the foramen rotundum area. Ocu-lar motor signs and diplopia may result either from nerveinvolvement or displacement of the globe, due either tointraorbital extension of the tumor or to intracranial in-vasion through the foramen lacerum into the middle fossa,cavernous sinus, and/or superior orbital fissure. Trismususually indicates involvement of the pterygoid fossa.

RADIOGRAPHIC EVALUATIONInvaluable clinical information may be gained by theradiographic evaluation of the paranasal sinuses, skullbase, and adjacent brain. Visualization of soft tissuedetail relative to osseous walls and air spaces in thesinuses, as well as delineation of disease processes be-yond the boundaries of the sinuses, are indispensablein determining operability and in planning radio-therapy. Radiographically, it is important to assess theextent of the soft tissue mass in the infratemporal fossa,in the parapharyngeal space, and in the frontal andtemporal fossae. Critical to determining operabilityare the pterygoid region, nasopharynx, sphenoid si-nus, cavernous sinus, orbital apex, and cerebrum.Edema in the frontal lobes, in the absence of an en-hancing tumor, suggests that the dura is involved© 1991 The American Association of Neurological Surgeons

208


and in some instances bridged by microscopic foci oftumor.

Before the decision is made as to operability, in caseswhere the tumor originated in the paranasal sinuses, pa-tients routinely undergo biopsy to determine the histol-ogy of the tumor. Treatment planning and discussion withthe patient are facilitated by this information, as the prog-nosis depends largely on the type of tumor present.

The indications for combined craniofacial resectionare still somewhat subjective. Cases that should be con-sidered include those in which the tumor originates inthe nasal cavity, maxillary antrum, or orbit, extends tothe frontal ethmoid or sphenoid sinus complex, and mayor may not involve the base of the skull and dura. Theobjective is to achieve a gross monoblock resection ofthe tumor with clear surrounding margins. When this goalis impossible—e.g., because of invasion of the cavernoussinus or cerebrum beyond the frontal lobes-craniofacialresection is contraindicated.

SURGICAL TECHNIQUESBefore surgery, nasal cultures are obtained. Patients arealso evaluated by a prosthodontist so that prostheses toreplace parts of the palate, maxilla, or orbit can be madeif necessary. Antibiotics and glucocorticosteroids arestarted preoperatively.

The operation is performed under general anesthesiawith endotracheal intubation. The tube is placed in themouth on the side contralateral to the procedure, and theoral cavity is packed with gauze to avoid the subsequentpostoperative aspiration or swallowing of pooled blood.Arterial and venous, lines are placed, as is a central venouspressure line. The patient is turned on his side, and twospinal needles (18-gauge) are introduced for continuousintraoperative drainage of CSF. After these preparations,the patient is returned to the supine position. The needlesare secured through the gap in the operating room tableand mattress, and connected to an intravenous extensionline, which in turn is connected to a large glass syringe.The entire drainage system is sealed in sterile plastic bags.This closed system allows for continuous drainage, whilesaving CSF, which can be used at the end of the proce-dure for reexpansion of the brain.

Following initial preparation, the head is fixed in athree-pin Mayfield headrest in a straight supine posi-tion, with a mild elevation and a 15-20° extension. Thisfixed position permits visualization of the anterior fossaas far back as the anterior clinoid processes, while pro-viding adequate working exposure of the cribriform plateand planum sphenoidale with minimal brain retraction.

The eyes are protected with corneal shields and the headis shaved and prepared with Betadine. While the head isbeing prepared, a split thickness skin graft is obtainedfrom the thigh and preserved in antibiotic and saline so-lution, for later use in covering the mucosal defect.

The scalp incision-usually a bicoronal scalp inci-sion-is outlined with a marking pen (Fig. 1A). After infil-tration with an epinephrine/Xylocaine solution, the skinis incised as deep as the galea, carefully preserving theintegrity of the underlying pericranium. After incision ofthe galea, a subgaleal plane is developed, the posteriorscalp flap is retracted as far as possible, and the pericra-nium under this posterior part of the flap is incised andelevated with a periosteal elevator (Fig. 1B). The anteriorscalp flap, with attached pericranium, and the temporalmuscle on one or both sides, are elevated from the skulland from the superior temporal lines. This subperiostealdissection proceeds to the supraorbital ridges and gla-bella (Fig. 1C).

At this point, a free cranial bone flap is raised (Fig.2). Usually this is a bifrontal flap; however, for smallerlateral lesions, we have used a one-sided flap that scarcelycrosses the midline. In raising the bone flap, we try toavoid burr holes in the low frontal area. The Midas Rexhigh-speed drill/knife is effective in sawing through theanterior wall of the frontal sinus (Fig. 2). To provide visu-alization and maneuverability in the posterior frontal re-gion with minimal cerebral retraction, it is important toraise a low flap, the inferior margins of which are thesuperior orbital ridges, regardless of the location of thefrontal sinus. After being elevated, the flap is removedfrom the operating field. The frontal sinus mucosa is ex-enterated with pituitary rongeurs, and the sinuses packedwith Gelfoam and antibiotic solution. The posterior wallof the frontal sinus can be sawed with a high-speed drillor removed piecemeal with rongeurs. While the bone flapis elevated, it is important to preserve the integrity of thedura, and all dural tears should be repaired immediately.

Because most craniofacial operations are performedfor extradural tumors, dissection in these instances isstrictly extradural. The dura is gently elevated off the or-bital roofs and the anterior fossa. It may be helpful duringthis maneuver to vent CSF, thereby shrinking the intradu-ral volume, and to use a self-retaining retractor (e.g.,Greenberg or Leyla retractor). Dural dissection of the or-bital roof is fairly simple. However, separation of the durafrom the multifidous crista galli at the midline can be moredifficult. We often elect to free the crista from its base withsharp instruments or narrow-tipped rongeurs, and to el-evate the dura with part of the crista. The olfactory

209

ARBIT AND SHAH : COMBINED CRANIOFACIAL RESECTION FOR ANTERIOR SKULL BASE TUMORS

Figure 1. A, scalp and facial skin incisions (dashed lines). Abicoronal scalp incision for craniotomy and a Weber-Fergussonincision for the facial exposure are shown. B. the scalp flap isreflected posteriorly to permit elevation of the pericranial (peri-osteal) graft. C, the pericranial flap is reflected anteriorly as far asthe supraorbital ridges.



210


Figure 2. Burr holes are made and the frontal bone overlying the sinuses is cut with a pneumatic bone scalpel.


211


rootlets are cut sharply from the cribriform plate (Fig. 3).If any holes were created during dissection in the duraoverlying the crista galli and cribriform plate, these arenow sutured meticulously.

In cases where the tumor bridges the dura and ex-tends intracranially, no attempt is made initially to sepa-rate the basal dura from the skull base. Instead, the dura isopened along the floor of the anterior fossa. The falxcerebri is transected, and frontal lobes are elevated andinspected. By careful dissection, the tumor usually canbe separated from the brain parenchyma and then removed.Ideally, the tumor is separated first from the brain paren-chyma, and then the entire circumference of the dura isincised, in order to include all of the involved dura in thespecimen. The intracranial tumor and the dura are thenseparated from the skull base and removed from the oper-ating field. Afterward, the remaining basal dura is elevatedfrom the base of skull, using the same steps as for tumorsthat do not invade the intracranial compartment. Whenthe dura has been elevated and exposed, the dural defectis sutured either primarily, if there is enough dural tissue,or using a pericranial graft or temporalis fascia graft.Meticulous closing of the dura is crucial in preventing aCSF leak and potential infection.

The next phase of the operation is to prepare forextirpation of the remaining specimen. This is accom-plished by an osteotomy through the planum sphenoidaleand the roof of the ethmoids on either or both sides (Fig.4). If the bone is thin, a fine osteotome is used. On theother hand, for thicker bone, the Hall microdrill or MidasRex knife is appropriate. After mobilization of the supe-rior surface of the specimen, the bone flap is temporarilyreplaced and secured loosely to the skull in order to pro-tect the brain during the facial resection.

The facial component of the operation begins with aWeber-Fergusson incision using either a Dieffenbach orLynch extension, depending on whether a medial, par-tial, or total maxillectomy, with or without orbital ex-enteration, is indicated (Fig. 1A). In cases where tumorextends into the antrum, a medial maxillectomy with ex-enteration of the nasal cavity and palatal fenestration, orpartial maxillectomy removing only the hard palate onthat side, is performed. The skin incision is deepenedfrom the facial soft tissues to the periosteum, and thecheek flap is elevated and retracted laterally with skinhooks (Fig. 5A). At the upper end of the incision, themedial canthal ligament is identif ied, detached, andmarked for reapproximation. at the end of the procedure.The infraorbital nerve is sectioned inferiorly, and the an-terior wall of the maxilla is exposed (Fig. 5, B and C). Thehard palate is preserved whenever possible, in order tospare the patient impairment in swallowing and speech.

The nasal septum, however, is frequently removed to pro-vide access to the ethmoid sinuses.

For tumors requiring palatal resection, a dissectionis made through the hard palate either at or just off themidline, in order to spare as many teeth as possible forfixation of a dental prosthesis. The incision is carriedthrough the midline of the hard palate and the ipsilateralsoft palate. Laterally, the osteotomy proceeds throughthe zygomatic arch. If the orbit is preserved, the medialorbital walls are carefully freed from the periorbita. Byreflecting the nose and the nasal septum laterally, it ispossible to remove the entire specimen with the ethmoidsinuses of the opposite side.

After the facial specimen has been freed, the neuro-surgeon rejoins the operation and removes the specimenen bloc from the facial and cranial sides. The resectedspecimen usually includes the tumor, the entire cribri-form plate, the superior and middle turbinates on one orboth sides, and occasionally the orbit. During the separa-tion of the specimen, difficulty arises at the point wherethe maxilla is attached to the pterygoid plates. Here thespecimen must be freed by blunt, and often blind, dissec-tion from the sphenopalatine fossa. In cases where anorbital exenteration is indicated, the optic nerve and theophthalmic artery are exposed at the orbital apex, theophthalmic artery is ligated and cut, and the optic nerveis sharply transected at the optic foramen.

After the specimen has been removed, hemostasis isestablished. Irregular bone edges are rongeured orsmoothed for reconstruction. All redundant mucosa iscarefully trimmed and removed to prevent formation ofpostoperative polyps.

Reconstruction of the skull base takes place in threestages. The first stage is the closure of the dura, eitherprimarily or using a free graft, either of temporalis fasciaor of pericranium. In the second stage, the pedicled peri-osteal flap is rotated, laid over the bony defect on thefloor of the anterior cranial fossa, and secured both to thebasal dura and the bony skull base (Fig. 6). In cases wherethe orbit has been exenterated, it is simpler to secure thepedicled flap to the basal dura and bone through theorbital defect. The last stage of reconstruction includesthe application of a split thickness skin graft to the exte-rior of the periosteal flap. The skin graft is applied fromthe facial aspect and meticulously positioned into thebony crevices and over the soft tissue surfaces. It is im-portant to achieve a good approximation of the graftagainst the pericranial flap. The graft is sutured peripher-ally to the undersurface of the skin flaps and held inposition with a snug packing of Xeroform gauze.

The craniotomy and facial incisions are closedusing standard techniques. However, before the bone

212


Figure 3. The dura is dissected off the orbital roofs and crista galli.

Figure 4. The frontal floor is separated along the planum sphenoidale and the roof of the ethmoid aircells; the use of an osteotome or microdrill is appropriate.



213


Figure 5. A, the facial skin is incised and the cheek flap is elevatedand retracted with skin hooks. B and C, the medial canthal liga-

ment and infraorbital nerve are identified and transected. Amaxillectomy is performed.


214


Figure 6. Reconstruction of the anterior skull base. After closureof the dura, the pedicled periosteal flap is applied and secured to thebony margins of the skull defect and to the basal dura. A splitthickness skin graft is applied from the facial aspect to themaxillectomy defect and the undersurface of the periosteal flap.

flap is replaced, CSF is reintroduced through the lumbardrains to expand the brain and fill the subarachnoid spaces.Bone grafting may be necessary for cosmetic reasons, es-pecially to fill defects in the glabellar region (Fig. 7, A-C).After the bone flap is replaced, the epidural space is mini-mized with tenting dural sutures. An epidural suction drainsuch as a Jackson-Pratt drain is routinely used and thescalp is closed in two conventional layers.

The spinal needles are removed. The patient is trans-ferred to an intensive care unit, where vital signs and neu-rologic status are monitored for 24 hours. Intravenousbroad-spectrum antibiotics, anticonvulsants, and steroidsare routinely prescribed in the postoperative period. Thedrains are kept in place for 2-3 days, and the antral packingis kept in place for 510 days, during which time the patientremains on antibiotics. When the packing is removed, thedefect is either repacked loosely or covered with an in-terim dental prosthesis. The skin grafted area is irrigatedon a daily basis after the packing is removed, and, if neces-sary, necrotic shreds of the skin graft are debrided to expe-dite epithelialization of the exposed surfaces.

For the first three days after surgery, the patient iskept in the recumbent position, in order to avoid theegress of CSF or ingress of air, which would cause tensionpneumocephalus. Fluids should be restricted to preventbrain edema.

COMPLICATIONSCombined craniofacial resection is a fairly extensive pro-cedure, in which two distinct compartments, the intradu-ral-subarachnoid space and the paranasal sinuses, areentered into and made contiguous. Given these circum-stances, the rate of complications associated with the pro-cedure itself is quite low, and may reasonably be consid-

ered acceptable. Careful patient selection, preoperativeplanning, and meticulous adherence to basic surgical andneurosurgical techniques all reduce the associated risks.The reported mortality, including that observed in ourpatients, ranges from 3 to 5%. Morbidity, including mi-nor and major complications, ranges from 15 to 20%.

Complications regarded as minor include localseromas, sloughing of the skin graft, and pneumocephalus.Major complications include cerebral edema, contusion,or hemorrhage, infection, meningitis, tensionpneumocephalus, and CSF fistula. Although little can bedone to prevent minor complications, these usually re-solve spontaneously.

To prevent the more serious complications, severalimportant measures can be taken. Prevention of brainedema and contusion is of the utmost importance. Work-ing lumbar drains to vent CSF, and comprehensive use ofhyperventilation and diuretics (mannitol and Lasix)throughout the long procedure are paramount. CSF fistu-lae can be prevented by achieving watertight dural clo-sure and skull base repair. It is recommended that duralopenings be addressed as they occur, and closed eitherprimarily with 4-0 Nurolon sutures or by tissue grafting.If a CSF fistula does occur, we recommend placing thepatient in the supine position with one or two indwellinglumbar drains, and reducing the intracranial pressure fora few days. In most cases, the CSF fistula will resolvespontaneously. However, CSF leaks lasting a few days arenot uncommon. If a leak persists, the insertion of an ind-welling lumbar drain for continuous CSF venting is indi-cated. The need to reexplore the wound to repair a CSFleak is exceedingly rare.

Tension pneumocephalus maybe a life-threateningcondition unless it is promptly addressed. Usually it isdue to enlargement of a trapped pocket of intracranial airfrom the time of surgery, compounded by a CSF leak.Meticulous dural repair, reexpansion of the brain at theend of the operation, placement of subdural and/or epi-dural drains intraoperatively, and maintenance of the re-cumbent position for at least two days postoperativelyall reduce the risk of tension pneumocephalus. If it doesoccur, however, drainage must be promptly begun, to ventair and decrease the intracranial pressure. After verifica-tion by computed tomography scan or skull films thatthere is no underlying brain parenchyma, a polyethylenevenous catheter placed through one of the burr holesprovides effective drainage.

Infection may become a life-threatening situation,whether in the form of meningitis, infection of the boneflap, or empyema. Preventive measures are essential.Prior to surgery, nasal cultures should be taken and an-tibiotics begun; antibiotics should be given continu-


215


Figure 7. A, in this case, tumor involved the frontal sinus, frontalbone, and medial aspect of the orbit. The involved bone has beenremoved to expose the nasal cavity and the orbital adipose tissue.B, to reconstruct the skull defect in the glabellar region, a splitthickness bone graft has been harvested from the right posteriorfrontoparietal area and has been carved to provide for a cosmeti-cally acceptable repair. C, the view of the cranium prior to closure.The bone graft harvested from the right posterior frontoparietalarea is in position at the glabellar region and has been fixed in placewith Synthes miniplates and screws. A Jackson-Pratt drain hasbeen left in the epidural space.


216


Figure 9. A and B, coronal and sagittal magnetic resonance images after resection. Note the extent of the resection and the reconstruction.

Figure 8. A and B, a typical case for craniofacial resection. Coro-nal and sagittal magnetic resonance images demonstrate a paranasal

sinus tumor eroding the anterior skull base; note the absence ofintracranial tumor and edema.



217


ously until the nasal packing is removed. Any rise in thepatient’s temperature should be regarded as serious. Ifpyrexia develops, meningitis should be considered im-mediately and ruled out by lumbar puncture.

RESULTSThe prognosis for patients who undergo combined cran-iofacial resections is generally favorable (Figs. 8 and 9).The overall median survival in our series is five years.Survival time depends largely on the histological typeand grade of the tumor, and on the presence or absence ofintracranial involvement. Excellent local control can beachieved in well-differentiated adenocarcinoma and squa-mous cell carcinoma. However, successful control is lesslikely with anaplastic epithelial tumors and sarcomas.Although intracranial extension adversely affects prog-nosis, limited intracranial invasion is not necessarily asurgical contraindication. In patients with neurologicdysfunction for whom the prospect of local control bychemotherapy or radiation therapy is poor, palliation maybe sufficient reason for surgery, since the morbidity ofthe procedure itself is fairly low.

Radiation therapy and chemotherapy are important

adjuncts to surgery. Even for patients with well-differen-tiated tumors that have been removed with clear margins,we advocate postoperative radiation and perhaps che-motherapy. At this stage, control of potential microscopicresidual disease may depend on these treatments. In tu-mors that are of higher grade, have less clearly definedmargins, or invade the intracranial space, we often usepreoperative radiation and chemotherapy to shrink thetumor first to make surgical resection possible.

SUMMARYTumors involving the base of the skull can be resectedsuccessfully using a combined craniofacial approach,with minimum morbidity and acceptable mortality. Un-favorable prognostic factors include intracranial exten-sion, high-grade histology of the tumor, and previoustreatment failure. With the sophisticated diagnostic tech-niques now available, particularly brain imaging meth-ods, tumor margins can be defined preoperatively withincreasing clarity. As methods of adjuvant therapy be-come more effective, craniofacial procedures will be ap-propriate in a wider variety of cases, and their benefitswill increase accordingly.

218

DIAGNOSTIC OPEN BRAIN ANDMENINGEAL BIOPSY

RICHARD P. ANDERSON, M.D.HOWARD H. KAUFMAN, M.D.SYDNEY S. SCHOCHET, M.D.

INTRODUCTIONThere are circumstances in which patients with obviousbrain disease may require specific diagnosis, and thiscan only be accomplished through the evaluation of braintissue. This occurs in two circumstances:

1. Chronic bilateral cerebral symptoms and signs asso-ciated with progressive dementia and/or profound be-havioral change; or

2. Unilateral, often localizable findings which are fre-quently associated with other signs of meningoen-cephalitis.

In addition, biopsies should be performed only whenthe results of these studies have the potential for modify-ing therapy or are useful in genetic counseling. A list isprovided below to give a clearer understanding of thosedisease categories in which biopsy is useful and/ or nec-essary in arriving at a final diagnosis.

InfectionsBacterial cerebritis (Whipple’s disease, nocardiosis,

mycobacterial infection)Fungal cerebritis (aspergillosis, mucormycosis)Parasitic disease (toxoplasmosis)Viral encephalitisChronic leptomeningitis

Collagen Vascular DisordersSarcoidosisGranulomatous angtitisVarious vasculitidesVarious connective tissue diseases

Dementias of ChildhoodAlexander’s diseaseCanavan’s diseaseLafora’s diseaseNeuroaxonal dystrophyCeroid lipofuscinosis-rarely indicated

Dementias of AdulthoodPick’s diseaseCreutzfeldt-Jakob diseaseAlzheimer’s disease-rarely indicated

NeoplasmsPrimary and metastatic

Acute problems requiring brain biopsy would includecertain cases of encephalitis and suspected vasculitis. Opin-ions vary as to the need for biopsy in the diagnosis ofherpes simplex encephalitis now that relatively less toxicdrugs such as acyclovir are available. Nevertheless, occa-sional biopsy specimens from patients clinically diagnosedas having herpes simplex encephalitis have disclosed otherdisease processes such as fungal infections that requireddifferent therapy. Granulomatous or isolated angiitis canbe diagnosed only from cerebral biopsy specimens. Theprognoses for these disorders have been improved mark-edly by the use of immunosuppressive therapy.

Biopsy may be appropriate in certain subacute andchronic diseases. These include patients suspected of hav-ing chronic bacterial encephalitis (Whipple’s disease andpossibly Lyme disease), or chronic meningitis, in whomthe diagnosis cannot be established from the cerebrospi-nal fluid. In rare instances, biopsies maybe appropriate inchronically ill patients if the results are of importance inproviding accurate genetic counseling. Very rarely, biop-sies maybe justifiable as a source of tissue for research ondiseases of unknown etiology and pathogenesis.

Another category of brain biopsy that has receivedless attention is the sampling of additional tissue at thetime of a therapeutic procedure. This would include re-moval of brain tissue or meninges at the time of ventricu-lostomy, cerebrospinal fluid shunting, or evacuation of ahematoma. These tissues should be handled with the samecare as other diagnostic biopsy specimens because muchvaluable information may be obtained from these speci-mens. For example, cortex obtained when a hematoma isevacuated should be examined by techniques appropri-ate for the detection of amyloid angiopathy, arteriovenousmalformation, or neoplasm.© 1991 The American Association of Neurological Surgeons

219

ANDERSON ET AL : DIAGNOSTIC OPEN BRAIN AND MENINGEAL BIOPSY

The initial step in the performance of brain biopsy isdiscussion among the individuals requesting the biopsy,performing the biopsy, and examining the biopsy speci-mens. This discussion will help in selecting the most ap-propriate biopsy site and utilizing these tissues optimally.

The timing of biopsy should also be considered. Whenpossible, the biopsy should be performed during a work-ing day to ensure that the technical staffs in the variouslaboratories are available and are thoroughly familiar withthe specialized diagnostic procedures that may be neededfor optimal utilization of the tissue. Even in the case ofpatients suspected of having herpes simplex encephalitis,antiviral therapy can be initiated and the biopsy performedthe next day when proper laboratory support has beenmobilized, as this will not impair the test results. Flexibil-ity tailored to the individual case with communicationamong all is the most important factor.

SURGICAL TECHNIQUE

Anesthesia and MedicationGeneral endotracheal anesthesia is used most commonly,but if the patient is cooperative, biopsies may be donewith light sedation and local anesthesia. If increased in-tracranial pressure is a factor, mannitol and hyperventila-tion may be utilized. Prophylactic anticonvulsants toavoid a known risk of seizures should be given preopera-tively, but despite this, seizures are a potential complica-tion. Prophylactic antibiotics are used, but if an infec-tious process is suspected, should not be given until after

the specimen is taken. As previously mentioned, in casesof herpes simplex encephalitis, pretreatment withacyclovir will not alter the findings significantly for sev-eral days to weeks.

Operative TechniqueFor a temporal lobe biopsy, the patient is positioned su-pine on the operating table, and, when it is to be used,general endotracheal anesthesia is induced. With a dough-nut pad under the head, the head is turned to the side at alevel above the heart, and a towel roll is placed under theshoulder. We flex the patient slightly at the hips and knees(Fig. 1). A small area of scalp is shaved, and the operativesite is prepared. A 5-cm curvilinear incision is drawn 1 cmanterior to the external auditory meatus which extendsup from the zygoma and turns posterior above the ear toapproximately 1 cm above the pinna (Fig. 2). The regionis draped with paper drapes and a Betadine-impregnatedadhesive plastic sheet.

The proposed incision is infiltrated locally with 0.5to 1.0% lidocaine with 1/100,000 epinephrine. It takesfrom 5 to 15 minutes to achieve maximum vasoconstric-tion with epinephrine. The incision is then made with aNo. 10 blade through skin, fascia, and muscle down toperiosteum. The bone is cleared with a periosteal eleva-tor. Retraction is maintained with a Weitlaner retractor. AHudson brace and McKenzie perforator are employed tocreate a burr hole. (In an awake patient, power drills areunnecessarily loud, potentially frightening, and ratherexpensive.)

Figure 1. With the patient supine, a towel roll is used to positionthe body and a doughnut pad to support the head and protect theear. Always keep the head above the heart and position with slightflexion at hips and knees.


220

Figure 2. The incision is 1 cm anterior to the external auditory meatus from the zygoma to 1 cm above thepinna and about 5 cm long.


Using rongeurs, the craniectomy can be enlarged toa 2-2.5 cm diameter (Fig. 3). Cautery is subsequentlyavoided until the tissue has been removed. A dural hookis used to lift the dura. and a No. 11 blade to make eithera cruciate or a curvilinear incision. A sample of meningesmay be removed if needed. Sufficient brain tissue shouldbe obtained so that the appropriate studies can be under-taken. Generally, this can be done from a cube measuring1.0-1.5 cm in each dimension. Specimens should includecortex and white matter, and, whenever possible, a sul-cus. Using a No. 11 blade, a four-sided incision is made1.0-1. 5 cm on a side, and at least 1.5 cm deep (Fig. 3,inset). This is undermined with a Penfield No. 3 dissectorand lifted out with a cupped forceps.

The tissue for histopathologic studies and electronmicroscopy must be handled as gently as possible. Evenwhen manipulated with great care, neurons at the pialsurface and near the margins of resection will be shrunkenand angular (Fig. 4). Conversely, neurons deep in theinterior of a large specimen will often become abnor-mally swollen. When they are mild, handling artifactscomplicate but do not preclude accurate diagnosis. Vacu-olar artifacts from improper handling (and suboptimaltissue processing) are especially troublesome whenCreutzfeldt-Jakob disease is being considered. The mor-phologic diagnosis of this condition is based in part onthe recognition of subtle intracytoplasmic vacuoles thatmust be distinguished from artifactual changes.

When viral and other cultures are indicated, tissue forthese studies should be placed immediately in sterile con-tainers and sent directly to the appropriate laboratoriesfrom the operating room so as to avoid the risk of contami-nation. Although isolation of the etiologic agent remainsas the most definitive procedure for infectious disease, theuse of immunofluorescent and immunoperoxidase tech-niques, as well as in situ hybridization has extended therole of the morphologist in the diagnosis of these condi-tions. Some hint of the diagnosis should be derived fromthe clinical data or from gross morphologic features so thatonly necessary and appropriate stains are used.

After the specimen is removed, electrocautery andGelfoam are used to achieve hemostasis. It may be diffi-cult to close the dura; indeed, if a dural specimen is taken,this may be impossible. The wound is then irrigated andthe layers above the dura and bone are closed tightly. Arunning locked skin suture extending just beyond theends of the incision will ensure a watertight closure.

Specimens from other sites may be desirable, depend-ing on such things as disease process, imaging character-istics, and electroencephalographic localization. Biopsyof silent areas should be carried out using appropriateskin and bone flaps. in small peripheral lesions, com-puted tomography localization and intraoperative ultra-sonography are very useful to minimize the size of theincision and the craniectomy and the amount of brainmanipulation.


221

ANDERSON ET AL : DIAGNOSTIC OPEN BRAIN AND MENINGEAL BIOPSY

Figure 3. A 2.0-2.5-cm diameter craniectomy is performed andthe dura opened to expose the lower aspect of the temporal lobe.Inset, remove a cube 1.0-1.5 cm on a side and 1.0-1.5 cm deep. If

possible, include a sulcus. Undermine with a No. 3 Penfield dissec-tor and remove with cupped forceps.


222


Figure 4. Artifactual changes include shrunken angular neurons withpyknotic nuclei seen at the margin of a biopsy specimen. H & E × 650.

Lastly, highly infectious agents carry risks to thesurgical team and those who come in contact with theequipment. Extreme caution with complete body cover-age of team members and care to protect all team mem-bers from “sharps” should be followed. Since every op-erative case is potentially infectious, universal bloodprecautions are employed as standard practice. In patientssuspected of having a highly contagious disease such asAIDS or Creutzfeldt-Jakob disease, additional precau-tions may be taken. All unnecessary equipment is re-moved from the room. Disposables are used if possible,cabinet and room entry are limited, and an outside circu-lator should be present to allow persons in the room toremain within the room. For slow viruses, special clean-ing of nondisposables with steam autoclaving at 132°Cfor one hour or soaking in 1 N sodium hydroxide for onehour is recommended.


223

VENTRICULOPERITONEAL SHUNTINGDAVID C. McCULLOUGH, M.D.

INTRODUCTIONVentriculoperitoneal shunting is the preferred methodfor cerebrospinal fluid (CSF) diversion in hydrocephalusof children and adults. The technique was conceived andfirst attempted early in the 20th century and was revivedin the 1950s, with indifferent results. Introduction of sili-cone conduits with elastic properties eventually contrib-uted to clinical success. After 1970, ventriculoperitonealshunting supplanted various other diversionary methodsincluding ventriculoatrial shunting.

SELECTION OF PATIENTSBecause the majority of cases of hydrocephalus are notrelieved by direct surgical intervention, CSF diversionwith implantable devices has become the conventionaltherapy for this condition. The common use of ultrasound,computed tomography (CT). and magnetic resonanceimaging (MRI) has disclosed increasing numbers of po-tential candidates for shunt procedures. Some surgeonsrecommend that any patient with large ventricles shouldbe treated, but most conservatively advocate shuntingonly when there is a potentially reversible deficit or aprogressive condition. Infantile congenital hydroceph-alus is almost always progressive. Patients with acquirednoncommunicating hydrocephalus and symptomatic ac-quired communicating hydrocephalus will usually re-quire intervention.

The determination of progressive hydrocephalus maybe difficult in other circumstances. Serial developmentaland neurological examination in infants, neuropsycho-logical studies in older children and adults, or radio-graphic evidence of progressive ventriculomegaly canaffirm the need for CSF shunting. If a block is correctableby definitive surgery, or, if there is uncertainty as to theprogressive nature of hydrocephalus, shunt treatmentshould be deferred.

Active ventriculitis, fresh ventricular hemorrhage,or peritoneal or systemic infection present specificcontraindications to early shunting. Spontaneous com-pensation of the hydrocephalic condition may follow the

removal of obstructive neoplasms and certain posteriorfossa cysts. Posthemorrhagic hydrocephalus followingintraventricular hemorrhage in neonates is often evanes-cent. Temporizing measures such as serial lumbar punc-tures or the administration of osmotic diuretics may suf-fice for communicating hydrocephalus in prematureinfants.

Occasionally, radionuclide or contrast medium ven-triculography and cisternography assist in the detectionof progressive hydrocephalus. Abnormal pressure wavesduring continuous intracranial pressure monitoring sup-port the diagnosis of a progressive condition in olderchildren and adults.

Permanent CSF bypass is seldom an emergency pro-cedure. However, in the cases of acute acquired hydro-cephalus or a rapidly decompensating congenital condi-tion with increased intracranial pressure, delay may bedetrimental. When neurological and radiographic dataraise questions of fresh hemorrhage, active infection, orthe possibility of direct surgical cure (e.g., posterior fossatumor or cyst), temporary ventriculostomy may stabilizethe situation and permit careful preoperative investiga-tion and planning.

PREOPERATIVE PREPARATIONA single ventricular tap is recommended before surgeryto exclude active infection and examine CSF contents inposthemorrhagic or postinfectious hydrocephalus of in-fancy. Although shunts may function satisfactorily in theface of extremely high CSF protein levels, it is prudent todelay until the protein concentration is below 200 mg %.

Selection of ComponentsSimplicity should prevail in the selection of devices forimplantation in humans. I favor a specific system thateliminates some potential complications and makes oth-ers easier to manage. The flanged ventricular catheterallows for a certain margin of error in placement of theproximal tip of the device. Furthermore, it seldom be-comes entangled with choroid plexus and, therefore,produces a lower incidence of bleeding when extrac-tion is necessary. Spring-wire reinforcement of the ven-tricular catheter wall eliminates the need for insertionof a flange to ensure patency as the catheter© 1991 The American Association of Neurological Surgeons

224


angles sharply downward from its dural insertion site.The reservoir valve device permits percutaneous needleaccess to CSF. The diaphragm valve incorporated in thedistal part of the reservoir is quite predictable in its per-formance. It provides unidirectional flow for functionalevaluation on rare occasions when this is needed. A singleproximal valve is used so that the surgeon can expose thesystem, survey flow from all components, and, if neces-sary, change the resistance with one incision. A flat-bot-tom device secured to the periosteum inferior to the burrhole eliminates the annoying possibility of disconnec-tion between the reservoir and the (then inaccessible)ventricular catheter upon extraction of a burr hole reser-voir during a revision. The open-ended peritoneal cath-eter with three series of side slits avoids an additionalresistance in the system while allowing CSF egress fromproximal apertures if the distal end of the catheter be-comes occluded.

SURGICAL TECHNIQUEThe procedure is performed under general endotrachealanesthesia after intravenous thiopental induction. Spe-cial warming lights and airway humidification with warm-ing counteract hypothermia in small infants. Prophylac-tic vancomycin (10 mg/kg) is administered as soon as theintravenous line is inserted in the operating room.

Correct patient positioning is critical to the smoothimplantation of the shunt. The patient is placed at the topedge of the operating table to the operator’s side and thehead is turned sharply to the side opposite the insertion.The head is elevated for occipital access and the neckand trunk are extended with supporting cushions underthe shoulders to facilitate subcutaneous tunneling (Fig.1A). A meticulous 10-minute sequential skin preparationwith povidone-iodine scrub and paint solution is thenperformed. Proposed incisions are outlined with a surgi-cal marking pen. After palpating to exclude hepatome-galy, an abdominal incision is marked over the right up-per quadrant. Surgical drapes include an adherent plasticthat is impregnated with povidone-iodine over incisionand tunneling sites (Fig. 1B), excluding the anterior fon-tanelle if ultrasound is to be used to guide the insertionof the ventricular catheter.

When inserting the shunt from the occipital site (Fig.1), judgment of the appropriate distance from the skullsurface to a point in the anterior horn just in front of theforamen of Monro is guided by measurement from preop-erative CT scans. In older children and adults who tendto have smaller occipital horns, a frontal burr hole is pre-ferred. Then the length of the ventricular catheter is de-termined by measurement of the distance from the pro-posed burr hole at the skull surface in a straight line to a

point just superior to the midpoint of the bregma-interaural line (locus of the foramen of Monro).

The shunt is inserted from the right side unless theleft lateral ventricle is significantly larger (Fig. 2). Acurved incision with its long arm vertical is made 2.54cm superior to the inion and 2.5-3.5 cm lateral to theoccipital midline. If a frontal burr hole is used, the inci-sion site is slightly anterior to the coronal suture and 2.5-3.0 cm lateral to the midline. A small scalp flap is turned.Clips or hemostats are used at the wound edges and careis taken to avoid violation of the periosteum during theinitial incision. The scalp flap may be secured with a self-retaining retractor. A small cruciate periosteal incision isperformed just inferior to the scalp incision line and aburr hole is drilled. The dura is coagulated with bipolarforceps and a 2-mm round dural opening is producedusing a hook and a No. 11 knife blade. The arachnoidlayer is lightly coagulated and knicked.

Shunt components are removed from sterile pack-ages only immediately prior to placement. Skin incisioncontact with shunt parts is avoided and the gloves of thesurgical team are changed and washed prior to handlingthe shunt components. The ventricular catheter is intro-duced with a wire stylet toward the inner canthus on aline directly between the horizontal planes of the su-praorbital ridge and the bregma (Fig. 2). I prefer a flangedcatheter which is situated with its tip in the anterior hornto avoid choroidal attachment. Intraoperative ultrasound(using the anterior fontanelle as an ultrasonic window) isuseful for direct guidance in infants. Alternatively the tipis advanced to the premeasured point 1 cm anterior to theforamen of Monro (or 1 cm above the foramen on frontalplacement) as determined from the CT scan. After inser-tion to the appropriate length, CSF flow is checked andthe distal end of the catheter is immediately clamped tothe drapes.

A 4-5-mm abdominal incision is made in the rightupper quadrant passing through the subcuticular tissue.A peritoneal trocar is passed posteriorly and medially inthe direction of the umbilicus as the anesthesiologist in-duces a Valsalva effect to tighten the abdominal muscles(Fig. 3). After the stylet is removed 30-60 cm of an open-ended peritoneal catheter (with three sets of side slits) isintroduced through the trocar sleeve (Fig. 4) and thesleeve is removed.

A semisharp stainless steel tunneling device maybepassed through the abdominal incision from the scalpincision subcutaneously (Fig. 5). In older patients atiny cervical incision is often performed. Then, the tun-neling device is passed from that level taking care totunnel anterior to the clavicle. The proximal end of theperitoneal catheter is backed through the sub-

225

McCULLOUGH : VENTRICULOPERITONEAL SHUNTING

Figure 1. A, positioning of an infant for ventriculoperitoneal shunting. B, the prepared and draped surgical field with marks at incision sites.


226


Figure 2. Insertion of a ventricular catheter from a right posterior entry site.

Figure 3. Introduction of a peritonea trocar at a right upper quadrant incision site.



227


Figure 4. Introduction of a peritoneal catheter via the trocar sleeve.

Figure 5. A tunneling device introduced from the scalp inci-sion site to the abdominal incision site. The proximal end of

the peritoneal catheter is threaded through the slot in thetunneler tip.



228


cutaneous tunnel. After the subgaleal space under the scalpflap is dissected to accommodate a proximal reservoir, thetunneling device is used to bring the proximal end of theperitoneal catheter from the cervical incision to the scalp(Fig. 5). Excess tubing is trimmed from the proximal end ofthe catheter. The catheter is then attached to a proximalflow control reservoir valve (Fig. 8). Low resistance reser-voirs with diaphragm valves are used for infants less thanone year of age. Larger or standard size medium pressureflow control reservoir valves are used for older infants,children, and adults. The reservoir and distal catheter areinjected with a bacitracin irrigating solution (50,000 unitsin 200 ml of injection saline). Then, the distal end of theventricular catheter is trimmed to its appropriate lengthand attached to the proximal end of the reservoir valve.Silk ties (2-0) are firmly teased into the connector groovesand tied tightly. The reservoir base is sutured to the perios-teum with 4-0 silk. Occasional leakage around the exitpoint at the dura may require the application of Gelfoamwhich is secured by suturing the periosteum over the burrhole. After the reservoir is secured to the periosteum anyexcess loop of tubing at the peritoneal incision site is“stuffed” in (Fig. 6), making certain there is no subcutane-ous coil (Fig. 7). Synthetic absorbable sutures are used toclose the periosteum at the burr hole as well as the galea.Nonreactive synthetic sutures are used to close the skinand a compression scalp dressing is applied to discourageleaking until ventriculoperitoneal flow is well established.The cervical and abdominal incisions are closed in a simi-lar fashion.

POSTOPERATIVE CAREPostoperatively the patient is maintained in a horizontalposition for 12 to 24 hours. Vancomycin prophylaxis (10mg/kg, every 6 hours) is continued for 36 hours. Gradu-ally, the head is elevated depending on the condition ofthe anterior fontanelle or the tolerance of the older pa-tient. Alimentation is usually initiated within 12 hoursunless there is abdominal distension and hypoperistalsis.Patients maybe discharged from the hospital on the sec-ond postoperative day.

COMPLICATIONS AND PROGNOSISIn spite of potentially frequent and varied complications,mortality and morbidity can be surprisingly low if a me-ticulous follow-up system is provided. This includes regu-larly scheduled office visits, periodic CT scanning, andconvenient access to a neurosurgical service for symp-tomatic patients. Ventriculoperitoneal shunts are not elec-tively lengthened for axial growth when parents are ob-servant and a responsive care system is available.

Early postoperative complications include obstruc-

tion due to malposition of the device or inaccurate selec-tion of components with excessive resistance in the sys-tem. Older, taller patients often require very gradual mo-bilization because of postural head pains related tointracranial hypotension.

Immediate infection is rare, but most septic incidentsoccur within the first two months of insertion or revision ofa shunt. They usually manifest with obstructive symptoms,fever, cellulitis, and leukocytosis. Blood cultures are oftenpositive. Most organisms can be cultured from percutane-ous reservoir taps. Staphylococcal organisms predominate.In centers where shunts are frequently inserted in children,procedural infections rates approach 2%. The most reli-able therapy is complete shunt removal with temporaryexternal drainage and appropriate systemic antibiotics. Anew device can be inserted after obtaining at least twoconsecutive negative cultures 48 hours apart.

Injury to an abdominal organ occurs in 0.4% of pa-tients. This often presents as an infectious-obstructiveepisode, but extrusion of the catheter via the intestinaltract has been observed.

Although older patients with very large ventriclesare at risk for overdrainage leading to subdural effusionsand hematomas, less than 5% of pediatric patients de-velop this complication. All asymptomatic subdural ef-fusion should be monitored clinically and with serialimaging. Most will resolve spontaneously.

Obstructive complications will inevitably occur. Themajority involve the distal portion of the device. Patientswith high CSF protein values at the initiation of therapymay require several shunt revisions during the first post-operative year. After about the sixth year, patients treatedin infancy may present with obstruction due to axial so-matic growth and relative shortening of the peritonealcatheter. Obstruction may also result from rarely encoun-tered peritoneal pseudocysts, ascites due to peritonealabsorptive failure, peritoneal adhesions, internal debris,or breakage and disconnection of shunt components. Theuse of modern kink-resistant silicone tubing with avoid-ance of older, delicate, spring-reinforced peritoneal cath-eters has virtually ended the occurrence of disconnec-tion and catheter migration.

A small cohort of pediatric and adolescent pa-tients with CSF diversion receives considerable at-tention because of frequent bouts of vomiting, head-ache, and lapses of consciousness. The patients tendto have small ventricles, diminished extraventricularCSF spaces, and a thick skull. In my experience lessthan 4% of treated patients were suspected of havingthis “syndrome.” The majority of these actually hadtrue proximal shunt obstruction solved by appropri-ate catheter replacement in the frontal horn. In the re-

229


Figure 6. Stuffing the excess peritoneal catheter via a small abdominal incision.

Figure 7. A, the correct positioning of the peritoneal catheterafter stuffing. B, an incorrect position of the peritoneal catheter

which is coiled in an extraperitoneal location after attemptedstuffing.



230


Figure 8. A properly inserted ventriculoperitoneal shunt in aninfant. Note the anterior position of the proximal tip of the ven-tricular catheter, the location of the proximal reservoir valve

under the skin flap, and the generous coil of open-end peritonealcatheter within the abdomen.

mainder, corrective procedures such as augmenting valveresistance and subtemporal craniectomy may be requiredto manage symptoms and prevent decompensation.

Current mortality and morbidity data reveal that95% of children with nontumoral hydrocephalus sur-vive for over 10 years and about 65% are intellectu-ally normal. For infants with overt hydrocephalus at

birth, the outcome is less favorable with about halfdemonstrating normal intelligence. Shunt-independentarrest is seldom observed. In patients followed an aver-age of 15 years, 75% have required at least one revi-sion and several have required as many as eight repeatoperations. The mean number of revisions has beentwo procedures per patient.


231

VENTRICULOATRIAL SHUNTINGPAUL J. CAMARATA, M.D.STEPHEN J. HAINES, M.D.

INTRODUCTIONThe ventriculoatrial (VA) shunt was introduced as amethod of treatment for hydrocephalus in the 1950s andwas the first predictably successful valve-regulated cere-brospinal fluid (CSF) shunt. Technical developments havesince made ventriculoperitoneal shunts reliable and, be-cause they are easier and quicker to insert and revise, andhave less propensity to systemic bacteremia if infected,they are the initial treatment of choice for hydroceph-alus. for most patients. There are some patients in whom aVA shunt is the procedure of first choice, such as thosepatients with fibrosis or inflammation in the peritoneumfrom remote or recent infections or multiple previousabdominal operations. The peritoneal cavity of some pa-tients, particularly small infants, may occasionally nothave sufficient absorptive capacity to handle the neces-sary amount of cerebrospinal fluid.

INDICATIONSVentriculoatrial shunts are indicated for the treatment ofhydrocephalus, either obstructive or communicating,which is not transient in nature. Other indications forshunting, which may occasionally require VA shunting,include the treatment of pseudotumor cerebri and drain-age of arachnoid cysts and subdural hygromas unrespon-sive to other therapeutic measures.

CONTRAINDICATIONSBacteremia or infection of the CSF or proposed shunt tractare absolute contraindications to the placement of a VAshunt. In the presence of infection elsewhere in the body, ashunt should be inserted only in very unusual circum-stances. Congestive heart failure and pulmonary hyper-tension may both interfere with shunt function and be ag-gravated by the additional fluid load delivered to the heart;these are relative contraindications to the procedure, espe-cially in infants. Abnormal venous anatomy and previous

jugular or subclavian vein thrombosis are relativecontraindications. Where both VP and VA shunting arecontraindicated, shunts to the pleural space, the gall blad-der, the ureter, the bone marrow, the subarachnoid space,and other areas have been reported to be successful.

PREOPERATIVE PREPARATIONPreoperatively the patient or his or her parents are in-formed that the major risks of the procedure are those ofinfection and shunt malfunction, either of which wouldnecessitate revision or replacement of the shunt. When aVA shunt is placed in an infant, malfunction due to growth-related migration of the atrial catheter into the superiorvena cava is so predictable that elective revision at abouttwo years of age has been recommended by some. Thereis a slight risk of intracranial hemorrhage (which may beincreased in patients with marked hydrocephalus). Re-mote risks of air embolism, cardiac rupture and tampon-ade, and thromboembolism are mentioned, as are the at-tendant risks of general anesthesia.

Where possible, an antiseptic bath or shower is ad-ministered preoperatively. The hair should be shavedimmediately preoperatively. An appropriate dose of ananti-staphylococcal antibiotic is administered upon an-esthetic induction.

Anesthetic ConsiderationsMost patients with hydrocephalus can be presumed tohave some degree of increased intracranial pressure (ICP).Because of this a gentle anesthetic induction is preferred,being careful to avoid any manipulations that would in-crease ICP, i.e., Valsalva maneuvers, coughing, prolongedhypoventilation. Appropriate inhalational or intravenousanesthetics that decrease ICP and preserve cerebral auto-regulation are used, combined with hyperventilation ifdeemed necessary. Succinylcholine is avoided becauseof its propensity to increase ICP.

Special EquipmentA C-arm fluoroscopic unit is extremely helpful in veri-fying correct catheter placement. Operating room per-© 1991 The American Association of Neurological Surgeons

232


sonnel must remember to don lead aprons prior to scrub-bing. It is also useful to have intravenous contrast mate-rial available. Heparinized saline is necessary to flushthe atrial catheter, as well as an antibiotic saline solutionwith which to irrigate the shunt system and wounds. Theappropriate shunt system is chosen preoperatively, and acentral venous pressure monitoring device must be avail-able to monitor pressure waves as the atrial catheter isadvanced.

PositioningThe basic principle of positioning the patient is to pro-vide clear access to the head for ventricular puncture andto the neck for cannulation of the venous system. There-fore, following induction of satisfactory general endotra-cheal anesthesia, with the patient in the supine positionthe head is turned to the appropriate side. In the preferredsetting, the head is turned to the left to provide access tothe right neck. Because of the vascular anatomy, right-

sided cannulations are often easier than left-sided ones.The ear maybe taped forward and thereby easily drapedout of the field. Soft padding is placed beneath the shoul-ders to expose the anterior triangle of the neck (Fig. 1, Aand C). The skin in the operative field and surroundingarea is then prepared with an appropriate antiseptic soap.Prior to draping, the landmarks for ventricular access andaccess to the venous system are drawn on the skin (Fig.1B). A mark is placed approximately 2.5-3 cm from themidline and 11-12 cm posterior to the nasion in the adult(or approximately one-seventh of the distance from thecoronal suture to the nasion in the child). A line is thendrawn from this mark toward the inner canthus of theipsilateral eye. Another line is drawn that passes throughthis point and a spot 1 cm anterior to the tragus of theipsilateral ear. If these lines are taken to represent imagi-nary planes in partial sagittal and coronal directions, theirintersection forms a line that should pass through theforamen of Monro (Fig. 2).

Figure 1. Lateral (A) and vertex (B) views of incision land-marks. Note the intersecting lines drawn to represent twoplanes. The ventricular catheter should be passed along the

line that is the intersection of these two planes. C, side viewof padding placed under the head and shoulders for optimalpositioning.


233

CAMARATA AND HAINES : VENTRICULOATRIAL SHUNTING

Figure 2. A ventricular catheter traveling along the intersection of the two drawn planes will be directed at the foramen of Monro.


234


Sterile drapes are then applied in such a manner as toallow for access to head and neck sites. In draping thehead, it is helpful to place a sterile towel with one borderon the midline to use as a landmark in placing the burrhole. If intraoperative ultrasound is to be used in the caseof an infant, the anterior fontanelle should be draped intothe operative field.

The shunt hardware, including ventricular and atrialcatheters, valve system, reservoir, and connectors, shouldbe chosen before the procedure, opened, and placed in anantibiotic-saline solution on the instrument table beforethe incision is made. We avoid the use of bacitracin be-cause its foaming action may interfere with the functionof some valves. It is prudent to have several valves withdifferent pressures readily available. The valve system tobe used should be tested according to the manufacturer’srecommendations to ensure its appropriate function. Itshould be filled with saline solution, all air bubbles re-moved, and clamped on one end with a rubber shod mos-quito clamp to keep it full of fluid. To the greatest extentpossible, one should avoid touching the skin and shuntsystem with the gloved hand to minimize the risk of post-operative infection.

SURGICAL TECHNIQUEThe operation is carried out in a logical, stepwise fash-ion. We prefer to place the ventricular and vascular cath-eters as the last part of the operation because of the risk ofdislodging either catheter before it is secured in finalposition.

Step One—Isolation of the Common Facial VeinA 2-cm transverse neck incision is made one finger breadthbelow and parallel to the ramus of the mandible in anadult, and a proportionally smaller distance below thejaw in children. This should be centered just medial tothe medial border of the sternocleidomastoid muscle (Fig.3). The platysma is split in the direction of its fibers, andusing a combination of blunt and sharp dissection alongthe avascular plane medial to the sternocleidomastoidmuscle, the common facial vein is identified. The vein isisolated for about 1 cm of its length, and two ligatures ofa size appropriate to the size of the vessel are placedaround it. The vein is then ligated using the most ceph-alad tie.

Recently, we prefer to use a percutaneous techniquesimilar to that used for placing central venous catheters.Usually the internal jugular vein is used for access, thoughthe subclavian may be used, and the procedure is identi-cal to cannulation for standard central venous catheters.For the internal jugular vein, a thin-walled, 18- or 20-gauge needle which is contained in most central venous

access kits is used. The skin is punctured approximately4-6 cm (2-3 finger breadths) above the clavicle along theposterior border of the sternocleidomastoid muscle or aproportionate distance in a child. The tip of the needleshould be aimed at the sternal notch (Fig. 4). Alterna-tively, the carotid artery can be palpated and the needleaimed lateral to it. For the subclavian vein, the needle isinserted just below the junction of the middle and innerthirds of the clavicle and aimed at the sternal notch.

Once the vein is punctured, the syringe is removedand a flexible J-wire (0.021 inch in diameter for olderchildren and adults) is inserted through the needle; itsposition can be checked with fluoroscopy. The needle isremoved when the wire is verified to be in the superiorvena cava. A small 1-cm incision is made at the point ofentry to facilitate passage of the introducer and to con-nect the shunt system with the atrial catheter and bury it.A standard, tear-away introducer sheath over a vessel di-lator is then passed over the wire through the subcutane-ous tissue and into the vein. The wire and dilator areremoved and a finger is placed over the sheath to preventthe introduction of air. The atrial catheter can now bethreaded through the sheath to its proper position de-scribed in Step 4 below.

Step Two—Placement of a Cranial Burr HoleThe ventricular catheter may be placed into the frontal oroccipital horn of the lateral ventricle depending on thepreference of the surgeon and the individual patient’sventricular anatomy. We prefer the frontal approach be-cause our experience indicates that such shunts functionbetter than parieto-occipitally placed ones. The patient’snondominant hemisphere is selected because of the re-mote risk of parenchymal damage caused by hemorrhageat the site of insertion. A curved incision is made with itsbase directed inferiorly so that no portion of the incisionwill be overlying the shunt apparatus. The burr hole isthen made using a standard hand or power drill at thepredetermined site, bone wax is applied, and the exposeddura is covered with a saline-soaked sponge or cottonoidpatty until later in the procedure.

Step Three—Placement of the Valve SystemAt this point the valve system is passed subcutaneouslyfrom the cranial incision to the neck incision, passing infront of the parietal. boss and behind the ear. Care must betaken to avoid the very thin skin just behind the ear and notto have the bulky portion of the valve or reservoir over abony prominence where they may cause pressure erosion ofthe skin. Depending on the size of the patient, a small trans-verse incision may be necessary behind the ear so that thedistal tubing may be passed the remainder of the distance to

235


Figure 3. Complete ventriculoatrial shunt in place.


236


Figure 4. Guidelines for percutaneous placement of the atrial catheter.


237


the neck incision. We prefer using the Cordis-Hakim shuntsystem, which is easily passed in this manner using acurved passing device included with the system. Othershunt systems may require minor technical variations.The proximal shunt system tubing is then cut to the ap-propriate length and a right-angled connector is attachedusing a permanent suture. Both ends of the system areclamped with rubber shod clamps in order to maintainthe system fluid-filled and free of air bubbles, and toavoid inadvertently pulling the tubing out of the inci-sion, and are covered with sponges moistened in an anti-biotic solution. The purpose of placing the valve-reser-voir system f irst is to minimize the number ofmanipulations once the vein and the ventricle have beencatheterized. When the venous or ventricular catheter isplaced early in the procedure, it maybe dislodged inad-vertently. Because of this possibility, we then place theatrial catheter, and the ventricular catheter last.

Step Four-Placement of the Atrial CatheterThe goal in placement of the atrial catheter is to positionthe catheter midway between the superior vena cava(SVC)-right atrium junction and the tricuspid valve. Avariety of techniques are available to confirm this place-ment, the least reliable of which is the chest x-ray, uponwhich we do not rely. A more accurate, yet simple, way ofcorrectly placing the catheter utilizes continuous pres-sure wave recordings as the catheter is advanced.

A small venotomy is made in the common facial vein,and the atrial catheter is introduced through it into thejugular vein. Alternatively, the catheter may be intro-duced directly into the internal jugular vein. In this casea purse string suture should be placed prior to making thevenotomy. The catheter has been filled with an isotonicheparinized saline solution with the proximal end con-nected to a pressure transducer for continuous monitor-ing in a closed system, to prevent air aspiration. The liga-ture is cinched around the catheter just enough to preventback bleeding yet allow the catheter to advance easily. Acharacteristic change is noted as the catheter is advancedfirst from the superior vena cava into the right atrium andthen into the right ventricle (Fig. 5). The catheter is thenwithdrawn until the atrial pressure tracing is again identi-fied. Fluoroscopy is often used during advancement ofthe catheter. If necessary, the catheter may be made vis-ible by the injection of an appropriate quantity of con-trast material. When the proper position has been identi-fied, the remaining ligature is tied around the catheter sothat it cannot move, but still will permit free flow of CSF.

This should be ensured by aspirating to remove air bubblesand then flushing the distal catheter with heparinizedsaline. A rubber shod clamp is placed proximally, and astraight metal connector is tied to the catheter which isthen covered with an antibiotic moistened sponge.

Step Five-Placement of the Ventricular CatheterThe goal in placement of the ventricular catheter is toplace the tip in the frontal horn just anterior to the fora-men of Monro. In an infant, ultrasound through the ante-rior fontanelle can often be used to direct the catheterplacement in this fashion.

The center of the exposed dura is coagulated with aneedle-point monopolar cautery to create an opening witha diameter equal to that of the catheter. The catheter shouldbe marked in some fashion so that it is not advanced exces-sively (beyond 7 cm in the adult). A standard Silastic ven-tricular catheter with multiple side holes at its tip and aninternal stylet in place is then advanced through the duralopening utilizing the guidance lines and planes describedabove until CSF return is evident. (We do not use flangedventricular catheters because of the tendency of the chor-oid plexus to grow into the interstices of the catheter. Thiscan lead to intraventricular hemorrhage when the catheteris removed during a revision.) Usually the resistance en-countered when the catheter pops through the ventricularependymal surface is easily felt. The stylet is then removedand the catheter advanced approximately 1 cm. A depth of5-6 cm is usual in the adult and is proportionally less inchildren. A manometer may be connected to the catheter,and the ventricular pressure may be measured and recorded.The catheter is then clamped with a rubber shod clamp,cut, and connected to the valve system with the right-angled connector. The connector should then be securedto the periosteum with a single stitch.

The system is now inspected to ensure spontaneousflow of CSF from the valve, and the atrial catheter is againaspirated and flushed and connected to the valve systemwith a straight connector. The sutures are tied across theconnector, to further protect against disconnection.

The position of the atrial catheter is again checkedfluoroscopically. The tip of the catheter should be seenoverlying the seventh thoracic vertebra. All wounds areinspected and irrigated with antibiotic saline solutionand then closed in the standard fashion. The galea is firstclosed with interrupted 3-0 or 4-0 absorbable sutures,and the skin is closed according to the surgeon’s prefer-ence. The posterior auricular and cervical incisions areclosed in two layers, and sterile dressings are applied.

238


Figure 5. Typical pressure wave form readings seen with advancing catheter positions.


239


INTRAOPERATIVE PROBLEMS

Difficulty in Locating the VentriclesWe find the above method of passing the catheter mostsuccessful in cannulating the anterior part of the frontalhorn of the lateral ventricle. If unsuccessful, three passesare made at slightly different angles, at no time passingthe catheter to a depth greater than 7 cm in an adult.Passing the catheter perpendicular to a plane tangent tothe skull at the point of entry is often helpful. As men-tioned, ultrasound can be used in a child with an openanterior fontanelle.

Difficulty Passing the Atrial Catheter ina Patient with Normal Vascular AnatomyOnce the catheter enters the vein, it usually passes straightinto the superior vena cava and right atrium. If any diffi-culty is encountered, a standard J-wire commonly used forcentral vascular access can be passed into the vein andalmost always easily directed into the atrium under fluo-roscopy. The catheter can then be threaded over the wire.

Difficulty Passing the Atrial Catheter ina Patient with Abnormal Vascular AnatomyIf a problem with venous access is anticipated, such as apatient who has had multiple VA shunts, previous SVC orjugular thrombosis, or multiple previous central venousaccess lines, it is advisable to obtain preoperative venog-raphy to define the anatomy. Where the internal jugularhas been previously sacrificed or thrombosed, the exter-nal jugular may be of use. It is also possible with the aidof an interventional radiologist to catheterize a vein in aretrograde fashion from the groin using an angiographiccatheter. One may then cut down directly on the wire, andthis may be used to pull the catheter into the atrium.

Difficulty Positioning the Atrial CatheterWe find the method of using continuous pressure record-ings the easiest and most accurate way of placing the tipof the catheter correctly in the atrium. The use of contrastmaterial under intraoperative fluoroscopy is anothermethod. Or, using the saline-filled atrial catheter as anECG lead, the characteristic biphasic P-wave changes seenin the atrium may be identified. Two-dimensional intra-operative echocardiography has also been used success-fully for this purpose.

POSTOPERATIVE CAREAnteroposterior and lateral x-ray films of the skull andthorax should be obtained within the first two postop-erative days to verify catheter positions and continuityof the system. A computed tomographic scan should bedone within the first few weeks after surgery in the caseof a first-time shunt placement to document ventricularsize at a time when the shunt is known to be functional.This can be invaluable in assessing shunt function inthe future.

We monitor the electrocardiogram (ECG) for the first24 hours postoperatively because of the possibility ofarrhythmia. One may wish to nurse the patient in thesupine position for the first day if the intracranial pres-sure was particularly high to protect against collapse ofthe cerebral mantle.

COMPLICATIONSMalfunction of the shunt system and infection are thetwo most common complications of the procedure andmay occur at any time after operation, including in therecovery room. The former maybe a consequence of anindolent CSF infection, obstruction of any part of thevalve or tubing by proteinaceous debris or fibrosis, ordisconnection. We believe the technique of securing allconnections with a nonabsorbable suture tied across theconnector can help prevent disconnection.

Infection can be minimized by the use ofperioperative antibiotics, and by keeping the handlingof the shunt system to a minimum. When at all possible,the system should be manipulated with instruments ratherthan the gloved hand. When tapping the shunt reservoirthe skin should always be adequately prepared with theappropriate shave and antiseptic wash. Patients with aventriculoatrial shunt should be advised to follow stan-dard bacterial endocarditis antibiotic prophylaxis beforeany surgical or dental procedure.

CONCLUSIONSVentricular shunting procedures are commonly thoughtof as “minor” neurosurgical procedures. Given short shriftin training and low priority in practice, the operationmay give suboptimal results. However, with meticuloustechnique and skillful execution, the ventriculoatrialshunt can be a safe and effective tool in the neurosurgeon’sarmamentarium for the treatment of hydrocephalus.

240

EXCISION OF ACOUSTIC NEUROMASBY THE MIDDLE FOSSA APPROACH

DERALD E. BRACKMANN, M.D.

INTRODUCTIONThe middle fossa approach for the removal of acoustictumors was developed by Dr. William F. House in theearly 1960s. It has been shown to be a safe approach witha minimum of mortality and morbidity. In this chapter,the approach for removal of an acoustic tumor will bedescribed.

INDICATIONS AND PATIENT SELECTIONThe middle fossa approach offers several advantages forthe removal of small, laterally placed acoustic tumors.First, the majority of the dissection is extradural, therebylowering morbidity. Second, the lateral end of the inter-nal auditory canal is exposed which ensures removal ofall of the tumor. With the retrosigmoid approach, the mostlateral end of the internal auditory canal cannot be ex-posed safely without entering the labyrinth. Third, posi-tive identification of the facial nerve is possible at thelateral end of the internal auditory canal. This facilitatestumor dissection from the facial nerve in this area.

For middle fossa acoustic tumor removal, we selectpatients who have a tumor that extends no further than Icm into the cerebellopontine angle. If the tumor is medi-ally placed and does not extend to the fundus of theinternal auditory canal, the retrosigmoid approach is pre-ferred. A tumor that involves the distal end of the internalauditory canal is better approached via the middle fossa.

Candidates for hearing preservation surgery are thosewho have serviceable hearing, i.e., usually no greater thana 40-dB pure tone loss with residual speech discrimina-tion of at least 80%. Preservation of wave form with onlya slight increase of latencies on the auditory brain stemresponse is a favorable prognostic sign. Loss of functionof the superior vestibular nerve as indicated by a reducedvestibular response on electronystagmography is also afavorable sign indicating a tumor in the superior com-

partment of the internal auditory canal. Tumors in thesuperior compartment are less likely to intimately in-volve the cochlear nerve and are also more likely to dis-place the facial nerve anteriorly rather than be locatedbeneath the facial nerve.

There are also disadvantages to the middle fossa ap-proach. The first is that with this approach, the surgeonmust work past the facial nerve to remove the tumor. Thissubjects the facial nerve to more manipulation than doesthe translabyrinthine approach. A second problem some-times encountered is postoperative unsteadiness result-ing from partial preservation of vestibular function. Thisproblem also occurs with the retrosigmoid approach, butit rarely occurs with total vestibular denervation with thetranslabyrinthine approach. Careful section of the remain-ing vestibular nerve fibers reduces the incidence of post-operative unsteadiness but also increases the risk of hear-ing loss.

The final potential problem with the middle fossaapproach is limited access to the posterior fossa in theevent of bleeding either at surgery or postoperatively.Although this has not occurred in our series, a problemcould arise if significant bleeding occurred from the an-terior inferior cerebellar artery.

OPERATIVE MANAGEMENT

AnesthesiaThis operation is performed under general endotrachealanesthesia using inhalation agents. Facial nerve monitor-ing is routinely used so that muscle paralysis is not usedexcept for the initial induction of anesthesia. Diureticsand mannitol are usually used to promote diuresis.

Positioning and PreparationThe patient is placed supine on the operating table withthe head turned so that the operated ear is facing up. Noexternal fixation is utilized. The surgeon is seated at thehead of the table. The remainder of the operating roomsetup is shown in Figure 1.© 1991 The American Association of Neurological Surgeons

241

BRACKMANN : EXCISION OF ACOUSTIC NEUROMAS BY THE MIDDLE FOSSA APPROACH

Figure 1. The operating room arrangement. Thesurgeon is seated at the head of the table.

A large area of hair removal is required because theincision extends far superiorly. The area of preparationextends nearly to the top of the head and far anteriorlyand posteriorly. The skin is prepared with Betadine scruband self-adhering plastic drapes are applied.

IncisionThe middle fossa incision begins within the natural hair-line just anterior to the base of the helix and extendssuperiorly approximately 7 to 8 cm, curving first anteri-orly and then posteriorly (Fig. 2). Curving the incisionallows the surgeon to spread soft tissue widely to gainmore access anteriorly. Bleeding vessels are controlledwith cautery. The surgeon often encounters a branch ofthe superficial temporal artery which is ligated with non-absorbable sutures to avoid late postoperative bleedingand hematoma formation.

The initial incision extends to the level of thetemporalis fascia. Finger dissection develops the planeof the temporalis fascia along the temporal line. An inci-sion is then made posterosuperiorly along the insertionof the temporalis muscle onto the squamous portion ofthe temporal bone. The temporalis muscle is freed fromthe temporal bone and retracted anteroinferiorly. Eleva-tion of the temporalis muscle in this fashion preserves itsnerve and blood supply so that it could be utilized laterfor a temporalis muscle transfer to the lower face in caseof persistent facial paralysis. The temporalis muscle iselevated to the temporal line and held in place with self-retaining refractors.

Elevation of the Bone FlapA craniotomy opening is made in the squamous portionof the temporal bone (Fig. 3). The opening is approxi-mately 4 cm square and is located two-thirds anterior andone-third posterior to the external auditory canal. It isimportant to place the craniotomy opening as near aspossible to the floor of the middle fossa. This usuallyrequires hand retraction of soft tissue by an assistant.

A medium cutting burr and continuous suction-irri-gation are used. The bone flap is thicker superiorly as thesquamoparietal suture is approached. It is important not tolacerate the dura during the bone removal for this couldallow herniation of the temporal lobe. Herniation can bestbe avoided by leaving a thin plate of bone over the dura.The bone can then easily be fractured and removed. Weprefer to make a bone flap rather than a burr hole androngeur enlargement so that the temporal bone flap can bereplaced at the end of the procedure. This results in lesstissue retraction in the area of the incision, thus improvingthe cosmetic appearance. Replacement of the bone flapalso reduces the possibility of transmission of brain pulsa-tions to the skin, which is cosmetically undesirable.

Bone bleeders are commonly encountered and arecontrolled with bone wax. It is important to keep theedges of the bone flap parallel. This facilitates placementof the middle fossa retractor.

Once the surgeon has drilled nearly through the tem-poral bone flap, a joker elevator is used to separate theunderlying dura and the bone flap is removed. Sharp edgesare trimmed from the bone and it is placed in normalsaline solution during the operation.


242


Figure 2 (Top). The incision extends 7-8 cm superiorly from thenatural hairline and is 0.5 cm anterior to the helix.

Figure 3 (Lower left). The bone flap is made two-thirds anteriorand one-third posterior to the external auditory canal.

Figure 4 (Lower right). A middle fossa retractor is firmly lockedin place, and dural elevation is begun.

© 1991 The American Association of Neurological Surgeons © 1991 The American Association of Neurological Surgeons


243

Elevation of the DuraThe dura is separated from the margin of the craniotomydefect with the joker elevator. Any sharp bone edges areremoved with the rongeur. Occasionally, the Inferior bonecut is above the floor of the middle fossa. This excessbone is removed with the rongeur. At times, there will bebleeding from the branches of the middle meningeal ar-tery on the surface of the dura. This is best controlledwith bipolar cautery with care being taken not to burn ahole in the middle fossa. dura. The level of cautery mustbe reduced to the minimum necessary to coagulate thevessel. It is important to maintain the integrity of thedura since any defect may allow herniation of the tempo-ral lobe. If a small tear is produced, it should be closedwith dural silk suture to prevent extension and hernia-tion. After separation of the dura from the edges of thecraniotomy defect, the House-Urban retractor is put inplace and firmly locked. The blade of the retractor is thenset in place and gentle elevation of the dura from thefloor of the middle cranial fossa. is begun (Fig. 4).

The House-Urban retractor contains three adjust-ments, allowing for the desired placement of the retractor.The first adjustment affords movement of the entire blademechanism in an inferior-superior direction. Once theblade is properly centered at the depth of the middlefossa. dissection, this adjustment is secured at a positionthat gives maximum exposure but does not impinge onthe superior edge of the craniotomy opening.

Dural elevation is carefully begun from posterior toanterior. As the dura is elevated, the tip of the blade isadvanced. The two other adjustments are arranged appro-priately. One adjustment allows for anterior-posterior

movement of the tip of the retractor blade. The other ad-justs the placement of the tip of the blade in a superior-inferior direction.

The structures within the temporal bone as viewedfrom above are shown diagrammatically in Figure 5. Thefirst landmark to be identified is the cranial entrance ofthe middle meningeal artery at the foramen spinosum.This marks the anterior limit of the dural elevation. Fre-quently, venous bleeding is encountered in this area. Itmay be necessary to control this bleeding by placing afirm pack of Surgicel into the foramen spinosum.

The surgeon’s attention is then directed posteriorlyand the medial elevation of the dura is accomplishedfrom posterior to anterior. First the petrous ridge is iden-tified posteriorly. Care is taken in this area because thepetrous ridge is grooved by the superior petrosal sinuswhich the surgeon must avoid entering. If the sinus isinadvertently entered, bleeding can usually be controlledby extraluminal packing with Surgicel. However, smallpieces of Surgicel must not be placed in the lumen of thesinus because they can produce a pulmonary embolus.

The dura is then elevated from the floor of the middlefossa medially from posterior to anterior. In approximately5% of cases, the geniculate ganglion of the facial nervewill not be covered by bone. Blind or rough elevation ofthe dura in such cases can result in damage to the facialnerve. It is best to gently elevate the dura from the tempo-ral bone rather than to scrape the elevator along the boneof the middle fossa.

The posterior to anterior elevation avoids rais-ing the greater superficial petrosal nerve. If this nervewere elevated and the dissection carried posteriorly,

Figure 5. The relationship of the structures within the temporal bone as viewed from the middle fossa.



244


the geniculate ganglion and facial nerve would again besubject to injury. In the literature on the middle fossaapproach to the gasserian ganglion an up to 5% inci-dence of facial paralysis has been reported. With carefuldural elevation performed with the aid of the surgicalmicroscope, this complication can be avoided in all cases.

As dural elevation proceeds, the arcuate eminence isencountered. At times, this is an obvious landmark butsometimes it is indistinct. The positive landmark is thegreater superficial petrosal nerve which passes parallelto the petrous ridge, anteriorly from the geniculate gan-glion. This nerve lies medial to the middle meningealartery. Once the greater superficial petrosal nerve has beenidentified, it is carefully followed to the hiatus of thefacial nerve and the blade of the middle fossa retractor isreadjusted (Fig. 6).

A word of caution is necessary regarding pressure onthe bone of the floor of the middle fossa. Both in the area ofthe tegmen and over the internal carotid artery the bonemay be very thin and rarely even dehiscent. Although in-jury to the internal carotid artery has not occurred in ourcases, care must be taken to avoid this complication. Atthis point the major landmarks of the middle fossa ap-proach have been identif ied. These are the middlemeningeal artery, the arcuate eminence, the greater super-ficial petrosal nerve, and the facial hiatus. Therefore, thesurgeon is ready to begin bone removal over the internalauditory canal. There is often considerable bleeding fromsmall vessels on the surface of the dura and the floor of themiddle fossa. This bleeding is particularly troublesomesince it pools into the most dependent portion of the woundwhere the bone removal is to begin. Considerable bleed-ing from any one vessel must be controlled. It is usual,however, for oozing to occur from multiple sites. We havefound that such bleeding will stop spontaneously and it isbest to proceed with the operation at this point.

Exposure of the Internal Auditory CanalA large diamond burr and continuous suction-irrigationare brought in and careful removal of bone to identifylandmarks of the temporal bone is begun. It is importantto use the largest burr early in the bone removal since itoffers the most protection against accidental injury tothe geniculate ganglion, facial nerve, or superior semi-circular canal. Attention is directed to the greater superfi-cial petrosal nerve and bone is gently removed from thearea of the hiatus until the geniculate ganglion is identi-fied. A thin shell of bone is usually left over the genicu-late ganglion but the ganglion itself is readily apparentthrough the bone (Fig. 7). By this time, bleeding hasusually subsided. If not, it is advisable to spend the timenecessary to control the bleeding before proceeding.

The labyrinthine portion of the facial nerve is thenfollowed from the geniculate ganglion to the internalauditory canal (Fig. 8). The labyrinthine portion of thefacial nerve courses parallel to the plane of the superiorsemicircular canal. A smaller diamond burr is necessaryfor this bone removal since the ampullated end of thesuperior semicircular canal lies only a few millimetersposterior to the facial nerve at this point and the cochlealies only a few millimeters anteriorly.

Some surgeons prefer to proceed with removal ofbone over the internal auditory canal following the iden-tification of the superior semicircular canal. The internalauditory canal makes an angle of 45 to 60° with the supe-rior semicircular canal. We have found it easier to posi-tively identify the internal auditory canal by followingthe facial nerve. We have not experienced problems withfacial nerve paralysis by using this technique. Continu-ous monitoring of the facial nerve alerts the surgeon whenthe facial nerve is exposed.

Once the internal auditory canal has been identified,bone removal is continued medially until the entire su-perior surface of the internal auditory canal is exposed.As the surgeon proceeds medially it is possible to en-large the exposure because the superior semicircular ca-nal courses posteriorly and the dissection is medial to thecochlea (Fig. 9). Bone removal is continued until theporus acusticus is removed. The superior petrosal sinusgrooves the petrous ridge and care is taken not to enterthe superior petrosal sinus.

Great care must be taken to remove the bone withoutentering the dura because the facial nerve lies directlyagainst the dura. It is best to leave an eggshell thicknessof bone over the entire surface of the internal auditorycanal until all of the bone removal has been completed.

Bone removal must be extensive for middle fossaacoustic tumor surgery. Medially, bone removal is car-ried far anterior and posterior to the internal auditorycanal. The superior petrosal sinus will be lying free inthe dura following removal of the petrous ridge. Caremust be taken to avoid bleeding but should it occur itmay be controlled with extraluminal packing of Surgicelor with clips.

The dissection is limited posteriorly by the supe-rior semicircular canal. Bone is carefully removed fromthe entire extent of the internal auditory canal whichallows exposure of approximately three-quarters of thecircumference of the canal at the porus. It is not pos-sible to achieve this degree of exposure in the lateralportion of the internal auditory canal because of therestricting position of the ampulla of the superior semi-circular canal and the cochlea. When bone removalhas been completed medially, the lateral end of

245


Figure 6 (Top). The dura is elevated and the greater superficialpetrosal nerve is identified medial to the middle meningeal artery.

Figure 8 (Bottom). The labyrinthine portion of the facial nerveis uncovered from the geniculate ganglion to the internal audi-tory canal.

Figure 7 (Middle). Bone is removed from the greater superficialpetrosal nerve until the geniculate ganglion is identified.




246


the internal auditory canal is dissected and the verticalcrest of bone separating the facial nerve from the superiorvestibular nerve (Bill’s bar) is identified.

This completes the bone removal. The fine eggshelllayer of bone is then removed and the dura is opened alongthe posterior aspects of the internal auditory canal (Fig.10). The facial nerve lies anteriorly and the first exposureof the internal auditory canal should be away from thefacial nerve. The dural flap is carefully elevated from theunderlying tumor and the facial nerve is identified at thelateral end of the internal auditory canal where the verticalcrest of bone (Bill’s bar) allows positive identification. Thesuperior vestibular nerve lies posteriorly at this point anda fine hook is used to begin the separation of the superiorvestibular nerve and tumor from the facial nerve.

Tumor RemovalThe vestibulofacial anastomotic fibers are cut and thesuperior vestibular nerve is cut at the end of the internalauditory canal. Separation of the tumor from the end of

the internal auditory canal and the facial nerve is begunnext (Fig. 11). The principle of the tumor removal is thatthe tumor is freed from the facial nerve and the internalauditory canal and is delivered posteriorly out from un-der the facial nerve. For this reason, it is most importantto remove all of the bone from the posterior aspect of theinternal auditory canal.

Freeing of the lateral end of the tumor, particularlyin the inferior compartment of the internal auditory ca-nal, is one of the most difficult parts of the dissection.This difficulty is attributable to the poor view of themost lateral aspect of the inferior compartment of theinternal auditory canal. The removal is accomplished witha long hook that is used to palpate the end of the internalauditory canal. It is best to totally section both the supe-rior and inferior vestibular nerves to prevent postopera-tive unsteadiness (Fig. 12). A partial vestibular denerva-tion is more likely to result in unsteadiness than is totalremoval of both vestibular nerves.

The tumor is gently teased out of the lateral end of

Figure 9 (Upper left). Bone is removed from the entire length ofthe internal auditory canal, including the porus acusticus.

Figure 10 (Lower right). The lateral end of the internal auditorycanal has been dissected and the vertical crest (Bill’s bar) identified.The dura is then incised at the posterior margin of the internalauditory canal.



247


Figure 11 (Lower left). The tumor is separated from the facialnerve at the end of the internal auditory canal. It is carefully freedfrom the facial and cochlear nerves and delivered posteriorly.

Figure 12 (Upper right). The tumor has been removed. It is bestto section the inferior vestibular nerve to minimize the risk ofpostoperative unsteadiness.

the internal auditory canal. The blood supply to the co-chlea usually runs between the facial and cochlear nervesand in most cases is preserved with those nerves. At times,however, the arterial supply to the cochlea is interrupted,resulting in hearing impairment even though preserva-tion of the cochlear nerve was achieved.

Once the lateral end of the tumor has been freed, theplane between the cochlear and facial nerves and tumorbecomes apparent. This plane is then carefully devel-oped through the use of fine hooks and the Rosen separa-tor. Tumor dissection is continued medially and a searchfor the anterior inferior cerebellar artery is begun. Theartery may loop up into the internal auditory canal infe-rior to the cochlear nerve or the tumor may displace itinto the cerebellopontine angle. Great care must be takento identify and not injure this most important artery.

After the anterior inferior cerebellar artery has beenidentified, it is freed from the surface of the tumor bycareful blunt dissection with the Rosen elevator. The fi-nal problem is freeing the most medial aspect of the tu-mor. Before this, it is often necessary to partially removethe tumor with small cup forceps. This prevents stretch-

ing of the facial nerve. During the course of the dissec-tion, the surgeon must be careful not to injure the facialnerve with suction. This possibility is reduced by the useof the fenestrated neurolologic suction tip. Freeing of themedial end of the tumor with small hooks allows its re-moval (Fig. 12). Continuous intraoperative monitoringof facial nerve activity greatly facilitates dissection ofthe tumor from the facial nerve and is utilized routinely.

After total tumor removal, the tumor bed is irrigatedprofusely. Bleeding from small vessels usually subsidesduring the irrigation. At times, larger vessels will requirebipolar cautery for control of bleeding.

Inadvertent injury to the anterior inferior cerebellarartery is always a possibility during removal of the medialaspect of the tumor. Control of bleeding would be extremelydifficult in this situation because of limited access. Fortu-nately, this has not occurred in our experience; if it should,it might be necessary to remove more bone in the area ofthe superior semicircular canal to expose more of the pos-terior fossa dura in order to gain access for the applicationof a clip. If bleeding should still not be controlled, it mightbe necessary to perform a postauricular approach and



248


translabyrinthine exposure of the cerebellopontine angleto achieve this control.

Wound ClosureClosure of the defect in the internal auditory canal isaccomplished with a free graft of temporalis muscle. Thetemporal bone flap is then replaced and the temporalismuscle resutured to its insertion. The subcutaneous tis-sue and skin are closed in layers and a sterile dressing isapplied. If there is excessive oozing, a Penrose drain isutilized.

COMPLICATIONS AND MANAGEMENTAn epidural hematoma is an uncommon early postopera-tive complication. The incidence of this may be reducedby the use of a drain where there is excessive oozing.Meticulous attention to hemostasis is a necessity. Pa-tients with this complication will exhibit signs of in-creasing intracranial pressure. Treatment is immediateevacuation of the hematoma on the intensive care unit.The patient is then taken to the operating room wheremore definitive control of the bleeding is accomplished.

Other complications are those that are common toany intracranial procedure, such as meningitis. Temporallobe injury from retraction on the temporal lobe was anearly concern but has not been a problem in our series.We have had no patients who have had signs of corticalinjury such as hemiparesis or aphasia. Hearing loss andfacial paralysis are expected complications in some ofthese patients as with any acoustic tumor removal.

DISCUSSION OF SERIESAs of June 1989, over 2500 acoustic neuromas had beenremoved at the Otologic Medical Group (Los Angeles,CA). The middle fossa approach had been used to remove106 of these tumors as of December 1986. The size ofthese tumors varied from 0.4 to 2 cm. Hearing was pre-served in 63 patients (59%). In 37 patients (35%), hear-ing was the same as before surgery. There was a partialloss of hearing in 26 patients (25%). In the remainder, atotal sensorineural hearing loss occurred despite preser-vation of the cochlear nerve in 89% of patients. Hearingroughly correlated with tumor size: hearing preservationis better for smaller tumors.

Eighty percent of patients had normal facial nervefunction one year after middle fossa surgery for removal ofan acoustic tumor. Another 9% had Grade II function andthe remainder a greater degree of weakness. No patient hada total facial nerve paralysis. These statistics demonstratethe increased risk to the facial nerve in the middle fossaapproach. In the translabyrinthine approach, with tumorsof this size, 88% have normal facial function. The attemptat hearing preservation offered by the middle fossa ap-proach does increase the risk to the facial nerve.

Other than hearing loss and facial weakness, therewere no other serious complications in this series. Therewere no deaths.

CONCLUSIONThe middle fossa approach is the preferred method forremoval of small, laterally placed acoustic neuromas.

249

UPPER THORACIC SYMPATHECTOMYBY A POSTERIOR MIDLINE APPROACH

PREM K. PILAY, M.D.ISSAM A. AWAD, M.D.

DONALD F. DOHN, M.D.

INTRODUCTIONCurrent indications for upper thoracic sympathectomyinclude intractable palmar hyperhidrosis and, to a lesserextent, Raynaud’s disease, reflex sympathetic dystrophy,and other sympathetic mediated pain syndromes. A vari-ety of approaches for upper thoracic sympathectomy havebeen employed. These include anterior transthoracic,preaxillary transthoracic, supraclavicular, posterior tho-racic, thoracic endoscopic, and percutaneous/stereotaxicapproaches. The posterior approach is the most advanta-geous to neurosurgeons because it incorporates an expo-sure similar to that of thoracic laminectomy, and it allowsbilateral exposure of the sympathetic chains and gangliaat a single sitting and via a single incision. The standardapproach developed at our institution is a posterior mid-line approach, with bilateral 3rd rib costotransversectomyallowing a one-stage bilateral excision of the 2nd tho-racic sympathetic ganglia. This is a modification of thetechnique described by Cloward and is based on the ob-servation that the 2nd thoracic sympathetic ganglion pro-vides exclusive innervation of sweat glands of the palm.If axillary sweating is also a prominent complaint, boththe T2 and T3 ganglia are removed.

POSITIONINGOur preference is to perform this procedure in the sittingposition because we believe that it affords a better ex-posure, with easier instrument handling than in the proneposition. The possibility of air embolism is less of aconcern because the operative wound is nearly at thelevel of the heart. The use of a central venous line isusually not necessary. After endotracheal general anes-thesia, the patient is placed in the sitting position on aGardner chair (Fig. 1) or by appropriately flexing theoperating table. The head is held by elastic straps to a

padded horseshoe rest applied to the face and is helderect by means of a metal frame slotted into the sides ofthe chair. The use of head-pins is avoided as these maybe a source of air embolism.

Localization of the T2 spinous process is done byplacing a radiopaque marker on the skin and obtaining alateral cervical/upper thoracic x-ray film. The skin isscratched over the T2 spinous process, which is approxi-mately at the level of the T3 costotransverse junction(Fig. 1). The patient’s back is then shaved, cleaned, pre-pared, and draped.

OPERATIVE PROCEDURE

ExposureA midline linear skin incision is made centered over theT2 spinous process approximately 10-15 cm in length.This will allow exposure of the T1 -T3 spinous processesand laminae. The subcutaneous fat is cleanly divided toexpose the underlying deep fascia (Fig. 2). A monopolarelectrocautery knife is then used with a periosteal re-tractor to expose the spinous processes in the midline,followed by subperiosteal dissection of the paraspinalmuscles off the laminae and costotransverse junctionsfrom medial to lateral. The wound is held open withangled D’Errico refractors. Confirmation of the correctlevel is done by placing metal clips on either side of thetransverse process and obtaining an intraoperative pos-teroanterior chest roentgenogram (Fig. 3). Compulsionabout correct localization will avoid the undesirablecomplications of inadvertent stellate (T1) ganglionec-tomy with associated Horner syndrome, or a more cau-dal (below T2) ganglionectomy without adequate pal-mar sympathectomy.

Using a Leksell rongeur, the costotransverse junc-tion bone is thinned out. Bone removal is then com-pleted with Kerrison rongeurs after stripping the un-derlying pleura away with a Penfield instrument. Inthis manner a 3-cm section of the costotransverse© 1991 The American Association of Neurological Surgeons

250


Figure 1. Operative positioning. The T2 spinous process is markedusing lateral cervicothoracic radiographic guidance. The incisionextends from T1 to T3 and is centered on the spinous process ofT2. Because of downward slanting of the spinous processes, thecenter of the incision is at the level of the 3rd costotransverse

junction. Because the incision is approximately at the level of theheart, no central venous line is placed for air embolism monitor-ing. To further minimize this risk, skull pins are not used, and thehead is fixed instead to a padded horseshoe cerebellar rest usingelastic bands.


251

PILLAY ET AL. : UPPER THORACIC SYMPATHECTOMY BY A POSTERIOR MIDLINE APPROACH

Figure 2. Paraspinal muscle dissection. The midline inci-sion is extended to the spinous processes. Using electrocau-

tery, the paraspinal musculature is stripped in a subperi-osteal fashion.

Figure 3. Intraoperative localization. Metal clips are placed aboveand below the 3rd costotransverse junction bilaterally. An intraop-erative posteroanterior x-ray film of the chest allows verificationof the correct level (3rd rib). These clips are subsequently removedso they will not be confused on postoperative films with othermetallic clips placed directly on the sympathetic chain.



252


Figure 4. Bone removal. Using Kerrison rongeurs, bilateral T3costotransversectomies are performed. The area of resection shouldextend to the pedicle medially and should include approximately 3

cm of the head of the rib (inset). The spinal canal is not entered bythis bony removal.

junction is resected (Fig. 4). Care is taken to avoid too me-dial a bony exposure as inadvertent damage to the nerveroots and dura. may occur with entry into the spinal canal.This procedure is carried out bilaterally. The bone edges arewaxed and hemostasis carefully secured before the next stepof identification of the sympathetic chain and ganglia.

Identification of the SympatheticChain and GanglionectomyA pair of Smithwick sympathectomy dissectors are usedin the next step of this procedure. The spatula end of theinstrument is used to depress and sweep the pleura in amedial to lateral fashion, while the hook end is used toprobe for the chain. This is usually visualized as a finegray glistening strand under loupe (× 2.5) magnification.This strand is usually followed superiorly to the T2 cos-totransverse joint level. The T2 ganglion can be identi-fied as a prominent ovoid bulb along the sympatheticchain, with rami communicans tethering it to the adja-cent intercostal nerve (Fig. 5).

A tonsillar hemostat is used to retract the gangliondownward (caudally), while sharply isolating the gan-glion from the adjacent intercostal nerve. Theneuroganglionic segment is then resected en bloc be-tween two clips applied to the sympathetic chain ceph-alad and caudad. The clips provide a permanent radio-graphic marker of the sympathectomy and may helpprevent regeneration of sympathetic innervation. The T3ganglion may be included in the caudal end of theneuroganglionic segment if axillary sweating is a promi-nent complaint (Fig. 6). Prior to closure, the wound isirrigated with saline solution and Valsalva’s maneuverinitiated by the anesthetist to identify any pleural airleak which would predispose to a pneumothorax.

Wound ClosureClosure is carried out in a standard fashion. A fewdeep muscle stitches are placed using 1-0 Nurolon orsilk sutures, followed by a watertight fascial closurewith 2-0 Nurolon or silk sutures. The subcutaneous


253


Figure 5. Dissection of the sympathetic chain. The Smithwickdissector is used (the hook end in one hand, and the spatula endin the other hand) to free up the epipleural adhesions and re-tract the visceral pleura away from the spinal column. Thehook end is then used to find and deliver the sympathetic chain.

Further dissection with the Smithwick instrument cephalad al-lows the identification of the T2 ganglion. Its communicationswith the intercostal nerve are sharply divided. The ganglion isfurther delivered downward to visualize the sympathetic chainabove it.


254


Figure 6. A, metal clips are placed above the T2 spinous pro-cess cephalad (making sure not to include the next higher gan-glion which is the stellate ganglion), and another clip caudad ator below the 3rd sympathetic ganglion. B, the specimen is sentfor rapid frozen pathologic confirmation of sympathetic gan-

glion, while the epipleural space is explored using irrigation andValsalva’s maneuver to ensure the absence of a tear in the vis-ceral pleura which would predispose to postoperative pneumotho-rax. The management of a visceral pleural tear is discussed inthe text.

layer is apposed with 3-0 Dexon or Vicryl and the skin closedwith interrupted 3-0 Nurolon or nylon sutures. In the pres-ence of an air leak from a pleural tear, this is located and asmall red rubber catheter (No. 8 French) is placed within thepleural space. The wound is then closed in layers around thered rubber catheter until just prior to the last fascial stitch.The other end of the catheter is then placed in a basin ofsaline solution and positive intrapleural pressure is used bythe anesthetist to drive air out through this underwater seal.The catheter is withdrawn during Valsalva’s maneuver andwound closure is completed.

After application of the wound dressing, the headstraps are removed, the patient transferred to the supineposition, and anesthesia is reversed and the patient extu-bated. The palms of the hands should be warm and com-pletely dry immediately following surgery.

POSTOPERATIVE CAREAn x-ray film of the chest (upright expiration view) isobtained in the recovery room and again the next day toexclude a pneumothorax. Vigorous incentive spirometryis encouraged in the postoperative period.

Parenteral narcotics and muscle relaxants are typi-

cally required for 1 or 2 days, followed by oral analge-sics as needed. The patient is ambulated 24 hours aftersurgery and is usually sent home by the 3rd or 4thpostoperative day. The skin sutures are left in for 14days and the patient is restricted from lifting and strenu-ous activities for 4-6 weeks. A transient period of pal-mar moisture can be noted a few days after surgery(after initial palmar dryness); this has been attributedto denervation hypersensitivity of the sweat glands.This is never of the same intensity as preoperativelyand only lasts 1 to 2 days; it warrants patient reassur-ance. Occasional neuritic discomfort is encounteredin the axillae and interscapular region, especially fol-lowing excessive manipulation of the intercostalnerve. This typically resolves spontaneously.

From 1960 to 1988, 440 patients with severe hy-perhidrosis were selected for upper thoracic sympath-ectomy at the Cleveland Clinic. There was no opera-tive mortality. The main complications were woundinfection (2.9%), and pneumothorax requiring chesttube insertion (<1% of patients). Air embolism has notbeen a problem in any patient despite the use of thesitting position. All patients operated upon for hyper-


255


hidrosis had immediate relief of palmar sweating. Ninety-two percent of patients had long-term satisfaction withsurgical results after a mean follow-up period of 7.2 years.A major cause of dissatisfaction was the occurrence ofcompensatory sweating. This occurred in other body ar-eas in up to 35% of patients in the postoperative period,but was felt to be a persistent troublesome complaint inless than 5% of patients on longer follow-up. Frank recur-rent palmar hyperhidrosis occurred in only three patients,3, 5, and 6 weeks post-surgery, respectively. One of thesepatients with persistent unilateral hyperhidrosis hadreexploration and ganglionectomy (missed at the firstprocedure) and was left with dry hands.

The results of this operation for Raynaud’s diseaseand sympathetic mediated pain have been less gratify-ing, with less than half the cases deriving significantlasting benefit.

In conclusion, the upper thoracic (T2) ganglionec-tomy performed via the posterior approach is a safe, effec-tive, and durable operation for the treatment of intractableessential palmar hyperhidrosis. It has allowed patients ofall walks of life to improve social and occupational inter-actions without the hindrance, stigma, and embarrassmentof excessive palmar sweating. The operation has also hadlimited application in other pathophysiologic situationsmediated by sympathetic hyperactivity.

256

CAROTID ENDARTERECTOMYDANIEL L. BARROW, M.D.

CHRISTOPHER E. CLARE, M.D.

INTRODUCTIONAtherosclerotic lesions of the cervical carotid artery usu-ally occur at or near the common carotid bifurcation.These lesions may result in cerebral ischemia or infarc-tion due to hemodynamic impairment of carotid bloodflow or distal embolization. Although both of these mecha-nisms are important, embolization appears to play thedominant role in carotid system ischemia. Carotid endar-terectomy is the surgical procedure for removing theseatherosclerotic lesions from the region of the carotid bi-furcation. Although the operation has significant intui-tive appeal, the appropriate indications for carotid en-darterectomy are the subject of ongoing debate. Severalstudies have demonstrated that patients with ischemicneurologic symptoms referable to the distribution of asigniftcantly stenosed and/or ulcerated carotid arteryexperience a decreased risk of stroke following a success-ful carotid endarterectomy.

INDICATIONS AND ASSESSMENT OF RISKThe decision to perform a carotid endarterectomy shouldbe based on a number of factors other than the mere pres-ence of an atherosclerotic lesion of the carotid bifurca-tion. Important factors in the decision-making processinclude the patient’s symptoms, physiologic state, col-lateral circulation, and underlying health. Carotid endar-terectomy is primarily indicated for patients with clearischemic neurologic symptoms ipsilateral to a signifi-cant carotid stenosis and/or ulceration. In certain situa-tions, after consideration of all variables, a patient willbe treated with systemic anticoagulation or antiplateletagents rather than surgery. Patients with amaurosis fugax,central retinal artery occlusion, venous stasis retinopa-thy, transient ischemic attacks (TIAs) in the carotid dis-tribution, prolonged reversible ischemic neurologic defi-cits, and mild-to-moderate fixed ischemic neurologicdeficits are at risk for ischemic hemispheric damage.

The indications for carotid endarterectorny in a largegroup of patients with asymptomatic neck bruits, asymp-tomatic carotid stenosis, and vertebrobasilar symptoms

associated with carotid artery stenosis are quite contro-versial. We do not recommend further work-up of patientswith asymptomatic bruits. Because the natural history ofasymptomatic carotid lesions is not well defined, we re-serve endarterectomy for only those asymptomatic pa-tients with severe stenosis (<2 mm residual lumen) ordeep ulcerations. We have not performed a carotid endar-terectomy for vertebrobasilar symptoms. However, after acritical analysis of the competence of the circle of Willisand collateral blood flow, some surgeons have thoughtthat blood flow to an insufficient vertebrobasilar circula-tion can be augmented by removing a hemodynamicallysignificant carotid lesion.

In all patients, the decision must be individualizedafter considering the patient’s age, risk factors, and surgi-cal risks for the particular institution and surgeon. Thegrading system developed by Sundt and associates groupspatients according to predetermined risk factors and isuseful in predicting the risks of surgical intervention:

Group 1. Neurologically stable with no major medical orangiographically defined risks, with unilateral orbilateral ulcerative-stenotic carotid disease;

Group 2. Neurologically stable with no major medicalrisks but with significant angiographically definedrisks;

Group 3. Neurologically stable with major medical risks,with or without significant angiographically definedrisks;

Group 4. Neurologically unstable, with or without associ-ated major medical or angiographically defined risks.

Patients in groups 3 and 4 are at greatest risk fornonneurologic and permanent neurologic compli-cations, respectively.

PREOPERATIVE ASSESSMENTAll patients with cerebrovascular symptoms should beevaluated to determine the cause of the symptoms and toassess the risk of stroke. Those patients with frequentTIAs, stroke-in-evolution, or acute onset of a mild deficitare at greatest risk and should be evaluated urgently.

We do not rely heavily on noninvasive tests of thecarotid artery in clinical decision making. Good qual-© 1991 The American Association of Neurological Surgeons

257

BARROW AND CLARE : CAROTID ENDARTERECTOMY

ity angiography is essential in the evaluation of patientsbeing considered for carotid endarterectomy. Visualiza-tion of the origins of the cranial vessels in the thorax andof their extracranial and intracranial distributions are allimportant. In the symptomatic patient, angiography isusually performed after the patient has had appropriatemedical evaluation and computed tomography (CT) ormagnetic resonance imaging (MRI) of the head. The CTor MRI will usually rule out the presence of a mass le-sion, such as a tumor, that may cause symptoms mimick-ing cerebral ischemia. These studies will also reveal oldor new infarcts that may alter the timing of surgical inter-vention or show a “silent” lesion in the distribution of anotherwise asymptomatic carotid lesion.

Prior to surgery, the patient’s cardiopulmonary statusis carefully evaluated. Carotid atherosclerosis is an impor-tant indicator of systemic atherosclerosis. In patients withsignificant cardiac disease, a Swan-Ganz catheter may as-sist in fluid management in the perioperative period.

Patients with a recent transient ischemic attack orprogressing focal cerebral ischemic event will frequenttype placed on heparin while awaiting surgery. If symp-toms continue despite adequate anticoagulation, thesepatients are considered surgical emergencies. Other pa-tients selected for endarterectomy are scheduled for sur-gery as soon as possible unless they have had a recentmajor stroke or recent myocardial infarction which ne-cessitates a delay. Patients awaiting an elective endarter-

ectomy are placed on 325 mg of aspirin per day beforesurgery; this is continued postoperatively.

SURGICAL TECHNIQUE

AnesthesiaGeneral endotracheal anesthesia is induced with thio-pental sodium (3-5 mg/kg) and paralysis is obtained withpancuronium or vecuronium. bromide (0.1 mg/kg).Lidocaine (1.0 mg/kg) or fentanyl (0.05-0.1 mg) is usedto diminish a cardiovascular response to intubation. An-esthesia is maintained with a nitrous oxide/ oxygen mix-ture and isoflurane (0.5-1.5%). Respirations are controlledto maintain an end-tidal CO2 between 35 and 38 mm Hg.The patient is given glucose-free fluids during and afterthe operation. Anesthesia monitoring includes urine out-put, arterial blood pressure measurements, arterial bloodgas analysis, endtidal CO

2 determinations, and electro-

cardiographic monitoring. Monitoring of cerebroelectricalfunction is carried out with conventional 16-channelelectroencephalography (EEG).

Operative PositioningThe patient is placed in the supine position with thehead on a foam rubber “doughnut.” The head is ex-tended and turned slightly away from the side of theoperation (Fig. 1). One or two rolled sheets are placedunder the shoulder blades to facilitate extension of the

Figure 1. The patient is placed supine with the head on a foamrubber doughnut and turned slightly away from the side of theoperation. The skin incision (dashed line) is made along the ante-

rior border of the sternocleidomastoid muscle. It is curved superi-orly to facilitate distal exposure without injuring the marginalmandibular branch of the facial nerve.


258


head. Rotation of the head will bring the internal carotidartery to a more accessible position laterally than its nor-mal position behind the external carotid in an anteropos-terior plane. The amount of head turning needed for opti-mal exposure can be determined from the preoperativeanteroposterior angiogram. To diminish the risk of dis-lodging embolic material from the atherosclerotic plaque,the neck is not scrubbed but is prepared with povidone-iodine solution. If there is any suggestion from the pre-operative angiogram that a venous patch graft will beneeded or if the operation is a repeat procedure, a distallower extremity is shaved, prepared, and draped for expo-sure of the saphenous vein for obtaining the graft.

Skin IncisionThe skin incision is placed along the anterior border of thesternocleidomastoid muscle (Fig. 1). This may be curvedmedially at the lower end to a point just above the sternalnotch. The superior limb is curved posteriorly from a pointabout 1 cm below the angle of the mandible toward themastoid process. This superior curve is designed to avoidinjury to the marginal mandibular branch of the facialnerve. The exact length of the incision is dictated by theposition of the carotid bifurcation and the morphology ofthe plaque as demonstrated on the angiogram. The skinincision is carried down to the platysma muscle, dividingthe transverse cervical nerve, which results in unavoidablenumbness anterior to the skin incision.

Operative ProcedureThe platysma is divided sharply, and meticulous hemo-stasis is maintained. A self-retaining retractor is placedinto the wound with the medial side more superficial toavoid injury to the laryngeal nerves (Fig. 2). Dissectionis carried out along the anterior border of the sternocleido-mastoid through the loose areolar tissue that lies betweenthis muscle and the strap muscles overlying the trachea.The plane of dissection is followed down to the internaljugular vein, which, as a key landmark for the exposure,lies lateral, parallel, and slightly anterior to the internaland common carotid arteries. Extreme care is taken toavoid manipulation of the carotid bifurcation and proxi-mal internal carotid artery to minimize the risk of dis-lodging embolic material from the plaque. If the artery ispalpated, as in a patient with a thick neck, it should beperformed only on the common carotid artery and faraway from the bifurcation.

Once the jugular vein is identified, dissection is alongthe medial border of the vein (Fig. 3). The common facialvein, which crosses the carotid at the level of the bifurca-tion, is doubly ligated and divided. It is usually neces-sary to section the omohyoid muscle to obtain adequate

proximal exposure. The carotid sheath is incised over thecommon carotid artery. Tacking the sheath to the exter-nal cervical fascia helps elevate the carotid in the wound.Two self-retaining refractors are placed into the externalfascia, further elevating the carotid from the wound.

As dissection is carried superiorly, the descendenshypoglossi nerve may be divided as it joins the hypoglo-ssal nerve proper to prevent undue traction on the latterstructure and to mobilize it for distal dissection of theinternal carotid artery (Fig. 3). The superior thyroid ar-tery is identified and dissected circumferentially. It ismandatory to gain access to the distal internal carotidartery. To do so, it may be necessary to divide the digas-tric muscle and mobilize the hypoglossal nerve. Injec-tion of a local anesthetic into the carotid body and sinusis not routinely performed. However, if the anesthesiolo-gist notes any change in vital signs during dissection ofthe bifurcation, 2-3 ml of 1% lidocaine is used to tempo-rarily block the effects of carotid sinus stimulation.

The common and external carotid arteries are dis-sected free from their underlying beds only in those areaswhere umbilical tapes or clamps are placed around them.The umbilical tapes placed around the internal and com-mon carotid arteries are threaded through rubber tubingto fashion a Rummell tourniquet. These should be placedwell above and below the plaque as determined by theangiogram, observation of thickness in the vessel wall,or gentle palpation.

A No. 9 French malleable multiperforated suctiontube is placed adjacent to the common and internal ca-rotid arteries and fixed into position by stapling it to thesurgical drapes.

Prior to carotid artery clamping, 100 units/kg hep-arin is given intravenously. Bolus doses of thiopentalsodium (150-250 mg) are given until 15- to 30secondburst suppression is seen on the EEG recording. The bar-biturate is continued by bolus injections or constant in-fusion to maintain burst suppression until internal ca-rotid artery flow is reestablished. A phenylephrine infusionis occasionally required to maintain systemic blood pres-sure in the normal range. Hypotension must be avoided.

Once the heparin and barbiturates have circulated,the internal carotid artery is occluded first with an aneu-rysm clip. The common carotid artery is immediately oc-cluded with a Fogarty vascular clamp and temporary an-eurysm clips are placed on the external carotid andsuperior thyroid arteries (Fig. 4).

The arteriotomy is begun in the common carotidartery with a No. 11 scalpel. Potts arterial scissors areused to extend the arteriotomy distally beyond thetermination of the plaque in the internal carotid ar-tery (Fig. 5). The true lumen of the vessel is entered,

259


Figure 2 (Top). The loose areolar tissue anterior to the sterno-cleidomastoid is separated and a self-retaining retractor placed nodeeper than the platysma muscle. The omohyoid muscle has beendivided to improve proximal exposure.

Figure 3 (Middle). The common facial vein has been doubly ligatedand divided, and the carotid sheath has been opened. Division of thedescendens hypoglossi allows greater mobilization of the hypoglossalnerve and reduces the risk of traction injury to the latter.

Figure 4 (Bottom). The common and external carotid arterieshave been dissected free from their underlying beds where clampsare placed around them. Rummell tourniquets have been fashionedfor the common and internal carotid arteries in the event a shuntis necessary. Once heparin and barbiturates have circulated, theinternal carotid artery is occluded first with an aneurysm clip.Next, the common carotid artery is occluded with a Fogarty clampand aneurysm clips are placed on the external carotid and superiorthyroid arteries. The arteriotomy is initiated with a No. 11 scalpel.


260


and no attempt is made to perform an extraluminal dissec-tion of the plaque from the artery. Great care must be takento ensure that the back wall of the carotid is not damaged.The arteriotomy incision should extend up the midline ofthe vessel to facilitate later closure of the arteriotomy. Thearteriotomy into the common carotid artery is continueduntil the plaque ceases to be ulcerated and the wall of thevessel becomes more normal in texture. The atheroscle-rotic plaque is carefully dissected from the internal carotidartery with a microdissector or small spatula (Fig. 6). Theplaque is removed by following the pathological cleavageplane created by the atherosclerosis between the intimaand the fine layer of media. During the dissection, a vascu-lar pickup is used to hold the wall of the artery as theassistant holds the edge of the plaque. The spatula is moved

from side to side, developing the plane described above,which is usually readily separated. Dissection proceedshalfway around the wall before repeating the separationfrom the other side. A clean feathering away of the plaqueis usually possible in the distal internal carotid artery butnot in the common carotid artery. The plaque is transectedsharply in the common carotid artery and elevated to visu-alize the orifice of the external carotid.

In removal from the external carotid artery, theplaque is elevated from the wall through the orifice ofthe vessel (Fig. 7). As it thins, the vessel is everted, andthe plaque is grasped with a hemostat and pulled infe-riorly. Although this is a blind procedure, it usuallybreaks cleanly from the wall of the external carotidartery. Patency of this artery may be determined by

Figure 5 (Top). The arteriotomy is extended distally with Pottsarterial scissors. The opening should extend up the midline of thevessel and within the true lumen of the artery. The arteriotomy isextended distally until the limit of the plaque on the posterior wall ofthe internal carotid artery is identified. The arteriotomy into thecommon carotid is continued until the plaque ceases to be ulceratedand the wall of the vessel becomes more normal in texture.

Figure 6 (Bottom). The plaque is removed by developing andfollowing the pathological cleavage plane created by the athero-sclerosis between the intima and media.



261


intraoperative angiography or by palpation of the super-ficial temporal artery pulse following restoration of bloodflow.

The cut edges of the proximal end of the commoncarotid artery are everted, and the plaque is cutcircumferentially 1-2 cm below the end of the arteriotomy(Fig. 8). This allows plaque removal below the end of thearteriotomy incision, making closure of the vessel easierand avoiding proximal stenosis. Once the plaque has beenremoved, the operating microscope is positioned to allowboth the surgeon and assistant to have clear binocularvision. Loose fragments of atherosclerotic material andabnormal intima are meticulously removed under con-tinuous heparinized saline irrigation. The irrigation fluidand blood are cleared from the wound by the Microvacsuction tube. With the operating microscope, the distalinternal carotid artery is carefully inspected, and any el-evation of the intima beyond the end of the plaque iscarefully trimmed with microscissors. If there is an abruptstep off or if the intima is loosely adherent to the media,double-armed sutures of 7-0 Prolene are placed verticallyfrom the inside of the vessel outward so they traverse theintimal edge and are tied outside the adventitial layer.The arteriotomy is closed using a running 6-0 Prolenesuture, starting at the distal end of the incision (Fig. 9).The use of the operating microscope to place small, closelyspaced stitches results in a tight, nonleaking suture linewithout compromising the lumen of the vessel.

Prior to final closure, back bleeding from all the ves-sels is allowed to expel air and debris from the lumen ofthe repaired segment. The superior thyroid artery is notreclipped, allowing continuous back bleeding duringplacement of the final sutures to avoid air being trappedin the vessel. Following final closure of the arteriotomy,the arteries are reopened in a specific order. The externalcarotid artery is opened initially, followed by the com-mon carotid artery. The common carotid artery is brieflyreclosed just before the internal carotid artery is openedto allow any embolic material to be washed into the ex-ternal carotid artery. The common carotid artery is thenunclamped.

Placement of a ShuntA shunt is used when there are changes in the EEG thatdo not immediately respond to a trial of induced hyper-tension. We prefer the Sundt internal shunt. The plaque isremoved from the distal internal carotid artery prior toshunt placement (Fig. 10). The distal end of the shunt isfirst placed into the internal carotid artery. Back bleed-ing from the internal carotid artery will fill the shunt withblood. Then the proximal end is placed into the common

carotid artery and secured with the Rummell tourniquet.Plaque removal is completed by working around the shunt.

An alternate method of shunt placement is first toinsert the shunt into the common carotid artery and se-cure it with the Rummell tourniquet. The shunt is heldclosed at its midportion with vascular forceps, brieflyopened to confirm blood flow and evacuate any debris inthe shunt tubing, and then placed into the lumen of theinternal carotid artery.

Regardless of the technique, the shunt should easilythread up the internal carotid artery, and no force shouldbe used to advance the shunt or intimal damage and dis-section may result.

In its removal, a suture is placed loosely around theshunt to facilitate extraction. Prior to completely closingthe arteriotomy, the shunt is withdrawn through a smallopen segment of the arteriotomy. The arteriotomy closureis then completed in a routine fashion.

Wound ClosureThe refractors are removed and sufficient time is spent toobtain excellent hemostasis because the heparin is notreversed. The platysma and subcutaneous tissue arereapproximated and the skin closed with a 4-0 Vicrylsubcuticular suture and Steristrips.

POSTOPERATIVE CAREWith the use of barbiturates, many patients are obtundedfor a short period of time following recovery from anesthe-sia and may require continued intubation until they arealert. However, these patients may be examined, as brainstem function is readily tested and the patients will movetheir extremities in response to noxious stimulation.

Blood pressure is carefully monitored both in therecovery room and in the intensive care unit. Patientsgenerally spend 24-48 hours in the intensive care unitfollowing surgery. Once the patient tolerates liquids, as-pirin is restarted and is continued indefinitely.

ALTERNATIVES IN SURGICAL TECHNIQUEThere are a number of technical variations in performinga carotid endarterectomy that are quite acceptable. Ex-cellent surgical teams have advocated the routine use ofvenous patch grafts, routine shunting, local anesthesia,and many other personal preferences. It is important thatthe surgical team becomes comfortable with its methodand frequently assesses its results.

Closure with a Venous Patch GraftIt is our custom to make liberal use of a venous patchgraft in the closure of an endarterectomy, although it

262


Figure 7 (Top). The plaque has been removed from the internalcarotid artery initially so that a shunt may be placed more safely ifnecessary. The plaque has been sharply excised from the wall of thecommon carotid artery with Potts scissors. Next, the plaque iselevated and dissected from the external carotid artery by workingthrough its orifice with a spatula. Once the plaque becomes thin, itis grasped with a hemostat and pulled inferiorly.

Figure 8 (Middle). The cut edges of the proximal end of thecommon carotid artery are everted and the plaque cutcircumferentially 1-2 cm below the end of the arteriotomy. Thismaneuver makes closure of the proximal arteriotomy easier andavoids stenosis below the end of the arteriotomy incision.

Figure 9 (Bottom). The arteriotomy is closed under the operat-ing microscope with running 6-0 Prolene sutures.




263


Figure 10. In the event that a shunt is required, the plaque is initiallyremoved from the distal internal carotid artery. The shunt is grasped inthe center with hemostats and placed into the internal carotid artery.

The hemostats are temporarily opened to allow back bleeding to fillthe shunt with blood. The proximal end is then placed into the com-mon carotid artery and secured with the Rummell tourniquet.

is not employed routinely. A venous patch graft is used inall cases of reoperation or if there is a suggestion from thearteriogram that the internal carotid artery is small. Webelieve that use of the operating microscope allows forclosure of the arteriotomy with small suture bites andlittle compromise of the vessel lumen, thus reducing theneed for a patch graft.

The venous graft is harvested from the saphenousvein just anterior to the medial malleolus at the ankle. Asegment of vein 6-8 cm in length is removed, and thedirection of flow is maintained. The vein is opened alonga longitudinal seam by initially creating a small incisionwith the Potts scissors and advancing the partially openedscissors along the vein (Fig. 11). The distal end of thepatch graft is preshaped and the vein stored in heparin-ized saline until ready for use.

The saphenous vein patch graft is sewn into placeunder the operating microscope with a 6-0 runningdouble-armed Prolene suture (Fig. 12). The initial sutureis critical and is placed through the apex of the graft andmost distal point of the arteriotomy. This suture shouldbe placed from the exterior of the graft to the interior andfrom the intimal surface of the internal carotid artery tothe adventitial surface. Once the distal half of the grafthas been sewn into place on both sides, the proximal endof the graft is shaped to conform to the proximal portionof the arteriotomy.

ShuntingAttitudes toward the use of shunts during endarterectomyrange from those surgeons who routinely place shunts inall patients to those who do not monitor and never use ashunt. Our preference is to monitor patients with EEG,

use barbiturates to increase the brain’s tolerance for is-chemia, and selectively employ shunts in those patientswith asymmmetry in the EEG. Sundt and associates havehad outstanding results using intraoperative cerebralblood flow measurements and EEG to determine the needfor a shunt without the addition of barbiturates.

Barbiturate AnesthesiaMany surgeons perform carotid endarterectomies underlocal or general anesthesia without the use of barbitu-rates. The advantage of local anesthesia is the ability toneurologically monitor the patient during surgery. Bar-biturates reduce the metabolic requirements of neural tis-sue and have been shown to modify or prevent cerebralinjury from focal, reversible ischemia. They are most ef-fective if given prior to the period of temporary focalischemia and at a dosage that achieves burst suppressionof the EEG. For these reasons, we routinely administerthiopental sodium prior to carotid cross-clamping andmaintain burst suppression of the EEG with thiopentaluntil the flow is reestablished. Spetzler and associateshave advocated the routine use of barbiturates for carotidendarterectomy and reported a 1.5% morbidity and mor-tality in a series of 200 consecutive endarterectomies.

Despite the barbiturate-induced burst suppressionof the EEG, asymmetry of the recordings between the twohemispheres can be readily appreciated. This allows forthe placement of a shunt if a trial of induced hyperten-sion does not correct the asymmetry.

In our opinion, the added cerebral protection ofbarbiturates will provide the time necessary for a me-ticulous, unhurried endarterectomy. The often-re-ported barbiturate hypotension has not been a prob-


264


Figure 11 (Top). To prepare a saphenous vein graft, the vein isopened with Potts scissors. The Inferior margin of the vein isgrasped with forceps and the partially opened Potts scissors areadvanced to produce a truly linear incision (dashed line).

Figure 12 (Bottom). In closure of the arteriotomy with a saphe-nous vein patch graft, the distal end of the graft has been preshapedand is sewn into place under the operating microscope with a 6-0running double-armed Prolene suture. Sutures should be placed fromthe exterior of the graft to the interior and from the intimalsurface of the internal carotid artery to the adventitial surface.



265


lem in our patients. Although barbiturates may delay adetailed postoperative neurologic evaluation, we thinkthat the benefits far outweigh the risks.

Operating MicroscopeMany excellent surgical teams perform carotid endarter-ectomies using only loupe magnification and headlightillumination. We have found that the superior illumina-tion and magnification afforded by a microscope pro-vides exceptional visualization of the operative field,especially the distal internal carotid artery. After remov-ing the atherosclerotic plaque, the surgeon brings themicroscope into the field to complete a meticulous en-darterectomy by removing small fragments of plaque anddiseased intima. With the microscope one can inspect thecritical region of the distal internal carotid to be certainthere is a smooth internal surface. Arteriotomy closureperformed under the microscope is unequaled in our opin-ion. The sutures are placed with small close bites of aprecise thickness of the vessel wall. Such a closure avoidsstenosis of the internal carotid, reduces the need for apatch graft, and invariably results in a completely dryarteriotomy closure.

Intraoperative AngiographyTechnical complications of carotid endarterectomy includeresidual plaque, intimal flaps, stenosis, and thrombosis.Such defects may cause a perioperative stroke throughocclusion of the carotid artery, distal embolization, or de-layed recurrent stenosis. Some surgeons elect to performpostoperative digital or conventional angiography to de-tect such defects. Others use noninvasive carotid testing todetermine the presence of technical deficiencies.

With the availability of compact, portable, digital in-traoperative angiography equipment, we have found in-traoperative angiography to be simple and quite helpful.Technical errors are identified during the operation andcan be corrected immediately. Once the arteriotomy isclosed and flow reestablished, a 19-gauge butterfly needleis placed into the common carotid artery proximal to theendarterectomy site. After the fluoroscope is put into place,about 10 ml of 60% meglumine iothalamate is manuallyinjected as rapidly as possible. If intraoperative angiogra-phy equipment is not available, a standard, single-shotconventional x-ray film exposed during contrast injectionwill provide adequate detail. If a large x-ray plate is posi-tioned under the neck and head, an adequate film of thecervical and intracranial vessels can be obtained.

COMPLICATIONS OF CAROTIDENDARTERECTOMYAny therapeutic advantage of a carotid endarterectomyis easily negated if the morbidity and mortality are ex-

cessive. A low perioperative morbidity and mortality ratein the range of <4% is mandatory for any potential ben-efit of the operation to be realized. Unfortunately, theserates range from 1.5% to more than 20% in the literature.

Complications may result from technical errors inperforming the operation or as a consequence of the un-derlying systemic atherosclerosis these patients frequentlyhave. Unfortunate consequences of carotid endarterec-tomy may be divided into non-neurological and neuro-logical complications.

Non-neurologic Complications

Cardiac IschemiaThe atherosclerosis necessitating carotid endarterectomygenerally reflects systemic vascular disease, includingthat of the coronary arteries. Myocardial infarction is themost serious systemic complication following carotid en-darterectomy and accounts for the majority of postopera-tive deaths. In the study by Fode and associates of 1234carotid endarterectomies, the incidence of myocardialinfarction was 4.9% in those patients who had a signifi-cant cardiac history, although it was much lower in thosewho had no such history. Thus, close perioperative moni-toring of cardiac function, including pulmonary arterycatheterization. as well as strict avoidance of severe hyper-or hypotension, is necessary in those with a significantcardiac history.

Wound InfectionWound infection following endarterectomy is uncommon.Staphylococcal organisms are usually the offendingagents. This complication is best managed byreexploration of the wound with irrigation and debride-ment. If the infection appears to involve deeper fasciallayers, an arteriogram should be performed to excludethe possibility of a false aneurysm.

False AneurysmThis rare complication should be considered wheneverthere is a persistent hematoma or infection. The diagnosisis made by arteriography, and treatment consists of exci-sion of the aneurysm wall with repair of the artery, using asaphenous vein patch graft. This complication is more com-mon when synthetic patch grafts are used to close the arte-riotomy, especially if a deep wound infection occurs.

Wound HematomaPostoperative swelling of the wound may be due tohematoma or lymphatic fluid. This is usually a self-limiting process that will resolve in a few weeks. How-ever, if airway compromise (usually due to hematoma)is present, the wound should be reexplored, the fluid

266


collection evacuated, and the source of bleeding controlled.

Neurologic Complications

Embolization or ThrombosisDistal embolization may occur at any time during or afterthe operative procedure. The highest risk for this compli-cation is during dissection and manipulation of the ar-tery and placement of a shunt. Minimizing manipulationand dissection in the region of the plaque will reduce therisks of embolization. Placing the clip on the internalcarotid artery prior to occluding the common and exter-nal carotid arteries also reduces the chance of distal em-bolization, as does appropriate back bleeding prior tocompletion of the arteriotomy closure.

Another potential source of intraoperative or post-operative emboli or thrombosis is the newly exposedmedia of the artery, which is quite thrombogenic. Avoid-ing reversal of the heparinization may allow the forma-tion of a platelet “pseudoendothelium” and may decreasethe incidence of this problem. Perioperative treatmentwith aspirin may also reduce the incidence of embolicand thrombolic complications.

Ischemia during Carotid OcclusionMost patients are able to tolerate the temporary carotidocclusion necessary for performing an endarterectomy. Inthe Mayo clinic series, where cerebral blood flow was mea-sured during carotid endarterectomy, 8% of patients hadlevels below 10 ml/100 g/min. This level of cerebral bloodflow is below the threshold necessary for maintenance ofcellular integrity and will result in cerebral infarction.

We use barbiturates during the period of time thecarotid is occluded to protect the brain from ischemia. Ifthere is EEG evidence of ischemia, a shunt is placed toprovide collateral blood flow during the operation.

Postoperative HemorrhageThis rare complication may result from poorly controlled

postoperative hypertension, especially in the setting ofa high-grade carotid stenosis that has been opened.Sundt and associates have suggested that a normal per-fusion pressure breakthrough phenomenon, similar tothat observed in some patients undergoing resection ofan arteriovenous malformation of the brain, may be amechanism of postoperative hemorrhage. It is postu-lated that patients with a high-grade carotid stenosis arechronically hypoperfused; the intracranial vessels be-come dilated to maintain perfusion and lose the abilityto autoregulate. These chronically dilated vessels arethen unable to handle the restored perfusion pressureafter endarterectomy.

Reperfusion of a recent stroke with hemorrhage intothe infarct is another potential mechanism for this com-plication. Solomon and associates reported that 0.41%of patients undergoing carotid endarterectomy experi-enced a postoperative hemorrhage.

SeizuresThis rare complication of carotid endarterectomy is usu-ally seen within 1-2 weeks of surgery. It is postulated thatthis is due to edema secondary to relative hyperperfusionof the hemisphere. This complication is more common inpatients who have had a preoperative stroke.

Cranial Nerve InjuriesA variety of cranial nerves are exposed to potentialinjury during a carotid endarterectomy, including thegreater auricular, the marginal mandibular branch ofthe facial, the hypoglossal, and the superior and recur-rent laryngeal nerves. Most cranial nerve injuries aredue to traction and, unless inadvertently severed, willusually recover. If bilateral endarterectomies areplanned, it is advisable to assess the function of thevocal cords prior to the second operation, as bilateralvocal cord paralysis due to bilateral recurrent laryn-geal nerve injury is a serious complication that neces-sitates a tracheostomy.

267

TRANSSPHENOIDAL EXCISIONOF MACROADENOMAS OF THE

PITUITARY GLANDGEORGE T. TINDALL, M.D.

ERIC J. WOODARD, M.D.DANIEL L. BARROW, M.D.

INDICATIONSTranssphenoidal surgery is the preferred approach for thesurgical management of the majority of pituitary tumors,including macroadenomas. Indications fortranssphenoidal surgery over transcranial approaches inlarge pituitary tumors include extension into the sphe-noid sinus, associated cerebrospinal fluid (CSF) rhinor-rhea, and invasion and/or destruction of the sphenoidbone with multidirectional intracranial extensions. Thisapproach is also indicated when an intracranial opera-tion would carry excessive risk for a patient (e.g., an eld-erly person in poor health or a patient with severely com-promised vision). Occasionally, pituitary apoplexywarranting rapid decompression of the optic chiasm isanother indication for a transsphenoidal approach.

CONTRAINDICATIONSTranssphenoidal surgery is contraindicated when thepatient has an infectious process involving the sphenoidsinus, a suprasellar mass associated with a normal sellaturcica, or a “bottleneck” constriction between anintrasellar tumor and the suprasellar extension. Atranscranial approach may also be considered in patientswith signif icant intracranial tumor extension to thesubfrontal, retrochiasmatic, or middle fossa regions.

A conchal sphenoid sinus is not a contraindicationto the transsphenoidal operative approach as the use of ahigh-speed, angled drill allows the pituitary to be ex-posed safely.

PREOPERATIVE PREPARATIONAs with any operation, preoperative assessment of pa-tients undergoing transsphenoidal surgery is of paramount

importance. To determine general health and tolerance ofanesthesia, the patient’s cardiovascular, respiratory, re-nal, hepatic, and endocrine systems should be evaluated,and a history of allergies, present medications, and prioranesthetic complications should be taken.

In patients with normal pituitary-adrenal function,100 mg of hydrocortisone is routinely administered justbefore surgery; 50 mg is added to each liter of intrave-nous fluid during surgery, and 50 mg is given by mouthor intramuscularly in the recovery room and continuedevery 8 hours through the second postoperative day.Thereafter, the steroid dosage is tapered and subsequentlydiscontinued or maintained at the level necessary for thepatient’s individual needs.

To reduce the risk of infection, a second-generationcephalosporin (e.g., cefuroxime) is also given just priorto surgery and continued for 24 hours postoperatively,

After the patient is anesthetized, an indwelling lum-bar subarachnoid catheter or lumbar puncture needle isinserted and connected to a closed sterile drainage sys-tem. Pituitary tumors with suprasellar extension usuallyrequire withdrawal of CSF or the infusion of sterile salineto aid in removing the portion of the tumor above thesella. In the case of an intraoperative tear of the diaphragmssellae, the drain will aid in keeping the operative fielddry during closure.

POSITIONINGIn transsphenoidal operations, the endotracheal tube,esophageal stethoscope, and temperature probe are tapedtogether and brought out the left side of the mouth. Theendotracheal tube must be taped securely to the face. Nooral airway is used.

Figure 1 provides an orientation for subsequentillustrations of the transsphenoidal approach. In the© 1991 The American Association of Neurological Surgeons

268


usual case, the skull clamp is placed with the pins in thetemporal bones above the squama just below the parietalbossing (Fig. 2). If intraoperative x-ray examination orfluoroscopy of the sella is anticipated, e.g., with a presellaror conchal type of sphenoid sinus, a single pin is placedfrontally, and two rear pins are placed in the occipitalarea. This allows free access to intraoperative x-ray ex-

amination should the surgeon want to verify intraopera-tive location and monitor position of instruments in thesella and the suprasellar area (Fig. 3).

The head is tilted to the patient’s left and elevatedabout 10° from the operating table with the surgeon stand-ing on the patient’s right side.

Figure 1. Orientation for illustrations of transsphenoidal ap-proach for macroadenoma resection. The view is axial, with

some of the subsequent insets depicting the procedure in a coro-nal plane.


269

TINDALL ET AL. : TRANSSPHENOIDAL EXCISION OF MACROADENOMAS OF THE PITUITARY GLAND

Figure 3. Lateral intraoperative skull film show-ing the position of a metal marker identifyingthe floor of the sella.

Figure 2. The patient is positioned with the head tilted to the leftand elevated slightly. The skull clamp is placed just below theparietal bossing.



270


TRANSSPHENOIDAL SURGICAL PROCEDUREAntiseptic solution (e.g., povidone) is applied to thenose and mouth, and the area is draped with steriletowels. The left lower quadrant of the abdomen is pre-pared and draped to obtain adipose tissue for later in-sertion into the sella turcica and sphenoid sinus. Aseparate set of sterile instruments is used to obtain theadipose tissue graft.

A small transverse incision is made in the upper gin-gival mucosa with a cutting cautery (Fig. 4). A cuff ofmucosa on the inferior gingival edge should be left to beused for later closure. The incision is carried down to themaxilla, and a Freer dissector is used to separate the softtissue in an upward direction to expose the piriform aper-ture, the floor of the nares, and the nasal mucosa.

With the cutting cautery in contact with the dissec-tor, the tissue over the superior cartilaginous nasal sep-tum is incised vertically (Fig. 5). Extending the incisionslightly into the cartilage facilitates a subchondral sepa-ration of the mucosa from the nasal septum on one side.

Once the mucoperichondral plane between the sep-tum and nasal mucosa is identified, it is developed infe-

riorly to the nasal spine using a small suction tip in theleft hand and a dissector in the right hand (Fig. 6). At thatlevel, the mucosa often adheres to the spine and consid-erable care should be taken to avoid a tear. The piriformaperture may have to be enlarged inferi-

orly and laterally with a small Kerrison punch.Next, the nasal septum is fractured at its base anteri-

orly with a small osteotome, allowing it to be displacedto the opposite side. One of the palatine arteries, which islocated anteromedially, may require coagulation at thispoint. Continued separation of mucosa from the nasalseptum in a posterior direction is accomplished using asmall sucker held in the right hand and a nasal speculumwith long, thin blades held in the left hand. The bladesare advanced and opened just enough to admit the tip ofthe sucker, which performs the actual separation of themucosa from the nasal septum. At the posterior limit ofthe cartilaginous septum, a thin bony septum comprisedof the superior portion of the vomer and inferior parts ofthe perpendicular plate of the ethmoid comes into view.

When the bony septum is in view, a speculum isintroduced between the septum and mucosa and

Figure 4. Incision in the gingival mucosa with electrocautery. The upper lip is retracted by an assistant.


271


Figure 5. The mucosa overlying the septal cartilage is cauterizedallowing a plane to be developed between the nasal septum andmucosa.

Figure 6. Method of separation of the nasal mucosa from theseptum using a sucker tip and Freer dissector. Separation is madeonly on one side of the septum, usually the left.



272


opened (Fig. 7). This opening will fracture the thin perpen-dicular plate of the ethmoid near or at its junction with thesphenoid crest allowing one blade of the speculum to swingacross the midline, thus exposing both sides of the anteriorwall of the sphenoid sinus. The anterior wall of the sphe-noid sinus with its characteristic “boat keel” appearance isa distinctive structure and is easily recognized. Althoughthe speculum is inserted unilaterally, displacement of thenasal septum to the patient’s right side makes the orienta-tion midline (if the surgeon is using the left side). Fromthis point on, the operating microscope is used with a300mm objective lens and × 12.5 oculars. The anteriorwall of the sphenoid sinus is usually thin and can be openedwith either Jansen-Middleton rongeurs, osteotomes, orKerrison punches. The opening into the sinus can be startedthrough the ostia of the sphenoid, located laterally at the 2and 10 o’clock positions.

The size of the opening in the sinus is made slightlylarger than the width of the speculum blades. The lateralextensions of the sphenoid opening are carried just farenough to visualize the most lateral aspect of the sella.The sphenoid sinus mucosa can be cauterized and shrunkusing an insulated suction cautery. The speculum tip is

advanced into the sinus and gently opened by hand only.This maneuver brings the surgeon closer to the sella andbrings the long axis of the speculum directly in line withthe center of the sella turcica (Fig. 8). It should be empha-sized that although advancing the tips of the speculuminto the sinus offers technical advantage, there is alsopotential danger of fracturing the sphenoid bone if forceis used to open the speculum. Thus, one should not useexcessive force (and certainly not a speculum-spreadinginstrument) when the tips of the speculum are positionedwithin the sphenoid sinus.

When the interior of the sphenoid sinus is visual-ized, recognition of the position of the sellar floor as itbulges into the sinus is straightforward. However, shouldany doubt exist, a lateral x-ray film of the skull can beobtained with a metallic marker such as a Kirchner wirelightly impacted into the anticipated position of the sellarfloor (Fig. 3).

An opening is made in the anterior wall of the sellawith a small osteotome placed on the sellar floor. Lightmallet taps will fracture the floor and initiate the open-ing. Slight twisting of the impacted osteotome facili-tates this maneuver. The opening is enlarged with a

Figure 7. Insertion and opening of the bivalved speculum posteri-orly fractures the thin, bony septum near or at the sphenoid crest

and exposes the anterior wall of the sphenoid sinus on both sides ofthe midline.


273


Figure 8. The anterior wall of the sphenoid sinus is opened and the flanged tips of a Cushing-Landolt speculum are gently placed within the sinus.


274


small punch laterally to the cavernous sinus and carotidbulge, inferiorly to the sella floor, and superiorly to theintercavernous sinus (Fig. 9). Care must be taken not totear redundant dura with the rongeur when opening thesella. A vertical incision is made in the dura with a bayo-net-handled scalpel having a No. 1 blade (Fig. 10), and aportion is excised in an elliptic manner using angledmicroscissors (Fig. 11). If possible, a blunt hook is usedto establish a plane between the dura and pituitary glandor tumor. The dural opening must not be extended too farsuperiorly or laterally as entry into the suprasellar recessor cavernous sinuses can occur.

Macroadenomas frequently erode the floor of thesella, appearing in the sphenoid sinus or extruding intothe sinus upon opening the dura. Usually the tumor isreadily visible and is removed easily with blunt ringcurettes, enucleators, and/or suction (Fig. 12). With largetumors, a thin, flattened pituitary gland can be identi-fied and spared. It is generally displaced posteriorlyand superiorly.

In cases of suprasellar extension, the tumor can beremoved after evacuation of the intrasellar portion. Pre-serving a cuff of tumor in the anterior and superior aspectof the sella provides a handle to manipulate the remain-ing suprasellar portion of the lesion. After the bulk of theresection has been completed, the diaphragma sellae andtumor capsule can be inverted into the sella by carefullyraising the intracranial pressure with a slow infusion ofRinger’s solution through the lumbar subarachnoid cath-eter. The tumor is then carefully peeled off the diaphragmwith gentle suction.

The tumor bed generally stops bleeding once all neo-plasm is removed. In the absence of any CSF leak, absolutealcohol can be applied to the resection bed for about 5 min-utes. This is an optional maneuver designed to destroy re-maining tumor cells. This should not be done, however, ifthere has been a tear in the diaphragma sellae. The tumor bedand sphenoid sinus are then loosely packed with an adiposegraft, which may be held in place with fibrin glue. The specu-lum is withdrawn, and one or two absorbable catgut

Figure 9. Removal of the sella floor with a small Kerrison rongeur.


275


Figure 10 Incision in the dura covering the pituitary gland and tumor.

Figure 11. Removal of an ellipse of dura using microscissors.



276


Figure 12. Resection of the tumor using a ring curette.


277


sutures are used to close the gingival mucosa. Soft rubbernasal airway tubes are placed in each nostril toreapproximate the nasal mucosa and are left in place for24 hours. The nasopharynx and mouth are suctioned well,and the lumbar subarachnoid catheter is removed beforethe patient is awakened.

OPERATIVE MORBIDITY AND MORTALITYSerious complications in patients undergoingtranssphenoidal procedures are fortunately uncommon.In several large series, the incidence of operative deathhas been reported as 0.5%. Most fatal cases are associ-ated with additional extenuating circumstances. Like-wise, the incidence of nonfatal complications ranges from1.1 to 13%, with an average of 3.6%.

Complications of transsphenoidal surgery may beclassified anatomically as parasellar, intracranial, andsphenoid/nasofacial:

PARASELLARCSF leakageHypopituitarismDiabetes insipidusCavernous sinus damageIntracavernous cranial nerve damageIntracavernous internal carotid artery damageHemorrhageFalse aneurysmCarotid-cavernous sinus fistula

INTRACRANIALHydrocephalusIntracranial hemorrhageHypothalamic damageMeningitisOptic chiasm or nerve damageCerebral vasospasmEmbolization

SPHENOID AND NASOFACIAL,SinusitisMucoceleFracture of hard palateFracture of cribriform plateNasoseptal perforationExternal nasal deformityDevascularization or denervation of teeth

Because these have been described previously instandard texts, only a few of the more serious and/or fre-quent complications will be discussed.

CSF RhinorrheaThe incidence of postoperative CSF leakage followingtranssphenoidal surgery depends primarily on the expe-

rience of the surgeon and the intrasellar pathologic find-ings. CSF leakage is common if the diaphragma sellae isdisrupted during tumor removal, or in the case of an ex-tensive macroadenoma directly eroding this structure.Some delayed CSF rhinorrhea possibly occurs as a resultof postoperative rupture of the diaphragma sellae whereit herniates into the space left by removal of the tumor.

To prevent rhinorrhea, the intrasellar cavity and sphe-noid sinus are filled with fat held in place with fibringlue. The sellar floor may also be reconstructed with sep-tal bone or cartilage, which helps to maintain the posi-tion of the fat graft and thus reduce the incidence of CSFleakage. Fat must be packed loosely to avoid producinga mass in itself. If the diaphragma sellae is tom or is inten-tionally opened during surgery, hyperventilation and/orlumbar drainage will diminish the CSF leakage duringclosure and permit a watertight seal.

Significant leakage of CSF usually manifests as asteady dripping of clear fluid, especially when the patient’shead is placed in a dependent position. If only a fewdrops of clear to yellowish fluid are present, this likelyrepresents nasal secretions or breakdown of the fat graft.If an adequate amount of fluid can be collected for quan-titative analysis of glucose, a value greater than 30 mg/dlis diagnostic for CSF.

The initial approach to a persistent postoperativeCSF leak should be conservative. One or more lumbarpunctures are performed and CSF is removed slowly to aslow a pressure as the patient’s headache will tolerate. Ifthe CSF leak persists, an indwelling spinal subarachnoidcatheter is inserted percutaneously and left in place for 3days, while the patient remains in bed with the headslightly elevated. The catheter is connected to a sterilereservoir, which is placed no higher than the lumbar punc-ture site. Should the fluid continue to leak following re-moval of the catheter, the transsphenoidal wound is re-opened, and the sella and sphenoid sinus are repackedwith adipose tissue.

Diabetes InsipidusThis may occur transiently or permanently followingtranssphenoidal surgery. Severe dehydration and elec-trolyte imbalance may result if the disorder is not promptlyrecognized. Treatment is with parenteral vasopressin(Pitressin) or intranasal desmopressin acetate (DDAVP).

Visual ComplicationsDamage to the optic nerves and/or chiasm may occurduring removal of tumor from the suprasellar region.Attempts to remove adherent tumor or the tumor cap-sule from the optic apparatus may devascularize thesestructures and produce an irreversible visual deficitfollowing infarction. This may be more common in pa-tients who have been previously treated by craniotomyor radiation or who have poor vision preoperatively.Direct damage to the optic nerves may result

278


from a fracture of the optic foramen and orbit producedby over-enthusiastic spreading of the bivalved speculumin the sphenoid sinus. Overpacking the sella with fat maycause pressure on the optic nerves and result in visualreduction; yet failure to leave a prop in the sella follow-ing removal of a large tumor leaves a cavity that possiblywould allow prolapse of the optic nerves and chiasm.

Pressure from a hematoma may damage the optic nervesand/or chiasm and may be reversible following promptremoval of the mass.

If vision appears to worsen postoperatively, an im-mediate computed tomography scan should be performedto identify a possible remediable cause, such as a he-matoma or overpacked sella.

279

COMPUTER-DIRECTEDSTEREOTACTIC RESECTION

OF BRAIN TUMORSPATRICK J. KELLY, M.D.

INTRODUCTIONIn order to resect an intra-axial tumor, a surgeon mustvisualize the tumor as a three-dimensional volume inspace and know where surgical instruments are locatedwith respect to that volume. However, there are severalproblems in the resection of deep-seated tumors whichincrease the risk and decrease the efficacy of surgery onthese lesions. A deep-seated subcortical tumor maybedifficult to find at nonstereotactic craniotomy. In addi-tion, a glial neoplasm may have an irregular geometricconfiguration and it is difficult to stay oriented withinthese irregular extensions. Moreover, the histologicboundaries between a glial neoplasm and the normal oredematous gray and white matter may not always be clearat an open surgical procedure. Since deep-seated tumorsare usually located in nonexpendible brain tissue, a sur-gical resection which strays out of tumor into brain pa-renchyma can result in severe neurologic complications.Computer-assisted volumetric stereotactic methods canbe employed to help a surgeon localize the tumor andstay oriented within it. The following discussion con-cerns the instrumentation and methods presently usedfor computer-assisted volumetric resection.

INSTRUMENTATION

Stereotactic FrameThe COMPASS stereotactic system (Stereotactic Medi-cal Systems, New Hartford, NY) was specifically designedfor volumetric tumor stereotaxis. It evolved from a stan-dard Todd-Wells stereotactic frame. The system consistsof a fixed arc-quadrant, three-dimensional slide, and re-movable headholder (Fig. 1). It can be fixed onto a semi-permanent base unit as shown in Fig. 2A, or mountedonto the lateral support rails of a standard operating table(Fig. 2B). In addition, data acquisition hardware (local-ization systems for computed tomography (CT), mag-

netic resonance imaging (MRI), and digital angiography(DA)) and computer support hardware and software arealso considered part of the system.

HeadholderThe headholder consists of a base ring, four vertical sup-ports, and skull fixation system. The headholder is fixed tothe patient’s skull by means of flanged carbon fiber pins.

Detachable micrometers are used to measure the dis-tance between the end of the carbon fiber pins and theouter face of the vertical supports (Fig. 3). This provides amechanism for accurate replacement of the frame if thedata acquisition and surgery are not performed on the sameday or if further stereotactic procedures are contemplated.

Arc-QuadrantThe 160-mm radius arc-quadrant attaches to horizontalarms which extend from the baseplate of the three-di-mensional slide. Probes and refractors are directed by anattachment on the upper face of the arc. The arc-quadrantprovides two angular degrees of freedom for approachtrajectories (Fig. 4): a collar angle (from the horizontalplane) and an arc angle (from the vertical plane).

Three-Dimensional Positioning SlideThe headholder fits into a support yoke of a three-dimen-sional slide which moves the patient’s head within thefixed arc-quadrant. Each axis of the three-dimensionalslide is moved by a computer-controlled stepper motor orby hand crank if desired.

Stereotactic coordinates on the slide are detected byoptical encoders on the x, y, and z axes which transmitthe coordinates to digital readout scales and to the com-puter. In addition, Vernier scales on each axis for directreading of stereotactic coordinates are provided as abackup to the optical encoders.

Computer SystemData acquisition, treatment planning, and stereotac-tic tumor resections are possible utilizing manual© 1991 The American Association of Neurological Surgeons

280


Figure 1. Components of the COMPASS stereotactic system whichinclude a CT/MRI-compatible stereotactic headholder (A), a ster-eotactic arc-quadrant (B), and a three-dimensional positioning slidesystem, (C).

methods for calculation of stereotactic coordinates andcross-correlation of target points between the differentimaging modalities. The computer is therefore not abso-lutely necessary. However, the computer saves a surgeona great deal of time in calculating target points, interpo-lating imaging-defined tumor volumes, cross-registeringpoints and volumes between CT, MRI, and DA, and inreal-time interactive image displays during the surgicalprocedure. The computer makes volumetric stereotacticprocedures practical and time efficient. At present, theCOMPASS stereotactic frame is supported by a Vicomimage processing system and Sun host computer. A DataGeneral MV 7800 is used for backup and data storage.

LaserThe carbon dioxide laser is very useful in deep tumorstereotaxis and provides several advantages. First, thelaser is convenient for removing tissue from a deep cavityand is relatively hemostatic. Second, the laser removestissue by a narrow beam of light and thus there is one lessinstrument which must be inserted into a narrow surgicalfield. Finally, the laser beam can be computer-monitoredand computer-controlled if desired.

Stereotactic RetractorsThe stereotactic retractor system comprises cylindri-cal retractors, dilators, and an arc-quadrant adaptor.The retractor is a thin wall hollow cylinder 140 mm inlength and 2 cm in diameter. Indexing marks on theretractor shaft are provided for measurement of inser-tional depth with respect to the stereotactic arc-quad-

Figure 2. The stereotactic system can be mounted on a semipermanent base (A) or fixed onto the siderails of a standard operating table (B).



281

KELLY : COMPUTER-DIRECTED STEREOTACTIC RESECTION OF BRAIN TUMORS

Figure 3. The stereotactic headholder is mounted to the patient’shead by means of flanged carbon fiber pins which are inserted intoholes drilled through the outer table of the skull into the diploë. Notethe circular base ring of the stereotactic headholder with indexing

marks at 5° intervals. Micrometer attachments are used to measurethe distance between the outer face of the vertical support and thecarbon fiber pin. This provides a mechanism for reproducibly replac-ing the frame for subsequent stereotactic procedures.

Figure 4. The stereotactic arc-quadrant provides two angular de-grees of freedom. These are a collar angle or angle from the horizon-tal plane, and an arc angle which is an angle from the vertical plane.The attachment on the face of the arc directs a probe or a retractorperpendicular to the tangent of the arc-quadrant. Thus, these instru-ments will always arrive at the focal point of the arc-quadrant irre-spective of the arc or collar angle. In the COMPASS system thepatient’s head is moved in x, y, and z axes to place an intracranialtarget point at the focal point of the stereotactic arc-quadrant.© 1991 The American Association of Neurological Surgeons


282


rant. The retractor cylinder is directed perpendicular to thetangent and toward the focal point of the stereotactic arc-quadrant. Dilators which fit inside the retractor cylinder are1 cm longer than the retractor. The distal end of the dilator iswedge-shaped and spreads an incision to the diameter of theretractor so that the retractor cylinder can be advanced. Theretractor is used to maintain exposure by creating a shaftfrom the surface of the brain to the superficial aspect of adeep-seated tumor. In addition, the retractor itself provides afixed stereotactic reference structure in stereotactic space towhich computer-generated slice reconstructions of the CT/MRI-defined tumor volume are related.

Accessory InstrumentsExtra long bipolar forceps with a shaft length of 150 mmare required to control bleeding in the surgical field whenworking through the stereotactic retractor. In addition,150-160-mm-long suction tips, dissectors, and alligatorscissors are also used.

Heads-Up Displayfor the Operating MicroscopeWe developed a system by which the image output of asmall video monitor mounted on the operating micro-scope is optically superimposed on the surgical fieldviewed through the microscope (Fig. 5). The computer-generated image displayed on the video monitor is scaled

by a system of lenses to the desired size. Thus, the sur-geon sees the actual surgical field with the computer-generated rendition of that field based on CT and MRIsuperimposed.

PROCEDURAL ASPECTS

Data Base AcquisitionThe CT/MRI-compatible COMPASS stereotacticheadholder is applied under neuroleptic sedation andlocal anesthesia. Micrometer measurements are recordedat the application procedure. The headholder can then beremoved following the data base acquisition and replacedin exactly the same position for the stereotactic surgicalprocedure at a later date.

The patient then undergoes stereotactic computedtomography, magnetic resonance imaging, and digitalangiography. Separate localization systems for CT, MRI,and DA attach to the base ring of the COMPASS stereo-tactic headholder (Figs. 6 and 7). These create referencemarks on CT, MR, and DA images from which stereotac-tic coordinates can be developed.

Stereotactic CT ScanningThe stereotactic headholder secures to a CT table ad-aptation plate on a General Electric 9800 CT scanner.The CT localization system consists of nine carbon

Figure 5. A “heads-up” display system forthe operating microscope. The computerdisplay is transmitted to the video monitorand then optically superimposed on the sur-gical field.© 1991 The American Association of Neurological Surgeons

283


Figure 6. Principle of stereotactic localization systems for planardata such as computed tomography and magnetic resonance imag-ing and for projection images such as digital angiography. The CT(A) and MRI localization systems consist of “N”-shaped localiza-tion devices which create nine reference marks on each CT or MRI

slice from which stereotactic coordinates can be calculated. TheDA localization system (B) produces 18 reference marks on an-teroposterior and lateral images which allow calculation of magni-fication cross-correlation between DA and CT and MRI, and calcu-lation of stereotactic coordinates from DA.

Figure 7. The MRI 1ocalization system which creates reference marks on axial, transverse, and sagittal MR images.



284


fiber rods arranged in the shape of the letter “N” locatedon either side of the head and anteriorly which createnine reference marks on each CT slice (Fig. 6A). Intrave-nous iothalamate meglumine is administered for contrastenhancement. Five-millimeter slices are gathered throughthe lesion, utilizing a medium body format.

Stereotactic Magnetic Resonance ImagingStereotactic MRI examinations are performed on a Gen-eral Electric 1.5-tesla Signa unit. The MRI localizationsystem consists of plates containing capillary tubes filledwith copper sulfate solution arranged in the shape of theletter “N.” Plates are arranged bilaterally, superiorly, an-teriorly, and posteriorly (Fig. 7). This allows sagittal, coro-nal, and transverse image data acquisition. Nine refer-ence marks are created on each sagittal, coronal, and axialMR image. Recently, MRI studies have been performedutilizing gadolinium diethylenetriamine pentaacetic acidcontrast. In general, axial slices are useful for interpola-tion of tumor volumes. Sagittal and coronal images areuseful in surgical approach planning.

Stereotactic Digital AngiographyThe stereotactic head holder fits into a DF table adapta-tion plate on the General Electric DF 3000 or 5000 digi-tal angiographic unit. Lucite plates which contain nineradioopaque reference marks located on either side of thehead, anteriorly, and posteriorly create 18 reference markson each anteroposterior and lateral DA image (Fig. 6B).The mathematical relationships between the fiducialmarks and their locations on the DA images are the basisfrom which stereotactic coordinates for intracranial ves-sels can be calculated, and stereotactic target points de-rived from CT and MRI can be displayed on angiographicimages. Digital angiography is performed utilizing a stan-dard femoral catheterization technique. Orthogonal and6° oblique arterial and venous phases are obtained inorthogonal and 6° rotated stereoscopic pairs.

Tumor Volume InterpolationThe archived data tapes from the CT, MRI, and DA ex-aminations are transferred to the operating room com-puter system. Beginning with the lowest CT slice, thesurgeon traces around the boundary of the tumor on eachcontiguous CT and MRI slice using the computer’s cur-sor subsystem, and deposits multiple points around theboundary using the deposit key on the mouse. In addi-tion, a single point located in the approximate geographi-cal center of the lesion on one of the CT or MRI slices isdigitized and retained as the reference target point. Thedigitized CT and MRI defined tumor contours are sus-

pended in the computer’s image matrix. Interpolated slicesare then created at 1-mm intervals. The computer thenfills in each of the digitized and interpolated slices with1-mm cubic voxels. This creates a volume in space. Sepa-rate volumes for the CT- and MRI-defined contours arecreated. The interpolated CT- and MRI-defined volumesare constructed about the reference target point and cal-culations output which will center this point in the focalpoint of the stereotactic arc-quadrant frame.

The tumor volume can be reformatted for any de-sired rotation of the patient in the stereotactic frame (0 =supine; 90° = right shoulder down, lateral decubitus; 180°= prone; etc.). This facility allows a patient to undergodata acquisition in the most comfortable position, (i.e.,supine) and to be operated upon in a position which willbe comfortable and convenient for the surgeon. The ac-tual numbers for patient rotation and arc and collar anglesare determined during surgical planning.

Surgical PlanningSurgical planning is done at the computer console afterthe CT, DA, and MRI data tapes have been loaded on theoperating room computer system. In the most ideal situ-ation, data base acquisition and surgery take place ontwo separate days; the stereotactic headholder is removedfrom the patient following data base acquisition and isreapplied at the time of surgery. Surgical planning canthus take place in a relaxed environment.

All solid intracranial tumors are resectible. The solidportion of glial tumors is defined by the volume of con-trast enhancement on CT or MRI. However, brain tissueis always damaged in open surgical approaches to anytumor which is not directly on the surface of the brain.The object of surgical planning is to select the approachto a tumor which traverses the most expendable braintissue. The actual direction of this approach is the surgi-cal viewline. The surgical viewline is defined by andexpressed in terms of patient orientation, and arc andcollar trajectory settings on the stereotactic instrument.The computer can illustrate this approach by means of ashaded graphics display. The CT and MRI tumor volumedisplays can also be presented as slices cut perpendicu-lar to this viewline.

Figure 8 summarizes the various approaches used inthe resection of subcortical intra-axial tumors. These in-clude transcortical, transsulcal, transsylvian, and inter-hemispheric approaches. Tumors located near the brainsurface should be approached from the closest cranialentry point possible. Subcortical approach trajectoriesto deep tumors should traverse nonessential brain tissuein a direction parallel to major white matter projections.

285


Figure 8. Summary of the various surgical approaches utilized inthe exposure of subcortical tumors. These include transcortical(A), transsulcal (B), transsulcal (C), transsylvian (D), interhemi-

spheric transcingulate (E), and transcortical (F) which would uti-lize a deep white matter incision as well.


286


The brain surface anatomy must be established in or-der to plan a safe approach to centrally located and deep-seated lesions. In some instances, deep sulci may also beused as a route to expose some deep lesions. Sagittal MRimages are useful for the localization of specific fissuresand sulci. Stereotactic stereoscopic cerebral angiographyis also useful for finding the position of major sulci. Thesulci are localized by identifying the deep vessel segments(apparent on the stereoscopic view) on the orthogonal ar-terial and venous phases of the stereotactic angiogram.

In particular, the position of the central sulcus, pre-central sulcus, and precentral convolution must be estab-lished with respect to the tumor volume. The precentral

sulcus can be split microsurgically to approach laterallylying precentral lesions. More medially located lesionsunder the precentral convolution are approached from ananterior direction, transcortically through the superior ormiddle frontal convolution or through the superior fron-tal sulcus. Posterior approaches through the superior pa-rietal lobule are selected for the resection of lesions lo-cated behind the central sulcus. This approach isparticularly useful for exposure of lesions near the atriumof the lateral ventricle. In this latter situation, the patientis rotated to 180° (prone position).

Trajectories for deep-seated lesions located in thebasal ganglia or thalamus will depend in part on where

Figure 9. Three approaches are used for the exposure and resec-tion of thalamic tumors. The route chosen will depend on thelocation of the tumor with respect to the normal thalamic anatomy.A, anterior thalamic tumors are approached utilizing a coronaltrephination just behind the hairline in a trajectory which progresses

posteriorly and medially through the anterior limb of the internalcapsule. B, posterior dorsal thalamic tumors are exposed throughthe superior parietal lobule and lateral ventricle (patient rotation =180°). C, posterior ventral lesions are exposed through a corticaland white matter incision at the temporo-occipital juncture.


287


the lesion is located with respect to the noninvolved por-tions of these nuclear groups. The approach to variousthalamic lesions is illustrated in Figure 9. Anterior tha-lamic lesions are exposed by a retracted incision throughthe anterior limb of the internal capsule. Posterior dorsallesions are resected utilizing an approach through thesuperior parietal lobule and atrium of the lateral ven-tricle with the patient prone (180° rotation). Posteriorventral thalamic lesions are approached through a corti-cal and white matter incision at the temporal-occipitaljunction (patient rotation of 135° is used for left-sidedand 225° rotation for right-sided lesions).

Lesions in the anterior basal ganglia are approachedfrom anteriorly and superiorly. Lesions in the posteriorputamen are approached posterolaterally through theposterior temporal lobe.

Lesions in the posterior fossa are operated upon withthe stereotactic frame placed in an inverted fashion (Fig.10). Data acquisition is also performed with the headframe inverted. Midline lesions in the cerebellum or brainstem are exposed through the inferior vermis. However,midline approaches are uncomfortable for the surgeon ifthe patient is prone and the surgical approach is direct onthe midline (arc angle 0). It is therefore best to rotate thepatient 30°. This rotates the midline toward the surgeonso that he or she can stand to the patient’s side and oper-ate more comfortably (Fig. 11). Employing an arc angleof 30° returns the surgical approach to the midline. Com-

puter software recalculates the target point for patientrotation, so that the frame adjustments continue to placethe target point in the focal point of the arcquadrant.

Lesions of the cerebellar hemisphere or lateral ponsare approached in a lateral oblique trajectory from poste-rolateral to anteromedial. In this situation the patient isplaced prone (rotation 180°) in the stereotactic frame. Anarc angle of about 30-40° off the midline toward the sideof the lesion is used.

The surgical approach should be comfortable for thesurgeon. Patients can be rotated in the stereotacticheadholder in order to provide the most comfortable work-ing situation for the surgeon. Figure 12 illustrates thevarious patient rotations of the stereotactic headholderin the receiving yoke of the COMPASS three-dimensionalpositioning slide. The patient rotation selected will de-pend on the direction by which the surgeon intends toapproach the tumor.

At the end of the planning session, the computercalculates and outputs target stereotactic coordinates andarc and collar angles, and reformats tumor volumes whichaccount for patient rotation.

Figure 10. Inverted head frame position for resection of posteriorfossa neoplasms. This position is utilized for data acquisition andsurgery. It allows unencumbered access to the suboccipital area.

Figure 11. Setup for stereotactic resection of posterior fossa le-sions. Note the inverted position of the head frame and the posi-tion of the surgeon.



288


Figure 12. In the COMPASS stereotactic system the patient’shead can be rotated to any position which will provide a comfort-able working position for the surgeon. Only a few of these areillustrated here. Top left, the supine or 0° rotation is used foranterior approaches (shaded area). Top right, 180° rotation is usedfor posterior approaches. Bottom left, laterally lying lesions areapproached employing a 90° head frame rotation for left-sidedlesions and a 270° rotation for right-sided lesions. Bottom right,

any rotation angle can be selected. A 225° rotation is used forapproaches to the posterior medial temporal area as well as theposterior ventral thalamus. Rotations are also possible with theinverted frame position in the inverted head frame placement.The degree of rotation is aligned with an indexing mark on theyoke of the three-dimensional positioning system to confirm therotation. The computer rotates all target, volumetric, and trajec-tory calculations to account for the rotation used.

Surgical ProceduresGeneral endotracheal anesthesia is used. The stereotacticheadholder is replaced utilizing the same pin placementsand micrometer settings used during the data acquisitionphase. The actual position of the patient’s head depends onthe headholder rotation selected, and this will be fixed inthe stereotactic frame by placing the base ring of theheadholder into the receiving yoke of the stereotactic frame.The index marks on the receiving yoke should line upwith the degree index marks on the headholder at the de-sired rotation which was determined in surgical planning.

Four body positions are used to avoid compressionof the jugular veins by the angle of the mandible on thetransverse processes of the cervical vertebrae when thehead is turned to the desired rotation position.

Supine: The supine position is used for head frame (“pa-tient rotations”) between 60° and 300°.

Lateral Decubitus: This position is used for head framerotations between 60° and 120° (right side down) and

between 240° and 300° (left side down). The down-sidearm is extended forward. To maintain the position thebottom leg is bent at the knee, the top leg is straight. Apillow is placed between the legs.

Park Bench: The “park bench” or three-quarter proneposition is employed for all head frame rotations be-tween 180° and 240° (left side down) and between 120°and 180° (right side down). Here the patient lies on oneside of the chest with the dependent arm behind to theside and down. The knee closest to the operating table isbent and the other leg straight with a pillow separatingthe legs.

Prone: This is rarely used for head frame rotations of be-tween 150° and 210°. The patient’s chest is supported bybilateral laminectomy-type chest rolls. The patient’s bodymust be elevated by means of these rolls high enough sothat the neck is not hyperflexed. In most situations, how-ever, the park bench position is used instead of the fullprone position for posterior approaches.


289


Following positioning of the body and securing thehead frame in the receiving yoke of the stereotactic frame,the head of the operating table may be raised in order toplace the patient’s head above the heart to reduce thechance of a “tight brain.”

The patient’s scalp is prepared and draped as for aroutine craniotomy. The sterile arc-quadrant is then se-cured to the support arms of the stereotactic frame. Thetumor volume is positioned in the focal point of the ste-reotactic arc-quadrant. Entry and approach trajectoryangles are set on the arc-quadrant.

Intracranial shifts in the position of deep cystic tu-mors and tumors near the ventricular system may occurafter after skull and dura are opened, and cyst or ventricu-lar fluid drained. The following step is therefore recom-mended. The guide tube on the stereotactic arc-quadrantis advanced to mark the scalp; it is withdrawn and a stabwound is made in the scalp. The guide tube is advancedto rest on the outer table of the skull. The skull is thenopened with a stereotactically directed 1/8-inch drill andthe tumor is traversed with an open-ended biopsy can-nula. A series of (0.5 mm diameter) stainless steel ballsare deposited at 5-mm intervals along the viewline. Theposition of these markers is documented on anteroposte-rior and lateral stereotactic teleradiographs and will pro-vide reference points for any subsequent intracranial shiftsthat may occur. The position of the tumor volume is trans-lated accordingly within the computer image matrix totake this intracranial shift into account for subsequentimage displays and stereotactic coordinate calculations.

The scalp is then opened with a linear incision whichincludes the stab wound. The incision follows parallel tothe hairline in frontal approaches, is vertical in temporal,parietal, and posterior fossa approaches, and is usuallytransverse across the midline in parasagittal approaches.The skull is then opened using a 1.5-2-inch cranial tre-phine centered on the twist drill hole used to deposit thereference balls.

After removing the bone plug, the dura should bepalpated. If it is tight and nonpulsating, the anesthesi-ologist should ensure that the PaCO2 is 25 or less. Man-nitol and/or barbiturates may be necessary to reduce in-tracranial pressure. The dura is then opened in a cruciatefashion. At this point, the procedure varies for superficialand for deep lesions.

Superficial LesionsThe location of the circular trephine opening in space isknown. The edges of the cranial defect are used as a refer-ence to which the computer-generated slices of the CT-

and MRI-defined tumor volume can be referred.A section of cortex having the same size and con-

figuration of the most superficial tumor slice is removedwith bipolar cautery and scissors. We have found thatbrain tissue is nonviable when tumor extends to within 5-7 mm of the cortical surface. Resection of this overlyingbrain tissue will result in no new neurologic deficit. Aplane is then created around the tumor with bipolar cau-tery and suction. The computer displays the configura-tion of the trephine opening in relationship to the refor-matted tumor outlines. The depth of the surgicalinstruments below the trephine opening is measured andrelated to the “slice distance” of the image displayed onthe screen. Specifically, the “slice distance” is the dis-tance of the image plane which is perpendicular to thesurgical viewline from the zero point of the stereotacticarc-quadrant. The depth may be determined in the surgi-cal field by measuring the distance between the outerface of the probe holder on the arc-quadrant and the ex-ternal edges of the trephine opening.

The “heads-up” display on the operating microscopeis very helpful in the stereotactic resection of superficialtumors (Fig. 13). The microscope field of view is moveduntil the image of the trephine opening projected intothe microscope by the “heads-up” display unit is super-imposed over the actual trephine opening in the surgicalfield. The tumor outline images are used as a template sothat a plane between tumor and surrounding brain can becreated. The positions of surgical instruments in the sur-gical field viewed through the microscope are also posi-tioned in reference to the computer-generated imagealigned to the surgical field.

In this way, high grade gliomas, some low grade glio-mas, metastatic tumors, and vascular malformations canbe removed as intact specimens with minimal bleeding.

Deep LesionsAs with the procedure for superficial tumors, the stereo-tactic coordinates which position the tumor in the focalpoint of the stereotactic frame, arc, and collar trajectoryangles are set. A linear scalp incision is made, a 1.5-inchcircular cranial trephination is performed, and the durais tacked up and opened in a cruciate fashion. At thisstage a cortical incision at least 2 cm long must be made.This can be made in the crown of a gyrus or in the depthsof a sulcus. The incision is deepened into the subcorti-cal white matter.

The retractor cylinder is mounted in the arc adap-tor between the gently retracted gyral banks. The dila-tor is inserted and advanced until the dilator rests inthe depths of the incision. The retractor cylinder is

290


Figure 13. Method for stereotactic resection of a superficial tu-mor. The trephine opening has been placed stereotactically. Thecomputer displays the position of a tumor volume slice at a speci-fied distance along the viewline on a display monitor and into the“heads-up” display unit of the operating microscope (A). The

image is scaled until the configuration of the trephine opening inthe image display is exactly the same size and aligns to the actualtrephine opening. The surgeon then uses the tumor slice image asa template which will aid In the isolation of the tumor from sur-rounding brain tissue.

advanced 5 mm over the dilator. Then the dilator is re-moved. The operating microscope is used to view thedepths of the incision through the retractor. The subcorti-cal linear incision is extended 5 to 10 mm deeper utiliz-ing a focused CO

2 laser beam. The dilator is then rein-

serted and the cylindrical retractor advanced 5 mm further.The dilator is removed and the incision deepened further(Fig. 14, A-F).

Using this method a long subcortical incision fromthe surface of the brain to the tumor is made in small stepswhich consist of deepening the incision with the CO

2

laser and spreading the incision with the dilator overwhich the retractor is advanced and secured. At the outerborder of the tumor, the incision is undercut mediallyand laterally, reflecting the laser beam off a stainless steelinstrument inserted to the base of the retractor (Fig. 14, Gand H). The dilator is then inserted, rotated 90°, and theretractor is advanced and secured. Thus a shaft is createdfrom the stereotactic arc-quadrant to the superficial bor-der of the intracranial lesion (Fig. 14I)

The retractor maintains the surgical exposure andprovides a convenient reference for the depth of the ster-eotactic procedure since the radius of the arc-quadrantand the length of the retractor are known.

In addition, the operating room computer systemdisplays the configuration of the cylindrical retractor as

viewed by the surgeon (circle) in reference to reformattedCT/MRI-defined tumor slice outlines at a specified depthwhich usually corresponds to the insertional depth of theretractor. A calibrated millimeter reference grid is alsodisplayed (Fig. 15).

The computer calculates a range along the viewlinewhich indicates where tumor should first be encounteredand where it should end along that given trajectory. Slicesbeyond the target point (at the focal point of the arc-quadrant) are given positive numbers (e.g., slice distance= +5).

The stereotactic removal of the tumor is performeddifferently than in conventional tumor surgery. In con-ventional surgery the center of the lesion is decompressedand then the walls of the cavity are resected. However, instereotactic resection a plane of dissection between tu-mor and surrounding brain is established before decom-pressing the interior of the lesion. This prevents the wasof the cavity produced from closing in, rendering subse-quent stereotactic computer images inaccurate.

Small deep tumors having the same size incrosssectional area as the retractor are easily resected.with good postoperative results. A plane of dissectionaround the tumor is developed using a slightlydefocused laser beam (power about 25 to 30 watts, spotsize 1 mm). The dissection proceeds circumferentially


291


Figure 14. Use of the stereotactic cylindrical retractor. The sulcusis opened microsurgically. A, the sulcus is dilated utilizing gentlepressure on the dilator which is inserted through the retractorcylinder. B, the retractor is advanced to the depths of the sulcusand a cortical and subcortical incision is made with bipolar cauteryand microscissors. C, the incision is deepened using a carbon diox-

ide laser. D, this incision is spread with the dilator. E, the retractoris advanced over the dilator. The dilator is removed. F, the incisionis deepened with the laser. G, the superficial incision extends to thesuperficial border of the tumor. H, the brain tissue is dissected offthe tumor with laser or bipolar forceps. I, the retractor is advancedto the superficial aspect of the tumor.


292


Figure l5. Stereotactic resection of a deep tumor requires a stereo-tactically directed cylindrical retractor, operating microscope, andsurgical laser. The computer displays the configuration of the re-tractor with respect to its end and slices through the tumor volume

cut perpendicular to the surgical viewline. This is displayed on thecomputer monitor as well as in the “heads-up” display of theoperating microscope (A).

around each tumor level, progressing from the most su-perficial levels of the tumor to the deepest. The retractoris advanced as each level of 2 or 3 mm is dissected free ofsurrounding brain tissue until the widest cross-sectionaldiameter of the tumor is encountered. Then laser power isincreased to 50-80 watts, the beam is defocused (spotsize 2-3 mm), and the tumor is vaporized down to thelevel corresponding to the deepest insertional point ofthe retractor. Then, using the computer-generated sliceimage as a guide, the deeper portions of tumor are sepa-rated from surrounding brain using the laser beam at alower power and on a smaller spot size as described above.

Tumors much larger than the retractor opening canalso be removed with good postoperative results. Theretractor is first advanced to the most superficial aspectof the tumor and using the slice image as a guide, a planeis established between tumor and brain tissue. The retrac-tor position with respect to the tumor boundary is thenset so that a plane of dissection between tumor and sur-rounding brain can be established on one side of thetumor. This is accomplished by first translating the im-age of the retractor on the computer image display of thetumor in order to have one side of the tumor within theretractor. The computer then calculates new stereotacticcoordinates which will provide this situation with respectto the retractor. These new coordinates are executed auto-matically by the computer which activates the stepper

motors on the three-dimensional slide in order to providethese new stereotactic coordinates. The retractor mountis loosened so that the retractor can move freely duringthe movement of the patient’s head within the arc-quad-rant in order to place this selected part of the tumor at thefocal point of the arc-quadrant. When the movement hasbeen completed, the surgeon carefully tightens the setscrews of the retractor mount on the arc-quadrant and thecollar. This maneuver then orients the stereotactic retrac-tor toward the focal point of the stereotactic arc-quad-rant. The configuration of that edge of the tumor withrespect to the edge of the retractor is noted on the com-puter-generated image display before creating a planebetween tumor and brain tissue. The actual position ofthe laser in the surgical field is known in reference to thesuperimposed computer-generated tumor slice viewed inthe “heads-up” display unit of the operating microscope.After this side of the tumor has been separated from braintissue, attention is now turned to another quadrant of thetumor slice. The relationship between tumor and retrac-tor is again translated on the screen, new coordinatescalculated, the other side of the tumor shifted under theretractor, and this side separated from brain.

In this manner, a plane of dissection is developedentirely around the tumor to a depth of about 10 mmbeyond the end of the stereotactic retractor. Tumor isthen vaporized using 60-100 watts of defocused laser


293


power down to this level at which the plane of dissectionbetween the tumor and brain tissue had been established.The retractor is then advanced 10 mm into the cavityproduced and the process described above which devel-ops a plane around the tumor is repeated.

Intraventricular landmarks can be used to maintainsurgical orientation in conventional approaches to intra-ventricular tumors in patients having large lateral ven-tricles. More difficulty is encountered when attemptingto localize lesions in patients having small or normalsized ventricles. A more limited but direct approach tointraventricular lesions can be made stereotactically.Brain and ventricular incisions need to be only largeenough to remove the lesion. The surgical approach to alesion within the lateral ventricle will depend on its loca-tion with respect to the central sulcus. In general, anteriorlesions are approached anteriorly utilizing a scalp inci-sion behind the hairline, a 1.5-inch diameter trephine,and the superior frontal sulcus. Lesions located posteriorto the central sulcus are approached from behind throughthe superior parietal lobule.

Third ventricular lesions are approached through theright lateral ventricle. Colloid cysts can be removedthrough the foramen of Monro. Here it is important toapproach the more obstructed foramen of Monro. Coro-nal MRI is most useful in determining if the colloid cystis leaning forward and dilating one foramen of Monromore than the other. Very large third ventricular lesionscan be exposed through the frontal horn of the lateralventricle. One fornix must be sacrificed to gain the expo-sure. An internal decompression of the lesion is performedwith the laser until only a thin rim of the capsule remains.The computer display of the cross-sections of the digi-tized tumor volume are extremely useful in this step asthe surgeon can be quite aggressive within the tumorwith no risk of extending through the capsule and dam-

aging the walls of the third ventricle. Following this in-ternal decompression, the retractor is withdrawn to thelevel of the roof of the third ventricle and the capsule iscarefully dissected from the walls of the third ventricle.The tumor capsule can be contracted utilizing thedefocused laser which facilitates the dissection of thecapsule from the wall of the third ventricle.

CONCLUSIONComputer reconstruction of CT and MRI data and intra-operative display provides a mechanism by which a sur-geon is able to orient himself to the global lesional vol-ume and have precise feedback information as regardsthe location of the surgical instruments (stereotactic re-tractor and CO

2 laser) in relationship to planar contours

of the lesion displayed on a monitor in the operatingroom. With this method and instrumentation, aggressiveresection of subcortical lesions is possible with minimaldamage to surrounding brain tissue.

Three hundred sixty-two computer-assisted volumet-ric stereotactic resections for 196 gliomas, 104 nonglialtumors, and 62 nonneoplastic mass lesions were per-formed at the Mayo Clinic by the author in the five-yearperiod between August 1984 and August 1989. From thisexperience it is clear that patients having histologicallycircumscribed lesions such as pilocytic astrocytomas,metastatic tumors, intraventricular lesions, and vascularlesions derive the greatest benefit from this procedure.Postoperative morbidity also depends more on the de-gree of histologic circumscription than on the locationof the lesion. However, in high grade glial neoplasms amaximal reduction of tumor burden can also be achievedby computer-assisted stereotactic resection with betterpostoperative neurologic results than would be associ-ated with conventional procedures for lesions in centraland deep-seated locations.

294

SAGITTAL SYNOSTOSISA. LELAND ALBRIGHT, M.D.

PATIENT SELECTIONSagittal synostosis restricts transverse skull growth, andinfants develop a narrow, elongated skull (dolichoceph-aly, scaphocephaly) because the normal lambdoid andcoronal sutures permit growth in the anterior-posterioraxis. Most infants with sagittal synostosis have an un-usual head shape at birth and are diagnosed when theyare 2-4 months old. The typical child with sagittal synos-tosis has an elongated head with a palpable bony ridge inthe posterior half of the sagittal suture, biparietal narrow-ing, and occipital bossing. Narrow heads are also seen inpremature infants who lay with their heads turned to theside for long times, but these infants lack the typicaloccipital bossing of sagittal synostosis and the charac-teristic radiographic finding: skull fusion at the sagittalsuture, seen best on the Towne’s view. The diagnosis ofsagittal synostosis can usually be made on clinicalgrounds and can be confirmed by skull radiographs. Com-puted tomography scans, especially those in the coronalprojection, demonstrate the sagittal fusion and show com-pressed cerebrospinal fluid spaces lateral to the cerebralhemispheres, but are not needed to make the diagnosis.

Although a few infants with sagittal synostosis havesuch mild deformity that an operation is not indicated,almost all have an obviously abnormal head shape andan operation is indicated to normalize that shape. With-out an operation, their head shapes subject them to ridi-cule and name-calling. The optimal time for operation is3-4 months; at that age, their blood volumes are greaterthan in newborns, they are past the 8-10-week nadir ofphysiologic anemia, their skulls have greater pliabilitythan at 6-12 months, and their subsequent brain growthfurther improves the postoperative skull shape.

Neurologic development is usually normal, with orwithout an operation. The goal of operations for sagittalsynostosis is to normalize skull shape, not to preventneurologic damage. There is some evidence that intracra-nial pressure (ICP) is increased in infants with sagittalsynostosis: ICP measurements have been higher than

normal in several of the few monitored patients, infantsare often less irritable postoperatively than they werepreoperatively, and in nonoperated children with sagit-tal synostosis, skull radiographs later in childhood usu-ally demonstrate a beaten-copper appearance.

Operative risks include anesthetic risks, blood loss,infection, dura/sinus tears, and cortical injuries. The like-lihood of requiring a transfusion is related to the patient’sblood volume, to the compulsiveness of the surgeon touse techniques that minimize blood loss, and to the he-matocrit chosen for transfusion (usually 21-24; belowthat level infants develop tachycardia, tachypnea, dia-phoresis, and poor feeding). In my experience with 50operations for sagittal synostosis, approximately 25% ofthe patients needed transfusions intraoperatively orwithin 48 hours postoperatively. The risk of infection is<1% (unless bone edges are lined with Silastic). The riskof dura/sinus tears varies with the technique used but isabout 1-2%, and the risk of cortical injury is <1%.

PREOPERATIVE PREPARATIONNeurosurgeons should plan their operations to correctthe specific deformities harbored by the individual pa-tient. In my experience, it has not been possible to nor-malize skull shape-correcting the occipital boss and wid-ening the biparietal diameter-with only a midline stripcraniectomy. With current procedures, the use of Silasticand other interposition materials is unnecessary.

The infant’s hair can be shampooed and the scalpscrubbed the night before operation. A preoperative bloodcount should be obtained and blood should be typed andcross-matched. Antibiotics and anticonvulsants are notneeded.

OPERATIVE TECHNIQUE

AnesthesiaCommunication and trust between the neurosurgeonand the anesthesiologist are imperative because ofindividual preferences and abilities in these opera-tions. General anesthesia is used. There is no prefer-© 1991 The American Association of Neurological Surgeons

295

ALBRIGHT : SAGITTAL SYNOSTOSIS

red method of induction but anesthesia is usually main-tained with inhalation agents because they are more eas-ily reversed than narcotics at the end of the operation.Blood pressure can be monitored adequately with a cuff;arterial cannulation for continuous monitoring is notgenerally required for sagittal synostosis operations.These operations last less that 2 hours so an indwellingcatheter to monitor urinary output is unnecessary.

It is helpful to have two intravenous lines, usually22- or 24-gauge, so that blood can be infused rapidly ifneeded. The indications for transfusion are not alwaysclear cut; neurosurgeons and anesthesiologists shoulddiscuss the indications preoperatively, considering fac-tors such as preoperative hematocrit, estimated bloodvolume, ability to adequately oxygenate tissues, the na-ture of the planned operation, and the risks and benefitsof transfusion. Operative blood loss in these operationsis difficult to quantitate and transfusions are best begunwhen the estimated blood loss approaches that volumeneeded to decrease the hematocrit to 21-24. The risk ofclinically important venous air embolism is small andthe risk is not worth the additional time (often 45 min-utes) required to insert a central venous catheter. The useof Doppler monitoring, etc., is optional.

Operative PositioningInfants are positioned prone, with the chest elevated bychest rolls and the head resting face down on a foam-padded pediatric horseshoe head holder, with careful in-spection to ensure that there is no pressure on the eyes(Fig. 1). Antibiotic ointment should be applied to thecorneas and the lids taped closed; tarsorrhaphy suturesare not needed for these operations. The head is posi-tioned with the neck slightly extended, so that there isaccess anteriorly to the fontanelle and access posteriorlyto the undersurface of the occipital prominence. Neuro-surgeons who perform only a midline strip craniectomycan position the infant’s head in either the prone or lat-eral position.

Draping

The amount of hair to be shaved and the closeness ofdrapes to the incision is a matter of personal preferenceand ranges from shaving and exposing only a 1-cm mar-gin on either side of the incision to a total head shave andexposure of the entire dome of the calvarium.

Skin IncisionThe incision depends on the planned procedure. If theneurosurgeon plans to remove the sagittal suture and toalter the abnormal parietal or occipital regions, a trans-verse incision is used, positioned halfway between the

anterior fontanelle and the bottom of the occipital promi-nence (Fig. 1). If a midline strip craniectomy alone isplanned, a midline sagittal incision is used, extendingfrom the anterior aspect of the anterior fontanelle to justbehind the posterior fontanelle. Prior to incising the scalp,0.5% bipuvacaine containing 1:200,000 epinephrine canbe injected intradermally to diminish bleeding and de-crease postoperative, pain.

Operative ProcedureThe scalp edges should be compressed and retractedbackward while the epidermis is incised with a knife.Below the epidermis, the scalp can be opened with aneedle tip on the unipolar coagulating cautery or with

Figure 1. Patient positioning for correction of sagittal synostosis,vertex (top) and lateral (bottom) views. The Incision is markedhalfway between the anterior fontanelle and the bottom of theoccipital prominence (arrows).


296


the Shaw scalpel, with minimal blood loss. Most neuro-surgeons apply hemostatic clips to the scalp margins,and clips are needed if the scalp is not opened with thecautery. The scalp flaps are then elevated away from theunderlying periosteum, using the cautery to divide theareolar tissue and vessels between galea and periosteum.Blood loss is increased if the scalp flaps are elevated bythe traditional technique of digital dissection. Scalp flapsare mobilized anteriorly until the anteriorly fontanelle isseen and posteriorly until the bottom of the occipitalprominence is exposed (Fig. 2).

After the flaps have been mobilized, the first stage ofbone work is the sagittal strip craniectomy. Although thesagittal suture is not fused along the entire length of thesuture in some infants, it is advisable to remove the entiresuture during these operations; it is difficult to normal-ize head shape if a portion of the suture is left in place.The unipolar coagulating cautery is used to score theperiosteum where the skull is to be cut (Fig. 2). Perios-teum over the sagittal suture should be left in place; itsremoval gives no benefit and causes additional bloodloss. An Adson periosteal elevator is inserted at the poste-rior margin of the anterior fontanelle to separate the durafrom the bone. Parallel parasagittal cuts are then made inthe parietal bones. The width of the strip craniectomyvaries inversely with age, from approximately 2.5 cm in a2-month-old infant to 1.5 cm in a 12-month-old. Bleed-ing is less if the sagittal strip craniectomy is performedwith a high-speed craniotome, e.g., the B-5 attachment ofthe Midas Rex drill, than by piecemeal removal withrongeurs. The sagittal strip is elevated from anteriorly toposteriorly and the dura, which is usual invaginated intoa shallow cleft in the under surface of the bone, is dis-sected off with an Adson periosteal elevator (Fig. 3A).This dissection needs to be accomplished gently butquickly so that bleeding points on the dura, usually smallemissary veins, can be coagulated with the bipolar cau-tery. The bone edges are then waxed.

The second stage of the procedure addresses the bipa-rietal narrowing. That narrowing can be corrected by bilat-eral parietal wedge-shaped craniectomies and sutures.Wedge-shaped craniectomies are removed with the cran-iotome at the sites of maximal parietal narrowing, usuallyremoving wedges 1-1.5 cm at the base and 4 cm in length(Fig. 3B). Two holes are drilled on either side of the craniec-tomy edges and 2-0 polyglactin sutures are inserted throughthe holes and tightened to draw the bone edges towardeach other enough to visibly widen the biparietal diameterto normal and to shorten the anterior-posterior diameter by5-10 mm (Fig. 4, A and B). If the infant has no significantoccipital bossing, only these two stages are needed, and

the wound can then be closed. Alternative techniques toimprove biparietal narrowing include the reverse-pi craniec-tomy and, parietal morcellation.

Most infants have occipital bossing, however, and athird stage of the operation is indicated. I incise the oc-cipital periosteum with bilateral semicircular incisionsand then elevate it off the bone with the cautery, leavingperiosteum attached at its base below the occipital promi-nence. An Adson or Cushing periosteal elevator is theninserted at the junction of the sagittal and lambdoid su-tures and the dura is separated from overlying occipitalbone. The dura is adherent along the lambdoid suturesbilaterally, but rarely adherent at the torcular herophili. Acraniotome is then used to remove a circular disc of bonethat encompasses the occipital prominence and usuallymeasures 3-4 cm in diameter. Occipital dura should bedepressed away from the bone while it is being cut, tominimize the risk of a sinus tear (Fig. 3B). After duralemissary veins are coagulated and the bone margins arewaxed, the dura. is plicated with either interrupted su-tures or by coagulating the dura with the bipolar cauteryin parallel vertical rows 5-8 mm apart (Fig. 3C). The oc-cipital bone is then reduced in size by removing a 510-mm strip around its periphery and is repositioned with 2-0 polyglactin sutures (Fig. 4A). Occipital periosteum isthen drawn back up over the occipital bone and looselysewn in place with 4-0 polyglactin sutures. The alterna-tive technique, the reverse-pi procedure, removes bilat-eral rectangular parietal craniectomies just anterior tothe lambdoid sutures and parasagittal strip craniectomiesand then draws the occipital bone forward with suturesinto the parietal bone, anterior to the craniectomy chan-nels. That technique has the advantage of not removingbone over the sinus, but the disadvantage of leaving theoccipital prominence.

For the occasional child with marked preoperativefrontal bossing, that bossing can be reduced by removing1-cm rectangular strip craniectomies immediately poste-rior to the coronal sutures and inserting sutures to drawthe frontal bones posteriorly toward the parietal bones.

The traditional parasagittal craniectomy procedureis outdated; it is often followed by refusion and does notcorrect either the biparietal narrowing, the occipital promi-nence, or the occasional frontal prominence. With thecurrent procedure, there has been no refusion and thehead appearance has been normal in all children oper-ated on at less than 1 year of age. Children operated on at1-4 years have far less skull mobility, and the neurosur-geon must make additional cuts in the calvarium so thatthe bones can be put where they should be, without rely-ing on additional brain growth to move them there.

297


Figure 2. Scalp flaps have been elevated and the periosteum has been scored with the coagulating unipolar cautery where the skull is to be cut.


298


Figure 3. A, parasagittal cuts have been made in the parietal bonesand the strip craniectomy has been dissected off the dura. Smallemissary veins have been coagulated with the bipolar cautery. B,the dura has been dissected away from the occipital prominence

and the parietal bones. Parietal wedge craniectomies are removedat the sites of maximal biparietal narrowing. The occipital circularcraniotomy is removed and reduced in size as indicated by thedashed line. C, plication of the occipital dura. with bipolar cautery.


299


Figure 4. A, repositioning of the reduced occipital bone. Parietalsutures have normalized the biparietal diameters and shortened the

anteroposterior diameter. B, lateral appearance after cranial re-modeling is complete.


300


Closure TechniquesSubcutaneous tissues are irrigated and the scalp isthen closed with subcutaneous polyglactin suturesand external nylon sutures. A strip dressing is ap-plied along the incision and the head is wrapped witha 2-3-inch compression bandage to decrease scalpoozing. The scalp always swells substantially for 2-4

days. The use of subcutaneous drains is optional;their use is thought to possibly increase the risk of awound infection but no data support that postulate.The duration of the operation is generally 60-90 min-utes. The sagittal craniectomy channel f ills in withbone in approximately 2 months and a protectivehelmet is not needed postoperatively.

301

GLOSSOPHARYNGEAL RHIZOTOMYBURTON M. ONOFRIO, M.D.


INTRODUCTIONGlossopharyngeal neuralgia is characterized by parox-ysms of pain in the sensory distribution of the ninth cra-nial nerve. Except for the location of the pain and sen-sory stimuli which induce it, the attacks are identical totrigeminal neuralgia. The attacks are typified by a seriesof lancinating electric-like jabs of pain In the region ofthe tonsil or posterior third of the tongue. Radiation tothe external auditory meatus or angle of the mandible maymake it difficult to differentiate from trigeminal neural-gia involving the third division and pain arising from thenervus intermedius. Occasionally glossopharyngeal neu-ralgia and trigeminal neuralgia of the third division maycoexist and require surgical manipulation of both fifthand ninth cranial nerves. Glossopharyngeal neuralgiaoccurs with oneseventieth the frequency of trigeminalneuralgia and although trigeminal neuralgia may occa-sionally be bilateral, this author has never seen a case ofbilateral glossopharyngeal neuralgia.

The glossopharyngeal nerve is a mixed nerve. Thespecial visceral efferent fibers, which innervate the sty-lopharyngeus muscle of the pharynx, originate in thenucleus ambiguus. The general visceral efferent fiberswhich supply the parasympathetic innervation to the pa-rotid gland arise in the inferior salivatory nucleus and ter-minate in the otic ganglion. The general somatic afferentfibers supply the sensation to the back of the ear, theircell bodies are in the superior ganglion, and the centralconnections terminate in the spinal nucleus of the trigemi-nal nerve. The general visceral afferent fibers supply sen-sation to the carotid sinus, carotid body, eustachian tube,pharynx, and tongue. The cell bodies are in the inferior(petrosal) ganglion, and the central connections termi-nate in the tractus solitarius. The special visceral afferentfibers from the taste receptors of the posterior one-thirdof the tongue, in like manner, have cell bodies in the infe-rior (petrosal) ganglion and terminate in the tractussolitarius.

The glossopharyngeal nerve emerges from the me-dulla, dorsal to the inferior olivary nucleus, and passes

through the jugular foramen, in its cephalic portion, be-ing separated from the fibers of the tenth and eleventhcranial nerves by a distinct dural septum. The ganglia ofthe glossopharyngeal nerve lie within the jugular fora-men (Fig. 1A).

When dilemmas arise as to the nerve or nerves oforigin of the neuralgia, differential temporary blocks maybe employed. Cocainization of the pharynx alleviates theninth nerve component of the pain while cocainization ofthe pyriform fossa relieves neuralgia of the superior la-ryngeal branch of the vagus. Blocking the foramen ovalewith bupivacaine determines the component of the paindue to the third division of the trigeminal nerve. A tetra-caine block of the jugular foramen will block all afferentimpulses via the ninth and tenth cranial nerves and helpto discover that rare patient suffering from pain mediatedby the nervus intermedius component of the seventh cra-nial nerve. Obviously, for the blocks to be reliable theymust be done when the frequency of the attacks and ste-reotypic triggers are dependable enough to allow the phy-sician to appreciate an interruption in their occurrence inorder to determine the efficacy and dependability of theblock in sorting out which nerve or nerves are mediatingthe pain.

Although the medical treatment of trigeminal neu-ralgia may give gratifying results for years, this authorhas not observed the same degree of efficacy of medicaltreatment in patients suffering from glossopharyngealneuralgia. While intermittent repetitive alcohol blocks maybe very effective in control of the pain of trigeminal neu-ralgia, the extracranial anatomy of the ninth nerve makesalcohol block of the ninth cranial nerve impossible with-out incurring unacceptable tenth nerve dysfunction.

PREOPERATIVE CONSIDERATIONSIn determining the type of surgical treatment, the surgeonmust be aware of the clinical phenomena associated withthe hypersensitivity of the dorsal motor nucleus of thevagus which include cardiac arrest, syncopy, and seizures.Section of the ninth and upper fibers of the tenth cranialnerves causes little or no defineable neurologic deficit.Microvascular decompression risks intraoperative cardiacabnormalities

302


Figure 1. A, the head is turned 15° toward the side of the desired ninthnerve visualization. The inset to the left shows the initial craniectomywhich is then enlarged laterally to, but preferably not into, the mastoid

air cells. B, the ninth nerve is isolated with a ball-tip dissector. C, aftersectioning the ninth nerve, the upper one-sixth of the tenth nerve fila-ments are sectioned.


303

ONOFRIO : GLOSSOPHARYNGEAL RHIZOTOMY

and invites recurrence of pain postoperatively and seemsunwarranted. Authors have described hypertensive crisisand intra- and extracerebellar hemorrhage from ninth andtenth nerve manipulation during microvascular decom-pression.

Radiofrequency procedures for extracranial destruc-tion of the ninth nerve are unacceptable in that, like ex-tracranial alcohol blocks of the jugular foramen, they in-vite unacceptable tenth nerve dysfunction. Most surgeonsbelieve, as Dandy did, that there is no other treatment worthmentioning than intracranial section of the ninth and up-per one-sixth of the tenth cranial nerves at the jugularforamen.

SURGICAL TECHNIQUE

AnesthesiaThe sitting position offers some surgical advantages: easeof surgical exposure and less blood pooling in the opera-tive field. Some anesthesiologists believe that access to theendotracheal tube, the reduction of facial swelling, and theability to observe facial nerve function are notable advan-tages of the sitting position in anesthetic management.Hazards to the patient in the sitting position include venousair embolism, arterial hypotension, vital sign changes dueto brain stem manipulation, specific cranial nerve stimula-tion, airway obstruction, and position-related brain stemischemia. Preoperative identification of a patent foramenovale cordis may be a relative contraindication to the sit-ting position. Management should be directed at the pre-vention, early detection, and treatment of these problems.

At least two intravenous access sites are recom-mended for posterior fossa procedures. Whenelectroneurophysiologic monitoring is used and an awakebaseline is sought, the patient is not sedated prior to thecollection of those data. The only exception to that situa-tion is when an awake fiberoptic intubation is done andthere is a need to perform somatosensory evoked poten-tial monitoring after the endotracheal tube has been placed.

The anesthesia is induced, and an arterial line isplaced. Then, a central venous line is placed from the arm,using electrocardiographic localization, to the high rightatrium. Adequate anesthesia is administered for pinionplacement, with supplementation by local anesthesia ifnecessary. If transesophageal echocardiography is to beused, it is placed at this point.

The patient is then placed upright with care to avoidexcessive neck flexion in the sitting position. The head isthen turned 15° to the side of desired jugular foramenvisualization. The awake range of motion tolerated by thepatient is now used as a positioning guideline. The armsare carefully padded and, equally important, are carefully

supported. With the use of muscle relaxants the weight ofthe arms themselves is enough in some instances to causestretch injuries of the brachial plexus. The knees are flexedto avoid sciatic tension and the buttocks are padded toprevent pressure injury to the sciatic nerves. The Dopplermonitor is placed and the right atrial catheter is flushedvigorously with saline to confirm correct catheter place-ment. The blood pressure transducer is placed at the headlevel. Blood pressure requires some support with vaso-pressors or lighter anesthesia in about 25% of the cases.

Monitoring for air embolism can be approached fromseveral aspects. Monitors include a precordial Dopplermonitor, a right atrial catheter, a capnograph or mass spec-trometer, an esophageal stethoscope, transcutaneous 02assessment, and echocardiography (ECHO). The mostsensitive of these are the ECHO and Doppler, followedby end-tidal CO

2, transcutaneous O

2, expired N

2, right

atrial catheter, and (least sensitive) the esophageal stetho-scope.

Many surgeons use the park bench position for pos-terior fossa surgery including that for cerebellopontineangle tumor removal and microvascular decompressionof the fifth cranial nerve. Orientation from an anatomicpoint of view and familiarity with a specific positioningtechnique are an integral part of minimizing the risk ofany operation. For those surgeons familiar with the parkbench positioning, the description of the surgical anatomyand orientation of the illustrations are equally as valid asdescribed below for the sitting position by merely rotat-ing the illustrations 90°.

Operative TechniqueThe patient is placed in the upright sitting position in thepinion headrest, with the head flexed and rotated to theside of the glossopharyngeal neuralgia. A central rightatrial catheter for monitoring and aspiration of possibleair emboli is placed before the patient is placed in thesitting position. Either an S-shaped or a hockey stick-shaped incision is made over the ipsilateral occipital bone.The 3-cm craniectomy is done inferiorly to incorporatethe portion of the occipital bone which lies directly adja-cent to the foramen magnum and which is oriented in atransverse plane directly above the lamina of the first cer-vical segment. If a rongeur is used for the craniectomy,there is increased risk of epidural hemorrhage (Fig. 2, A-C). The high-speed drill obviates the need to introduce arongeur beneath the inner table of the occipital bone sothat there is less chance of a dissecting epidural clot in-tra- and postoperatively (Fig. 2D). The dural margin re-mains adherent to the edges of the craniectomy.

The dura is then opened in a cruciate fashion andtacked back over the craniectomy margin to the peri-

304


Figure 2. A, introduction of a rongeur dissects dura away beyond thelimits of the craniectomy. B and C, this may allow a dissecting epiduralhemorrhage to occur. D and E, a diamond drill (not shown) avoids thedura laceration potential of a cutting burr and allows tight dural bonecontact immediate to the craniectomy.

cranium or occipital fascia or muscle layer, preventing adissecting epidural hemorrhage (Fig. 2E). The cerebellarhemisphere is elevated to expose the arachnoid of the cis-terna magna, which is opened, allowing the egress of cere-brospinal fluid and allowing relaxation of the cerebellum.By identifying the sigmoid sinus as it traverses the poste-rior fossa floor, the surgeon can achieve precise retractorposition for identification of the jugular foramen. Oncethe self-retaining retractor has been fixed in position, il-lumination by the overhead surgical lights and operatingloupes is usually sufficient. The operating microscope mayenhance visualization at this point. The ninth cranial nervein the jugular foramen is always separated by a dural sep-tum from the tenth and eleventh cranial nerves and jugu-lar vein (Fig. 1B).

The ninth nerve and upper one-sixth to one-eighth ofthe filaments of the tenth nerve are sectioned with the aidof a black spatula or blunt hook and bipolar coagulation(Fig. 1C). The dural opening may then be closed by usingpericranium, fascia lata, or homologous dura as a graft,or by closing the dura primarily.

INTRA- AND POSTOPERATIVECONSIDERATIONSAlthough sensation is diminished over the pharynx andthe gag reflex is abolished on the side of the dividednerve, and although discrete neurological testing revealsabsence of taste over the ipsilateral posterior one-thirdof the tongue, we, like Dandy, have never noted morethan a transient disturbance in swallowing. With section-ing of the ninth and upper rootlets of the tenth cranialnerves, auricular flutter, tachycardia, hypertension, ec-topic ventricular contractions, and cardiac arrhythmiashave been noted. Most of these events are transient in-traoperative events.


305

OCCIPITOCERVICAL AND HIGHCERVICAL STABILIZATION

VOLKER K. H. SONNTAG, M.D.CURTIS A. DICKMAN

INTRODUCTIONArthrodesis of the atlas and axis is indicated when insta-bility of these structures becomes clinically or radiographi-cally significant. Atlantoaxial subluxation (AAS) canoccur suddenly from traumatic injuries such as upper cer-vical vertebral fractures or ligamentous disruption, or maydevelop progressively from rheumatoid arthritis, neo-plasm, infection, or os odontoideum. Occipitocervicalinstability occurs less frequently than AAS and is due tosimilar pathologic processes.

PATIENT SELECTIONThe decision to perform an atlantoaxial or occipitocervicalarthrodesis should be individualized, based on the neuro-logic symptoms and signs, the radiographic extent of AAS,the presence of fractures, the patient’s age and medical con-dition, the specific pathology, and the level of instability.Symptoms or signs of myelopathy from AAS are the mostclear indications for arthrodesis. Patients with occipitalradicular pain and persistent or progressive radiographicAAS are also surgical candidates. The least well-definedindication is isolated neck pain. In the absence of neuro-logic deficits, specific pathological and clinical featuresbecome more important to the operative decision. Patientsat high risk for nonunion of fractures or those with neuro-logic deficits from progressive AAS may be consideredfor arthrodesis.

The anatomical features of the atlantoaxial complexshould be specifically evaluated in each patient. The od-ontoid process, spinal cord, and subarachnoid space eachnormally occupies approximately 1 cm within the spinalcanal at the C1 level. These relationships are altered withAAS. The subarachnoid space may be effaced by thesubluxed vertebral segments and may be additionally com-promised by fracture fragments, inflammation, bony cal-

lus, basilar invagination of the dens, or other mass le-sions. These features enhance the vulnerability to injury,so that even minor trauma can precipitate deficits. Neu-rologically intact patients with more than 6 mm of AASor radiographic progression of AAS are at high risk forneurologic injury or sudden death. Individuals with TypeII odontoid fractures with more than 6 mm dens displace-ment have a high risk of nonunion (60-85%). In theseinstances, neurologically intact patients may benefit fromatlantoaxial arthrodesis.

Radiographic evaluation of patients with AAS shouldinclude anteroposterior, lateral, open-mouth odontoid, andoblique roentgenograms. If acute neurologic deficits oracute fractures are absent, lateral roentgenograms in flex-ion and extension will indicate the extent of AAS. Flexion-extension views are contraindicated with acute unstablefractures because neurologic deficits may be precipitatedor exacerbated. Thin-section computed tomography (CT)defines the static relationships of bony abnormalities andthe extent of fractures more precisely than plain x-ray films.In selected cases, three-dimensional CT enhances the vi-sualization of osseous pathology. Magnetic resonance im-aging (MRI) has replaced myelography for the evaluationof neural compression because of its superb anatomicalresolution, high sensitivity, and multiplanar graphic dis-plays. Patients with unstable osseous lesions maybe placedin an MRI-compatible halo brace. MRI is most useful forassessing compressive pathology that requires treatmentbeyond arthrodesis. Flexion and extension MRI (Fig. 1)delineates neural compression and the extent of positionalreduction in patients with rheumatoid arthritis andcraniovertebral settling, Chiari malformations, or ligamen-tous AAS.

Patients with irreducible or partially reduciblecompression of the cervicomedullary junction shouldundergo decompression. If instability exists after thedecompressive procedures, arthrodesis is indicated.Anterior compression from basilar invagination of thedens or from other ventral lesions at the cer-© 1991 The American Association of Neurological Surgeons

306


Figure 1. Magnetic resonance imaging study of the craniovertebral junc-tion and upper cervical spine. Neutral, extension, and flexion viewsdelineate the extent of atlantoaxial subluxation. Note the effacement of

the subarachnoid space and the widening of the predental space withflexion.

vicomedullary junction is best approached transorally.Posterior decompressive approaches for Chiari malfor-mations or other lesions may influence the type of arthro-desis needed. Decompression and fusion may be per-formed simultaneously if necessary.

In most instances of AAS, we prefer to limit the fu-sion to the atlantoaxial vertebrae. C1-C2 fusion preservesmore mobility of the cervical spine compared withoccipitocervical fusion, and the rate of nonunion increaseswhen the occiput is included in the fusion. In most casesof AAS, the fusion can be confined to the unstable verte-bral segments. If C1 cannot be incorporated into the fu-sion, then occipitocervical arthrodesis becomes necessary.Other indications for occipitocervical fusion includeoccipitoatlantal instability or instability at multiple ver-tebral segments.

Surgery for combination fractures involving the at-las and axis is indicated only if the C2 fracture is unstableor has a high risk of nonunion. In these instances, theextent of the fusion is dictated by the type of C1 fracture.If the atlas fracture involves the anterior portion or lateralring of C1, then atlantoaxial arthrodesis may be performed.If the C1 fracture involves the posterior ring, then the at-las cannot be wired; an occipitocervical arthrodesis or aC1-C2 screw fixation is required.

PREOPERATIVE PREPARATIONThe halo brace is the most rigid form of external immo-bilization for stabilizing the upper cervical spine. Preop-

erative halo placement ensures optimum alignment ofsubluxed segments and avoids the risk of neurologic in-jury from cervical manipulation during intubation. A pa-tient in a halo vest should be intubated while awake withnasotracheal, bronchoscopic, or fiberoptic-guided tech-nique. Somatosensory evoked potentials are monitoredperioperatively. If an occipitocervical procedure isplanned, brain stem evoked potentials are also monitored.Individuals with neurologic deficits are begun on a corti-costeroid preparation and an H2-receptor antagonist atleast 24 hours preoperatively.

A patient in the halo brace is placed prone on theoperating table. Rolls are positioned to support thechest, to allow adequate respiratory excursion, and toprevent abdominal compression to avoid epiduralvenous distention. With a Mayfield headholder adap-tor for the halo brace (Fig. 2), patients can be securedto the operating table with minimal difficulty. Alter-natively, stacked weights can support the anterior barsof the halo brace (Fig. 3). The back plate of the halovest may be removed to access the posterior cervicalregion or iliac crest; however, this is often not neces-sary. The posterior bars of the halo vest function asexcellent hand rests when left in place. Securing pa-tients to the operating table with wide cloth tape per-mits intraoperative table adjustments. Padding avoidscompressive neuropathies. The occipital and cervicalregions are shaved. Waterproof adherent plastic drapesprotect the evoked potential monitoring


307

SONNTAG AND DICKMAN : OCCIPITOCERVICAL AND HIGH CERVICAL STABILIZATION

Figure 2. The Mayfield headholder adaptor for the halo brace facilitates patient positioning.

Figure 3. The patient is positioned on the operating table using weights to support the halo brace anteriorly.



308


leads from moisture. Lateral cervical roentgenograms areobtained to confirm alignment after positioning, and astandard skin preparation is performed.

OPERATIVE TECHNIQUE

Atlantoaxial ArthrodesisA posterior cervical linear midline skin incision is made,extending from the inion to the spinous process of thevertebra prominens. The incision is deepened to the levelof the nuchal fascia. Sharply dividing the fascia and pos-terior cervical muscles in the midline plane allows a rela-tively avascular dissection, which is deepened until theposterior arches of C1 and C2 are exposed. The ring ofC1 and large spinous process of C2 are palpable land-marks; however, the exploration should be gentle to avoidtranslocation of C1. The periosteum of the ring of C1 andthe tip of the spinous. process of C2 are sharply incised,and the paraspinous cervical muscles are swept laterallyusing careful subperiosteal dissection with lightweight pe-riosteal elevators. This maneuver proceeds from the mid-line laterally and avoids exposing the vertebral arteries(Fig. 4). Self-retaining angled refractors are inserted tomaintain operative exposure.

The soft tissue is removed from the superior and infe-rior margins of C1 and from the superior margin of C2with a curette. Along the inferior ring of C1 and the supe-rior surface of the spinous process and laminae of C2, thepoints of contact of C1 and C2 with the graft are decorti-cated with a high-speed drill or Kerrison rongeur (Fig. 5).A Kerrison rongeur is used to notch the inferior margin ofthe spinolaminar junctions of C2 to seat the wire (Fig. 6).

An autologous bone graft (4 cm long × 3 cm high) isobtained from the posterior iliac crest as a curvedtricortical segment. Additional cancellous bone is obtainedfrom the iliac crest for use in the fusion.

The graft is prepared by removing its rounded corti-cal edge with a Leksell rongeur, creating a bicortical cur-vilinear strut (Fig. 7). The graft is then fitted between C1and C2 to approximate the curve of the ring of C1. Anotch in the inferior margin of the bone graft in the mid-line matching the contour of the spinous process of C2enables the graft to fit securely. The graft is temporarilyremoved to obtain wire placement.

Twenty-four-gauge, double-stranded wire (3 turns/cm) is halved, looped, and passed beneath the posteriorarch of C1 in the midline, directed superiorly (Fig. 8). Awire-passer or heavy silk may facilitate the passage. Thewire must be simultaneously fed and pulled using a “two-handed” process to avoid traction or manipulation of theatlas and to avoid displacing the wire anteriorly. The graftis then replaced into position between the atlas and axis.

The loop of the wire is passed over the ring of C1 andsecured under the base of the C2 spinous process. Onefree end of wire is passed below the spinous process ofC2, and the wires are tightened snugly with a wire twister(Fig. 9). The graft is affixed by wire anteriorly and poste-riorly and seated between the posterior-inferior arch ofC1 and the posterior-superior arch of C2.

Small areas of the ring of C1, laminae of C2, andcortex of the bone graft are decorticated with a high-speeddrill and covered with fragmented cancellous bone (Fig.10). The wound is closed securely, obliterating all deadspace. Muscle and fascia are closed with heavy absorb-able sutures in a watertight fashion. Skin closure withnylon suture is preferable.

The halo brace is maintained for 12 weeks postop-eratively. The halo ring is then disconnected from thebrace. If flexion-extension lateral roentgenograms indi-cate clinical and radiographic stability, the halo brace isremoved and a Philadelphia collar is applied. The collaris weaned according to patient tolerance.

Occipitocervical ArthrodesisPatient positioning, preparation, and incisions are identi-cal to those described for atlantoaxial fusion. The osseousexposure is more extensive, involving subperiosteal dis-section of the squamous portion of the occipital bone,rim of the foramen magnum, and C3 spinous process andlaminae. Two burr holes are created in the occipital bone,0.5 cm superior to the rim of the foramen magnum. Theburr holes are waxed for hemostasis, and the dura is sepa-rated from the skull using dural elevators, connecting eachhole with the foramen magnum. The foramen magnum isenlarged posteriorly with Kerrison rongeurs. Twentygaugewires are passed between each of the burr holes and theforamen magnum, and sublaminar at the levels to be fused(C1, C2, and/or C3).

A Steinmann pin, bent into a “U” shape, is fashionedto approximate the contour of the occipitocervical region.The wires are twisted, securing the Steinmann pin to theocciput and cervical laminae (Fig. 11). The threads of thepin prevent excessive mobility of the construct. The boneis decorticated with a high-speed drill, and bone fragmentsare laid upon the levels to be fused.

When the posterior ring of C1 has been removedor a suboccipital craniectomy has been performed, aplate of cortical iliac crest bone may be wired to thecentral portion of the Steinmann pin. This plate recre-ates the osseous contour of the occipitocervical regionand provides a template for the subsequently develop-ing fusion mass (Fig. 12). Closure techniques and post-operative care are identical to those described for at-lantoaxial fusion.

309


Figure 4. A, anatomical relationships of the suboccipital andposterior cervical regions. Note the course of the vertebral ar-tery at the C1 and C2 levels. B, illustration of the incision and

muscular dissection. The bony exposure is not carried later-ally enough to expose the vertebral arteries. (© 1990. B.N.I.)


310


Figure 5. (Top) Decorticating the surfaces of contact of the atlas andaxis with the bone graft. (© 1990, B.N.I.)

Figure 6. (Bottom) Notches are created at the inferior spinolaminarjunctions of C2 for wire seating. (© 1990, B.N.I.)



311


Figure 7. Preparation and fitting of the bicortical strut graft. (© 1990, B.N.I.)


312


Figure 8. (Top) Sublaminar wire passage at C1. The loop is passedsuperiorly beneath the atlas.

Figure 9. (Bottom) The fusion construct, demonstrating the relation-ships of C1, C2, bone graft, and wires.



313


Figure 10. Decortication of the fusion construct. The prepared regions are covered with bone fragments (indicated with dashed outline).


314


Figure 11. Occipitocervical fusion construct. The Steinmann pin is se-cured to the occiput and cervical laminae with wires. (© 1990, B.N.I.)

Figure 12. A plate of cortical iliac crest bone wired to the pin can beused as a template to support bone fragments. (© 1990, B.N.I.)


315


COMPLICATIONSWire breakage, inadvertent extension of the fusion mass,or nonunion may occur. The halo brace maintains rigidexternal immobilization, provides optimal conditions forformation of the fusion, and minimizes the risk of wirebreakage. Wire breakage is monitored with intraoperativeand postoperative x-ray films. Extension of the fusion isprevented by limiting the subperiosteal dissection to theosseous segments to be fused. If the occiput, C3, or othervertebrae are exposed, they may be inadvertently incorpo-rated into a C1-C2 fusion. This most commonly occurs inchildren. Osseous union is maximized by sharp subperi-osteal dissection, segmentally decorticating small areas of

the bone to be fused, and precisely fitting the bone grafts.Monopolar cautery devascularizes bone and may contrib-ute to nonunion. Excessive decortication may weaken thevertebrae. Fibrous union may convert to delayed osseousunion with additional halo immobilization. Nonunion mustbe managed with additional internal fixation.

Neural and dural injuries are prevented by a knowl-edge of each patient’s normal and pathologic anatomy, byavoiding traction or manipulation of unstable osseous seg-ments, and by looping the wires before their sublaminarpassage. Evoked potential and postoperative intensive careunit monitoring allow early recognition of neurologicdysfunction.

316

PETROCLIVAL MENINGIOMASOSSAMA AL-MEFTY, M.D.MICHAEL P. SCHENK, M.S.

ROBERT R. SMITH, M.D.

PATIENT SELECTIONPetroclival meningiomas remain challenging lesions,owing to their rarity, crucial location, insidious growth,and relentless natural progression leading to a fatal out-come. With the advent of advanced imaging, refined ap-proaches, microsurgical techniques, intraoperative moni-toring, and modern anesthesia and postoperative care, thepreviously dismal surgical outcome has been largely over-come. Meningiomas are benign tumors, and the treatmentobjective is total resection. This is best achieved duringthe first operation because dissection of neurovascularstructures is facilitated by intact arachnoid membranes.Although surgeons should pursue total removal with skilland zeal, their judgment should not be skewed from thegoal of preserving or improving neurologic functions. Thismay require surgeons to accept subtotal removal at times.

The size of the lesion is a significant factor influenc-ing surgical morbidity and mortality. Because these tu-mors are slow-growing, watchful waiting is justified inelderly patients who are asymptomatic, until evidence ofbrain stem compression appears. Recently, stereotaxicradiosurgery has offered an alternative for small or re-sidual tumors. The effectiveness and long term results ofthis modality, however, await further reports.

PREOPERATIVE PREPARATIONDetailed radiological studies are crucial for surgical plan-ning. Computed tomography (CT) scans and magneticresonance imaging (MRI), in coronal, sagittal, and axialviews, are obtained preoperatively for complete defini-tion of the tumor: its location, extension, relation to thebrain stem, encasement of vessels, and involvement ofthe cavernous sinus and the temporal bone (Fig. 1). An-giography remains a necessary preoperative diagnosticstudy, to identify a vascular lesion, to demonstrate cere-brovascular anatomy and displacement, to outline the

tumor’s blood supply, and to confirm the patency and theconnection between the two transverse sinuses (Fig. 2).MRI angiography may soon replace conventional angiog-raphy in this role.

Dexamethasone is administered preoperatively, andantibiotics are given intraoperatively. The authors’ choiceof antibiotics is a combination of vancomycin and a thirdgeneration cephalosporin.

ANESTHESIA AND MONITORINGGood anesthetic technique is essential for successful re-moval of petroclival meningiomas. Premedication is usu-ally withheld; induction is rapid and smooth and shouldbe accomplished with an agent that reduces intracranialpressure. Lidocaine, given intravenously after thiopentalinduction, decreases the intubation-induced hypertensiveresponse. The choice of anesthetic agents should be flex-ible and tailored to suit the circumstances of each case.Intracranial hypertension should be avoided and adequatecerebral perfusion maintained. The use of intraoperativemonitoring of cranial nerve and brain stem evoked po-tentials necessitates the use of certain anesthetic agentsor switching intraoperatively from one to another. Nor-motension is the goal, and hypotension should be avoided.Should temporary vascular occlusion be necessary dur-ing the surgical resection of the tumor, a barbiturate isgiven for its known cerebral protective effect.

Systemic monitoring is accomplished by means ofan arterial line, a double-lumen central venous line (withconfirmation of the location of the tip in the right atrium),a Foley catheter, a Doppler monitor, and an oximeter. Inaddition, end-expiratory PC%, arterial blood gas, elec-trolyte, hemoglobin, and hematocrit values, and coagula-tion profile are determined as necessary.

Brain stem auditory evoked potentials are recordedbilaterally as are somatosensory evoked potentials. Fa-cial nerve function is monitored by recording an elec-tromyogram (EMG) from several groups of facial muscleson the ipsilateral side.© 1991 The American Association of Neurological Surgeons

317

AL-MEFTY ET AL. : PETROCLIVAL MENINGIOMAS

Figure 1. MRI of a large petroclival meningioma. Various views assistin preoperative planning by delineating the tumor’s relationship to thesurrounding neural and vascular structures, and in the postoperativeperiod by confirming total removal. A, contrast enhanced coronal MRIof a petroclival meningioma depicting distortion of the brain stem. B,sagittal MRI of the same patient. C, postoperative contrast-enhancedcoronal MRI of the same patient demonstrating total tumor removaland the expansion of the previously compressed brain stem.


318


Figure 2. Angiographic views in three different patients revealing im-portant information for surgical planning. A, a vertebral angiogram, ofa large petroclival meningioma showing marked displacement of thebasilar artery to the opposite side, stretching of the AICA (arrows) andencasement of the basilar artery (open arrow). B, a lateral carotid arte-riogram demonstrating vascular feeders to a meningioma from theintrapetrous segment of the carotid artery. C, an anteroposterior angio-gram. in the venous phase in a patient with a petroclival meningioma.Notice the small left sigmoid sinus that is not connected at the torcularHerophili. Also notice the tumor blush.


319


OPERATIVE APPROACHSeveral approaches or a combination of approaches havebeen used to remove petroclival meningiomas. These in-clude the frontotemporal, occipital-transtentorial,subtemporal-transtentorial, suboccipital, combinedsubtemporal and suboccipital, combined subtemporaland translabyrinthine, transcochlear, and the combinedsuboccipital translabyrinthine approaches. The petrosalapproach described herein evolved from refinements ofseveral other techniques. It is centered on the petrousbone, allowing exposure of the tumor extending fromthe middle fossa to the foramen magnum. This approachis preferred because of the following advantages: (a) thecerebellum and temporal lobes are minimally retracted;(b) the operative distance to the clivus is shortened by 3cm; (c) the surgeon has a direct line of sight to the le-sion and the anterior and lateral aspects of the brain stem;(d) the neural and otologic structures, including the co-chlea, labyrinth, and facial nerve, are preserved; (e) the

transverse and sigmoid sinuses, as well as the vein ofLabbé and the basal occipital veins, are preserved, (f)the tumor’s vascular supply is intercepted early in theprocedure: and (g) multiple axes for dissection are pro-vided.

Patient Position (Fig. 3)The patient is placed supine with the patient’s head at thefoot of the operating table, allowing space and ease ofmovement for the seated surgeon. The table is flexed toallow 20-30° elevation of the head and trunk. The patient’sipsilateral shoulder is slightly elevated. The head is turnedaway from the side of the tumor, inclined toward the floor,and tilted toward the opposite side. The position of theneck is inspected to avoid compression of the contralat-eral jugular vein. The head is fixed in a three-pointMayfield headrest. During the operation, the surgeon’sline of sight can be altered by rotating the table from sideto side or up and down.

Figure 3. The position of the patient and the skin incision for a right-sidedpetrosal approach. EMG needle electrodes (arrows) are inserted into the

muscles innervated by the facial nerve. Inset, a skull model depicting theposition of the burr holes and outline of the bone flap.


320


Operative ProcedureCraniotomy Flap (Fig. 4)A reverse question-mark incision is made starting at thezygoma in front of the ear, circling above the ear, anddescending 1 cm medially to (behind) the mastoid pro-cess (Fig. 3). The skin flap is elevated and retracted ante-riorly and inferiorly. A large triangular pericranial flapwith an intact vascular base is elevated and retracted overthe skin flap to the level of the external ear canal. Thisflap is used to cover the drilled surface of the temporalbone at the time of closure. The temporal muscle then isretracted anteriorly and inferiorly while the sternomas-toid insertion is detached and retracted posteriorly andinferiorly. The bony surface of the temporal fossa, mas-toid, and lateral posterior fossa are thus exposed.

Four burr holes are made, two on each side of thetransverse sinus. A hole made just medial and inferior tothe asterion opens into the posterior fossa below the trans-verse sigmoid sinus junction, while a hole located at thesquamous and mastoid junction of the temporal bone,along the projection of the superior temporal line, opensinto the supratentorial compartment (Figs. 3 and 4). Theburr hole at each of these points will exactly flank thesigmoid sinus. The temporal bone and a portion of theoccipital bone above the tentorium, as well as the occipi-tal bone below the tentorium, are cut between burr holesusing the foot attachment of the Midas Rex drill. The burrholes flanking the lateral sinus are then connected usinga thin rongeur, or drilled with the B-1 attachment of the

Figure 4. The temporalis muscle (TM) and sternomastoid muscle (SM) areelevated and retracted on the right side. A pericranial triangular flap (F) iselevated and saved for later coverage of the drilled surface of the temporalbone. The position of the burr holes flanking the transverse sigmoid sinus isshown. A craniotome with foot attachment (1) is used to make the bony cutin the temporal and posterior fossae, while a drill (2) is used to cross over

the sinus. Inset, the bone flap has been removed, the dura of the temporalfossa (TD) and posterior fossa (PFD) have been exposed, the right sigmoidsinus (SS) has been skeletonized, and the petrous bone has been drilledextensively. The anatomical landmarks in the temporal bone (the facial ca-nal (FC) and the semicircular canals (SC)) are demonstrated.


321


Midas Rex drill. Particular attention should be paid toavoid tearing the wall of the venous sinus, which domesinto a bony impression on the inner surface of the skull.

The single bone flap is elevated, exposing the trans-verse and sigmoid sinuses. The bone adheres tightly tothe dura at the junction of the sigmoid and transverse si-nuses and requires careful dissection and elevation. Analternative to this method of skull-bone removal is to per-

form a temporal craniotomy, followed by a posterior fossacraniectomy extending over the transverse sinus.

Temporal Bone Drilling (Fig. 4, inset)The second stage, drilling of the temporal bone, re-quires a thorough knowledge of the anatomy of thepetrous bone and surrounding structures. A Zeiss op-erating microscope, mounted on a Contraves stand

Figure 5. Initial exposure via a presigmoid (retrolabyrinthine) route.The temporal dura is incised along the floor of the temporal fossa. Theposterior fossa dura anterior to the sigmoid sinus is incised toward thesuperior petrosal sinus. The sectioning of the petrosal sinus (area a) andthe dissection of the vein of Labbé (area b) are magnified in Insets aand b (bottom). Inset a, clipping and sectioning of the superior petrosal

sinus and the beginning of the tentorial incision. Inset b, dissection ofthe vein of Labbé in order to retract the temporal lobe and preserve thevein. SS, sigmoid sinus; PS, superior petrosal sinus; T, tentorium; TL,temporal lobe; PF, posterior fossa; TD, temporal dura; PFD, posteriorfossa dura; VL, vein of Labbé.


322


and equipped with an adjustable-angle eyepiece, is used.The angle of the eyepiece is changed as the field alter-nates between the subtemporal and suboccipital routes.The surgeon performs a complete mastoidectomy using ahigh-speed air drill. A diamond bit should be used whendrilling is close to vital anatomical structures. The sig-moid sinus is skeletonized down to the jugular bulb. Thesinodural angle, Citelli’s angle, which identifies the posi-tion of the superior petrosal sinus, is exposed. The super-ficial mastoid air cells behind the posterior wall of theexternal ear canal, as well as the deep (retrofacial) aircells, are drilled out to expose the facial canal and thelateral and posterior semicircular canals. Drilling is con-tinued along the pyramid to thin the petrous bone toward

its apex. The facial canal as well as the middle and innerear structures are kept intact, while opened air cells areobliterated with bone wax.

Exposure of the Tumor (Figs. 5-7)When the surgeon needs a shorter and more lateral access tothe petrous apex and clivus, the posterior fossa dura anteriorto the sigmoid sinus is opened along the anterior margin ofthe sinus. The incision is then extended upward toward asupratentorial dural incision made on the floor of the tempo-ral fossa (Fig. 5, Inset b). The temporal lobe is gently re-tracted. The vein of Labbé is preserved by dissection fromthe cortical surface, allowing retraction of the temporal lobewithout tension on the venous wall (Figs. 5 and 6).

Figure 6. Exposure of the tumor via a presigmoid route. The sigmoidsinus (SS) and cerebellum (C) are retracted medially while the temporallobe (TL) is retracted superiorly. The tentorium. (T) is incised along the

pyramid, through the incisura. The brain stem, cranial nerves (III-XI ),and tumor (Tu) are visualized. Inset, demonstration of tentorial section-ing along the pyramid toward the incisura. M, middle fossa floor.


323


Figure 8. Artist’s enhancement of an operative photograph showingremoval of a tumor via the petrosal approach after the tentorium hasbeen incised. The debulked tumor lies in front of the midbrain and pons.The IVth nerve is seen on the lateral surface of the brain stem and tu-mor. P, pons; M, midbrain; SCA, superior cerebellar artery; T, tumor; R,retractor on the temporal lobe.

cerebrospinal fluid (CSF). The tumor is furtherdevascularized by coagulating its insertion on the pyra-mid and the meningeal feeders over the tentorium. Whenthe tumor is small or moderate in size, the seventh andeighth cranial nerves are usually stretched posteriorly andthus are easily identified (Fig. 6). When the tumor reachesa large size, however, these cranial nerves may well beengulfed in the tumor.

A suitable area on the tumor surface is selected andthe arachnoid over the tumor is opened. The tumor is thendebulked with extreme caution since the VIIth and VIIIthnerves, as well as the posterior inferior cerebellar artery(PICA) and anterior inferior cerebellar artery (AICA),maybe embedded in the tumor. Debulking is performedusing suction, laser, Cavitron ultrasonic aspirator, and/orbipolar coagulation and microscissors.

The tumor capsule is then dissected free from thesurrounding structures. Maintaining dissection within thearachnoidal planes is crucial to preservation of the vitalneural and vascular structures. Cranial nerves and thebasilar artery and its branches may, however, be embed-ded in a large tumor, demanding meticulous and tedious

dissection (Figs. 10 and 11). A cut edge of the tumor shouldnot be allowed to slip away lest the plane of cleavage belost. Furthermore, gutting the tumor may be necessarybefore dissection of the thinned-out capsule can be con-tinued.

The lower cranial nerves are dissected off the infe-rior pole of the tumor. Gentle dissection is required toavoid hypotension and bradycardia from vagal stimula-tion. The VIIth and VIIIth cranial nerves are carefully dis-sected from the tumor. The sixth nerve, stretched anteri-orly and inferiorly, is dissected free from the tumor andfollowed distally (Fig. 12). It helps to visualize alternatelybetween the supra- and infratentorial routes while dis-secting the tumor capsule carefully away from the brainstem (Figs. 8, 9, and 12). If it is not embedded in thetumor, the basilar artery is usually displaced to the oppo-site side (Fig. 13). The preservation and careful dissec-tion of the main and small branches of the basilar arterycannot be overemphasized (Fig. 14).

Once the tumor has been excised, all neurovascu-lar structures in the posterior fossa are covered withwet surgical patties, and the area of tumor attachmentis vaporized extensively with the laser. If the tumorhas extended into the internal auditory meatus, themeatus wall is drilled and the tumor removed. Exten-sion through the jugular foramen is likewise removed.

Figure 9. Artist’s enhancement of an operative photograph showingtumor exposure via the retrolabyrinthine approach. Notice the directexposure of the lateral brain stem. P, pons; Pv, petrosal vein; T, tumor;V, trigeminal nerve; R, retractor; C, cerebellum.



324


Figure 10. Artist’s enhancement of an operative photograph showingthe removal of a meningioma encasing the basilar artery, vertebral ar-tery, and cranial nerves VII-XI. T, tumor; B, basilar artery; V, vertebralartery; VII-IX cranial nerves.

Figure 12. Artist’s enhancement of an operative photo graph showing theupper portion of the tumor dissected from the brain stem and cranial nervesV, VI, and VII. The pons and basilar artery are clearly seen. The lowerpart of the tumor is still present. P, pons; B, basilar artery; T, tumor; S,suction tip; V-VII, cranial nerves.

Figure 11. Artist’s enhancement of an operative photograph showingcranial nerves (VII, VIII) which were encased by the tumor dissectedfree with their accompanying nutrient vessels. Cranial nerves IX, X,

and XI have been dissected free from the lower portion of the tumor. T,tumor; VII-XI, cranial nerves. Arrow points to the PICA.




325


Figure 13. Artist’s enhancement of an operative photograph showingthe basilar artery displaced by the tumor to the opposite side. Cranialnerves V, VII, and VIII are displaced posteriorly and are easily visual-ized. Notice the tumor extension across the clivus. B, basilar artery; T,tumor; S, suction tip; V, VII, cranial nerves; SP, surgical patty.

Figure 14. Artist’s enhancement of an operative photograph demon-strating dissection of AICA off the tumor capsule. The VIth nerve andpons are clearly seen. T, tumor; A, AICA; P, pons; VI, abducens nerve;H, surgical hook; S, suction tip.

ClosureThe dura is closed in a watertight manner. The perios-teum is turned over the petrous bone to avoid a CSF leak,the temporal muscle is rotated over the defect and sewedto the stemomastoid muscle, and the soft tissues are closedin multiple layers.

COMPLICATIONSSurgery of petroclival meningiomas may entail all poten-tial complications of intracranial surgery. Surgical mor-tality, usually resulting from manipulation of the brainstem or interference with its blood supply, is high. It wasparticularly so in the premicroscopic era. Infarction ofthe lateral tegmental region of the pons results from anoccluded AICA. This complication is more likely to oc-cur when the collateral circulation is poor, and the risk iscompounded when the AICA loops deeply into the inter-nal auditory canal. Occasionally the appearance of defi-cit may be delayed during the postoperative period.

Temporal lobe swelling or hemorrhage is a gravepotential consequence of the subtemporal exposure, par-ticularly on the dominant hemisphere. It is caused by co-agulation or tearing of the vein of Labbé or the basilaroccipital veins. Every effort should be made to preservethese veins and minimize temporal lobe retraction. Like-

wise, cerebellar swelling and an intracerebellar hernatomamay follow excessive retraction of the cerebellum. Theretrolabyrinthine presigmoid avenue alleviates cerebellarretraction and minimizes the risk of cerebellar swelling.Cerebellar resection is seldom needed nowadays. Poste-rior fossa hematomas remain a frightening complicationbecause of the resulting rapid and direct brain stem com-pression. Venous hemostasis is deceptive when the headis elevated and the veins are collapsed. Thus, an inducedValsalva maneuver and jugular compression prior to clo-sure are essential to assure meticulous hemostasis.

Any of the cranial nerves (III to XII) are at risk duringsurgery of petroclival meningiomas. Injury to the trochlearnerve is a frequent hazard during tentorial splitting, be-cause of its fineness, fragility, and close relationship to theincisura. Morbidity resulting from its paralysis, however,is minimal compared to paralysis of other cranial nerves.Trigeminal nerve deficit is more morbid because of theresulting corneal anesthesia and subsequent keratitis, par-ticularly if the facial nerve is also paralyzed. In these cases,immediate tarsorrhaphy should be performed followed byreconstructive surgery to restore facial nerve function.Trigeminal nerve trauma may result in facial pain, anes-thesia dolorosa, and trigeminal neuralgia. Facial pain maydevelop months or years postoperatively.



326


Justifiable emphasis has been placed on facial nerveinjury. Anatomical preservation, however, does not nec-essarily mean functional preservation. Tumor size is oneof the most decisive factors in preserving facial nervefunction. Removal of large tumors is more likely to causeparalysis. The facial nerve is usually displaced posteri-orly by petroclival meningiomas, and may actually traversethe tumor. Intraoperative endto-end direct anastomosis ofthe facial nerve is feasible, with recovery occurring in40-80% of cases. Intraoperative nerve grafts may also beperformed; however, delayed hypoglossal-to-facial anas-tomosis is a more common procedure.

Although hearing loss usually exists preoperatively,normal hearing is not infrequent. In the latter instances,loss of hearing becomes a potential complication of sur-gery. Thus, the VIIIth cranial nerve and the inner earshould be preserved, particularly since improvement inhypoacusis has been reported. Lateral-to-medial retrac-tion of the cerebellum during posterior fossa. surgery ismore dangerous to hearing than retraction in a caudal-to-rostral direction. When preserving hearing, sparing thecochlear blood supply is as important as preserving thenerve itself.

While emphasis is given to facial nerve preserva-tion, a greater emphasis should be given to preservationof lower cranial nerves. Their deficit is a significantcause of morbidity and mortality. Injury to the lowercranial nerves may be troublesome both intra- and post-operatively. Intraoperatively, dissection of these nervesmay produce bradycardia and hypotension. Postopera-tively, dysphagia, vocal cord paralysis, and depressedcough and gag reflexes may lead to grave pulmonarycomplications which could be fatal. Careful dissectionof the inferior pole of the tumor and use of nonadherentpatties help protect the nerves.

Complications related to disturbed CSF dynamicsinclude CSF leakage, hydrocephalus, andpseudomeningocele formation. Although hydrocephalusmay be present prior to surgery and persist despite totalremoval of the mass, it may also develop postoperatively.Acute postoperative hydrocephalus is usually obstructiveand related to mass effect (tumor, edema, hemorrhage)whereas delayed hydrocephalus is usually communicat-ing and related to poor absorption of CSF or scarring ofthe basal cisterns. A CT scan is diagnostic, and the treat-ment is shunting.

CSF leakage is a significant risk in the petrosal ap-proach, occurring via the skin or through the middle ear.The leak is best avoided with the following maneuvers:watertight closure of the dura; applying bone wax to thedrilled surface of the temporal bone; and placing the peri-cranial flap over the drilled temporal bone surface.

When a CSF leak occurs, the incidence of meningi-tis is one in five. Antibiotic coverage during CSF leakageis controversial. Initial management of a CSF leak includeshead elevation, repeated spinal taps, or continuous spinaldrainage. If the leak does not cease in a few days, or if thearea has been previously heavily irradiated, cisternographywith water-soluble contrast can determine the leakage site,and may be followed by watertight dural closure with afascial graft. If hydrocephalus is an underlying factor,shunting is required.

Ligation of the sigmoid sinus is a step used by othersurgeons. Although thought to be inconsequential, it hasbeen associated with fatal complications. Assurance ofthe patency of the opposite sigmoid sinus and normalconnection through the torcular herophili is a prerequi-site to ligation of the sigmoid sinus. The above describedapproach preserves the sinus and eliminates the risk as-sociated with its ligation.

327

FACIAL REANIMATION WITHOUTTHE FACIAL NERVE

MARK MAY, M.D.STEVEN M. SOBOL, M.D.

INTRODUCTIONCombining a clinical experience with over 1430 facialreanimation procedures with well-establishedneuroscientific principles, the authors have establishedguidelines for the rehabilitation of facial paralysis whichstress restoration of facial symmetry and facial function.The choice of the most appropriate rehabilitation proce-dure is dependent on a number of factors, including: 1)cause of paralysis; 2) extent of paralysis and functionaldeficits; 3) duration of paralysis; 4) likelihood of recov-ery; 5) presence of concomitant cranial nerve deficits; 6)life expectancy; and 7) patient needs and expectations.

It is generally accepted that restoration of facial nervecontinuity is the preferred approach whenever possible,provided it can be established ideally within 30 days andnot greater than 1 year following injury. When the centralstump of the facial nerve is not available, and the facialmuscles can still be innervated, a hypoglossal-facial anas-tomosis is the favored approach, albeit not without draw-backs. This procedure is most effective when performedwithin two years of injury and not later than four years.However, when the facial nerve is unavailable, or whenthe facial muscles are deficient, other methods of facialreanimation must be selected. This might include situa-tions in which an attempt to preserve the facial nerve orgrafting has failed, and the critical time frame of two yearshas been exceeded. After this period, suff icientcollagenization of the distal facial nerve may prevent anysignificant amount of useful axonal reentry regardless ofthe proximal nerve’s ability to regenerate.

In the absence of the facial nerve, both dynamic andstatic procedures may prove useful. Dynamic procedureswhich we have found most valuable include regionalmuscle transposition using the temporalis muscle and oc-casionally the masseter, and certain eyelid reanimationprocedures including gold weight lid loading and

eyespring implantation. Static procedures serve predomi-nantly an adjunctive role, and will thus not be discussedin any great detail in this chapter. Those which we havefound most helpful include fascial and alloplastic slings,brow lift, rhytidoplasty, canthoplasty, and lid-tighteningprocedures. Our overall experience suggests that an ap-proach which 1) divides the face into upper and lowersegments (i.e., looks at the eye and mouth separately), 2)individualizes patient needs, and 3) combines both staticand dynamic procedures provides the best results in termsof optimizing regional reanimation.

EYELID (UPPER FACIAL) REANIMATIONThe goals of an ideal procedure aimed at reanimating theparalyzed facial muscles about the eye are to: 1) providecorneal protection; 2) avoid restriction of the visual field;3) be cosmetically acceptable; 4) restore dynamic blink;5) be reversible in the event of nerve recovery; and 6) betechnically reproducible. Our preferred procedures are thegold weight lid implant and the open eyelid spring. Bothtechniques are thought to be superior to traditional tars-orrhaphy which is often cosmetically unappealing, re-stricts the visual field, and fails, to reanimate dynami-cally. Our experience with over 280 gold weights and 250eye springs has allowed us to formulate criteria for pa-tient selection, and develop techniques which provide re-producible results.

Gold Weight ImplantationGold weight implantation is best for patients with paralysisassociated with minimal lid retraction, and those with re-versible or partial paralysis. Eyelid closure is augmented bythe force of gravity acting on the gold weight. Eyelid closureis therefore best when the head is in an upright or semiuprightposition. Implant extrusion and migration are rare if theweight is properly inserted and secured. The procedure ispotentially reversible and may be performed under local an-esthesia. Success in terms of satisfactory eyelid closure andpatient satisfaction is better than 90%. Occasionally,© 1991 The American Association of Neurological Surgeons

328


mild ptosis or bulging of the implant will be problematic.The operative field is prepared and draped in the usual

manner, topical anesthetic is placed in the eye, and a scleralshield is placed over the cornea for protection. One per-cent Xylocaine with 1:100,000 of adrenalin is injectedinto the tarsal-supratarsal fold. An incision is made witha razor blade knife in the fold about 1 cm in length (Fig.1A). The incision is extended through skin, subcutaneoustissue, and orbicularis oculi muscle and through the leva-tor aponeurosis down to and on top of the tarsus. A pocketis created to accommodate the gold weight (Fig. 1, B andC). A 1-g weight is sufficient for the majority of patients.At times a lighter (0.75 g) or heavier (1.2 g) gold weightis used. The optimal weight is determined prior to sur-gery by pasting the gold weight on the eyelid and select-ing the proper weight by trial and error. The gold weightis approximately 1 mm thick, 5 mm high, and 1 cm long;it has three holes in it. The gold is 24 carats and polishedso that there are no rough edges. The gold weight is placedin the pocket and secured with an 8-0 permanent monofila-ment suture through each of the three holes in the goldweight (Fig. 1D). The two upper openings are sutured tothe orbital septum and the lower one is sutured to the tis-sue just lateral to the tarsus. It should be noted that thesuture is not passed through the tarsus. Figure 1E demon-strates the gold weight in its position about 3 mm fromthe lash line and camouflaged under the supratarsal foldwhen the eyelid is open (Fig. 1F).

Open Eyelid Spring ImplantationThe open eyelid spring is most effective in patients withcomplete and/or permanent facial paralysis associatedwith significant lid retraction and/or a poor Bell’s phe-nomenon. When properly designed and positioned, thespring affords the best degree of restoration of normalcyto rapid, spontaneous, and voluntary blink, compared tothe gold weight. Potential problems with the spring in-clude: 1) the technical mastery required; 2) ptosis if thespring is open too far; 3) the bulge created by the Dacroncuff over the distal wire in thin-skinned individuals; 4)the increased difficulty associated with spring removalcompared with removal of the gold weight; and 5) theincreased extrusion rate in inexperienced hands. Despitethese potential drawbacks, success has been achieved inover 85% of carefully selected patients.

The operative field is prepared and draped in the usualmanner and 1% Xylocaine with 1:100,000 of adrenalin isinfiltrated along the supratarsal-tarsal upper lid fold. Theinjection is extended along the lateral orbital rim to theperiosteum. A razor blade knife is used to make an inci-sion through the skin, subcutaneous tissue, and orbicu-

laris oculi muscle to the tarsus (Fig. 2A). Bleeding is con-trolled with a bipolar forceps and the tarsus is uncoveredwith scissors. Next, the periosteum over the lateral or-bital rim is exposed through a separate incision (Fig. 2B).A spring is fabricated prior to surgery, when the patient isevaluated in the office setting. The spring is made of 0.01-inch round orthodontic wire; it is prepared as illustratedin Figure 2, C-F, using orthodontic instruments. The finalconfiguration is determined by making the spring con-form to the natural curvature of the patient’s orbit (Fig. 2,G-H). The proper tension is adjusted in order to provideadequate opening and closing of the eyelid. A 19-gaugespinal needle is passed just lateral to the tarsus within atunnel of soft tissue and brought out at the lateral orbitalrim pocket just lateral to the periosteum (Fig. 2, I-J). Thestylet from the spinal needle is removed (Fig. 2K). Thewire is then placed into the spinal needle and the spinalneedle is withdrawn (Fig. 2L). The lower limb of the springthat is now in place is secured with a 5-0 Supramid sutureto the periosteum (Fig. 2, M and N). Two or three suturesare placed around the fulcrum to the periosteum in thearea of the lateral canthus. This fixes the fulcrum to theperiosteum just above and lateral to the lateral canthus.The upper limb of the wire is looped and sutured withtwo 5-0 Supramid sutures along its shaft (Fig. 2O), andthen the lower limb is looped (Fig. 2P) and enveloped inDacron as noted in Figure 2Q. The Dacron is closed overthe loop with 8-0 monofilament sutures. The wound isclosed in layers using 7-0 Vicryl for the deep layer and 6-0 chromic catgut for the skin (Fig. 2R).

CommentsBoth the gold weight and eye spring implantation are pre-ferred techniques because they allow for eyelid closure,independent of mouth movement. Since neither the goldnor the spring relieves the lack of lower lid tone, paralyticectropion, and brow ptosis associated with the paralysis,each may be combined with static procedures aimed at cor-recting these defects. Our preferred method of lower lidtightening has been the Bick procedure or a modificationknown as a lateral tarsal strip. In either of these proceduresa lateral wedge of the lower lid is removed and a tongue oftarsus is developed, resutured, and anchored to the lateralorbital rim, thus shortening and tightening the lower lidto correct the paralytic ectropion. Care must be exer-cised in either technique not to excise too much lowerlid tissue since this can result in pulling the puncta awayfrom the globe, resulting in increased epiphora. In casesin which the lower lid tends to pull away from the globeresulting in increased epiphora, we have foundthat implantation of autogenous auricular car-

329

MAY AND SOBOL : FACIAL REANIMATION WITHOUT THE FACIAL NERVE

Figure 1. Gold weight implantation technique. (Modified with permission from May M. The Facial Nerve. New York: Thieme-Stratton, 1985).


330


Figure 2. A-R, eyelid spring technique. (Reprinted with permission from May M. The Facial Nerve. New York: Thieme-Stratton, 1985).


331



332



333


tilage to the midportion of the lower lid deep to the tarsusin a subconjuctival plane is a very useful procedure. Browptosis and supratarsal fold hooding may be corrected witha brow lift (with or without blepharoplasty) in an effort tofurther improve the quality of eyelid reanimation and cos-metic appearance. Caution must be exercised not to ex-cise too much tissue in this area, however, for fear of com-promising eyelid closure.

MOUTH (LOWER FACIAL) REANIMATIONIn the absence of a usable facial nerve, reanimation of themouth and lower face has been achieved using thetemporalis muscle transposition. It is useful in patients withlong-standing facial paralysis and has also been employedin selected cases to augment the results following a nervegraft or hypoglossal-facial nerve anastomosis. Our experi-ence with over 200 temporalis muscle transpositions hasyielded satisfactory results in approximately 90% of thecases. Early improvement in symmetry and a limited voli-tional smile may be achieved in three to six weeks withcontinued expected improvement with proper exercises overa period of one year. Motivated patients may further en-hance their results through motor sensory reeducation. Inapproximately 10% of patients, facial movements in re-sponse to emotional stimuli have been observed.

Temporalis Muscle TranspositionThe patient is prepared and draped in the usual fashion.The hair is parted and, using scissors, hair is removedthrough a narrow path over the region of the proposedincision in the scalp (Fig. 3A). The areas of the incisionsin the vermilion and in the scalp are infiltrated with 1%Xylocaine and 1:100,000 of adrenalin for hemostasis.After waiting 5-10 minutes, the incision in the scalp ismade with a cutting cautery down to the superficial mus-culoaponeurotic system (SMAS). SMAS is divided withscissors, exposing the fascia over the temporalis muscle.The temporalis muscle is elevated in its midportion in astrip approximately 4 cm wide extending from above thefasciaperiosteal attachment superiorly to the level of thezygomatic arch inferiorly (Fig. 3B). The cutting cauteryis used to outline the flap and an elevator to lift the muscle-periosteum from the skull (Fig. 3, C and D). The layerbetween the subcutaneous tissue and SMAS is identifiedand a pocket is made between the two with scissors (Fig.3E). This ensures that any residual function that might bepresent via the facial nerve will be preserved since thefacial nerve fibers lie in or deep to SMAS. Then an inci-sion is made in the vermilion exposing the orbicularisoris muscle (Fig. 3F). The layer just deep to the subcuta-neous tissue and lateral to SMAS is established using scis-

sors (Fig. 3G). The pocket started in the scalp lateral toSMAS and that in the Vermillion region are connectedwith scissors and then enlarged to accommodate two fin-gers (Fig. 3H). The temporalis muscle that was elevatedis bisected and 3-0 Prolene sutures are passed througheach of the pedicles in a figure eight fashion (Fig. 3I).The sutures are then brought through the tunnel using aKelly clamp (Fig. 3J). With the muscle pulled throughthe pocket (Fig. 3K), it is sutured to the submucosa (Fig.3L) and subcutaneous layer (Fig. 3M). This double clo-sure sandwiches the muscle between the submucosa andthe subcutaneous layers. The comer of the mouth is over-corrected in order to ensure a pleasing result (Fig. 3N).The overcorrected smile will begin to normalize over aperiod of three to six weeks. The defect created by trans-posing the temporalis may be refilled by implanting a softSilastic implant. A drain is placed through the scalp intothe cheek and hooked to wall suction for two days.

CommentsAlthough the temporalis muscle has been used by othersto reanimate the entire face including the eye region, wehave found that using it exclusively for the mouth offersbetter regional reanimation by separating eye and mouthfunction. A variety of modifications have been introducedto eliminate many of the problems encountered by othersusing this technique, which may have contributed to thelack of uniformly high-quality results. The prominentbulge in the cheek often noted by many performing thisprocedure has been reduced by creating an adequate tun-nel which allows the transposed muscle to lie flat. More-over, since only the middle one-third of the temporalismuscle is used, there is less bulk in the cheek and lesstendency toward temporal depression. The depressionwhich is created is remedied using a soft Silastic implantplaced at the time of transposition. Facial symmetry isfurther enhanced by close attention to proper positioningof the muscle slips to the mouth in an effort to achieve amirror image of the muscle pull seen on the normal con-tralateral side. Using Conley’s modification of raisingperiosteum attached to the superior aspect of thetemporalis muscle, rather than repositioning and sutur-ing the temporalis fascia to the muscle as proposed byRubin, has given more consistent results. The method ofattachment to the oral region has also been modified.Access is gained through a vermillion-cutaneous incisionrather than a nasolabial incision which further reducesthe visible scar. Multiple sutures placed in the muscle-submucosal layer at the level of the superior andinferiorlateral aspect of the orbicularis oris, creating asubstantially overcorrected smile have afforded the most

334


Figure 3. A-N, temporalis muscle transposition for lower facial reanimation. (Reprinted with permission from May M. The Facial Nerve. NewYork: Thieme-Stratton, 1985).


335




336


consistent results. The importance of overcorrection can-not be overemphasized.

Although the masseter muscle has been used alone orin combination with the temporalis muscle by some sur-geons, our experience has been less satisfactory. We havefound it less optimal in terms of vector forces, and, more-over, it adds little to the effect of the temporalis muscle.

Complications of the temporalis muscle transposi-tion have included hematomas, seromas, infection, andsuture granulomas. Using both a temporal-facial suctiondrain as well as a small suction lip-cheek drain reducesthe incidence of these complications. By avoiding pen-etration of the oral mucosa with the suspension sutures,the incidence of the suture granulomas is reduced. Pro-phylactic antibiotics are given routinely. Using the softMentor prefabricated temporal implant has reduced ourearly experience with implant extrusion noted with thehard carved Silastic implants.

The temporalis muscle itself creates a moderate de-gree of static suspension in addition to providing dynamicreanimation to the mouth area. However, if sufficient over-correction is not achieved at the primary procedure, or ifthe anchoring sutures pull out, the overall static suspen-sion effect may diminish with time. The temporalis muscle

may be tightened by placation utilizing the same vermil-lion incision or augmented with the use of fascia oralloplastic material (i.e., Gortex). When prestretchedGortex is utilized, the suspensory effect should be per-manent if the Gortex is securely anchored to a fixed bonystructure cephalad. We prefer a mini lag screw technique.A rhytidoplasty (face lift) may be combined with atemporalis muscle transposition in patients withlongstanding facial paralysis accompanied by loss of skintone and sagging.

CONCLUSIONSWithout a useful facial nerve, facial reanimation can onlybe achieved using methods other than nerve grafting andnerve substitution procedures. Separate eyelid reanima-tion using the gold weight implant or eyelid spring, com-bined with regional temporalis muscle transposition tothe mouth, provides the best opportunity for independenteye and mouth movement. By combining appropriate staticprocedures aimed at correcting brow ptosis, lower lid ec-tropion, epiphora, and facial sagging, facial symmetry andfunction are improved dramatically in the majority ofpatients. As innovative surgeons strive for perfection,improvement in overall results can be expected.

337

OMENTAL AND MUSCULOCUTANEOUSFREE FLAPS FOR COVERAGE

OF COMPLICATEDNEUROSURGICAL WOUNDS

DANIEL L. BARROW, M.D.FOAD NAHAI, M.D.

INTRODUCTIONExtensive areas of traumatic or neoplastic tissue loss in-volving the face, scalp, skull, and dura present formidableproblems in reconstruction. Successful reconstructivesurgery should protect vital structures and restore formand function. In planning any reconstructive effort, thesurgeon should utilize the simplest method to provideadequate coverage and aesthetic results. The least com-plex means of repair on the “reconstructive ladder” is di-rect closure. When direct closure is not feasible, one mustconsider more complex methods of wound closure. Split-thickness skin grafts are usually not sufficient to coverexposed cranium. When the wound includes loss of cra-nium, the problem is compounded by the risk of cere-brospinal fluid leakage and subsequent infection. Localrotation of scalp flaps is often sufficient if the extent ofthe wound permits use of this method. With more exten-sive loss of scalp, local rotation flaps cannot provide ad-equate coverage. Staged distant flaps have been used inthis situation but require multiple operations carried outover long intervals.

Following the loss of a large area of full-thickness scalp,including periosteum, the bare bone of the cranium requirescoverage as soon as possible. Often the outer table of theskull will have to be removed or perforated and time al-lowed for granulation before applying a flap.

The development and refinement of microvascularsurgical techniques has led to the use of heterotopic trans-fer of free vascularized grafts to cover distant areas of thebody that are devoid of skin and subcutaneous tissue. Earlyuse of musculoskeletal flaps involved transposition ofmuscle locally on its vascular pedicle. A prerequisite for

the use of a transposed flap is the presence of a dominantvascular pedicle on which the flap can be rotated throughan arc. This arc of rotation is based on the length of thedominant pedicle.

With the advent of microvascular techniques, the areato be reconstructed is no longer limited by the arc of rota-tion of the vascular pedicle. Instead, the muscle and over-lying skin that is best suited for adequate coverage maybe dissected out as a free flap, with microvascular anasto-mosis of an arterial supply and venous drainage to locallyexisting vessels. In certain clinical situations involvingprimarily soft tissue loss, replacement by a free omentalgraft is, in our opinion, the procedure of choice. Distantflaps, examples of which are musculocutaneous and omen-tal free flaps, represent the highest rung on the recon-structive ladder. Wounds of the face and head, especiallythose involving the cranium or dura, frequently requirethese more complex methods of reconstruction.

Optimal results are achieved in these complex re-constructive procedures through the cooperative effortsof the neurosurgeon and the plastic and reconstructivesurgeon.

INDICATIONS

Latissimus Dorsi MusculocutaneousFree FlapIn the majority of situations, when a distant compos-ite flap is required for reconstruction, the musculo-cutaneous free flap is ideal. The latissimus dorsimuscle provides the most versatile musculocutaneousflap for reconstructive microsurgery. The muscle andthe cutaneous territory overlying it represent the larg-est single flap available. Furthermore, its sacrifice pre-sents minimal functional def icits. The latissimus© 1991 The American Association of Neurological Surgeons

338


dorsi is supplied by a consistent proximal vascular pedicleof generous length and caliber. The vascular supply al-lows the muscle to be split to individualize each recon-struction or to preserve some function at the donor site,where a viable, innervated segment of the muscle may beretained.

This flap is excellent for providing coverage undercircumstances when the wound requires a flat, broad,malleable flap. As such, It is ideal for the coverage ofscalp, cranial, and dural defects as well as those that re-sult from orbital exenteration with maxillectomy.

Omental Free FlapIn certain clinical situations involving primarily soft tis-sue loss, replacement by a free omental graft is, in ouropinion, the most appropriate procedure. When the de-fect involves primarily subcutaneous tissue loss resultingin an abnormal contour, an omental free flap provideseffective coverage and restoration of contour. In addition,the inherent capability of this tissue to combat infectionand furnish an ideal bed for establishment of skin or bonegrafts provides further indications for its use. Therefore,there are three basic uses for omentum in reconstructionat a distant site: 1) to provide contour when there is softtissue loss (in this circumstance, especially if the face isinvolved in the wound, a musculocutaneous flap is toobulky and muscle atrophy unpredictable); 2) to fill a cavi-tary wound and establish an appropriate defense againstinfection; and 3) to provide a vascular bed for grafts ofbone and skin.

The greater omentum has a characteristic ability torepair defects through cellular proliferation, fibrous tis-sue formation, and adhesion production. It has a rich vas-cular and lymphatic supply, which probably accounts forits ability to rapidly absorb exudate or edema fluid andcombat infection. Another major reason for the ability ofomentum to cope with infection is related to the largenumber of macrophages within its areolar tissue. The in-herent ability of omentum to control infection and pro-vide a vascular bed for skin or bone grafts is retainedwhen omentum is used as a transplant.

SURGICAL PROCEDURELatissimus Dorsi Musculocutaneous FlapSurgical AnatomyThe latissimus dorsi muscle takes its origin from the lowersix thoracic, the lumbar, and the sacral vertebrae and fromthe posterior crest of the ilium (Fig. 1). There is also amuscular origin along the anterior lateral border of thelower four ribs. The insertion of the muscle is into theintertubercular groove of the humerus. The latissimusdorsi extends, adducts, and rotates the arm medially. It

also draws the shoulder downward and backward. Mathesand Nahai have classified the vascular anatomy of musclesbased on variables in the anatomical configuration of theirvascular pedicles. Muscles of Type I, II, III, and V areusually suitable for free-flap transplantation because oftheir dominant vascular pedicle. The latissimus dorsi is aType V muscle, the major vascular pedicle of which is thethoracodorsal artery, a terminal branch of the subscapu-lar artery. The thoracodorsal artery and nerve, along withone or two venae comitantes, enter the muscle at the neu-rovascular hilum on the deep surface, approximately 10cm from the insertion of the muscle.

The thoracodorsal motor nerve to the latissimusdorsi muscle is derived from the posterior cord of thebrachial plexus. The nerve is intimately associated withthe thoracodorsal vascular pedicle and branches intothe muscle in conjunction with the blood vessels. Themuscle also receives the segmental innervation of in-tercostal nerves T2 to T6. The thoracodorsal nerve istraditionally sacrificed during free tissue transfer to thehead and neck to prevent inappropriate motor functionpostoperatively. The denervation of the muscle also re-sults in a variable reduction in muscle bulk up to 50%over the ensuing 12 months, which yields an ever-im-proving wound contour.

Preoperative PreparationPreoperatively, the surgeon must ascertain the presenceof a functional latissimus dorsi muscle on the side fromwhich transfer is anticipated. This is most easily accom-plished by having the patient place his or her hands onboth iliac crests and apply pressure. The muscle shouldbecome visible and palpable in the posterior axillary fold.The night before surgery, the back, shoulder, face, neck,and head are washed with Betadine solution; also at thattime, a broad spectrum antibiotic or antibiotic appropri-ate to the flora of the wound is given parenterally.

Surgical ProcedureThe patient is securely positioned on the operating tablein a lateral decubitus position. The ipsilateral arm shouldbe included in the sterile field to facilitate dissection ofthe pedicle (Fig. 2). The exact position of the head mayvary according to the local pathology. Occasionally, thepatient may have to be repositioned after the latissimusdorsi is harvested.

After local excision of the scalp and cranial or in-tracranial pathology, the recipient vessels are exposed.Branches of the external carotid system such as the su-perior thyroid, facial, and occipital arteries are suitablerecipients for the thoracodorsal artery. The superficialtemporal artery may be used but is often much

339

BARROW AND NAHAI : OMENTAL AND MUSCULOCUTANEOUS FREE FLAPS

Figure 1. The latissimus dorsi is a large triangularmuscle that covers the lower posterior thoracic andlumbar regions. Its origin is primarily a broad apo-neurosis by which it is attached to the spinous pro-cesses of the lower six thoracic, the lumbar, and thesacral vertebrae, to the supraspinal ligament, and tothe posterior part of the crest of the ilium. It also arisesby muscular fasciculi from the external lip of the iliumand from the caudal three or four ribs by digitationsinterposed between similar processes of the externaloblique muscle. The muscle ends in a quadrilateraltendon, about 7 cm long, and is inserted into the bot-tom of the intertubercular groove of the humerus.

Figure 2. Patient position for a latissimus dorsi free flap to cover a cranial defect. The ipsilateral arm isincluded in this sterile field to facilitate dissection of the vascular pedicle.



340


smaller than the thoracodorsal artery and is frequentlydamaged by the local pathology or previous cranial sur-gery. Venous drainage may be accomplished by preserv-ing the external jugular vein during neck dissection or byanastomosis to the posterior facial vein or the internaljugular vein.

A limited incision is performed transversely acrossthe lower axilla and carried down along the posterior ax-illary fold. The subcutaneous tissue over the entire por-tion of the muscle to be transferred is elevated. For mus-culocutaneous flaps, the skin island is designed and thenincised down to the fascia of the latissimus dorsi muscle.

The thoracodorsal artery can be easily visualized onthe deep surface as the vascular pedicle enters the ventralaspect of the muscle approximately 10 cm below its hu-meral insertion. Here the thoracodorsal vessel divides intoa lateral branch to the anterior portion of the muscle anda medial branch to the posterior portion.

The anterior border of the latissimus dorsi muscle isidentified and the plane between it and the underlyingserratus anterior muscle is bluntly dissected toward theposterior midline. The muscle is then transected distallyand elevated to visualize the vascular pedicle on the deepsurface (Fig. 3). Crossing branches of the thoracodorsalartery to the serratus anterior muscle are divided as arethe muscular branches to the teres major. The vascularpedicle is dissected circumferentially and then protectedduring the transection of the tendinous humeral insertionof the muscle as well as when transecting the thoracodorsalnerve. Thus, the muscle remains attached only by the vas-cular pedicle and dissection proceeds toward the axilla toobtain the desired pedicle length. The vascular pediclemay be dissected all the way to the axillary artery andvein, if necessary (Fig. 3).

As dissection proceeds upward, care must be takennot to injure the intercostobrachial sensory nerve to theinner aspect of the upper extremity. This nerve coursesacross the axilla inferior to the axillary artery. Likewise,the long thoracic nerve which courses the length of thechest wall superficial to the serratus anterior muscle mustbe preserved; otherwise a winged scapula will result.

After preparation of the recipient vessels, thethoracodorsal artery and vein are transected and irrigatedwith heparinized saline. The flap is then transferred to thearea to be reconstructed and temporarily sutured in place.Under the operating microscope, the open proximal endsof the recipient vessels are held secure by microvascularclamps and the adventitia is excised. The thoracodorsalartery and vein are placed in microvascular approximating

clamps after the adventitia is removed. Under magnifica-tion of an operating microscope, end-to-end anastomosesare made using interrupted sutures of 9-0 or 10-0 Prolene.Once the vascular anastomoses are completed, the clampsare removed. The flap is sutured or stapled to the local tis-sue and, if necessary, the muscle may be grafted with asplit-thickness skin graft from the thigh.

A suction drain is placed in the back and the donorwound is closed primarily (Fig. 4). Any musculocutane-ous flap wider than 10-12 cm will not allow a primaryclosure of the donor site, and a skin graft will be required.To improve the aesthetic appearance of bulky flaps, themuscle is transferred without overlying skin and fat andis skin grafted. This is our preferred method as it leaves amore acceptable donor defect.

Postoperative CareThe suction drain is removed three to four days postop-eratively when drainage has decreased to less than 50 mlin 24 hours. The axilla should be checked daily for thedevelopment of a hematoma or seroma. Physical therapyconsisting of passive range of motion of the shoulder andupper arm is initiated on the third or fourth postoperativeday. Active exercises are initiated one week after surgery.

In the immediate postoperative period, the free flapshould be closely observed for evidence of vascular com-promise. Early signs of vascular impairment include flapcyanosis or the development of vascular petechiae associ-ated with sudden swelling of the flap. When the muscle iscovered by a skin graft, the most reliable check of vascularpatency is obtained by piercing the flap with a 21-gaugeneedle. Arterial blood should be seen. Rapid flow of venousblood may indicate venous obstruction. Early detection ofthrombosis of the vascular pedicle may enable successfulrevision of the vascular anastomoses, especially if imple-mented within the first 12 hours of the time of vessel throm-bosis. Success is less likely if thrombosis occurs more than48 hours after initial transfer of the free flap.

Illustrative CaseA 22-year-old man suffered a full-thickness electrical bumof the scalp and skull. After failure of a free flap per-formed at another institution, the exposed skull was cov-ered by expanding the existing hair-bearing scalp withrotational flaps. This coverage also broke down, leavingexposed cranium and dura (Fig. 5). The wound was cov-ered with a latissimus dorsi muscle free flap covered by asplit thickness skin graft, and the patient experienced goodfunctional and aesthetic results (Fig. 6).

341


Figure 3. The latissimus dorsi has been separated from the underlyingserratus anterior muscle. The vascular pedicle is identified on the infe-rior surface of the muscle. Vascular pedicles crossing to the serratus

anterior and teres major are divided. Dissection of the latissimus dorsiand the thoracodorsal artery is extended toward the axilla as far as nec-essary to obtain an adequate pedicle length.


342


Figure 4. Illustration of wound closure with a drain in place.

Figure 5. Preoperative photograph of full thickness loss of scalp andskull following an electrical bum. Note the breakdown of previous rota-tion flaps used to provide coverage.

Figure 6. Postoperative appearance following latissimus dorsi muscu-locutaneous free flap coverage.


© 1991 The American Association of Neurological Surgeons © 1991 The American Association of Neurological Surgeons

343


Omental Free Flap

Surgical AnatomyThe greater omentum is a double layer of peritoneumthat arises from the greater curvature of the stomach.It lies over the abdominal viscera and folds posteriorlyover the transverse colon. It is based on the right andleft gastroepiploic arteries, which join to form the gas-troepiploic arch (Fig. 7). From the main arch a variety

of vascular arcades then distribute the blood through-out the omentum. There are usually three main omen-tal vessels that arise from the gastroepiploic arcade.These vessels are 1.5-3.0 mm in diameter, providingadequate size for microvascular anastomoses. Althoughthe omental flap can be based on either pedicle, wegenerally prefer the right gastroepiploic pedicle becauseit is larger than the left and dissection in the region ofthe spleen is avoided.

Figure 7. Surgical anatomy of the omentum. Note the arterial supply by the right and left gastroepiploicarteries, which join to form the gastroepiploic arch.


344


Preoperative PreparationOne day before surgery the patient is placed on a clearliquid diet and the bowel is mechanically prepared. Theabdomen, face, neck, and head are washed with Betadinesolution the night before operation. Either a broad spec-trum antibiotic or an antibiotic appropriate to the flora ofthe wound is started by parenteral administration on thenight before surgery.

Surgical ProcedureThe patient is placed in the supine position. An incisionis made in the head or neck to isolate the vessels to beused for microvascular anastomosis. The incision will varywith the particular local wound or pathology. Branchesof the external carotid artery and internal or external jugu-lar vein are preferred as recipient vessels. Normal vesselsoutside the field of previous resection, trauma, or radia-tion therapy are selected.

A second team of surgeons with separate instru-ments and scrub team simultaneously performs an ab-dominal exploration through a midline celiotomy (Fig.8). The omentum is separated from the transverse me-socolon, proceeding from left to right, by dividing thetranslucent nonvascular attachments (Fig. 9). Theomental f lap is usually based on the r ight Figure 8. Incision used for harvesting an omental free flap.

Figure 9. Separation of the omentum from the transverse mesocolon by division of the translucent nonvascular attachments.



345


gastroepiploic vessels. The omenturn is then divided fromthe peripheral margin toward the gastroepiploic arcade ata point that would leave a suitable volume of tissue basedon the right gastroepiploic vessels. The short gastric ves-sels are isolated, clamped, cut, and ligated, proceedingalong the greater curvature of the stomach to the pylorus(Fig. 10). The right gastroepiploic vessels are isolated,

freed of surrounding fat, and dissected separately as ar-tery and vein toward the gastroepiploic artery, therebydeveloping a vascular pedicle 6-10 cm long with a diam-eter of 1.5-2 mm.

When the recipient vessels in the neck have beenprepared, the right gastroepiploic artery and vein areclamped and divided and the omentum is removed

Figure 10. The short gastric vessels are isolated, clamped, cut,and ligated along the greater curvature of the stomach to the py-

lorus. The omental flap is based on the right gastroepiploic arteryand vein.


346


from the peritoneal cavity. The vessels are tied and the ab-dominal wall, subcutaneous tissue, and skin closed. Nodrains are used. Once the omental tissue is harvested andthe closure is completed, that surgical team is excused.

Because no resorption of omentum occurs, it is nec-essary to trim all tissue and use only that amount neces-sary to reconstruct the wound. The vascular anatomy ofthe omentum permits the isolation of multiple small flapsor fingers within the major omental graft to restore con-tour to the forehead, eyelids, cheek, lips, or scalp. Anyportion of the exposed omentum that cannot be coveredwith adjacent skin undergoes immediate grafting with asplit-thickness skin graft.

The omentum is transferred to the area of reconstruc-tion, temporarily sutured or stapled in place, and coveredwith a moist sponge. Under the operating microscope,the open proximal ends of the recipient vessels are heldby microvascular clamps, and the adventitia is excised. Astandard microsurgical anastomosis of the vessels is per-formed. Once the vascular anastomoses are completed,the clamps are removed. There should be immediate flowinto the omentum with a visible pulse in the gastroepi-ploic arcade.

Postoperative CareAs with the muscle flap, the omental flap should be closelyobserved for evidence of vascular compromise. The pa-

tient is given clear liquids and the diet is advanced oncebowel sounds are present.

Illustrative CasesFigure 11: A 52-year-old man presented with a chronicopen frontal sinus infection following trauma. Multiplesurgical attempts to heal the wound had failed. The woundwas debrided and reconstructed with free vascularizedomentum. The sinus was filled with omentum and thewound closed, producing a good aesthetic result and ahealed wound.

Figure 12: A 34-year-old man developed an infiltrat-ing squamous cell carcinoma at the site of a 20-yearoldinjury to his scalp. At his initial operation, the tumor ex-tended through the cranium and involved the outer layerof the dura and sagittal sinus. A wide resection of thescalp and cranium was performed and free margins ob-tained. The surgical defect was filled with a latissimusdorsi free flap and skin graft. The patient did well for oneyear following this procedure but developed a recurrenceof his tumor that extended through the muscle flap. Atreoperation, the tumor was found to invade the dura, falx,sagittal sinus, and brain. The entire muscle flap was re-moved and the dura excised along with the occluded sag-ittal sinus. The tumor invading the brain was removed astotally as could be determined grossly. The defect wasthen covered with an omental free flap and skin graft.

347


Figure 11. A, the appearance of an open left frontal sinus infectionfollowing multiple attempts to close the wound. B, an intraoperativephotograph demonstrates the anastomosis of the gastroepiploic ar-

tery and vein to the superficial temporal artery and vein with theomental flap tunneled under the scalp to the area of the wound. C,front view of the omental flap before wound closure. D, final result.


348


Figure 12. A, an intraoperative photograph of the carcinoma and the cra-nial defect before wide resection. B, the defect following wide resectionof the scalp, cranium, dura, sagittal sinus, and involved brain. C, the anas-tomosis of the gastroepiploic vessels to the facial artery and vein. Theappearance of the vascularized omental flap before (D) and after (E) trim-ming to an appropriate size. F, the postoperative appearance.


349

REPAIR OF “GROWING”SKULL FRACTURE

TADANORI TOMITA, M.D.

INTRODUCTIONGrowing fracture is a rare complication of skull fractureoccurring in infancy and early childhood. This late com-plication of skull fracture is also known as a leptom-eningeal cyst. “Growing” fracture is somewhat of a mis-nomer, but it is characterized by progressive diastaticenlargement of the fracture line. Although skull fractureis a common occurrence in the pediatric age groups, theincidence of growing fracture is only 0.05-1 % amongskull fractures in childhood.

The usual presentation of the growing fracture is aprogressive, often pulsatile, lump on the head. Neurologicsymptoms such as seizure, hemiparesis, and mental re-tardation are less frequent. However, not infrequently,these patients may be perfectly asymptomatic, and a pal-pable mass or widening of the fracture line is the solesign noted incidentally by the parents. Usually a growingfracture develops within several months following theinitial skull fracture, but it may not be recognized for manyyears.

Growing skull fractures usually occur during the firstthree years of life (most often during infancy), and al-most never occur after eight years of age. Although frac-tures occur in any part of the skull, the most common sitefor growing fracture is over the skull vault in the parietalregion. Dural laceration is always present along the frac-ture line, and it is an essential factor for the developmentof a growing fracture. The dural laceration enlarges andgrows larger with the growing fracture. Computed tomog-raphy (CT) or magnetic resonance imaging (MRI) oftendemonstrates a focal dilatation of the lateral ventricle nearthe growing fracture. Lack of resistance of both dura andskull leads to focal amplification of the pulse wave of theintracranial pressure, causing herniation of the brain orsubarachnoid space through the fracture line and the du-ral defect. The “growth” of the fracture line is owing tobone resorption due to continuous pulsatile pressure at

the edge of the fracture line. A rapidly developing infan-tile brain and associated pathologic conditions such asbrain edema or hydrocephalus also contribute an outwarddriving force to cause brain herniation through the duraland skull defect. This pulsatile force of the brain duringthe period of its maximum growth produces the brainherniation through the dural laceration and fracture line,causing the enlargement of the fracture line of the thinskull.

One of the risk factors for the development of a grow-ing fracture is the severity of head trauma. A linear skullfracture with underlying hemorrhagic contusion of thebrain suggests a severe injury, significant enough to causea dural laceration. Initial CT scans for the evaluation ofhead trauma in patients who ultimately develop a grow-ing fracture usually reveal significant hemorrhage or con-tusion subjacent to the skull fracture. When a growingfracture is inspected at the time of surgical repair, theherniated brain is seen to be developing acerebromeningeal cicatrix. In some cases, loculated sub-arachnoid cerebrospinal fluid (CSF) cyst(s) maybe notedwith underlying gliotic, atrophic brain. Although the locu-lated subarachnoid space may become cystic (leptom-eningeal cyst), true leptomeningeal cysts are rare. Thecystic changes in the growing fracture usually representcystic encephalomalacia.

Depressed fractures usually do not cause growingfractures, but a linear fracture extending from the de-pressed fracture can lead to a growing fracture. The childwho on initial x-ray films of the skull has diastasis of thefracture more than 4 mm is considered to be at risk forfuture development of a growing fracture. Diastasis of acranial suture, however, is an unusual site for a growingfracture.

A growing fracture at the skull base can occur in anolder age group, especially where the bone is thin such asthe orbital roof, if a linear fracture is accompanied by adural laceration. Growing fracture and a meningoencepha-locele can develop with a similar mechanism as thoseoccurring in the skull vault of the young patient.© 1991 The American Association of Neurological Surgeons

350


RADIOLOGIC STUDIESX-ray films of the skull show a wide diastases of thefracture line. If initial skull films are available, one cancompare the films to confirm “growth” of the fractureline during the interval. When multiple fractures arenoted in the same patient, healing of the fracture in onearea may be noted as opposed to a growing fracture inanother area (Fig. 1, A and B). The fracture line can crossthe coronal or lambdoid sutures but is usually limited toone parietal bone.

CT scanning provides information regarding the con-tents within the growing fracture and any intracranialpathological changes. Furthermore, if CT scans are avail-able from the time of initial trauma, it should be possibleto demonstrate progressive changes. It is not unusual thatthe initial CT scans show hemorrhagic contusion, or sub-arachnoid or extraparenchymal hemorrhage. At the timeof discovery of the growing fracture, CT demonstratesthe diastasis of the fracture line and often a hypodenselesion near the fracture site. This hypodensity may repre-

sent encephalomalacia, a loculated arachnoidal cyst, orcortical atrophy. The ipsilateral ventricle tends to showfocal porencephalic dilatation with ipsilateral shift of themidline structure. This phenomenon may not only be dueto lack of dural resistance but due to cerebral atrophy.MRI provides further information as to pathological pro-cesses in association with the growing fracture.

MANAGEMENTSurgical intervention is indicated with a growing frac-ture line, seizure disorder, or progressive neurologicdeficits. A progressive cystic degeneration in the brainthat has herniated through the dural and cranial defectscan occur; therefore, surgical correction is recom-mended in young children even when seizures or otherneurologic symptoms or signs are absent. However,incidental, asymptomatic, and stable fractures in latechildhood or adulthood probably do not require sur-gery. The purpose of surgery for growing skull frac-ture is to repair the dural laceration and cranial defect,

Figure 1. A, an x-ray film of the skull made shortly after headtrauma showing multiple linear fractures in both parietal bones.B, a film obtained four months later showing a growing skullfracture with a resolving fracture on the contralateral side.


351

TOMITA : REPAIR OF “GROWING” SKULL FRACTURE

and to resect seizure foci. Growth of the growing fracturemay arrest after CSF diversion shunting by a decrease ofthe CSF pulse pressure, but this does not correct a sei-zure disorder. Placing a shunt for primary treatment ofthese patients is not advised unless hydrocephalus ispresent. Shunting for nonhydrocephalic patients createsundesirable shunt dependence.

Preoperative medications include anticonvulsant andantibiotic drugs. In a patient who is not already receivingan anticonvulsant, Dilantin should be given at a dosageof 15 mg/kg of body weight. Nafcillin in a dose of 50 mg/kg of body weight is given intravenously in the operatingroom just prior to surgery.

General anesthesia with an endotracheal tube is used.For a parietal or frontal lesion, the patient is in a supineposition with the ipsilateral shoulder elevated about 30°(Fig. 2). For an occipital lesion, a prone position is used.Because a young child has a thin skull, a U-shaped headholder is preferable to head pins.

SURGICAL TECHNIQUEThe scalp incision should be large enough to expose theentire length of the skull defect. An S-shaped or semicir-cular skin incision is made, and the scalp flap is turnedsubgaleally, leaving the underlying periosteal tissue in-tact (Fig. 3A). By palpation, the entire length of the cra-nial defect covered by pericranium is exposed to surgicalview. The site of the cranial defect is bulging and may be

accompanied by a bluish appearance due to an underly-ing subarachnoid cyst. As the cranial defect is covered bythe overgrowth of the pericranium, the edge of the cranialdefect is dissected by incising the pericranium along theedge of the bony defect (Fig. 3B). Soft tissues adherent tothe edge of the cranium defect are scraped off by a sharpdissector.

One should remember that the dural edge is usuallyretracted under and adherent to the inner table of the skull(the dural defect is invariably larger than the cranial de-fect), and that the pericranium is directly adherent to theunderlying cerebral tissue at the cranial defect. An effortto expose the dural edge by removing the cranial edgeshould not be undertaken as this procedure is often com-plicated by removing the dura simultaneously with theskull bone due to the adhesive nature of the dural edge. Inorder to identify the dura, several burr holes are madeaway from the skull defect with a distance of at least 50%of the width of the cranial defect. At this time, a largeenough amount of pericranium is removed from the neigh-boring skull in order to use it for repair of the dural de-fect. Once the dura is identified at each burr hole site, thedura is separated from the inner table of the skull towardthe defect (Fig. 3). A craniotomy is made around the skulldefect by connecting the burr holes with a craniotome.Two pieces of the craniotomy flap are obtained, one fromeach side of the growing fracture.

After the craniotomy is completed (Fig. 4A), reac-tive

Figure 2. Operative position. The patient is placed in a supine positionwith the ipsilateral shoulder elevated for a frontal or parietal lesion. Ifthe lesion is located in the occipttal region, a prone position is prefer-

able. The scalp Incision should be large enough to expose the entirelength of the fracture line and surrounding cranium.


352


Figure 3. A, the scalp flap is turned subperiosteally. The cranial defectis usually covered by the pericranium. B, the pericranium is incisedalong the edge of the cranial defect. Then, the edge of the cranial defectis exposed by scraping off the soft tissues adherent to it. C, the pericra-nium is removed from the surrounding skull surface and preserved fordural repair. Four burr holes arc made in the surrounding skull for a

craniotomy. After the confirmation of intact dura mater under the burrhole, the dura is separated from the burr hole toward the cranial defect.One should not attempt to identify the dura by removing the bone fromthe edge of the cranial defect. The craniotomy is carried out on bothsides of the growing fracture. The two bone flaps are removed and pre-served for autologous bone cranioplasty.


353

TOMITA : REPAIR OF “GROWING” SKULL FRACTURE

Figure 4. A, after the craniotomy, the intact dura mater is exposed aroundthe dural defect which is covered by the periosteum. Underneath theovergrowing periosteum is a cerebromeningeal cicatrix which is removedusing bipolar cautery and sharp dissection until healthy white matter isexposed. B, after all pathological tissues have been removed, the edgeof the surrounding dura is separated from the intact cortical surface. C,

the previously removed periosteum is used to repair the dural defect. Awatertight closure is achieved with 4-0 sutures. D, the bone grafts aresplit at the diploic space between the inner and outer tables by means ofan osteotome. E, the obtained split bone flaps are used to repair thecranial defect. The bone flaps are secured to each other and to the edgeof the cranial defect with nylon sutures or stainless steel wires.


354


periosteal tissue and the cerebromeningeal cicatrix at thecranial defect should be removed. There is a relativelywell demarcated plane between the cicatrix and the whitematter. Under magnified vision by means of surgicalloupes, the cicatrix including the periosteal tissue is lifted,and all abnormal tissue is separated and transected usinga bipolar cautery until normal white matter is exposed(Fig. 4B). The edge of the dura is separated from the cere-bral tissue, carefully avoiding trauma to the cerebral bloodvessels. In this region, abnormal tissue such as cysticchanges or xanthochromic discoloration due to previoushemorrhage is often noted.

After adequate debridement of the cicatrix at thegrowing fracture and freeing the intact dural edge fromthe cortical surface, the dural defect is closed using theperiosteal graft (Fig. 4C). Autologous pericranium is pref-erable to cadaver dura. A watertight closure of the dura isimportant to avoid a recurrence of the growing fractureor postoperative CSF leakage.

Each of the obtained craniotomy flaps is split at thediploic space with an osteotome, separating it into innerand outer tables (Fig. 4D). The cranial defect is thenrepaired by laying in the split autologous skull grafts.Usually four pieces are laid next to each other side byside to fill the cranial defect. These flaps are secured toeach other with either nylon sutures or stainless steelwires through drill holes (Fig. 4E). These flaps are fur-

ther secured to the craniotomy edge. If the defect of theskull is too large or the skull is too thin to separate intoinner and outer tables, one may consider autologous ribgrafts. These autologous bone grafts are well incorpo-rated and healing is excellent. Foreign materials such asmethyl methacrylate should be avoided for cranioplastyin the growing skull.

SPECIFIC CONSIDERATIONSThe growing fracture may extend toward a dural venoussinus such as the superior sagittal or lateral sinus. Althoughthese venous sinuses were spared from direct injury at theinitial trauma, direct exposure of them is not advised ornecessary. When the fracture line extends perpendicularlyto these sinuses, the closest end to the sinus does not needdural repair. However, if the growing fracture is paralleland near to the sinus, dural repair may be difficult due tothe lack of enough dural edge next to the sinus. In thesecases, one may repair the dural defect with a periostealgraft sutured to the periosteum of the skull above the sinus.

CSF diversion shunting has been recommended forpersistent postoperative CSF leakage from the cran-iotomy wound. It is justifted if coexisting hydroceph-alus is evident, or if CSF leakage occurs despite adequaterepair of the growing fracture. A lumboperitoneal shuntor temporary lumbar CSF drainage is to be consideredunder these circumstances.

355

OCCIPITAL ENCEPHALOCELESWILLIAM O. BELL, M.D.

INTRODUCTIONEncephaloceles are uncommon congenital malformationsof the central nervous system (CNS) occurring in approxi-mately 1-3 of every 10,000 live births in Western civili-zation. Seventy to eighty percent of all encephalocelesoccur in the occipital area, with the remainder locatedanteriorly or at the base of the skull. In the Far East, ante-rior and skull-base encephaloceles are the most common,and cranial CNS malformations are more common thanspinal malformations.

Occipital encephaloceles range in size from quitesmall (1-2 cm) to larger than the neonate’s head. Almostall are covered with partial-thickness skin, although theskin over smaller defects may be of full thickness. Identi-fication of these anomalies is usually easy, because theyare almost always quite obvious. However, a few may besmall, sessile, and nearly planar with the surrounding skin,thereby making their identification more difficult. Op-erative closure of these very small encephaloceles is notnecessary, but their presence can have broad prognosticimplications, especially if hydrocephalus is also present.

The occurrence of hydrocephalus in the presence ofencephalocele is due to aberrant development of the brainstem and cerebrospinal fluid (CSF) pathways. Hydroceph-alus may be present at birth or develop following repairof the encephalocele. Up to 50% of infants with an oc-cipital encephalocele will require a CSF shunt for controlof hydrocephalus.

PREOPERATIVE CONSIDERATIONSOperative closure of an occipital encephalocele is usu-ally straightforward, as long as certain important pointsare kept in mind. A computed tomography (CT) or mag-netic resonance imaging (MRI) scan should be obtainedpreoperatively in order to assess the intracranial contentsfor gross brain structure and ventricular size. I prefer aCT scan using 3-mm cuts, because the information soughtcan be obtained easily by this procedure and monitoring

the infant during an MRI scan may be problematic. Veryoften, there are striking brain abnormalities that will af-fect subsequent management, and these should be dis-cussed with the parents before the repair is begun.

The majority of occipital encephaloceles are locatedinfratentorially. The exact locations of the major venoussinuses and their relationship with the encephalocele canbe determined accurately with MRI if needed.

AU except the very smallest encephaloceles must berepaired, but because the majority are covered with skin,there is no emergent need to take the child to surgery within24-48 hours of birth. This delay allows adequate time forpreoperative planning and a full discussion of the implica-tions of an encephalocele with the infant’s parents. Althoughthe risk of seizures after repair is high, I do not begin pro-phylactic anticonvulsants preoperatively, but wait for sei-zures to occur before starting these drugs.

OPERATIVE TECHNIQUEA general anesthetic is required for this surgery. In thevast majority of instances, an inhalational anesthetic suchas halothane combined with nitrous oxide is sufficient.The anesthesiologist may or may not elect to use neuro-muscular junction blocking agents such as vecuronium.

The infant must be positioned prone (Fig. 1). If theencephalocele is large, this positioning will result in un-due pressure being placed on the globes unless appropri-ate care is taken to keep the area of the orbits free fromany encumbrance. I use umbilical tape around the pad-ded horseshoe headrest at the level of the orbits for thispurpose (Fig. 1). For lesions at the vertex, the neck maybe placed in a neutral position, but for lesions in the sub-occipital area, the neck must be flexed as much as pos-sible in order for the surgeon to work effectively. The an-esthesiologist must be made aware that the neck will beflexed and that this maneuver may change the position ofthe endotracheal tube. Extraordinary care must be takento ensure that the endotracheal tube is securely taped toavoid its dislodgement during the procedure.

The child’s torso is placed on soft rolls oriented© 1991 The American Association of Neurological Surgeons

356


Figure 1. The patient’s position for repair of an occipital en-cephalocele. Note the umbilical tape around the horseshoe

headrest at the level of the eyes and the rolls placed beneaththe child.

either vertically or horizontally and care is taken to avoidany pressure points. I usually do not place an arterial lineor a Foley catheter, because these operations are neitherbloody nor lengthy.

I use the 3M 1010 drapes, because they nicely estab-lish the perimeter of the area to be draped and reduce theamount of exposed skin, thereby allowing the infant toretain body heat during the procedure. In addition, thesedrapes prevent the skin preparation solution, blood, andirrigation fluid from dislodging the tape holding the en-dotracheal tube in place.

For skin-covered defects, I prepare the skin using di-luted Betadine scrub, tincture of iodine, and alcohol in thatorder. For defects with exposed tissue, I use Betadine scruband Betadine solution, followed by a normal saline rinse.This avoids applying alcohol to the exposed tissue.

I open the encephalocele sac either vertically or hori-zontally with a scalpel and then use Metzenbaum scissors(Fig. 2, A and B) in order to obtain a direct view of theinterior of the sac. The walls of the sac may be resected orthey may be everted with stay sutures (Fig. 2C), whicheverprovides the greater exposure. Frequently, there are mul-tiple concentric layers of arachnoid that need to be openedwith forceps in order to expose the neural tissue that islocated at the base of the encephalocele sac (Fig. 2C).

What to do with neural tissue located outside the cra-nial cavity remains controversial. Some have advocatedpushing it back inside the cranium; others have suggestedfirst performing electroencephalographic or evoked po-tential recordings to determine whether it is functionalneural tissue. Whether the externalized tissue is functionalor not, forcing it inside a cranium that has not been hous-ing it may exacerbate hydrocephalus or disrupt intracra-nial dynamics. Histologic sections of this external neuraltissue invariably show disorganized neural tissue withoutlayered cerebral or cerebellar cortex. The neurons are usu-ally interspersed in a glial background in what appears tobe a random manner.

My usual practice is to excise the exposed neural tis-sue (Fig. 3, A and B). Frequently, there are reasonably largevascular channels (both arterial and venous) coursingthrough the tissue, and these must be electrocoagulated care-fully with the bipolar forceps (Fig. 3A) before the scissorsare used to excise the tissue (Fig. 3B). If the location of themajor venous sinuses is known beforehand and reasonablecare is taken during the excision, the torcular herophili andtransverse sinuses are rarely encountered.

There is always a dural defect, and it is alwayssomewhat smaller than the associated skull defect. Inorder to obtain a watertight dural closure, I use the


357

BELL : OCCIPITAL ENCEPHALOCELES

Figure 2. The initial incision may be made vertically or horizontally(A) and then opened further with the scissors (B). Generally, there are

arachnoidal layers that must be opened with forceps so that the abnor-mal neural tissue at the base (C) can be identified.


358


Figure 3. After careful electrocoagulation with the bipolar forceps (A), the abnormal tissue is then excised with the scissors (B).

surrounding periosteum (Fig. 4, A-C). A No. 15 bladescalpel is used to incise the periosteum, which is thenreflected with an elevator, such as the Dingman peri-osteal elevator (Fig. 4, A and B). The periosteum/dura isthen closed with an interrupted or running absorbablesuture such as 4-0 Vicryl in a “vest-over-pants” fashion

(Fig. 4C and 5A). Because dura and periosteum havebeen used, this type of closure may allow some ossifi-cation of the skull defect, but complete ossif icationhas been rare in my experience. If the skull defect islarge, a piece of adjoining skull can be used to coverthe defect, sutured in place with absorbable 3-0


359


Figure 4. The periosteum is incised with a scalpel (A) and then reflected using a periosteal elevator (B). The dura is closed in a “vest-over-pants”fashion using absorbable suture (C).


360


Figure 5. Once this closure has been completed, the bone edges of the defect will be identified (A). Theexcess partial-thickness skin may then be trimmed (B).

or 4-0 sutures. The area from which the graft is taken willreossify in one to three months.

The skin may be closed in a vertical, horizontal,or oblique direction. The first step toward closure is totrim away excess partial-thickness skin (Fig. 5B) and

then to begin blunt dissection in the subgaleal space(Fig. 6A). The most distance for skin closure is obtainedin the cephalocaudal direction, and it is for this reasonthat I usually choose a horizontal skin closure (Fig.6B). The galea is closed with interrupted, buried


361


Figure 6. Undermining is done in the subgaleal space to allowfor skin mobilization sufficient for closure (A). After the galea

is closed, the skin is closed using a running monofilament su-ture (B).


362


4-0 Vicryl sutures and the skin with a running 4-0monofilament suture. I prefer Prolene for skin closurebecause of its low tissue drag and ease of removal.

POSTOPERATIVE CONSIDERATIONSPostoperative care is routine, with the only admonition be-

ing that the infant be kept from lying directly on the inci-sion until adequate healing has occurred, and that one bevigilant for developing hydrocephalus. Before the infant isdischarged from the hospital, if overt hydrocephalus hasnot developed, a follow-up CT scan or ultrasound studyshould be obtained for a baseline measurement.

363

FORAMEN MAGNUM MENINGIOMASAND SCHWANNOMAS:POSTERIOR APPROACH

CHAD D. ABERNATHY, M.D.BURTON M. ONOFRIO, M.D.

INTRODUCTIONApproximately one-third of all foramen magnum regionneoplasms are benign extramedullary tumors. The ma-jority are surgically removable meningiomas andschwannomas occurring in a ratio of roughly three to one.By definition, foramen magnum tumors include thoselesions which extend into both the posterior fossa and theupper cervical canal. Further subdivision into craniospi-nal and spinocranial types may be made based upon theprimary location of the tumor mass. Most schwannomasare of the spinocranial variety, with the majority arisingfrom the C-2 nerve root. Two-thirds of the meningiomasare also of the spinocranial type and occur in the antero-lateral aspect of the foramen magnum. Tumors to be ex-cluded from the designation of foramen magnum neo-plasms are those which arise in or adjacent to thecerebellopontine angle, jugular foramen, vermis, and highcervical spinal cord.

There is a female preponderance (3:2), with the me-dian age of onset occurring in the fifth decade. Due to thecapacious nature of the subarachnoid spaces in the vicin-ity of the foramen magnum, these tumors may reach im-mense proportions before producing clinically identifi-able symptomatology. Making the diagnosis by clinicalmeans is difficult due to the protean nature of the symp-toms and signs caused by tumors in this location. Themost common presentation is suboccipital or cervical painwith or without associated dysesthesias in the hands. Pro-gression usually involves weakness and atrophy of theintrinsic hand muscles, especially in anteriorly placed tu-mors. Other symptoms and signs include astereognosisand incoordination of the hands, sensory or motor defi-

cits of all four extremities, nystagmus, lower cranial nervepalsies, and long tract signs. There is typically a chronicprogression. However, a remitting course may occur whichcan confuse the clinical impression. Differential diagnosesbased on clinical grounds include cervical spondylosis,multiple sclerosis, syringomyelia, intramedullary tumor,type I Chiari malformation, normal pressure hydroceph-alus, amyotrophic lateral sclerosis, and subacute combineddegeneration.

PREOPERATIVE EVALUATIONPrior to the advent of computed tomography (CT) andmagnetic resonance imaging (MRI), the diagnosis of aforamen magnum meningioma or schwannoma was of-ten delayed or missed entirely. This was due to the pau-city of findings on routine radiographs and the fact thatPantopaque myelography was often not done in the su-pine position to diagnose posteriorly placed lesions or thatthe foramen was inadequately imaged in the prone posi-tion due to technical difficulties in holding the contrastmedium optimally across the foramen magnum.

At present, CT myelography and MRI are mutuallycomplementary studies (Fig. 1, A and B). CT myelogra-phy allows precise delineation of the tumor-cord inter-face and the cephalad-caudal extent of the pathologic pro-cess. However, bony artifact may occur at thecraniovertebral-junction. MRI is able to avoid this arti-fact and provide high resolution and detailed anatomicalinformation including vascular anatomy (Fig. 2, A andB). A tumor, especially a meningioma, may contain pro-ton density comparable to that of the nearby nervous tis-sue, and not be seen by routine MRI imaging techniques.If a tumor is expected, intravenous gadolinium-enhancedMRI imaging is recommended.© 1991 The American Association of Neurological Surgeons

364


Figure 1. A, a CT scan with intrathecal contrast shows a large posteri-orly placed meningioma extending from the foramen magnum to C-2

Figure 2. A, an anteriorly placed meningioma (arrows) at the foramenmagnum is well seen with a T1 weighted MRI midline sagittal section.

B, an axial MRI view accurately displays the tumor location anteriorly(arrows) and the degree of brain stem rotation and displacement.

(arrows). B, a T1 weighted non-contrasted MRI scan of the same pa-tient shows the tumor (arrows).



365

ABERNATHY AND ONOFRIO : FORAMEN MAGNUM MENINGIOMAS ANS SCHWANNOMAS

SURGICAL TECHNIQUEThe decision to administer steroids and diuretics is deter-mined by the severity of clinical findings and the amountof cord compression demonstrated on the neurodiagnosticstudies. Intraoperative electromyographic recording of thelower cranial nerves and somatosensory evoked potentialmonitoring of the posterior columns are warranted. In thesitting position, Doppler sonography is used to detect airembolism and preoperative placement of a right atrialcatheter is used to monitor cardiac function and for thewithdrawal of air emboli. A preoperative echocardiogramwill define a patent foramen ovale and allow the surgeonto determine the most prudent surgical position, eitherprone or sitting. If the sitting position is chosen, the de-gree of neck flexion from the neutral position is limitedby 1) the degree of osteoarthritic spurring of the cervicalinterspaces and associated risk of anterior spinal cordcompression, and 2) body habitus, i.e., short or elongatedneck with the risk of carotid and jugular compressionvaring with each individual patient. The head is securedin a pinion fixation device and the surgical area preparedand draped from the inion to the upper thoracic verte-brae. The attitude of the neck and lateral flexion are tai-lored to the position of the tumor as related to the brainstem and upper cervical cord. If it is anterior to the stemand cord, head rotation to the side of the lesion maxi-mizes tumor visualization and decreases the need for brainstem retraction.

A linear midline incision is made from the inion tothe level of the C-5 spinous process (Fig. 3). The incisionis carried to the fascia of the paraspinous muscles andhemostatic clips are placed along the skin margins. Themuscles are then separated from the spinous processesand swept laterally, being held in position by self-retain-ing refractors. The attachments of the trapezium, semi-spinalis capitis, and splenius capitis muscles are elevatedfrom the occipital squamae in a subperiosteal fashion.With the deep occipital triangle thus exposed, the remain-ing muscular attachments to the posterior rim of the fora-men magnum, arch of C- 1, and spinous process and lami-nae of C-2 are dissected free utilizing sharp, blunt, andelectrocautery dissection. The subperiosteal dissection ofthe arch of C- 1 is carried laterally until a pocket of looseareolar tissue is encountered, which is in close proximityto the vertebral artery. The dissection of the spino-occipi-tal musculature may be carried laterally on the occipitalsquamae until the mastoid processes are encountered (Fig.4).

The arch of C- I is then removed by rongeur resec-tion and/or the use of a diamond drill. Commonly, theinferior extension of the tumor necessitates a completelaminectomy of C-2 as well. Care must be taken to iden-tify the venous plexus that exists between the C- 1 andC-2 vertebrae. This plexus may be entered unknowingly,especially laterally, and act as an occult source of airemboli. The suboccipital craniectomy is performed byplacing multiple burr holes and rongeuring away the in-tervening occipital bone or by utilizing a diamond burr.The majority of tumors will arise in the anterior portionof the foramen magnum along the basilar groove. Fortumors in this location, the lateral extension of bony re-moval along the arch of C-1 and the rim of the foramenmagnum must be taken as far laterally as possible toallow adequate exposure of the tumor and its vascularsupply. Extensive lateral exposure also permits decreasedmanipulation of the medulla and cervical spinal cord asthe tumor is evacuated.

The dura. is opened in a standard Y shape with thesuperior flap reflected upward and secured to the bonymargin or muscle. The lateral dural margins are tacked tothe paraspinous musculature. After opening the arachnoid,the tumor will be identified as arising primarily from ananterior or posterior origin in relationship to the medullaand cervical spinal cord. The majority of tumors lying inthe anterolateral position will be meningiomas. Conversely,schwannomas will more commonly be found in the poste-rolateral location. Initial assessment of the vascular supplyto the tumor is imperative. Tumors arising in the posteriorportion of the foramen magnum typically derive theirvascular supply from the posterior meningeal artery whichis a branch of the vertebral artery just below the foramenmagnum. Anteriorly located tumors will ob-

Figure 3. The head is rotated to the side of the lesion and slightly flexedto visualize the tumor depicted in Fig. 2, A and B.


366


tain their blood supply from the anterior meningeal branchof the vertebral artery which arises immediately inferiorto the initial genu of the vertebral artery. This is roughlyat the level of the axis. This artery tends to course medi-ally and superiorly along the midline, supplying the duramater of the anterior foramen magnum. Great effort mustbe made to preserve the vertebral, posterior inferior cer-ebellar, anterior spinal, and posterior spinal arteries.

Posteriorly located tumors are approached directly.However, anteriorly positioned tumors tend to push and

rotate the cord and medulla posteriorly, stretching the den-tate ligaments, first and second cervical roots, and the elev-enth nerve over the tumor (Fig. 5). Division of the dentateligaments allows access to the anterior arachnoidal cisternsas well as providing a means of supporting the spinal cordduring the tumor dissection. On occasion it will be neces-sary to sacrifice the first or second cervical roots and pos-sibly the motor rootlets of the accessory nerve. The mass isdelivered piecemeal by means of sharp dissection and as-piration techniques. The ultrasonic aspirator and the

Figure 4. The muscles and fascia have been dissected, leaving the occiput and upper cervical laminae in view.


367


CO2 laser may be useful in accomplishing tumor removal.

If possible, the tumor-dural interface is developed withthe aid of bipolar coagulation, leaving the tumor as itsown retractor of the lower brain stem and upper cervicalcord (Fig. 6). When that is impossible or can be incom-pletely performed, then gutting of the tumor using bipo-lar coagulation and sharp dissection maybe necessary.Leaving the spinal cord/brain stem-tumor plane intactuntil the tumor has been devascularized and debulked

will allow separation of the tumor from the anterior spi-nal artery. Large tumors often parasitize blood supplyfrom the pial vasculature and care should be taken toidentify these arteries (Fig. 7). Undue traction on thesewith rupture will make identification of the source ofbleeding difficult or impossible and, if attendant withoccult subpial hemorrhage, will herald a potentiallytragic result.

Posterolaterally positioned meningtomas require a

Figure 5. In anteriorly placed lesions, the upper cervical roots andspinal portion of the eleventh cranial nerve are displaced over theposterior aspect of the tumor and often will need to be sectioned to

permit access to the tumor. The vertebral artery will be anterior tothe tumor in most instances.


368


Figure 6. Obtaining hemostasis by devascularizing the du-ral-tumor interface will allow debulking of the tumor at a

later stage while using the tumor judiciously as a brain stemretractor.


369


Figure 7. The last and most important post-debulking procedure is gen-tly retracting the tumor away from the brain stem while identifying and

controlling the arterial and venous supply common to the tumor and thebrain stem.


370


Figure 8. The tumor has been removed and the brain stem achieves a more normal position.


371


dural resection which encompasses the tumor’s origin. Thedura will necessarily be left intact on anteriorly locatedmeningiomas but must be cauterized to decrease thechances of recurrence. For foramen magnumschwannomas, the nerve root of origin should be sacri-ficed with adequate proximal and distal margins to en-compass the tumor extension along the root.

Upon completion of the tumor removal, complete he-mostasis must be obtained (Fig. 8). A watertight closureof the dura is mandatory, either primarily or with a fascialata graft. The capitis and trapezium musculature is thenreapproximated in the midline in layers with taut suturelines. One should begin at the base of the incision andwork superiorly with each successive suture line, ensur-ing that the muscles do not compress either the cerebel-lum, brain stem, or cervical cord. The suture line is car-ried superiorly until apposition of the musculature is nolonger possible. The most important layer is the ligamen-

tum nuchae inferiorly and the galea superiorly. This layermust be closed meticulously as a single structure. Thesubcutaneous tissues are then reapproximated and finallythe skin is closed.

COMPLICATIONSPerioperative complications are largely preventable withcompulsive attention to detail during the procedure. Com-plications include postoperative hemorrhage in the regionof the cervicomedullary junction and cerebrospinal fluidleakage from the wound. If either of these should occur,the treatment of choice is re-exploration of the wound withimproved closure technique. Infection is a potential sequelaof this operation and must be differentiated from asepticmeningitis. Finally, one must be prepared for the develop-ment of an air embolus which can occur at any juncture ofthe procedure, particularly during opening of the dura materand sectioning of the circular sinus.

372

PENETRATING WOUNDS OF THE SPINEEDWARD C. BENZEL, M.D.

INTRODUCTIONThe treatment of penetrating wounds of the spine is con-troversial. Some advocate surgery in the majority of caseswhile others advocate a nonsurgical approach in nearlyall cases. This controversy is fueled by a clinical situationin which neurologic improvement often ensues, regard-less of the treatment administered to the patient. Treat-ment efficacy is, thus, difficult to assess. Because themajority of penetrating wounds of the spine are gunshotwounds and because the majority of surgical indicationsare for gunshot wounds, the ensuing discussion will ad-dress them directly. Other penetrating wounds of the spineare relatively uncommon and seldom lead to surgical in-tervention. The indications for surgery, however, are simi-lar.

Complications of surgical therapy for gunshot woundsof the spine are predominantly related to infection, cere-brospinal fluid (CSF) leakage, and/or neurologic injuryincurred via surgical trauma. The potential for an adverseoutcome from surgery must be weighed against the chancethat surgery might confer a neurologic improvement tothe patient in addition to that expected with nonsurgicaltreatment alone. A number of authors have documentedthe relative safety of surgery for gunshot wounds. Theadvantages of surgery, however, are based upon anecdotalinformation and a few retrospective series only. The trueefficacy of surgery, therefore, is not known.

INDICATIONS FOR SURGERYThere are two fundamental indications for spine surgeryfollowing trauma: neural element compression and spinalinstability. Neural element decompression may be neces-sary under the following circumstances: 1) infection (in-cluding osteomyelitis, disc interspace infection, and epi-dural or intradural abscess) and 2) extra- or intraduralcompression from the missile, bone, soft tissue, or bloodclot. In many clinical situations, a combination of theseindications may exist simultaneously. In addition to theseindications, the location and the type of injury affect the

decisionmaking process regarding the utility of surgery.Several anatomical/clinical spinal gunshot injuries whichmay benefit from surgery, therefore, exist. These include:1) A through and through injury to the spinal cord whichhas resulted in a complete cervical myelopathy or an in-complete cervical, thoracic, or lumbar myelopathy (Fig.1A); 2) a dorsal injury to the spine which has resulted in acompressive mass lesion without dural penetration and anincomplete myelopathy (Fig. 1B); and 3) a ventral injuryto the spine which has resulted in a compressive mass le-sion without dural penetration and an incomplete myel-opathy (Fig. 1C). Obviously, clinical indications for anyplanned treatment supersede radiographic indications un-der most circumstances.

In this regard, patients who are candidates for sur-gery include: 1) patients in whom computed tomography(CT), myelogram, or magnetic resonance imaging (MRI)evidence of neural compression exists and in whom anincomplete myelopathy is present; 2) patients with a com-plete cervical myelopathy in whom the trajectory of themissile does not explain a higher neurologic (motor and/or sensory) level or loss of function with an accompany-ing radiographic evidence of neural compression; 3) thosepatients who harbor migratory or potentially migratoryintradural missiles or missile fragments and in whom in-termittent neurologic worsening is encountered or is prob-able; 4) those patients who have incurred an unstable spinesecondary to disruptive forces imparted to the spine bythe missile; and 5) patients in whom infection is present.

Patients with a complete myelopathy who 1) are notdeteriorating neurologically, 2) do not have an unstable spine,3) do not have a paraspinal or intraspinal abscess, 4) do nothave a significant neurologic deficit above that expected bythe trajectory of the missile alone, and 5) do not have radio-graphic evidence of neural compression above the level ofthe injury are not candidates for surgery. Therefore, regard-ing patients with a complete myelopathy, only those withcervical injuries who have a neurologic impairment whichis one or more segmental levels higher than that expectedby the missile’s trajectory should even be considered forsurgery. The remainder of the surgical can-© 1991 The American Association of Neurological Surgeons

373

BENZEL : PENETRATING WOUNDS OF THE SPINE

Figure 1. Illustrations depicting the major injury types and locations whichare amenable to surgery. A, a through and through injury to the cervicalspinal cord has resulted in a complete myelopathy. In this circumstance,surgery is indicated if spinal instability ensues or if the neurologic level ofinjury is significantly higher than the spinal level of injury and if therecoexists imaging evidence of either an extramedullary or intramedullarymass effect. B, a dorsal extrinsic mass effect has resulted from missile

Impingement upon the dural sac. If an incomplete myelopathy is present,surgical decompression via laminectomy is indicated. C, a ventral extrin-sic mass effect has resulted from missile impingement upon the dural sac.Ventral dural sac decompression is indicated if an incomplete myelopa-thy exists. Ventral decompression of the thoracic or lumbar dural sac re-quires a more thoughtful surgical approach.


374


didates have an incomplete myelopathy, combined witheither an extrinsic dural sac compressive lesion or an in-tradural mass lesion. It is emphasized that the indicationsfor surgery, therefore, are relatively uncommon.

Although surgery for spinal instability is included inthe discussion above, its treatment will not be addressedhere. The discussion, instead, will focus on the treatmentof injuries to or compression of the dural contents.

In view of the variability of injuries and injury pat-terns, a multitude of surgical approaches may be appliedto the treatment of spinal gunshot wounds. The surgeonmust use a rational and often creative approach and mustbe armed with the ability to exercise many of a multitudeof available options. The most common approaches willbe discussed and illustrated here.

TIMING OF SURGERYEarly surgery (surgery within the first five to seven daysfollowing injury) is associated with a relatively high com-plication rate, predominately related to infection and cere-brospinal fistula formation. It is therefore suggested thatfew, if any, patients undergo early surgery (within the firstweek following injury) for penetrating missile injuries ofthe spine. The rare exception to this policy pertains to thepatient with progressive neurologic deterioration. Thismight be caused by a migrating missile, infection, or verte-bral column instability or malalignment. The complicationsof early surgery are often secondary to contamination in-curred at the time of injury (infection) and difficulties re-lated to obtaining a tight wound closure following opera-tive intervention (CSF fistula formation). If surgery isdelayed, the contamination inoculum incurred at the timeof injury will have decreased signiftcantly by virtue of thepatient’s own defense mechanisms and via medical therapy,i.e., the administration of antibiotics. Furthermore, tissuefriability is increased immediately following missile inju-ries and tissue integrity is decreased. Wound healing ca-pacity is, therefore, diminished.

PREOPERATIVE PREPARATIONPatients are prepared for surgery in a routine manner.Broad spectrum antibiotic coverage should have been ad-ministered for three to five days immediately followingthe injury. Prophylactic antibiotic administration shouldagain be utilized prior to, during, and perhaps for one totwo days following surgery. Under most circumstances,the patient should be positioned in the prone position (al-though the lateral decubitus and the three-quarters proneposition may occasionally be used, depending upon theoperative approach used).

Care should be taken to avoid abdominal compres-

sion during the surgical procedure. The patient is preparedand draped in a routine manner with the draping includ-ing the upper and lower extremes of possible surgical ex-posure.

DORSAL APPROACHMost surgical approaches for the treatment of penetrat-ing wounds of the spine involve a dorsal exposure of thespine. For these patients, a midline incision and subperi-osteal exposure of the spine is performed. The surgicaltreatment of spinal gunshot wounds via this approach usu-ally is undertaken through a laminectomy or laminotomyapproach (Fig. 2, A and B). If an intradural exploration isindicated, a longitudinal dural incision is usually made,thus exposing the injured segment of the spinal cord (Fig.2C). A midline myelotomy of the spinal cord is appropri-ate when an intramedullary mass effect and a correspond-ing neurologic deficit are present (Fig. 3, A and B). Greatcare needs to be taken to not waver from the midline whenperforming the, myelotomy. Injury to longitudinal ascend-ing tracts (posterior columns) may result if deviation fromthe midline is excessive. Removal of blood clot whichshould be liquified at the time of surgery is then performed(Fig. 3B).

Dural closure should be attempted aggressively. Careshould be taken to not compromise the intradural contentsby constriction by the dural closure. Accessible dural holes,in addition, should be closed. Ventral dural holes which arenot accessible through the posterior approach are, in mostcircumstances, best left alone. Dorsal dural defects may beclosed with dural patch grafting (Fig. 3C). It is my prefer-ence that autogenous material be used: either locally obtainedfascia (such as thoracodorsal fascia) or fascia lata. Previ-ously reported experiences with fascia lata dural repairs haveillustrated its strength and utility, as well as its associationwith few complications. Patch grafting the dura mater shouldprevent spinal cord compromise.

Dorsal compressive injuries, which do not penetratethe dural sac, are best removed via laminectomy and gen-erous exposure of the injured region. Following such anexposure, the offending lesion may simply be removed(Fig. 4A). The exposure needs to be wide enough to allowfor a safe dural sac decompression. However, it shouldnot be so extensive that it compromises stability. This isparticularly the case with facet joint disruption. Careshould be taken to preserve the integrity of the facet jointswhen possible.

POSTEROLATERAL AND ANTERIORAPPROACHESThe posterolateral approach to spinal gunshot

375


Figure 2. A dorsal through and through injury to the cervical spine and itstreatment are illustrated. A, a subperiosteal exposure of the spine has al-lowed observation of the missile’s site of entrance into the spinal canal.Care should be taken during the subperiosteal dissection. Inadvertent plung-ing into the spinal canal may be avoided by simply being cognizant ofthis possibility and by simultaneously taking appropriate precautions. B,a generous laminectomy has been performed, thus exposing the site of

missile entrance through the dorsal dural sac. C, a dural opening has al-lowed visualization of the spinal cord. Note that the spinal cord was gen-erously exposed both superiorly and in a limited manner inferiorly. Thisallows for the decompression of the cervical spinal cord cephalad to theinjury site in the case of a patient whose neurologic level is higher thanthat expected by the trajectory of the missile. Also note the hole in thedural sac through which the missile passed.


376


Figure 3. An illustration of the tntradural and intramedullary decom-pression of a cervical spinal cord injury, as depicted in Figure 2. A, amidline dorsal myelotomy is made with a sharp knife, such as a No.11 blade. Bipolar cautery with a microforceps is emphasized so thatminimal posterior column injury is incurred. Care must be taken tonot waver from midline while performing the myelotomy for similarreasons. B, the exposure of an intramedullary blood clot and its re-moval are illustrated. Care must be taken to avoid spinal cord tissue

manipulation. The timing of the surgery to ensure that the clot is atleast semiliquified may help minimize surgical trauma. C, the duralsac closure is illustrated. A running suture closure of the dural inci-sion and a fascial patch grafting of the missile penetration hole havebeen performed. Fascia may be obtained locally (i.e., fromthoracodorsal fascia), by harvesting fascia lata, or by utilizingnonautogenous material. The ventral dural sac hole usually does notneed to be closed.


377


Figure 4. An illustration depicting two types of extrinsic dural sac com-pression. A, a dorsal injury has resulted in extrinsic compression of thedural sac. A simple generous laminectomy allows treatment of this prob-lem. B, a ventral injury has resulted in anterolateral compression of the

dural sac. In this case, a posterolateral approach has allowed an ad-equate exposure of the missile and should allow for its safe removal anddural sac decompression. It is emphasized, however, that oftentimes amore anterior approach may be more effective and simultaneously safer.


378


wounds is most appropriate for anterolateral and lateralextrinsic mass lesions. This approach utilizes an expo-sure gained by a wide lateral retraction of the paraspinoustissues, which, in turn, offers the surgeon a view of thefacet joints, transverse processes, and pedicles (follow-ing a laminectomy or laminotomy). This approach, there-fore, might be more appropriately termed a transpedicularapproach.

It is useful, almost exclusively, for ventral and/or lat-eral lesions which are compressing the dural sac. The ex-posure gained allows for a simple decompression. Figure4B illustrates the exposure gained by this approach, fol-lowing lateral soft tissue retraction. More extensive de-compressive operations, especially for lesions involvingthe ventral dural sac, might be best approached via a va-riety of available anterior approaches. These include theanterior approach to the cervical spine, as well as the an-terior, anterolateral, and lateral extracavitary approachesto the thoracic spine and lumbar spine. It is emphasizedthat these latter approaches are rarely indicated and thatthey do not offer a wide exposure of the dural sac. There-fore, injuries which are associated with dural sac pen-etration and which require intradural exploration and du-ral sac closure are fraught with difficulty when approachedfrom an anterior direction. It is also emphasized, how-ever, that no other approaches for truly ventral lesions arefeasible. Therefore, if such an approach is truly indicated,its degree of difficulty must be understood by both thesurgeon and the patient prior to surgery. The anterior ap-proaches are best suited for extrinsic lesions resulting indural sac compression such as is illustrated in Figure 1C.

WOUND CLOSUREClosure of the wound should be performed in multiplelayers, usually with absorbable sutures. A tight closure isemphasized. This should minimize the chance for cere-

brospinal fluid leakage. Patients with penetrating spinalgunshot wounds are at a relatively high risk for fistulaformation. It is my preference to not utilize drains, unlessabsolutely necessary. Careful attention to hemostasisshould, under most circumstances, minimize the indica-tion for drains. Intraoperative evoked potential monitor-ing should seldom play a role in the surgical managementof patients with spinal gunshot wounds.

COMPLICATIONSComplications of surgery for penetrating wounds of thespine are usually secondary to infection, iatrogenic exag-geration of a neurologic deficit, or cerebrospinal fluidleakage. Spinal instability may occasionally present as aproblem. These complications are minimized by payingcareful attention to surgical technique and the use of ap-propriate antibiotics. Patient selection for surgery shouldbe relatively strict, as well. This will minimize the chancefor complications resulting from surgical intervention.Abdominal injuries in which the bullet may have pen-etrated contaminated viscera prior to entering the spineshould be treated with antibiotics following general sur-gical consultation.

The proximity of the vertebral and carotid arteriesto the cervical spine is of significance. Patients with pen-etrating injuries of the cervical spine should almost al-ways undergo angiography shortly following admissionto the hospital. The passage of the vertebral arteriesthrough the foramina transversaria is of even greater sig-nificance. Bone or disc fragments, as well as missilefragments, may disrupt vessel integrity and alter flowthrough or occlude these vessels. It is emphasized thatneither the external visual examination of an artery norits palpation can define the presence or absence of en-dothelial injury, including intimal flap formation, dis-section, and even occlusion.

379

PERCUTANEOUS RADIOFREQUENCYRHIZOLYSIS FOR TRIGEMINAL

NEURALGIAJAMES FICK, M.D.

JOHN M. TEW, JR., M.D.


INTRODUCTIONMedically intractable trigeminal neuralgia, an incapaci-tating pain, can be eliminated by a percutaneous stereo-tactic rhizotomy. A review of 1000 procedures performedby the senior author documents that 91% of patients ob-tain complete relief of their pain with mild to moderatesensory loss.

The facial pain caused by this disorder is likened toan electrical current coursing through the nerve. This ex-cruciating sensation occurs in regions innervated by thetrigeminal nerve. The characteristic lancinating bursts ofpain are provoked by speaking, a change in facial expres-sion, or even a gentle breeze striking the face. The pain isintense. It may strike with the fury of a summer thunder-storm. It is common for patients to withdraw to a shel-tered environment, shunning food and conversation. Inthe worst cases, despair builds to the point that threats ofsuicide are common.

The initial treatment of patients with trigeminal neu-ralgia is pharmacologic. Drugs with anticonvulsive prop-erties rather than analgesic ones are most effective.Carbamazapine (Tegretol) can provide partial to completeresolution of the symptoms in most patients for severalmonths. Diphenylhydantoin (Dilantin) is a less effectiveoption. Patients treated with these drugs should be fol-lowed closely because hematologic abnormalities, hepaticdysfunction, and allergic reactions occur with variablefrequency. A patient who does not tolerate one of thesemedications should be tried on the other or on baclofen(Lioresal). Serum levels should be monitored especiallyif high doses of drug are needed to control the pain.

The pain of trigeminal neuralgia is characterized bya fluctuating course which, in the early stages, often spon-

taneously resolves. The medication can then be taperedand discontinued. Drug therapy should be reinstitutedduring periods of exacerbation. The pain is easily con-trolled initially, but symptoms become more intractableto medication as remissions become less frequent. Reliefcan be achieved only by larger volumes of drugs and drugcombinations. Unwanted side effects (lethargy and con-fusion) or toxic complications ultimately lead most pa-tients to consider other forms of treatment. For most pa-tients, selective interruption of the nociceptive fibers isthe best form of surgical therapy. Our experience docu-ments that radiofrequency thermocoagulation of the painfibers at the level of the posterior root is a very safe andeffective procedure. The percutaneous stereotactic rhizo-tomy (PSR) should be the first procedure considered forpatients with proven trigeminal neuralgia. Structuralcauses of trigeminal neuralgia should be excluded by animaging procedure. Magnetic resonance imaging or highresolution computed axial tomographic scanning are thepreferred techniques for evaluation.

Patients are frequently evaluated for this procedurewhile they are suffering from incapacitating pain. Theymay have been unable to eat or sleep for several days andoften appear exhausted and dehydrated. They may dem-onstrate symptoms of excessive narcotic consumption orof toxic levels of the anticonvulsant medication. In thesecircumstances, hospital admission for a careful assess-ment of the general medical condition may be required.

OPERATIVE PROCEDUREOral intake is restricted six hours prior to the procedure.Atropine (0.4 mg intramuscularly) to reduce the oral se-cretions and prevent bradycardia during sedation is ad-ministered one hour prior to the procedure. An intrave-nous line is required for the injection of an anestheticmedication.

380

NEUROSURGICAL OPERATIVE ATLAS, VOL 1

The procedure is performed in the diagnostic radiol-ogy suite. Figure 1 demonstrates how the equipment andpersonnel are arranged. The patient lies on the table in asupine position. The surgeon stands to the patient’s rightand must be able to control the radiofrequency generatorand view the fluoroscopy monitor. A qualified nurse oranesthesiologist stands on the opposite side of the sur-geon in order to administer the intravenous anestheticagent methohexital (Brevital) and observe the blood pres-sure utilizing an external monitor.

A reference electrode is placed over the deltoid regionof the arm. Alternatively, a 21-gauge spinal needle may beinserted in the subcutaneous tissue. Rare reports of skinburns occurring with the use of a needle has led some au-thorities to recommend grounding pads with an area of 150cm2 for use as the reference electrode. The cheek of theaffected side is prepared with an antiseptic solution.

The procedure is begun with a lateral cinefluorogra-phy (cine) radiographic view of the skull base. This viewshould clearly demonstrate the sella turcica and the cli-

vus. A bolus of methohexital (Brevital) (30-50 mg) is in-jected intravenously prior to placing the needle into theforamen ovale. A 19-gauge stainless steel needle insu-lated to the tip with Teflon is shown in Figure 7.

Three anatomical landmarks are used to guide theneedle placement. Marks are placed on the skin: 1) 3 cmlateral to the oral commissure; 2) on the lower eyelid atthe medial aspect of the pupil; and 3) 3 cm anterior to theexternal auditory meatus. These landmarks are shown inthe inset of Figure 2. In learning how to perform this pro-cedure, the junior author has found it helpful to use theselandmarks to plot the proper trajectory for the placementof the needle into the foramen ovale. This anterior ap-proach allows the surgeon to place the needle in the me-dial aspect of the foramen ovale by aiming at the inter-section of the coronal plane 3 cm anterior to the externalauditory meatus with the sagittal plane centered at themedial aspect of the pupil. Figure 2 depicts how theseplanes intersect at the foramen ovale at the base of theskull. On the inset of Figure 2 the vertical hatched line

Figure 1. Set-up of the radiology suite.


381

FICK AND TEW : PERCUTANEOUS RADIOFREQUENCY RHIZOLYSIS FOR TRIGEMINAL NEURALGIA

Figure 2. Artistic depiction of the method used to plot the location ofthe foramen ovale at the point of the intersection of the sagittal planeprojecting from the medial aspect of the pupil and the coronal planeprojecting from 3 cm anterior to the external auditory meatus. The ver-

tical hatched line is the intersection of the planes at the foramen ovalewhich is labeled 4. Inset, 1 needle entry point; 2, medial pupillary point;3, a point 3 cm anterior to the external auditory meatus; 4, foramenovale.


382


represents the intersection of these planes and depicts therelationship of the landmarks on the skin to the foramenovale. This helps a surgeon who has not performed a largenumber of these procedures to visualize where the needletip is to be directed.

Figures 3 and 8 illustrate how the needle is inserted.The cannula enters the skin 3 cm lateral to the oral commis-sure, as depicted in Figure 2, and is guided along the buccalsurface toward the target. The surgeon’s gloved index finger,placed in the patient’s mouth with its tip inferior to the lat-eral pterygoid wing, is used to guide the cannula into theforamen ovale. The finger prevents penetration of the oralmucosa by the tip of the cannula, the occurrence of whichwould introduce the risk of meningitis.

A lateral view of the skull base is obtained with animage intensifying device to assist in the needle placement.The tip is directed toward the clival line as depicted in Fig-ure 4. This trajectory permits the needle to be advancedinto the foramen ovale which is signaled by a brief con-traction of the masseter muscle. This results from a motorresponse when the needle stimulates the motor fibers ofthe mandibular nerve at the entrance of the foramen ovale.The needle placement is confirmed by a lateral image. Thetip of the needle should be located along the clivus at apoint 510 mm below the floor of the sella turcica. Figure 5is a demonstration of this fluoroscopic view with the needleproperly positioned. This approach of utilizing thelateral skull view to direct the needle tip toward a

Figure 3. Demonstration of the technique used to place the needle inthe foramen ovale. The surgeon’s finger is shown inferior to the lateral

pterygoid wing and guides the needle tip toward the foramen while pre-venting penetration of the oral mucosa.


383


Figure 4. This depicts how the needle should be directed at a point on the clival line 5-10 mm below the floor of the sella turcica.


384


Figure 5. Fluoroscopic images used to guide the electrode placement.The electrode tip is shown positioned in each of the three divisions:

mandibular division (A); maxillary division (B); ophthalmic division(C).

selected point along the clivus is preferred by the seniorauthor. The proper needle trajectory can be aimed and theneedle then reliably placed in the foramen ovale by usingthe landmarks obtained from the lateral skull image (Fig.6).

The stylet is removed to ensure that the carotid ar-tery has not been penetrated. If pulsatile blood flow isobtained, the needle is withdrawn and manual pressure isapplied over the posterior pharyngeal space. During thisphase of the procedure the carotid artery can be injured atthree locations: 1) In attempting to penetrate the foramenovale, a posteromedial deviation can result in the needlepiercing the cartilaginous covering of the foramenlacerum. 2) Upon entering the middle cranial fossa, aninferior and medial trajectory of the needle can penetratethe carotid artery where it lies behind the mandibularnerve. 3) Finally, if the cannula. is directed in the regionof the first 5 mm of the clival line below the floor of the

sella turcica, the carotid can be pierced in the cavernoussinus where a carotid cavernous fistula may be created.

The needle is advanced to a point even with the pro-file of the clivus. Removal of the stylet will allow the egressof cerebrospinal fluid (CSF) unless the subarachnoid spacehas been obliterated by a previous rhizotomy or chemicalinjection (Fig. 9). A flexible electrode is inserted into thecannula. The Radionics TEW KIT cannula permits the useof either a straight or curved-tip electrode. The curved-tipelectrode allows one to refine the lesion production duringthe procedure and has resulted in a substantive reductionof the side effects of this procedure. The curved electrodetip is a coil spring which carries a thermocouple, a stimu-lator, and a lesion-generating probe into the tissue at theend of the cannula. An insertion tool is used to insertthe electrode into the cannula (Fig. 7). The guide ofthe insertion tool is aligned with the slot in the cannula(Fig. 9). The curved-tip electrode is guided


385


Figure 6. A proper needle trajectory should place the needle tip 5-10 mmbelow the floor of the sella turcica along the clival line. One can appreci-

ate how the fibers of V3, then V

2, and finally V

1 are encountered in se-

quence as the needle is advanced toward the clivus.

Figure 7. Photograph of the components required for insertion of the flexible electrode.



386


into the cannula (Fig. 10). The insertion tool is removedand the electrode is pushed to the hub of the cannula (Fig.11). The exposure of the electrode tip can be controlledby a millimeter scale on the cannula’s hub. A plastic clipis attached to the cannula where it enters the skin to pre-vent advancement of the cannula into the cranium. Theelectrode is then connected to the radiofrequency genera-tor by a coaxial cable (Fig. 12).

Several models of the radiofrequency generator areavailable for use. The specific operation procedures foreach model should be obtained from the operator’smanual. Each model provides the ability to measurestimulation parameters (frequency, pulse wave) and le-sion data (voltage, amperage, and temperature) at thetip of the electrode during the administration of theradiofrequency current to the nerve. The older modelsrequire a thermocouple adapter if the Tew curved elec-

trode is to be used. The newest digital model, RFG-3C,is currently used (Fig. 13).

Physiologic localization of the electrode tip is deter-mined by the patient’s response to electrical stimulation.The needle is located, for descriptive purposes, in rela-tionship to the profile of the clivus. Generally, the man-dibular division of the root is encountered 5 mm proxi-mal to the clivus, V

2 is located at the clivus, and V

1 is 5

mm distal to the clivus (Fig. 6). The precise placement isconfirmed by stimulus-provoked paresthesia and provo-cation of pain. The pain and paresthesias originate fromthe characteristic trigger point and radiate into the der-matome where symptoms occur. This response can be elic-ited by a stimulus with the following parameters: 1 ms,100-400 mV, and 50 Hz. A higher voltage, 0.5-1.5 V. maybe required in patients who have undergone previous rhizo-tomy procedures or chemical injections. Increas-

Figure 8. The technique of needle placement.A gauze-padded bite block is shown.

Figure 9. Placement of the insertion tool intothe cannula. The inset demonstrates how CSFmay drip from the cannula when the stylet is re-moved.


387


Figure 10. Placement of the curved tip elec-trode into the cannula.

Figure 11. Removal of the insertion tool.

Figure 12. Connection of the coaxial cableto the electrode. The inset demonstrates howthe electrode is secured against the skin witha plastic clip.




388


Figure 13. Photograph of the RFG-3C radiofrequency generator (Radionics Inc., Boston, MA).

ing stimulus voltage can reliably reproduce both the char-acter and location of the pain. Pain or paresthesia shouldbe obtained, sequentially, in the third, the second, and fi-nally the first division of the nerve. This sequential re-sponse ensures the best chance of alleviating the painwithout producing undesired sensory loss in unaffectedregions of the face.

If paresthesia or pain is not provoked in the expecteddistributions when a low stimulus threshold is used, thecannula should be advanced, the electrode reinserted, andthe stimulation repeated. The flexible tip electrode pro-vides distinct advantages over the straight tip electrodeduring the needle localization for stimulation. Rotationof the flexible tip 180° along the long axis of the elec-trode enables one to reposition the tip to the V3 divisionfibers caudally or to the V1 division fibers by rotating thecannula cephalad. This maneuver can be accomplishedwithout requiring that the cannula be withdrawn or ad-vanced (Fig. 14).

Contractions of the masseter and pterygoid musclesmay occur at a low stimulation threshold. The needleshould be repositioned in a more lateral trajectory to limitthe involvement of the motor fibers. Stimulus-evoked fa-cial contractions indicate that the electrode is either toodeep or is inclined too low on the clivus, or the stimula-tion level is too high.

If movements of the eyes occur during stimula-tion, the needle is too deep and the oculomotor. tro-chlear, and abducens, nerves are at risk for injury if alesion is produced. Diplopia. can result from injuryto these nerves along their course in the cavernoussinus at the medial aspect of Meckel’s cave. This com-plication should be avoided if the needle is not ad-vanced more than 5 mm beyond the clival line. Move-ment of the eyes indicates that the cannula is too deepin the cavernous sinus or too near the brain stem. Theelectrode must be repositioned and the stimulationperformed again.

When an acceptable electrode placement is confirmedby stimulation, a second dose of intravenous anesthetic isinjected and a preliminary lesion is produced at 65°C for60 seconds. During the heating of trigeminal rootlets afacial blush will frequently appear in the trigeminal divi-sion which is being heated. Although this observation doesnot permit the precise localization of the site of sensorydenervation, the observation can help to avoid unwantedsensory loss in an adjacent division. This concept is par-ticularly important in preventing denervation of the cor-neal surface if the neuralgia has not involved the oph-thalmic division.

After the intravenous anesthetic agent dissipates in120-180 seconds, a careful sensory examination of


389


Figure 14. Artistic demonstration of the trigeminal ganglionwhich demonstrates how the curved-tip electrode can be posi-

tioned in each division of the nerve without changing the depthof the cannula.


390


the face should be performed. The targeted area for sen-sory impairment is determined by the division of thetrigeminal nerve involved, the severity of the symptoms,and preoperative discussions with the patient. These con-siderations enable us to determine the extent of deficit bestsuited for each clinical situation. The lesion should be ex-tended if pain persists after the initial coagulation, even ifsensory deficits have been produced. Sequential lesions,produced by 60- to 90 second applications, are performedby increasing the temperature by 5°C increments until thedesired result is achieved. This goal can frequently beachieved without additional intravenous anesthesia. Con-sequently, it is possible to have the patient’s cooperationwith the sensory examination in order to monitor the sen-sation of the cornea and face and to prevent the spread ofsensory loss to undesired locations.

The results of the cranial nerve examination are re-corded immediately after the procedure and at the time ofdismissal, which usually occurs within 24 hours. Patientsare allowed to resume normal activities following the pro-cedure. A diet consisting of soft food and liquids is advis-able for several days while the patient becomes accus-tomed to any change of oral and facial sensation orweakness of the muscles of mastication. Occasionally ahematoma of the cheek may occur at the needle insertion

site. Corneal sensitivity is protected with artificial tearsand frequent observation.

Detailed instructions concerning early symptoms ofcorneal abrasion and concerning eye care are given to eachpatient. Every patient and family must understand that anysymptoms of corneal abrasion require prompt ophthalmo-logic evaluation. Anticonvulsant medication should be gradu-ally withdrawn to avoid symptoms of agitation.

CONCLUSIONSPercutaneous stereotactic rhizotomy is a very effectiveprocedure for eliminating the painful symptoms ofmedically intractable trigeminal neuralgia. The goal isto coagulate the nerve fibers responsible for generat-ing the pain impulses while minimizing sensory lossover the areas of the face not involved with the neural-gia. Using technical developments such as the curvedelectrode and increasing experience, the complicationscan be limited to a very acceptable level. The precisionof lesion production and minimal surgical complica-tions allow this procedure to be offered as a valid op-tion for all patients with trigeminal neuralgia. The lowmorbidity and effective results justify selection of thisprocedure for most older patients for whom major in-tracranial procedures offer greater risk.

391

EXTENDEDCOSTOTRANSVERSECTOMY

EDDY GARRIDO, M.D.

INTRODUCTIONThoracic spinal cord compression due to lesions locatedanteriorly or anterolaterally in the spinal canal cannotsafely be managed with the standard laminectomy ap-proach because any significant manipulation of the spi-nal cord will likely lead to paraplegia. Other surgical ap-proaches have been developed over the years to enter thespinal canal from an anterior or posterolateral direction.Thoracotomy, costotransversectomy and its various modi-fications, and the transpedicular approach are the differ-ent ways in which these lesions can be removed with alow risk of spinal cord damage. Each one of these ap-proaches has advantages and disadvantages and its selec-tion depends on the experience of the surgeon. It is notwithin the scope of this paper to compare each of thesesurgical procedures.

I have had experience with a modification of the cos-totransversectomy approach to the thoracic spine in a se-ries of 18 patients. The lesions surgically treated were asfollows:

Herniated thoracic disc 8 casesFracture, vertebral body 4 casesLarge neurofibroma 2 casesMetastatic tumor 3 casesEpidural abscess 1 case

The radiographic diagnosis is made with x-ray filmsof the thoracic spine, computed tomography (CT) scans,and magnetic resonance imaging (MRI) scans. The MRIscan is probably the best imaging modality for the tho-racic cord and has, for the most part, replaced the use ofmyelography.

SURGICAL PROCEDUREThe procedure is done with the patient under general an-esthesia. A Foley catheter is inserted. Ancef (cefazolin),1 g intravenously, is given just prior to beginning surgery

as a prophylactic antibiotic. The patient is placed in theprone position on the operating table with the chest andabdomen resting on rolls. The arms are tucked along thesides of the patient-The procedure is usually done on thepatient’s left side unless the lesion is to the right of themidline which mandates a right-sided approach. The ap-proach from the left is technically easier for the right-handed surgeon. The operating table should permit theuse of anteroposterior x-ray films or fluoroscopy for lo-calizing purposes.

After the patient has been positioned on the operat-ing table, the appropriate rib to be resected is identifiedby radiography and/or fluoroscopy and marked with me-thylene blue. One needs to remember that the lower ribwill lead to the desired disc space. For instance, the headof the 8th rib will articulate with the T7-T8 disc space;however, the 11th and 12th ribs articulate just below thedisc space (Fig. 1).

Figure 1. This diagram shows a segment of the thoracic spine to indi-cate the anatomic relationship of the ribs with the vertebral bodies andintervertebral discs. The 11th and 12th ribs, however, attach to the ver-tebral body just below the disc space.© 1991 The American Association of Neurological Surgeons


392


Either of two types of skin incision can be used de-pending on the lesion to be dealt with (Fig. 2). An obliqueincision following the rib to be resected and extendingacross the midline is preferred for relatively localized le-sions such as herniated discs or neurofibromas or otherfocal neoplasms. In more extensive lesions and particu-larly in traumatic cases with fractured vertebral bodieswhere there is the need to do a fusion procedure with in-strumentation, then a T-shaped incision is best. This willallow the surgeon to resect several ribs and expose thespinal column at multiple levels.

With either incision, the surgeon continues bytransecting the trapezium or latissimus dorsi muscle de-pending on the level of the thoracic spine. The paraspinalmuscles are exposed, separated subperiosteally from thelaminae, and transected and retracted superiorly and in-feriorly. To minimize bleeding, the muscle section is bestdone with the electrosurgical knife. The ribs, lami-nae, facet joints, and transverse processes are visual-ized at this point (Figs. 3 and 4). The proximal 5-6 cmof the selected rib is resected after stripping off theperiosteal covering as well as any other soft tissues.The transverse process and the head of the rib

Figure 2. This diagram shows the patient positioned prone on the oper-ating table. Left, the T-shaped incision is used when a more extensiveexposure is needed such as for those cases requiring anterior decom-pression of the canal and a posterior fusion with instrumentation. Right,the oblique incision is used more frequently and gives adequate expo-sure for removal of herniated discs, dumbbell neurofibromas, and lo-calized epidural abscesses.

Figure 3. This drawing illustrates the initial approach with sec-t ion of the trapezium or lat issimus dorsi muscle and the

paraspinal muscles and exposure of the laminae, transverse pro-cess, and rib.



393

Figure 4. Operative photograph showing the exposed laminae (1,2) and the rib (3). A right-sided approach was made in this case.

GARRIDO : EXTENDED COSTOTRANSVERSECTOMY


394


are then removed. The spinal canal is exposed by doing acomplete unilateral hemilaminectomy of the two adjacentlaminae. The facet joint is removed entirely. The lateralaspect of the dural sac and spinal cord is seen at this point(Figs. 5 and 6). The disc space can be seen and palpated.The pleura is easily separated from the lateral wall of thevertebral bodies and disc space and is retracted laterally(Fig. 6). The nerve root and intercostal neurovascularbundle which has been separated from the rib is left in-tact whenever possible.

The operation then proceeds according to the patho-logical lesion (Fig. 7). For a herniated disc, it is best todrill out the pedicles and adjacent vertebral bodies nextto the disc, therefore creating a cavity where the discmaterial could be curretted down and extracted withoutmanipulation of the spinal cord (Fig. 8). Similarly, bonyfragments in the anterior aspect of the spinal canal can beremoved by thinning it down with the high-speed drilland by downward pressure with downward angledcurrettes. Tumors can also be easily removed under directvisualization of the dural sac. In cases of neurofibroma,the dura should be opened to remove the intraspinal por-tion of the tumor. With this exposure there is a clear viewof the anterior and lateral aspects of the dural sac, thenerve root, the disc, and the vertebral bodies. The operat-ing table can be rotated 10-15° away from the surgeon tofacilitate the view under the dural sac. After removalof the herniated disc or vertebral body fragments onecould use the ultrasound imaging equipment to check

Figure 5. Axial view of the thoracic spine which demonstrates the uni-lateral removal of the rib, transverse process, pedicle, laminae, facetjoint, disc, and part of the adjacent vertebral bodies.

Figure 6. This drawing shows the lateral wall of the dural sac, part of thevertebral bodies, and the intervertebral discs. At this stage of the operation

the laminae, facet joint, transverse process, and part of the rib have beenresected. The pleura is easily retracted from the lateral wall of the spine.



395

GARRIDO : EXTENDED COSTOTRANSVERSECTOMY

Figure 7. This illustration shows the three most common lesions that aremanaged with the extended costotransversectomy approach. Top, dumb-bell neurofibroma. Middle, fractured vertebral body with bone fragmentsin the spinal canal. Bottom, herniated disc.

the area to be certain that no disc or bone fragments havebeen missed. The arthroscope with an angle camera canalso be used to look under the dural sac. A fiberoptic head-light and ×3.5 magnifying surgical glasses are extremelyuseful for adequate visualization. The high-speed air drillis essential.

Spinal cord evoked potential monitoring has not beenused in any of these patients during surgery. The value ofsuch monitoring is somewhat controversial. This surgicalexposure gives such a clear, direct view of the dural sacthat the surgeon does not have to move or manipulate thespinal cord to remove the pathological lesion.

To close the wound, the different muscles and fasciallayers are approximated with heavy sutures. A Jackson-Pratt drain is used for 24 hours postoperatively. No prob-lem with wound healing has been encountered in any case.

COMPLICATIONSPossible complications are wound infection, pneumotho-rax, spinal cord damage from direct manipulation or fromischemia due to ligation of a radicular artery, and spinalinstability. The only complication in this group of 18 caseshas been that of a wound infection which was treated withdrainage and appropriate antibiotics.

If there is a significant tear in the pleura, it wouldbe best to use a chest tube for 24 hours. Neurologic defi-cits are avoided if one is careful not to manipulate

Figure 8. This illustration shows the defect left after removal of the disc and adjacent areas of the vertebral bodies.The nerve root is left intact in most cases.



396


the spinal cord. Ischemic neurologic complications couldoccur if a major radicular artery to the spinal cord is li-gated along with the nerve root. Given the extensive andmultiple radicular blood supply to the spinal cord, the po-tential for this complication to occur is low and it is notnecessary to sacrifice the nerve root and radicular bloodsupply in most cases. No instances of spinal instabilityhave occurred from the limited bony removal with thisprocedure.

SUMMARYThe extended costotransversectomy or posterolateral ap-proach to the thoracic spine is an excellent way to treatpathological lesions anterior to the spinal cord. There isno need to enter the thoracic cavity; therefore, a chestsurgeon’s assistance is not necessary. Herniated thoracicdiscs, large dumbbell neurofibromas, metastatic spinalcord tumors, abscesses, and traumatic bone fragments arethe usual pathologic lesions that can be managed withthis surgical approach. In traumatic cases where a poste-rior fusion with instruments is needed, it is done at thesame time as the decompression.

397

SURGICAL RESECTIONOF POSTERIOR FOSSA EPIDERMOID

AND DERMOID CYSTSLEE KESTERSON, M.D.

INTRODUCTIONIntracranial epidermoid and dermoid tumors are uncom-mon lesions that account for less than 2% of intracranialtumors in most series. The ratio of intracranialepidermoids to dermoids is 4:1 with epidermoid tumorstypically becoming symptomatic during the third and fifthdecades whereas dermoids tend to occur in the pediatricage group.

Aberrant closure of the dorsal neural tube is thoughtto account for the occurrence of these congenital, histo-logically benign lesions. Epidermoid tumors tend to oc-cur off the midline in a more lateral location while der-moid tumors occur near the midline. In the posterior fossathe typical location of epidermoid tumors is thecerebellopontine angle, petrous apex, cerebellum, andfourth ventricle. Dermoids are reported to occur more fre-quently in the region of the fourth ventricle and may com-municate with the skin via a sinus tract. The midline lo-cation of these tumors is due to the fact that theneuroectoderm separates dorsally along the midline. Thelateral occurrence of epidermoid tumors is possibly dueto proliferation of multipotential embryonic cell rests orthe transplantation of epithelial rests carried laterally withthe migrating otic vesicles or developing neurovasculature.

Dermoid and epidermoid tumors are frequently re-ferred to as inclusion tumors because their growth occursby progressive desquamation of the capsular components.Generally these are considered benign lesions; however,there have been rare instances of malignant degeneration.The only difference between these two lesions is the his-tology. The cyst lining of an epidermoid is composed of acapsule of stratified squamous epithelium. Dermoid cystsnot only contain this element but also dermal derivativessuch as hair and sebaceous glands. Because of the pres-ence of these latter elements, dermoids are firmer and

lack the pearly appearance that is characteristic of epi-dermoid tumors.

These lesions are confined to the extraparenchymalarea and can literally “flow” into any available subarach-noid space, crossing cisternal and compartmental bound-aries. These masses can insinuate themselves betweencranial nerves and vascular structures, and into the exitforamina of the fourth ventricle, sulci, and fissures. Be-cause the growth is slow and the included elements aresoft and pliable, an inclusion cyst tends to conform to theshape of the cavities it enters, with symptoms resultingfrom compression, distortion, and/or obstruction occur-ring long after the mass has attained a very large size.Furthermore, the neurologic symptoms initially may bevague and nonspecific with periods of waxing and wan-ing similar to those described for demyelinating disease.

Epidermoid and dermoid tumors of the posterior fossapresent with headaches, disequilibrium, and/or cranialnerve involvement. These lesions may also present withthe classic symptoms and signs of hemifacial spasm ortrigeminal neuralgia. Another form of presentation is thatof cyst rupture which allows spillage of keratinous mate-rial into the cerebrospinal fluid (CSF) spaces with thesubsequent development of aseptic meningitis.

PREOPERATIVE EVALUATIONThe main radiologic imaging modalities utilized now arecomputed tomography (CT) and magnetic resonance im-aging (MRI). Contrast agents are generally used as is thecase in the evaluation of most intracranial masses. Theuse of these modalities is complementary because of theeffectiveness of MRI in demonstrating anatomical detailin the posterior fossa. and CT for its value in delineatingbone anatomy.

With CT the attenuation value of an epidermoidcyst is usually very low (-20 to +30 Hounsf ieldunits); however, there have been some lesions reportedwith attenuation values somewhat higher (80-120© 1991 The American Association of Neurological Surgeons

398


Hounsfield units). This range of attenuation values hasbeen thought to be due to varying amounts of low-den-sity lipid and high-density keratin in the desquamativedebris of the tumor. Certainly, this debris can be visual-ized in the subarachnoid space if the cyst has ruptured.Very uncommonly, calcification is noted in the capsuleof an epidermoid tumor. Dermoid tumors have similarradiologic findings but have a greater range of attenua-tion values with more frequent association with calcifi-cation and with developmental anomalies of the skeleton.Neither tumor enhances to a great extent with contrastinfusion.

Epidermoids usually have prolonged relaxation timeson both T1 and T2 weighted images on MRI. On the otherhand, dermoid tumors have variable relaxation times de-pending upon the amount of fat present but commonlyhave short T1 and prolonged T2 relaxation times. Theextent of tumor growth is probably better detected by MRIbecause of the lack of bone artifact.

PREOPERATIVE PREPARATIONThe occurrence of postoperative aseptic meningitis fol-lowing resection of posterior fossa tumors, especially in-clusion tumors, is well known; consequently, the routineperioperative use of steroids for resection of these lesionsseems reasonable. However, the use of perioperative anti-biotics continues to be controversial in clean neurosurgi-cal procedures. Although the mastoid air cells are fre-quently entered when attempting to gain far lateralexposure, these air cells are usually considered to be ster-ile. The use of antibiotics is probably justified in thosecases involving reoperation, when there has been a his-tory of mastoid disease, or following radiation therapy. Ifan antibiotic is used, a third generation cephalosporin maybe considered because of its ability to cross the blood-brain barrier and also provide broad-spectrum antibacte-rial coverage.

OPERATIVE PROCEDUREThe best positioning for the surgical procedure dependson the experience of the surgeon and anesthesiologist,the location of the lesion, and the medical condition ofthe patient. I utilize the prone and lateral positions forresection of almost every posterior fossa. lesion exceptfor midline lesions for which I tend to use the sittingposition (Fig. 1A). In the sitting position it is importantto have a precordial Doppler monitor to detect venousair emboli as well as to have central venous access toaspirate such emboli.

When resecting lesions in proximity to the seventhnerve, it is important to consider monitoring seventh nervefunction intraoperatively. There are monitoring deviceson the market which make an audible tone when the nerve

is being manipulated. This provides an on-line monitor-ing system so that the surgeon can be warned on his prox-imity to the nerve.

The placement of the scalp incision and the craniec-tomy site is predicated by the location of the mass. Mid-line lesions are typically resected via a medial suboccipi-tal craniectomy whereas lesions occurring In thecerebellopontine (CP) angle can be resected via a lateralretromastoid suboccipital craniectomy. Because of thegrowth characteristics and the consistency of these lesions,a single-staged operation is all that is required for resec-tion. For example, as the tumor grows it expands the sub-arachnoid space; consequently, a “channel” is created asthe center of the tumor is removed which allows furtherexposure allowing easier resection even in regions notroutinely available via a specific approach. A single-stagedresection, however, requires careful preoperative planningwith the aid of modern imaging modalities.

Fourth Ventricular TumorsMidline lesions, particularly those located in the fourthventricle, are best approached by the classically de-scribed suboccipital craniectomy (a craniotomy can of-ten be performed easily instead of a craniectomy, espe-cially in younger patients). Because of the superiorlydirected angle of the axis of the fourth ventricle, fourthventricular tumors are frequently best approached withthe patient in the sitting position, which allows both sur-geon comfort and adequate visualization. The scalp in-cision of the midline suboccipital craniectomy extendsapproximately 3-4 cm superior to the inion with the in-ferior extension to the level of approximately C-4 (Fig.1B). To achieve scalp hemostasis, disposable Raney clipsare used. Placement of these clips above the level of thesuperior nuchal line is relatively easy; however, sharplyundermining the subcutaneous tissue below this levelaids in their application.

The median avascular plane is then dissected with theaid of a needle tip cautery. Although initial dissection ofthe soft tissues in this region is performed with a needlepoint cautery, a broad, flat periosteal elevator andMetzenbaum scissors are used in the latter aspects of theexposure. It must be emphasized that only 1.5 cm of lateralexposure is required at the level of C-1 and the foramenmagnum, which lowers the risk of vertebral artery injury.

During the dissection in the region of the occipitalbone, foramen magnum, and C-1 lamina, venous bleed-ing is commonly encountered, requiring bone wax,gentle tamponade, or meticulous use of the bipolar cau-tery for control. It is imperative to properly addresseach bleeding source because one cannot safely oper-ate in this region under a “pool” of blood. Furthermore,sources of air embolism must be elimi-

399

KESTERSON : POSTERIOR FOSSA EPIDERMOID AND DERMOID CYSTS

Figure 1. A, the various positions that are available for resection of a poste-rior fossa lesion. The prone position (a) is good for caudally located poste-rior fossa and craniocervical junction lesions; the sitting position (b) is es-pecially good for vermian lesions; and the lateral (“park-bench”) position(c) is good for laterally located hemispheric lesions and cerebellopontineangle lesions. In the lateral position, the vertex of the head should not beangled toward the floor too greatly because of the potential for venous ob-

struction. Note the importance of padding pressure points and avoidingexcessive neck flexion (the latter is also important in preventing venousobstruction). B, the various scalp incisions that are available for approach-ing the different regions of the posterior fossa. One incision not depicted isa vertical incision that is located midway between the inion and mastoid,which gives good visualization of laterally located lesions.


400


nated whether the patient is in the prone or sitting po-sition.

As this region is progressively exposed, Weitlaner(curved) self-retaining retractors are utilized to provideadequate exposure and the delineation of dissection planes.Once the soft tissue dissection is completed, oneMiskimmons cerebellar self-retaining retractor usuallyprovides adequate exposure. To begin the craniectomy,burr holes are created with either an air-powered perfora-tor or a Hudson brace-at least one being placed on eachside of the midline which will avoid encountering a per-sistent occipital dural sinus. Then double-action (Leksell)and/or single-action (Adson) rongeurs are used to per-form the craniectomy. The craniectomy is carried to thelevel of the transverse sinus superiorly, through the fora-men magnum inferiorly, and near the mastoid process oneach side. The removal of the bone in the region of thesuperior nuchal line is facilitated by thinning the craniumwith the use of a Midas Rex pneumatic drill. For the vastmajority of lesions located superior to the foramen mag-num, a C-1 laminectomy is not required.

A Y-shaped dural incision is then created with thesuperolateral incisions being carefully carried to the levelof the transverse sinus bilaterally and with the caudal ex-tent of the midline incision being carried to near the levelof the superior ring of the atlas (Fig. 2A). A persistentoccipital sinus may be encountered when making this in-cision; consequently, it must be appropriately suture-li-gated (with new imaging modalities, metal ligature clipsshould be avoided; furthermore, good suture techniquesare usually easier to use).

Once the dura has been adequately opened and re-tracted, the cisterna magna is incised to allow drainage ofCSF and brain relaxation. Once the incision of the arach-noid of the cisterna magna is completed, lateral retractionof the cerebellar tonsils will allow entrance into the caudalregion of the fourth ventricle (Fig. 2B). Not infrequentlythe lesion is directly visualized at this point; however,completion of the exposure requires the incision of the cau-dal vermis providing access to the cephalad aspect of thefourth ventricle. It must be emphasized that total circum-ferential exposure of the lesion is not required at this point,but the lesion’s relationship to the floor of the fourth ven-tricle must be identified. It is at this point that a cottonoidis laid onto the floor of the fourth ventricle to provide pro-tection and to maintain identification of this cleavage planeduring tumor dissection. At this point a “ribbon” retractor(0.5-0.75 inch in width) is attached to a Leyla retractorarm as a self-retaining retractor. Moreover, it frequently iseasier to perform resection of a midline tumor if two Leylaretractors are utilized, with each providing retraction of acerebellar tonsil and the adjacent lateral vermis to the ipsi-

lateral side. At this point the tumor capsule is opened gen-erously and the “debulking” of the lesion is initiated (Fig.3A). The debulking process allows the tumor mass to bemore mobile and allow for a more atraumatic circumferen-tial dissection. Debulking of an inclusion cyst can usuallybe performed adequately with a conventional sucker butthe use of an ultrasonic aspirator will further facilitate thisactivity. The subsequent routine is not unlike the removalof other tumor types located adjacent to other vital neuraland vascular structures in that the debulking process is re-peatedly alternated with gentle circumferential dissectionuntil the lesion is removed. In this particular instance,cottonoids are placed repeatedly as the cleavage plane withthe floor of the fourth ventricle is extended. It cannot beoveremphasized that the cottonoids are not to be used asdissecting tools; on the contrary, the lesion is either gentlypulled away or sharply incised from its attachments (Fig.3B).

Once the lesion is removed, the attainment of hemo-stasis is easily achieved with gentle tamponade with ei-ther cottonoids alone or in combination with Surgicel.When operating in the ventricular system, little or noSurgicel should be left in the tumor bed.

For lesions with lateral extension, a scalp incisionthat is a variant of the midline craniectomy incision isutilized. This involves an extension of the superior limbto a position at the level of the posterior aspect of thepinna with the cephalad extent being 1-2 cm superior tothe superior nuchal line (Fig. 1B). This scalp incision givesexcellent lateral exposure while maintaining access to themidline. A craniectomy is then created that extends fromthe midline to the sigmoid sinus laterally, with the supe-rior extent being the exposure of the transverse sinus.

Cerebellopontine Angle TumorsThe removal of a cerebellopontine angle lesion is achievedby utilizing a paramedian retromastoid incision located 2-3 cm posteromedial to the mastoid process. This is a verti-cal incision that extends 3-4 cm cephalad to the superiornuchal line with the cephalad extension forming a gentlecurve anteriorly to a level at the posterior aspect of thepinna. The caudal limb extends 7-8 cm inferior to the supe-rior nuchal line with the caudal extension forming a gentlecurve medially (Fig. 1B). The gentle curves in the incisioncreate an S-shape which allows more lateral exposurethan a conventional straight vertical incision. Again, thetissue is undermined approximately 1 cm on each sideof the incision to allow placement of Raney clips. Theperiosteum and superficial muscles are dissected withthe electrocautery and a periosteal elevator (a pericra-nial graft can be obtained easily at this time for later usefor dural closure). The superf icial muscula-

401


Figure 2. A, a suboccipital craniectomy has been performed and a Y-shaped dural incision is depicted with the superior limbs extending tothe inferior aspect of the transverse sinus. The ring of the atlas is intact,with the posterior aspect of the foramen magnum having been removed.Note that one Miskimmons retractor is all that is usually required to

provide soft tissue retraction. B, the dural incision has been completedand the dural leaves are retracted. The cisterna magna is opened withthe fourth ventricular tumor visualized between the cerebellar tonsils.


402


Figure 3. A, the debulking process has begun, with two retrac-tors in place that retract the cerebellar tonsils. This process al-lows the mass to be more mobile, making its later dissection offof the floor of the fourth ventricle more atraumatic. B, the

debulked mass is removed sharply from the floor of the fourthventricle. Note the presence of a cottonoid that has been laid inthe cleavage plane between the lesion and the fourth ventricularfloor.


403


ture below the superior nuchal line (splenius capitis, tra-pezius, and sternocleidomastoid) are all that need be in-cised, with dissection of the deeper musculature (obliquesand rectus capitis) being reserved for lesions at the levelof the foramen magnum. The bony landmark for the infe-rior aspect of the soft tissue dissection is a small ridgethat is referred to as the inferior nuchal line. The extrac-ranial vertebral artery is located in a fatty triangle poste-rior to the arch of the atlas and usually is not exposed. Asingle Weitlaner retractor is usually all that is requiredfor retraction of the soft tissues.

A craniectomy is created with the superior aspect ofbone removal being carried to the asterion, the bony land-mark of the junction of the transverse and sigmoid si-nuses. The lateral extent of the bony removal involvesexposure of the sigmoid sinus even if significant removalof the mastoid process is required. Generous use of bonewax is required to occlude the mastoid air cells and pre-vent CSF leakage. The medial extent of the bony removalis approximately 4-5 cm medial to the mastoid process.

A K-shaped dural incision is created with the base ofeach limb again based on a venous sinus (Fig. 4A). Thisincision is created in such a way that the cerebellar hemi-sphere remains covered with dura. The dura is sutured tothe surrounding musculature to provide retraction.

With the aid of the operating microscope, the infe-rior cerebellopontine cistern is easily visualized by plac-ing a retractor blade along the inferolateral aspect of thecerebellar hemisphere and retracting cephalad and slightlymedially (Fig. 4B). The inferior cerebellopontine cisternis readily identified by following the spinal accessorynerve to the jugular foramen. The arachnoid is initiallyincised with a Beaver 5910 knife, with the extension ofthe dissection with bayoneted microscissors (the use ofscissors is probably less traumatic than the use of the knifealone). There is no need to cauterize the arachnoid layerbecause it is avascular. The opening of this cistern pro-vides brain relaxation as well as early identification ofthe lower cranial nerves (IX, X, and XI).

The capsule of the tumor is incised and debulked in asimilar manner to that described for the resection of a fourthventricular lesion (Fig. 5A). It is important to systemati-cally identify the vascular and neural structures as the re-moval progresses because of the tendency of these lesionsto insinuate themselves among these elements, making theirlocation unpredictable (Fig. 5B). As the mass is made to beincreasingly more mobile as the debulking process is per-formed, these critical structures are protected by laying acottonoid onto them. It must be emphasized that the litera-ture is unclear as to how much tumor or capsule removal isrequired to prevent the patient from having recurrent clini-

cal problems. Even the most aggressive surgeons occasion-ally leave remnants of capsule behind. Consequently, ag-gressive dissection around the neural and vascular struc-tures (especially the basilar artery and its small perforators)should be tempered by this fact.

In closing the wound, a watertight closure of the pos-terior fossa dura should be attempted, which may requirethe use of a patch graft. A multilayer closure of the paraver-tebral musculature and fascia is performed with an inter-rupted suturing technique.

OTHER SURGICAL ALTERNATIVESThere are multiple approaches to the resection of CPangle lesions. One that has been quite popular for al-most three decades is the translabyrinthine approach, pri-marily used in the resection of acoustic neuromas. Thisis impractical, however, if the preservation of hearing isdesired. Variations of this approach, such as theretrolabyrinthine trans-sigmoid approach, may be quiteadvantageous in those instances where a significant por-tion of the mass is located anterior to the brain stem.One of the limiting aspects of this approach is the con-tinued need to dissect and remove tumor between cra-nial nerves VII and VIII.

Another modification that may be required of thestandard paramedian suboccipital craniectomy is in thesituation where there is significant tumor extendingthrough the tentorial incisura. This lesion can be readilydealt with by extending the scalp incision and craniec-tomy a few centimeters above the transverse sinus. Thedural opening will include incisions into both the supra-and infratentorial spaces, ligation of the transverse sinus,and incision of the tentorial incisura. Without a lot of work,this modification gives a tremendous amount of expo-sure to the incisura.

Another limitation of both of the previously describedapproaches is the sacrifice of the ipsilateral venous drain-age. Consequently, a preoperative angiogram must beperformed to delineate the collateral venous drainagebefore either of these maneuvers can be performed safely.

Petrous apex lesions may present a dilemma in re-gard to surgical approaches. The transcochlear approachcould be an appropriate consideration if hearing is ab-sent. If hearing is present, however, the transzygomatic-middle cranial fossa approach could be used, but the co-chlea and the internal carotid artery must be avoided.

POSTOPERATIVE MANAGEMENTIt should be noted that even in the best of circumstances,a CSF fistula (usually presenting with rhinorrhea) orpseudomeningocele may occur and require an

404


Figure 4. A, a lateral suboccipital craniectomy has been performed anda K-shaped incision has been initiated. The base of each limb is locatedat a venous sinus. Note that only one Weitlaner retractor is required toprovide soft tissue retraction. B, a self-retaining Leyla retractor is lift-ing the cerebellar tonsil cephalad and slightly medially to expose the

inferior cerebellopontine cistern. This is readily located by identifyingthe spinal accessory nerve and following it to the jugular foramen. Thiscistern is opened early in the dissection because It allows brain relax-ation and identification of the lower cranial nerves. Note the caudal tocephalad direction of the arachnoid incision.


405

Figure 5. A, the debulking process has been initiated with the cra-nial nerves and vascular structures already identified as much aspossible. The petrosal vein is usually cauterized and transected earlyin the procedure to prevent its inadvertent avulsion during the re-traction of the cerebellum. B, the late stage of tumor removal is

shown. At this point, the tumor has been debulked and is very mo-bile, allowing it to be sharply incised from the cranial nerves andarteries. Note that a cottonoid has been laid in the cleavage plane tomaintain anatomical orientation as well as to provide protection tothe neural and vascular structures.


406


open surgical repair. Whenever one of these two compli-cations occurs, the presence of hydrocephalus must beconsidered. An initial trial of CSF drainage may be uti-lized for either problem because the impairment of CSFcirculation that may occur after a posterior fossa tumoroperation can be transient. Because each of these compli-cations may allow enough CSF decompression to preventthe development of ventriculomegaly, the possibility ofhydrocephalus developing following repair should alsobe considered.

The most frequency source of CSF leakage is thefailure to completely occlude the mastoid air cells coupledwith the lack of a good dural closure. When operating torepair a CSF fistula, the surgeon should address both as-pects of the problem to minimize the need for further in-tervention.

Pseudomeningocele formation occurs most fre-quently because of a small dural rent where there is a“ball-valve” effect with the subsequent formation of anextradural collection of CSF. The repair of these lesionsusually requires an open operation but this can be de-layed if the suture line is not compromised.

Postoperative follow-up evaluations will always berequired, particularly for those patients with a tumor ofthe cerebellopontine angle where fragments of the tu-mor capsule were left attached to cranial nerves and ar-teries. Because of the ability of these lesions to attain alarge size prior to causing clinical symptoms, imagingstudies must be incorporated into the follow-up evalua-tions. There has been no proven benefit to the adjuvantuse of radiation therapy following resection of these in-clusion cysts.

407

LUQUE ROD SEGMENTALSPINAL INSTRUMENTATION

EDWARD C. BENZEL, M.D.

INTRODUCTIONThe instrumentation of the unstable thoracic and lumbarspine has been revolutionized by the introduction of uni-versal spinal instrumentation techniques. The surgeon’sability to rigidly instrument the spine in a relatively safemanner has thus been greatly enhanced. Specific circum-stances exist, however, in which neither the “traditionalapproaches” such as Harrington distraction rod instru-mentation techniques nor the “newer” universal spinal in-strumentation techniques are ideal.

Sublaminar wire augmentation of Harrington distrac-tion rod techniques has occasionally proven to be usefulin some of these situations. It has been successfully ap-plied in circumstances where the chance for instrumenta-tion failure was high and simultaneously where the riskof application was outweighed by the benefits of increasedstability. The sublaminar wire fixation allows for the ac-quisition of substantial “supplemental” stability and ri-gidity over that obtained with the Harrington distractionrod system alone. Anterior migration of sublaminar hooksinto the spinal canal induced by ventral migration of therod into approximation with the lamina by the sublaminarwires, however, can result in dural sac compression. Thisobviously decreases the impetus to place such a construct.Therefore, it may only be indicated when both distractionand multiple level segmental fixation are truly necessary.

Luque rod spinal instrumentation techniques, how-ever, may be utilized effectively in patients in whom sub-stantial spinal instability exists preoperatively and inwhom the intraoperative acquisition of significant struc-tural stability is desired. This technique usually requiresthe application of a long moment arm (lever arm) to theunstable segment in order to be effective. This dictatesthat the construct be fixed at least two to three spinal seg-ments above and two to three spinal segments below theunstable segment.

The operative procedure, therefore, involves signifi-

cant surgical manipulations at multiple levels. This, inturn, may require a prolonged operative time. These fac-tors must be considered in advance of the surgery.

OPERATIVE INDICATIONSPatients considered to be appropriate candidates for aLuque rod spinal segmental instrumentation procedureare those with thoracic and/or lumbar spine trauma orcancer who harbor a grossly unstable spine. Patients withsubstantial spinal deformities which defy correction byroutine techniques are also candidates. The patient witha complete myelopathy (loss of all motor and sensoryfunction below the level of the injury) and an accompa-nying unstable spine is a prime candidate for the Luquerod technique. This latter patient has both the structuralcriteria for the procedure (gross instability) and simul-taneously is subjected to minimal neurologic risk fromthe placement of sublaminar wires (a complete myel-opathy already exists). The risks of sublaminar wireplacement are small but, nevertheless, ever present. Thepassage of the wire under the lamina obviously exposesthe underlying dura mater to inadvertent compression.Furthermore, following the sublaminar passage of thewire, its manipulation prior to obtaining its security viatwisting around the rod may result in ventral protrusionof the wire. This may result in wire impingement uponthe dura mater. Great care, therefore, must be takenthroughout nearly the entire operation with regard to wiremanipulation and the potential for injury to the duralcontents.

A patient with a fracture that is difficult to reduce issimilarly an ideal candidate for the Luque instrumenta-tion technique because it offers a better advantage fordeformity correction than any other available technique.This advantage, however, is at the expense of its degreeof difficulty and risk.

PREOPERATIVE PREPARATIONThe patient is prepared for surgery and anesthesia isadministered in a routine manner. The patient is post-© 1991 The American Association of Neurological Surgeons

408


tioned in a prone position on the operating table (althoughthe three-quarters prone or lateral decubitus position mayoccasionally be used). Care should be taken to preventabdominal compression. The utilization of generous rolls,special frames, or a variety of alternative positioning tech-niques may be helpful. The back is scrubbed and drapedin a manner that allows both the exposure of the spinalmidline and of the bone graft donor site (if necessary).

It is recommended that the anesthetic technique notinclude pharmacologic muscle paralysis. This allows forthe detection of muscle contraction induced by iatrogenicneural trauma. If such is observed intraoperatively, thesurgical manipulation which induced the event (such asthe overaggressive passing of a sublaminar wire) shouldbe avoided.

OPERATIVE TECHNIQUE

Preparation for Sublaminar Wire PassageThe desired length of the dorsal spinal midline is exposedin a subperiosteal manner. The interspinous ligaments,both above and below each lamina to be instrumented,are removed with a rongeur. This exposes the underlyingligamentum flavum. This structure, at each intersegmen-tal level, is then penetrated. A curved curette is useful forthis purpose. The underlying epidural fat and dura materare thus exposed. A Kerrison rongeur is then used to en-large the interlaminar space. The bone removal shouldnot be extensive enough to weaken the construct by ex-cessively diminishing the amount of laminar bone avail-able for purchase by the construct. It, however, must beextensive enough to allow for a safe passage of the wireunderneath the lamina itself. All potentially interferingsoft tissue must be removed from the area so that the wiresmay be passed with ease and without interference fromrough bony edges or soft tissue. This minimizes the chancefor passing the wire onto the dural sac and its contents.Palpation for rough spots which may impede passage ofthe wire with a large nerve hook or Woodson instrumentis recommended prior to passing the wires.

Sublaminar Wire PassageA 10-inch length of 1.2-mm wire is looped to form two 5-inch double strands. The looped wire is then passed under-neath the lamina in a caudad to cephalad direction (Fig.1A). The passage of the wire is facilitated by using a sturdyneedle holder. As the wire emerges from underneath thesuperior aspect of the lamina, its continued passage maybe further facilitated by first gripping the wire with a sturdynerve hook or Woodson instrument and then by using asecond sturdy needle holder to pull the wire through. Thewire is then cut with a wire cutter (Fig. 1B). This allows the

utilization of one-half of the wire on each side of midline.One must remember that the cutting of wires and the leav-ing of jagged metal edges increases the chances of glovetears or perhaps even skin cuts.

Two double strands of wire may be passed at eachlevel so that the complete double strand may be used oneither side of midline. This may occasionally be indicatedif the wire is thought to be the weakest link in the con-struct complex.

It is my preference to always pinch the wires togetherover the lamina as illustrated in Figure 1C (center), sothat inadvertent bumping of the wire does not cause itsprotrusion ventrally into the dural sac (Fig. 1C, left). It isalso recommended that the surgeon develop a conventionwhen placing the wires around the rods. My conventionis that the cephalad aspect of the wire is located mostmedially and the caudal aspect of the wire most laterally,such that the wires go around the rods similarly at everylevel of the spinal column as illustrated in Figure 1D.

Fixation of the Rod to the LaminaThe rod on each side of midline is cut to the appropriatelength with a bolt cutter and placed along the lamina. Thewires are twisted around the rods (Fig. 1D). “L”-shapedrods are routinely used. The “L” shape aids in spine set-tling and prevention of rod rotation.

Reduction of Spinal DeformityIn fractures that are difficult to reduce, the Luque rodsystem is a particularly appropriate choice, especiallyin those situations where neither distraction nor com-pression is desired. Figure 2 illustrates a technique wherethe rod on one side of the spine is initially tightened atits caudal end and the rod on the contralateral side issimultaneously tightened at its cephalad end (the L endof the rod is the end which is initially tightened withthis technique). The rods are then gradually secured bysubsequently tightening the wires in sequence (Fig. 2B).As the wires are sequentially tightened, spinal alignmentis gradually attained (Fig. 2, C and D). The most diffi-cult of spinal deformities to reduce can be aligned withsuch a technique.

Special Considerations and TechniquesL-shaped rods should always be used. The L should passacross the midline though an interspinous space and un-derneath the straight aspect of the opposing rod (Fig. 3A).This prevents the dorsal protrusion of the L aspect of therod and minimizes the chance that it will subsequentlymigrate.

Following the completion of the tightening of allwires, the bone fusion is placed. A ventral interbody

409

BENZEL : LUQUE ROD SEGMENTAL SPINAL INSTRUMENTATION

Figure 1. An illustration of some fundamentals of the application ofLuque rod instrumentation. A, the passage of the looped 1.2-mm wire isfacilitated by using a blunt-nosed needle holder, while its retrieval isfacilitated by the use of a sturdy nerve hook or Woodson instrument. B,the cutting of the looped wire with a wire cutter is illustrated. This isperformed in order to allow one-half of the wire to be used on each sideof the midline. C, the potential for wire impingement onto the dural sacand its contents is illustrated on the left (arrow denotes the potential for

ventral wire migration prior to tightening). This potential for neural com-promise is present from the time the wire is passed underneath the laminauntil it is tightened (twisted) around the rod (right). The wire may becrimped around the lamina in order to minimize its chance for causinginjury during this aspect of the operation (center; arrows denote thecrimping of the wire). D, a dorsal view of the relationship of the twistedwires to the rods. Note that superiorly the wires arise medial to the rodsand caudally the wires arise laterally.


410


Figure 2. An illustration of the Luque rod technique as applied to thecorrection of gross spinal deformities. When a significant spinal de-formity is present, as illustrated in A, closed or even open reductionof the spine may be very difficult. The open reduction of the defor-mity with the Luque rod Instrumentation technique is facilitated, un-der these circumstances, by bone and soft tissue removal. This desta-bilizes the spine enough to allow for reduction. Subsequent fusionallows for the eventual acquisition of bony stability. Following theremoval of the impediments to reduction and the passage of thesublaminar wires, the L rods are placed along the spine and cut to the

appropriate length with a bolt cutter. The L end of each rod is placedthrough the most superior and Inferior interspinous spaces, respec-tively. The wires are then twisted around the rods as illustrated in B.Beginning at the L end of each rod, the wires are gradually tightened(C) until acceptable alignment is obtained (D). Usually, a bony fusionis then performed. In D, an anterior fusion is illustrated. In patientswith cancer, bony fusion may not be an acceptable alternative. In thesecases, the chance for the acquisition of a solid bony fusion may be nil.An acrylic strut may be placed in selected cases. In others, no fusionat all may be appropriate.


411

BENZEL : LUQUE ROD SEGMENTAL SPINAL INSTRUMENTATION

Figure 3. An illustration of some “final touches” with regard to theapplication of Luque rod instrumentation. A dorsal view of a Luque rodsystem in place is illustrated in A. Note the convention of the wirespassing from caudad to cephalad in a lateral to medial direction. Anopposite convention would be equally appropriate. Also note that the Laspect of each rod lies underneath the straight end of the contralateralrod. This prevents the dorsal rotation of the L end of the rod whichcould easily result in its cephalad or caudad migration. In B, the place-ment of a ventral interbody fusion is illustrated. The surgical approachused allowed for a more than adequate decompression of the dural sacprior to placement of the bony fusion. The extent of dissection alsoallowed for the destabilization of the spine prior to placement of the

Luque rod system, thus facilitating the acquisition of appropriate spinalalignment. Subsequent stability is initially gained via the acquisition ofbony fusion. In C, a posterior fusion has been placed. A posterior fusionis used when ventral decompression of the dural sac is not necessary orwhen a more aggressive surgical approach (i.e., a ventral dural sac de-compressive operation) is not warranted. In D, the placement of a rigidLuque Crosslink is illustrated. This allows for the acquisition of a morerigid construct. Of greater importance with the Luque rod technique,however, is that it virtually eliminates the chance for significant rodmigration or rotation. Collapse of the Luque rod construct with timemay occasionally occur secondary to the rods slipping through the wires.The Crosslinking technique prevents this.


412


fusion may be placed, either via an anterior, anterolateral,or lateral extracavitary approach (Fig. 3B). If a dorsalfusion is placed, the wires may be turned over the graft inorder to assist in the attainment of its position and secu-rity (Fig. 3C). For most neurosurgical applications, thefusion of a short segment of the spine is indicated. Onlyif multiple level instability is present should a more ex-tensive fusion be entertained. The wound is then closedin multiple layers, usually with absorbable sutures. It isthe author’s preference to not place a drain. A drain is nota substitute for the adequate acquisition of hemostasis.Furthermore, it encourages wound infection by allowingan avenue for the entrance of contaminating microorgan-isms. This is of particular significance in patients whohave undergone the placement of hardware.

Evoked potential monitoring may be useful duringthe application of Luque rods, especially in cases wheresignificant deformity correction is planned. It must beemphasized, however, that many surgeons (including theauthor) believe that the monitoring of intraoperativeevoked potentials does not confer an element of safety. It,in fact, may be the source of anxiety for the surgeon,which, in turn, may lead to a suboptimally performed sur-gical procedure. Furthermore, once an alteration or oblit-eration of the evoked potential is observed, the neuro-logic injury is almost always already incurred. Furtherintervention is, therefore, unlikely to alter the outcome.

High-speed burrs may be used to assist in denuding thecortex of the dorsal spine in preparation for bone graft place-

ment. Specialized wire twisters are useful for the uniformtwisting of the wire. A rongeur or high-speed burr may beused to perform facetectomies at the level(s) of planned fu-sion. The greater the surface area for bony fusion, the greaterthe chance for the acquisition of bony union.

COMPLICATIONSInfection, neurologic injury, non-union, and the inadequateacquisition of deformity correction are complicationswhich may occur with any spinal instrumentation proce-dure. Strict adherence to standard neurosurgical opera-tive techniques and to the specific techniques outlinedhere should minimize, but not eliminate, the incidence ofthese complications.

Loss of height of the construct is a complicationwhich is particularly unique to Luque rod spinal instru-mentation. The neutral nature of the Luque rod constructimplies that it applies neither distraction nor compres-sion to the spine. It, however, does allow for the slidingof the rods through the wire loops. This, in turn, allowsfor the sliding of each rod past its counterpart and a re-sultant settling of the spine. This latter phenomenon maybe prevented by the application of Crosslinks, (Fig. 3D).The Luque Crosslinking System is rigid and, therefore,eliminates the sliding action of the rods by making theright and left rod into a single functional unit. The ap-plication of Crosslinks also allows for the acquisition ofa quadrilateral frame construct which substantially aug-ments its rigidity.

413

EN BLOC ANTERIORTEMPORAL LOBECTOMY

FOR TEMPORALIMBIC EPILEPSYMICHEL F. LEVESQUE, M.D.

INTRODUCTIONThe “Falconer-Crandall” resection of the anterior tempo-ral lobe is performed in patients with medically intrac-table epilepsy of temporal lobe origin. This procedure hasbeen demonstrated previously to be sufficient to achieveseizure control independent of intraoperative interictalspiking activity. The validity of this approach is based onthe excellent seizure outcome after resection based onpresurgical criteria of focal structural and functional ab-normalities and, in selected patients, the results of inva-sive monitoring studies. The advantage of this techniqueis that it allows surgery under general anesthesia with thesurgical microscope using sharp dissection to provide a“standard” surgical specimen. This approach has permit-ted scientific analysis of pathophysiological changes intemporolimbic structures and correlation with neuropsy-chological and seizure outcome.

PATIENT SELECTIONThe majority of patients with medically intractable epilepsywill present with a focal seizure disorder. This group usuallyhas stereotypical complex partial seizures with rare second-ary generalization. Advances in neuroimaging and invasiveneurophysiological studies with stereoelectroenceph-alography (SEEG) have led to a better understanding of theunderlying focal epileptogenesis and focal pathology ame-nable to surgical treatment. Criteria for patient selection havethus been widened to include patients with several types ofpartial simple, partial complex, and secondarily generalizedseizures. However, patients with chronic psychosis or se-vere mental retardation are usually excluded from surgicalconsideration. Surgery is performed in patients whose sei-zures have been medically intractable for at least two yearsunless there is evidence of a structural abnormality on mag-

netic resonance imaging (MRI) which may indicate the siteof seizure origin.

The en bloc anterior temporal lobectomy is one ofseveral surgical approaches to temporolimbic and tem-poral lobe neocortical epilepsies. Patients with a docu-mented seizure onset within the mesial temporal lobe maybenef it from a smaller resection such as anamygdalohippocampectomy. Patients who present withevidence of an extrahippocampal structural lesion may insome cases benefit from resection of the lesion only. Inmost cases, however, the mesial temporal structures areinvolved in the seizure onset and propagation and presentevidence of pathological changes. The most significantdamage occurs in the hippocampus.

A potential risk of this surgery is to operate on a tem-poral lobe without sufficient evidence of seizure local-ization or without clearly documenting the ability of thecontralateral hemisphere to support memory function. Thepurpose of the initial noninvasive evaluation (Phase I) aimsat providing presumptive evidence of a focal cerebral ori-gin of the seizure disorder. The strongest evidence is pro-vided by recording stereotypical seizures byvideotelemetry with scalp and sphenoidalelectroencephalograms (EEGs). Several types of seizureonset are suggestive of temporal lobe involvement. TheEEG can show an initial focal onset (Fig. 1), a delayedfocal onset, or a regional pattern. A structural abnormal-ity within the same temporal lobe seen with MRI is also astrong factor in the surgical decision-making process.Evidence of a focal functional deficit suggested by anarea of hypometabolism using 18-fluorodeoxyglucosepositron emission tomography also strongly suggests alimbic seizure to originate within the temporal lobe (Fig.2). The results of neuropsychological tests demonstratinglateralized dysfunction have also been correlated with thesite of seizure origin. The intracarotid amytal test (Wadatest) helps to define the language representation in thecerebral hemispheres and also documents mem-© 1991 The American Association of Neurological Surgeons

414


Figure 1. A spontaneous seizure captured on scalp and sphenoidal EEG leads in a patient with complex partial seizures. There is an initial focalonset at the right sphenoidal electrode.

ory deficit within the lobe involved and the ability of thecontralateral lobe to support memory function. If the pa-tient fails the memory test he is at an increased risk ofdeveloping global amnesia after resection of the epilep-togenic focus.

If all evidence points to the same temporal lobe, thepatient is then a candidate for an en bloc temporal lobec-tomy. However, if there is conflicting evidence from elec-trophysiological, radiological, or psychological testing,then the patient may have to undergo invasive intracra-nial recording (Phase II) to confirm the lateralized tem-poral lobe dysfunction. Because there is frequently evi-dence of bilateral temporal lobe involvement in seizurepropagation and bilateral underlying pathological hippoc-ampal changes, one single investigation should not berelied upon as the basis for a surgical decision.

PREOPERATIVE PREPARATIONAND OPERATIVE PROCEDUREAfter a complete neurological examination has been con-ducted and informed consent has been obtained, all docu-mented evidence is reviewed with the patient and his orher family. Anticonvulsant levels are verified and main-tained at an appropriate therapeutic value. In some cases,an additional load of anticonvulsants becomes necessarydue to a decrease in levels postoperatively following an-esthesia or due to decreased intake. Diphenylhydantoinis not routinely used as a loading agent prior to surgery.

Once of the patient is called to the operating room,he is given a dose of 10 mg of dexamethasone to de-crease postoperative brain swelling. In addition, thepatient receives a course of prophylactic antibioticmedication for the next 48 hours. After induction, the


415

Figure 2. Interictal 18-fluorodeoxyglucose positron emission tomogra-phy in a patient with complex partial seizures of left temporal origin.

There is significant relative hypometabolism involving the left temporallobe and extending into the whole left hemisphere.

patient is given general anesthesia. An arterial line and Foleycatheter are inserted. The patient is positioned supine witha small bolster under the ipsilateral shoulder, and the headturned to the side contralateral to the temporal lobe to beremoved. The head is turned approximately 30' from thehorizontal and the vertex is tilted inferiorly to bring thezygoma slightly above the level of the vertex (Fig. 3). Thehead is also maintained above the level of the right atriumto facilitate unobstructed venous drainage. The patient’shair is shaved in the operating room over the temporal, fron-tal, and parietal regions. A 10-minute scrub using slow-releasing iodophor solution is then applied. Sterile drapingis placed in a standard fashion.

The incision line is then infiltrated with 1% lidocainewith 1/200,000 epinephrine. A question mark incision isused beginning 5 mm in front of the tragus at the level ofthe zygoma (Fig. 3). The incision is then curved at ap-proximately 1 cm above the ear and extends posteriorlyto the anterior aspect of the mastoid to curve superiorlytoward the parietal region at approximately 5 cm from

the midline. The incision is then carried anteriorly towardthe frontal region just above the temporal line.

At first the scalp and galea are dissected and an at-tempt is made to preserve the superficial temporal artery.A subgaleal plane is easily defined and a skin flap is el-evated and wrapped in a moist laparotomy sponge. Thescalp is reflected forward with the use of fishhooks. Thetemporalis muscle is then incised longitudinally along thelength of its fibers (Fig. 3). This allows a smaller amountof muscle bulk anteriorly at the area of the craniectomyand appears to minimize local jaw pain postoperatively.The temporalis muscle is detached anteriorly and poste-riorly along its attachment to the superior temporal line.The muscle flaps are then reflected anteriorly and poste-riorly and also retracted with fishhooks. Both frontal andtemporal origins of the zygomatic arch are exposed.

A craniotomy is then created over the temporal andinferior frontal and parietal regions using a highspeeddrill and craniotome. Two burr holes are usuallyplaced, above and below the pterion, respectively. An

LEVESQUE : EN BLOC ANTERIOR TEMPORAL LOBECTOMY


416


Figure 3. Positioning of the patient for a left frontotemporoparietal cran-iotomy. The nose is tilted 30° from the horizontal plane and the vertexis lowered 30° from the midline. The skin incision is shaped in a ques-

tion mark fashion within the hairline. The temporalis muscle is splitlongitudinally, detached from the temporal line, and reflected anteriorlyand posteriorly.


417


additional burr hole may be needed at the most superiorand posterior aspect of the craniotomy. The bone is freedfrom the dura using a No. 3 Penfield dissector and a freebone flap is created. Suture holes are placed through thebone along the perimeter of the craniotomy as well asthrough the bone flap. The inferior aspect of the temporalcraniotomy may be extended anteriorly and inferiorly withthe use of rongeurs. This allows satisfactory exposure ofthe anterior temporal area and usually reaches the floorof the middle fossa. Care is taken not to enter the air cellsalong the mastoid bone and bone wax is placed if such acell is entered. Dural retention sutures are placed at theperimeter of the bony opening using 4-0 Nurolon.

At this stage dural tension is assessed. Intravenousmannitol (1 g/kg) is given routinely at the time of incis-ing the skin, and hyperventilation is used to create an end-tidal CO2 of approximately 30 mm Hg. Before the durais opened, saline irrigation is used over the exposed areaand bone dust or dried blood is washed from the surgicalgloves. Bacitracin-soaked sponges are also placed at theperimeter of the exposure, followed by sterile towels. Asemicircular incision is placed through the dura which isreflected anteriorly and held with a suture. The posteriordura is incised perpendicular to the C-shaped incision andthe dural leaves are then reflected superiorly and inferi-orly. These leaves are covered with moist cotton to pre-vent desiccation, thus allowing an easy dural closure.

The cortical surface is at first inspected and any lo-calized atrophy or lesion is noted. The vascular drainageof the temporal lobe is also studied, in particular the loca-tion of the vein of Labbé. The sylvian fissure is easilyidentified as well as the sylvian veins. The extent of re-section is measured from the tip of the temporal lobe atthe level of the anterior wall of the middle fossa along thesecond temporal convolution. On the dominant side, theplane of resection lies about 4.5-5 cm posterior to the tip,and on the nondominant side, about 5-6 cm posterior tothe tip (Fig. 4). The posterior margin may vary with thelocation of the vein of Labbé which is usually behind theplanned resection plane.

The posterior (inferior) incision is made at approxi-mately 45° to the anterior (superior) incision to spare pri-mary auditory cortex and possible speech-sensitive cor-tex over the superior temporal gyrus (Fig. 4). The initialpial incision is placed approximately 5 mm inferior to thesylvian veins and continues anteriorly parallel to the cur-vature of the sphenoid wing until the floor of the middlefossa is reached. The pia and bridging vessels are coagu-lated and sharply divided with microscissors. The supe-rior aspect of the cut pia is then elevated and subpial dis-section is made with the use of a small-bore aspirator andsmall retractor. Dissection of the mesial aspect of the su-

perior temporal gyrus is then carried along the sylviancistern, preserving the arachnoid layer over the branchesof the middle cerebral artery (Fig. 5, Step 1). The cortexis progressively lifted until the level of the insula is reachedat the limen insulae (Fig. 5, Step 2).

The surgeon’s attention is then turned to the poste-rior margin of the dissection where the incision is deep-ened vertically with a bipolar coagulator and suction. Theplane is progressively deepened into the temporal stemand inferiorly along the floor of the middle fossa (Fig. 5,Step 3). At this level the pia is gently elevated to avoidinjury to the underlying fourth cranial nerve. Dissectionis then carried medially until the edge of the tentorium isidentified. An orthostatic Greenberg retractor may be usedat this point to elevate the anterior portion of the tempo-ral lobe. Retraction over the posterior aspect of the tem-poral lobe is avoided to decrease the amount of brainswelling and, on the dominant side, postoperative speechimpairment.

The draped surgical microscope is then brought overthe surgical field posteriorly and the ependyma of thetemporal horn is carefully opened (Fig. 5, Step 4). Agush of clear cerebrospinal fluid usually emerges fromthe opening and a micro-cotton is placed in the openingto avoid any entry of blood or debris into the ventricle.However, hemostasis has been achieved at each step ofthe dissection and the surgical field at this point is usu-ally very dry. Once within the ventricle, the dorsal as-pect of the hippocampus is clearly visible and the orien-tation of the ventricle is studied. The roof of the ventricleis identified in relation to the limen insulae. By pro-gressively thinning the roof of the ventricle, completeexposure of the anterior ventricular horn is achieved,until the tip is reached. A retractor is then positioned toelevate the temporal lobe laterally; the horn is then dis-sected off the lateral amygdala. which is located ante-rior to the tip of the temporal horn. This dissectionbridges the initial exposure of the first temporal gyrusand is made subpially at the level just inferior to thesphenoid ridge. This will allow exposure of the carotidcistern as well as the oculomotor nerve, which is pre-served. A small cottonoid is left at this site for futurereference.

The microscope is then brought back to the levelof the posterior exposure. The third temporal convolu-tion inferiorly as well as the parahippocampal gyrusare then dissected to reach the floor of the middle fossa.The pia is progressively dissected until the edge of thetentorium is reached. At this level the dissection is car-ried in a subpial fashion more mesially and anothercottonoid is left in place for future reference. The sur-gical microscope is then brought over the mesial

418


Figure 4. Lateral view of the left hemisphere showing the extent of resec-tion of a standard anterior temporal lobectomy. A distance of 4.5-5 cm ismeasured over the second temporal convolution from the anterior wall of

the middle fossa of the dominant side (5-6 cm for a nondominant resec-tion). The hippocampal resection extends at least to the level of the lateralincision and usually reaches 3 cm.


419


Figure 5. Coronal representation of the anterior temporal lobe showingthe different steps completing the en bloc anterior temporal lobe resec-

tion. A key landmark is the temporal horn of the lateral ventricle whichwill give access through the choroidal fissure to the ambient cistern.


420


aspect of the hippocampus where the choroid plexus iselevated to expose the choroidal fissure as well as thefimbria-fornix (Fig. 5, Step 5). This last structure is usu-ally very thin and can be dissected with a microprobe or amicrobipolar forceps. In some cases, it can just be peeledoff to give access to the ambient cistern. At this point, asecond retractor is positioned over a nonadhesive pledgetto maintain the elevation of the choroid plexus. Gentleretraction is applied. The fornix is dissected longitudi-nally for a distance of approximately 3-3.5 cm. At thisposterior level the hippocampus is transected in the coro-nal plane using microbipolar forceps and a microaspiratoruntil the mesial hippocampal pia is identified. The sub-pial dissection is then carried inferiorly and laterally toreach the posterior cotton left previously.

In the ambient cistern, perforating vessels originat-ing from the posterior cerebral artery are seen within thehippocampal sulcus and are dissected from the main feed-ing vessels (Fig. 5, Step 6). These small perforators, calledAmmon’s horn arteries, are then taken progressively fromposterior to anterior to allow access to the presubiculumand parahippocampal gyrus regions. The hippocampus isgently retracted laterally and the pia is incised longitudi-nally over the parahippocampal gyrus which is then peeledoff progressively from posterior to anterior (Fig. 5, Step7). Anteriorly, some arteries may originate from the ante-rior choroidal artery but the major medial branches arepreserved. No attempt is made to identify the superiorlyplaced optic tract to prevent injury. Because of the curva-ture of the hippocampus at the level of the uncus, the an-terior dissection is carried more mesially in front of thebrain stem. Subpial dissection is then completed and thehippocampus can be elevated off the pia within the ambi-ent cistern. The whole specimen is elevated superiorlyuntil the edge of the tentorium is identifted (Fig. 6). There,the anterior cottonoid is again identified and usually asmall remnant of the pia has to be cut over the tentorialedge to obtain an en bloc specimen. The specimen is im-mediately placed into a cold solution of artificial cere-brospinal fluid (CSF). The hippocampus will again bemeasured and dissected from the temporal lobe and sentfor further studies.

The resection cavity is then inspected and hemosta-sis is completed, usually at the margin of the pial resec-tion. Care is taken not to coagulate over the tentoriumitself or close to the cranial nerves. Layers of oxidizedcellulose may be left over the pial surface of the insulaand sylvian cistern. Hemostasis proceeds from inferior tosuperior levels of the dissection. The resection cavity isfilled with saline irrigation followed with bacitracin irri-gation. The surgical cavity is filled with fluid and the durais closed in a watertight fashion using interrupted andcontinuous sutures. The bone flap is sutured back in place

with large more absorbable sutures and central tack-upsutures. No drains are left in place. The temporalis muscleand fascia are then closed using 3-0 Vicryl sutures alongthe temporal line. The skin is closed in two layers and alocal dressing is applied.

The patient is sent to the recovery room after extuba-tion and is subsequently monitored in the intensive careunit overnight. Hourly neurologic checks and vital signmeasurements are carried out. A postoperative computedtomography (CT) or MRI scan is obtained the followingday (Fig. 7).

COMPLICATIONS AND OUTCOMEThe most frequent side effect of the temporal lobectomyis the production of a superior quadratic homonymousfield defect that is innocuous to the patient. This “pie inthe sky” defect varies with the posterior extent of the dis-section; a temporal lobectomy rarely produces a completehomonymous hemianopia. This is usually prevented bylimiting the posterolateral dissection just above the ven-tricle and avoiding the optic tract and optic radiation.

A second complication is the production of asepticmeningitis. This occurs in approximately 10% of oper-ated patients and typically begins with a spiking fever,neck stiffness, and photophobia. on the fourth to seventhpostoperative day. Typically the lumbar puncture will showxanthochromic CSF with significant leukocytosis and adecreased glucose content. However, all cultures remainnegative. These patients are treated by increasing the ste-roid dosage and using wide spectrum antibiotic therapyuntil it is determined that the cultures are negative. Thebest preventive measures are to achieve complete methodi-cal hemostasis and to leave the pia layer above the ambi-ent cistern.

Extraocular nerve paresis may also occur if care isnot taken with the dissection in the region of the third andfourth cranial nerves. This is usually transient and ex-traocular movements are back to normal within the firstsix weeks.

“Manipulation hemiplegia” has been described aftertemporal lobectomy. This has been attributed to tractionon the sylvian branches over the insula, or to delayed va-sospasm of one of these branches. Another possible causeis damage to the mesial branch of the anterior choroidalartery which supplies the lateral part of the cerebral pe-duncle.

Another frequent side effect is a short-termmemory impairment. This has been found to affect ver-bal memory on the dominant side and nonverbalmemory on the nondominant side. However, 30% ofpatients present with such a deficit prior to surgery,with transient worsening postoperatively which theni m p r o v e s

421


Figure 6. Surgeon’s view of an en bloc specimen after a left anteriortemporal lobectomy. The fimbria-fornix has been detached and the am-bient cistern is reached to expose the Ammon’s horn arteries reaching

the hippocampal sulcus. The lateral retractor lifts the roof of the ven-tricle and the medial retractor is gently positioned over the choroid plexusand insula.


422


Figure 7. Postoperative MRI views (T1-weighted images) showing themesial and posterior extent of the hippocampal resection on the sequen-

tial sagittal views. The lower right view is an axial image showing theposterior extent of a nondominant right anterior lobe resection.

during the following year. More severe global amnesia isusually prevented by documenting preoperative memoryfunction with the intracarotid Amytal test. Patients who failthis test usually have evidence of bilateral hippocampal dam-age and should not undergo mesial temporal lobe resection.Receptive aphasia or dysphasia may occur on the second tofourth day after an operation on the dominant side. This isrelated to the degree of brain edema and resolves when theedema disappears. When the epileptogenic lesion appears to

arise from the neocortex close to or within Wernicke’s area,surgery should be performed under local anesthesia to ob-tain a complete mapping prior to resection.

The overall morbidity of the procedure is below 5%and there have been no deaths following this procedure atour center. The outcome of this procedure has been ex-cellent (Class I and II of Engel) for at least 80% of pa-tients and with additional functional criteria, the successrate now reaches above 90%.


423

CINGULOTOMY FOR INTRACTABLEPAIN USING STEREOTAXIS GUIDED

BY MAGNETIC RESONANCE IMAGINGSAMUEL J. HASSENBUSCH, M.D., PH.D.

PREM K. PILLAY, M.D.

INTRODUCTIONPain relief in cancer patients can be difficult to achieve andmaintain. Oral narcotics remain the mainstay of treatmentbut can be limited by the development of tolerance to rela-tively large doses. Side effects, especially somnolence, nau-sea, and vomiting, often limit the degree to which the doseof systemic (e.g., oral, rectal, subcutaneous, intravenous)narcotics can be increased. These problems are particularlytrue of the terminal cancer patient with unresectable, dis-seminated disease. Frequently all treatment options, bothmedical and operative, will have been exhausted in thesepatients without provision of adequate comfort and theability to perform simple daily activities.

Operative options in these patients have included nerveblocks, implantable morphine (or other narcotic) pumps,and various surgical ablative operations. The list of neuro-surgical ablative operations in these patients is a lengthyone: neurectomy, rhizotomy, myelotomy, cordotomy, med-ullary tractotomy, mesencephalotomy, thalamotomy, andcingulotomy, among others. The usefulness of these opera-tive options is often limited by the cardiopulmonary risksof a major operation under general anesthesia in these de-bilitated patients. The added risk of a neurologic deficitwith these open operations is also an important consider-ation. For these reasons, major ablative operations are of-ten not recommended in such patients.

In search of a safe, effective operation for cancer pa-tients with intractable pain, the procedure of cingulotomyhas been modified using stereotaxis guided by magneticresonance imaging (MRI) (Fig. 1). Although lesions ofthe cingulate gyrus have been created in the past, the le-sions have been placed using ventriculogram guidance.This required ventricular puncture with the possible risksof meningitis and hemorrhage. Ventriculogram guidance

to the cingulate gyrus was also indirect and potentiallyinaccurate. The operation was somewhat cumbersome andusually required general anesthesia.

A new technique for the placement of these lesionsusing MRI guidance in conjunction with a stereotacticlocalizing system is presented. The operation can be per-formed under local anesthesia, is effective, and involvesminimal risk to the patient. Since most neurosurgeons aremore familiar with and more likely to have access to theBrown-Roberts-Wells (BRW) and Cosman-Roberts-Wells(CRW) systems, the technique was initially developed withthese systems. The technique can also be easily adaptedto the Compass Stereotactic (Kelly-Goerss) System whichallows a simple and reproducible ring reapplication.

PATIENT SELECTIONFollowing its development, this technique has been usu-ally applied to patients who: 1) have terminal cancer withmidline, bilateral, and/or diffuse body pain; 2) are ineli-gible for resection of any tumor; 3) have had inadequatepain control by non-narcotic medications (e.g., anti-depressants, nonsteroidal anti-inflammatory agents, non-narcotic analgesics); 4) have had inadequate pain controlby oral, rectal, and/or intravenous preparations of nar-cotics without significant side effects (e.g., nausea, vom-iting, somnolence, headache, respiratory depression); 5)have no significant pain components that would be ame-nable to special nerve block or denervation procedures;6) have an expected survival time of at least one month;7) have received maximal radiation therapy to the painsites or have such extensive disease that radiation therapyis not practical; 8) have no contraindications to an in-tracranial ablative operation (e.g., previous craniotomyand brain resection, significant untreated hydrocephalus,intracranial vascular malformation, signif icant ab-nor mal anatomy (as shown by computed to-© 1991 The American Association of Neurological Surgeons

424


Figure 1. Graphic illustration of the placement of the cingulotomy elec-trode through the cortex so that the electrode tip is located in the centerof the left cingulate gyrus. Note the relative positions of the 10-mm

exposed tip of the electrode (curved arrow), the distal portion of ananterior cerebral artery (closed arrow), and the lateral ventricle (openarrow).


425

HASSENBUSCH AND PILLAY : CINGULOTOMY FOR INTRACTABLE PAIN

mography (CT) or MRI) in the cingulate gyri, or infectionin the scalp, skull, or brain); and 9) have no significantpsychiatric illness that would interfere with the treatment.As can be seen from the above list, these patients are of-ten the type for whom long-term intraspinal (e.g., epidu-ral, intrathecal) infusions of narcotics (often with an im-planted programmable infusion pump) would beconsidered. In such a situation, both options are often of-fered to allow the patient to choose according to indi-vidual preference between an essentially one-time intrac-ranial ablative operation and an intraspinal operationrequiring postoperative pump refills and rate adjustments.It should also be noted that the use of one of these optionsdoes not preclude a later use of the other option shouldpain control become more difficult as the underlying neo-plasm progresses.

The patients are usually evaluated and treated by amultidisciplinary team composed of a neurosurgeon, psy-chiatrist, anesthesiologist, palliative care oncologist, andnurse clinician. Evaluation of the patient’s pain, both pre-operatively and postoperatively, is a very important as-pect of this technique. A pain scale (e.g., a visual analogor a verbal digital scale) is usually utilized to quantitatethe degree of pain preoperatively and at multiple pointsafter the operation. The verbal digital scale, which is morefacile for routine visits, consists of asking the patient torate the pain on a scale of 0 to 10, where 0 represents nopain and 10 the worst imaginable pain. The second partof pain assessment consists of the use of a McGill PainQuestionnaire to assess differential improvements in thesensory (Groups 1-10 on the questionnaire) or the affec-tive (Groups 11-15) categories.

PREOPERATIVE PREPARATIONBecause of the relatively simple and noninvasive natureof this technique, very little preoperative preparation isrequired. A preoperative CT or MRI scan with contrast(iohexol (300 mg of iodine/ml), or gadolinium DTPA(0.1 mmol/kg), respectively) is usually performed toconfirm the absence of significant vascular or anatomi-cal abnormalities that would preclude accurate and safestereotactic localization in the cingulate gyri. A prophy-lactic antibiotic is routinely given with one dose, usu-ally of cefamandole nafate, in the operating room im-mediately prior to the skin puncture for stereotacticelectrode placement. Postoperative antibiotic therapy iscontinued every six hours for four more doses.Anticonvulsants are not given unless there is a previoushistory and treatment of seizures. Steroids are also notgiven because the lesions, using the present technique,cause very little perilesional edema.

OPERATIVE PROCEDURE

Anesthetic TechniqueThe major advantage of this technique for pain control interminally ill patients with cancer is that it can be easilyperformed under local anesthesia with some intravenoussedation. Xylocaine (1% without epinephrine) is usuallyutilized for local anesthesia of the scalp where the pins ofthe stereotactic arc are screwed into the outer table of theskull. Similarly, 1% Xylocaine without epinephrine is alsosufficient for the scalp where the puncture site is madefor the insertion of the radiofrequency electrode. The useof a twist drill to make the small hole in the skull and thepuncture of the underlying dura mater are usually pain-less or minimally painful and do not require anesthesia.The actual passage of the electrode through the brain islikewise painless.

To facilitate patient cooperation and comfort, theoperation is routinely done with the presence of an anes-thesiologist who administers intravenous sedation. A com-bination of midazolam HCl and fentanyl citrate is usuallygiven as needed to ensure patient comfort. Because of themedical condition of the usual patient, the dosages of thesemedications are limited as much as possible although it isnot required that the patient be awake at any point duringthe operation.

In the MRI suite, monitoring of the patient consistsof an automated blood pressure monitor and a pulse oxime-ter with an anesthesiologist or anesthetist present. In theoperating room during creation of the actual lesion, anelectrocardiogram monitor is also added. Further special-ized monitoring is usually not necessary.

Specialized InstrumentationThis procedure utilizes standard BRW-CRW stereotacticequipment for all parts. Each of the pieces of equipmentis labeled by letter in Figure 2. Specifically, the MRI headring with MRI head ring posts and MRI head fixationscrews (A), MRI localizer cage (B), CT head ring (C),Mayfield adaptor and screws (D), and the BRW stereo-tactic arc (E) are all standard equipment for any stereo-tactic procedure under MRI guidance utilizing the BRWsystem. A standard block (F) is placed in the holder at thetop of the arc. A standard guide tube (G) is secured in theblock. The bone drill bit (H) and the BRW arc systempointer (I) which are used to make a hole in the skull andthe dura mater are also standard pieces. The actual elec-trode (J) is a thermistor (type TM) straight radiofrequencyelectrode with a 10-mm exposed tip. It is connected to astandard radiofrequency wire (K) which is then connectedto the radiofrequency generator.

426


Figure 2. Equipment used for stereotactic cingulotomy. Shown are: MRI-compatible head ring with head ring posts and head fixation screws(A); MRI-compatible localizer ring (B); CT-compatible head ring (C);Mayfield adaptor with attachment screws (D); BRW stereotactic arc

(E); standard block for the arc (F); standard needle holder (G); bonedrill bit (H); BRW arc system pointer (I); thermistor (type TM) straightradiofrequency electrode with 10-mm exposed tip (J); and standard elec-trode wire for connection to the radiofrequency generator (K).

Operative Technique

Target Data AcquisitionA standard MRI-compatible BRW head ring (with MRI-compatible head fixation screws) is applied to the patient’shead under local anesthesia (Fig. 3A). As the nylon headfixation screws are applied to the MRI-compatible headring, an assistant holds the head ring steady, making surethat the ring is placed symmetrically on the patient withminimal or no rotation or inclination to one side. It is im-portant to be sure that the ring is located so that the BRWarc has sufficient clearance over the top of the patient’shead to allow movement of the arc and the block holderover the head. The MRI localizer ring is then applied to thehead ring (Fig. 3B). This part of the operation takes placein a small induction room adjacent to the scanner suite.

The MRI scan is performed using a GE Sigma1.5 Tesla MR scanner with a 32-cm scan circle. A T1

signal (TR 500 ms, TE 15 ms) in the coronal plane,without intravenous contrast, is utilized with 5-mm-thick sections spaced every 6 mm. The location forthe coronal sections is chosen from a scout midlinesagittal scan with the most anterior section located infront of the corpus callosum (Fig. 4A). A total of ap-proximately 11 slices is required. The most anteriorcoronal slice containing any portion of the frontalhorns of the lateral ventricles is determined (Fig. 4B).The coronal slice that is four sections more posterioris then chosen as the target slice (Fig. 4C). This givesa cingulate gyrus target that is 24 mm posterior fromthe tips of the frontal horns.

On this target slice, the coordinates of the fiducialsare determined using a software program (resident onthe MRI scanner) with the ability to translate any cur-sor-labeled point on the MRI screen into cartesian coor-dinates. The f iducial coordinates are determined


427


Figure 3. A, placement of the MRI-compatible BRW/CRW head ringon the patient’s head. Note the head ring posts (small arrowheads) andhead fixation screws (large arrowheads) securing the ring to the outertable of the skull. This is usually performed using local anesthesia but

can be carried out under general anesthesia. B, the MRI-compatibleBRW/CRW localizer ring (arrowheads) is attached to the head ring priorto the performance of the MRI scan.


428


Figure 4. An MRI scan (without intravenous enhancement) using a T1signal (TR 300 ms, TE 15 ms). The scan was performed using a GE Sigma1.5 Tesla MR scanner with a 32-cm scan circle. A, a scout sagittal midlineview: the location for the coronal sections is chosen with the most anteriorsection located in front of the corpus callosum. A total of approximately 11slices is required. B, the most anterior coronal slice containing any portionof the frontal horns of the lateral ventricles (open arrows) is shown. Coronal

sections were 5 mm thick and spaced every 6 mm (i.e., 1 mm between eachslice). C, the coronal slice that is four sections more posterior to that shownin B is chosen as the target slice and gives a cingulate gyrus target that is 24mm posterior from the tips of the frontal horns. The cursor (shown) is placedon the center of the right cingulate gyrus. The fiducial points for the MRIlocalizer ring are seen on the scan. The hollow point fiducial (closed arrow)is the first fiducial point.


429


beginning with the hollow-point fiducial (Fig. 4C, arrow)and continuing to the next nearest fiducial. The center ofeach cingulate gyrus (right and left) is chosen as a targetand appropriate coordinates are likewise determined (Fig.4C, cursor). On completion of the MRI study, the patientis transferred to the operating room.

Calculation of Target Point CoordinatesThe coordinates for the fiducials of the MRI localizer cageand for the right and left cingulate gyrus targets are thenentered into the BRW system laptop computer, which isan Epson HX-20 portable computer. The actual coordi-nates obtained from the MRI scanner software are en-tered into the Epson computer using the TMRI-86A pro-gram and selecting the coronal MRI option in the mainmenu. This produces a set of BRW system coordinates(anteroposterior, lateral, and vertical numbers) for eachtarget (cingulate gyrus) site. These numbers are then en-tered into the Target Menu of the Epson computer usingthe STAX-84K program. By proceeding then to the Ap-proach Menu of the same program and using the Azi-muth/Declination Method, the actual approach parameters(alpha, beta, gamma, delta, and target depth) of the tar-gets (right and left cingulate gyri) can be obtained. Forthe right cingulate gyrus, an azimuth of 45°and a declina-tion of 45° is used. For the left gyrus, an azimuth anddeclination of 3150 and 45', respectively, are specified.The actual approach parameters are then set on the BRWstereotactic: arc. The phantom base is then used as adouble-check for correct arc settings with the BRW arc.BRW system coordinates (anteroposterior, lateral, andvertical) are set on the phantom base. The arc is then placedon the phantom base and the arc’s settings verified with aprobe placed through the block in the BRW arc. The depthof the probe (and subsequent electrode) is calculated fromthe depth given by the actual approach parameters addedto the height (usually 10 mm) of the block on the arc addedto the further height of the guide tube in the block for theprobe or electrode.

The CRW arc can also be used in a somewhat moresimple and shortened method. BRW system coordinates(anteroposterior, lateral, and vertical) can be directly seton the CRW arc without the need for using the STAX-84K program to obtain approach parameters. The CRWarc can then be placed on the patient’s head and the entrypoint adjusted to use a point located behind the hairline,in front of the coronal suture, and at least 3 cm lateral tothe sagittal sinus. For safety, it is recommended that thearc be tested against the phantom base, which is set up asdescribed above for the BRW arc. The depth of the probe(and subsequent electrode) is calculated from 160 mm (astandard value for the CRW system) added to the height

(usually 10 mm) of the block on the arc added to the fur-ther height of the guide tube in the block for the probe orelectrode.

Creation of LesionsIn the operating room, the patient is placed in the supineposition, with the head slightly flexed and the nose point-ing upward without rotation. The CT-compatible head ringis then placed on top of the MRI-compatible head ringthat is attached to the patient. This CT head ring is neces-sary both for stabilization of the patient’s head to the op-erating room table as well as for attachment of the actualBRW arc to the patient. The CT head ring is then con-nected by the Mayfield adaptor (an attachment plate) to aMayfield head holder apparatus (Fig. 5, A and B, closedarrow) that is secured to the operating table.

Both frontal scalp areas are then shaved, cleaned, andprepared with an iodine solution. A iodophorimpregnatedplastic craniotomy sheet (Ioban2 Antimicrobial Film) isdraped over the top of the patient’s head and over the headrings. For the purposes of this procedure, no further ster-ile drapes are needed over the patient. Openings are madein the Ioban sheet for the CT head ring’s three holes whichare used to secure the BRW arc to the patient. The BRWarc, after being sterilized, is placed over the patient’s headand the three pegs on the underside of the arc are securedto the CT head ring. The block and the guide tube (Fig.5A, curved arrow) are then added to the top of the arc(Fig. 5, A and B).

On the left side, the skull entry point is marked onthe scalp and a small skin puncture created down to theperiosteum with a No. 15 scalpel blade. A 3-mm twistdrill hole is made in the skull at the same site by placingthe drill bit through the guide tube in the BRW arc. Thedura mater is then gently perforated with a trocar or theBRW arc system pointer.

A thermistor (type TM) straight radiofrequency elec-trode with a 10-mm exposed tip is then passed throughthe guide tube in the BRW arc (Fig. 5, A and B, openarrow indicates electrode hub). It is directed through thehole in the skull so that the center of the exposed elec-trode tip is centered at the target, namely, the center ofthe cingulate gyrus (Fig. 1). This electrode is connectedby wire (Fig. 5B, double arrows) to a radiofrequency gen-erator (Fig. 6).

A lesion is made using an electrode temperature of75°C for 60 seconds. The exact same operation is thenperformed in the right hemisphere. A single stitch isplaced in each scalp puncture site. The patient is ob-served overnight and discharged the next morning. Fol-low-up MRI scans using a T1 signal (TR 600 ms, TE 30ms) and a spin density signal (TR 2000 ms, TE 32

430


Figure 5. Intraoperative pictures during the actual creation of the le-sions. The head ring is attached by a Mayfield adaptor (covered by thedrape) to a Mayfield holder (solid straight arrow) which is connectedto the operating table. A, the insertion of the electrode (open arrow)

into the guide tube (curved arrow) of the stereotactic arc. B, after inser-tion of the electrode and at the time of actual lesion creation, the hub ofthe electrode (open arrow) is connected by wire (double arrows) to theradiofrequency generator.

Figure 6. A radiofrequency generator showing the settings during the actual lesion production. The temperature is maintainedat 75°C for 60 seconds.



431


Figure 7. Postoperative MRI scans without contrast, using a T1 signal(TR 600 ms, TE 30 ms). This patient had no postoperative neurologicdeficits. A, coronal view: the location of the lesions (arrows) corre-

sponds closely with the stereotactic targets (center of each hemisphere’scingulate gyrus). B, sagittal view: the right-sided lesion (arrow) is shownin this slice.

ms) are performed in the coronal and sagittal planes with-out contrast one week later to evaluate placement of thelesions (Fig. 7).

COMPLICATIONSOne of the major strengths of this operation is the relativesafety of the cingulate gyrus area when creating lesions.In our experience, this operation using MRI guidance andthe radiofrequency settings described above has been with-out complications in these terminally ill cancer patientswith intractable pain. One early patient, treated with threeoverlapping lesions created in each gyrus using a higherelectrode temperature (85° C) for a longer time (90 sec-onds) than described above, did experience global apha-sia for five days which then resolved over the next fivedays without subsequent problems. Despite the close prox-imity of the electrode tip to the anterior cerebral arteriesand the lateral ventricles (Fig. 1), no complications haveoccurred in relation to either of these structures.

Cingulotomy using ventriculogram guidance has beenpreviously reported in many older studies involving a to-tal of more than 1000 patients. The studies included bothneurologic and cortical function (psychometric) testing.

In these reports, the mortality rate was 0.09% and one-sided paralysis rate (caused by bleeding) was 0.36%. Astudy from the Massachusetts Institute of Technology in-dicated that, in 137 patients undergoing bilateralcingulotomy, the only lasting behavioral deficit seen af-ter the operation consisted of difficulty in copying a com-plex drawing on the ReyTaylor Complex Figure Test. Thischange was seen in a small minority of patients and wasof minimal clinical significance to the patients.

Because these patients have usually been on longtermlarge doses of narcotics and because the operation oftengives marked, almost immediate, pain relief, great carehas to be exercised in ensuring an adequate and slow nar-cotic detoxification to avoid withdrawal symptoms in thepostoperative period.

CONCLUSIONSThe adaption of stereotaxis using MRI guidance forthe creation of cingulate gyrus lesions is an improve-ment that potentially allows the routine creation of theselesions in cancer pain management. The direct visual-ization on MRI scanning of the cingulate gyri, both forstereotactic localization of the lesions and for


432


postoperative evaluation of the placement of the lesions,is a major advantage. The technique is simple, uses localanesthesia, and has been effective and safe in patientstreated thus far. The use of local anesthesia is especially

important in the terminally ill cancer patient who is oftenmedically unstable. This improved technique may allowan increased use of this operation and better pain controlin patients with intractable neoplastic pain syndromes.

433

CEREBELLAR ASTROCYTOMASA. LELAND ALBRIGHT, M.D.

PATIENT SELECTIONCerebellar astrocytomas occur in the cerebellar hemi-spheres and in the vermis, and in either site frequentlycause hydrocephalus, which causes recurring headaches,vomiting, and drowsiness. Astrocytomas in the cerebel-lar hemispheres often cause appendicular ataxia or tremu-lousness; those in the vermis may cause truncal ataxia orno cerebellar signs. On computed tomography (CT) andmagnetic resonance imaging (MRI) scans, astrocytomasenhance brightly after contrast injection and are oftenpartially cystic (Fig. 1).

Operations are performed to diagnose and removethe tumor. Cerebellar astrocytomas are perhaps the mostsurgically curable tumors; there is no role for subtotalresection and postoperative irradiation unless the tumorcannot be excised safely. The timing of operation dependsprimarily on the severity of the associated hydrocephalus:the greater the hydrocephalus, the greater the urgency.Children who are lethargic from hydrocephalus second-

ary to a posterior fossa tumor should be operated on within48 hours.

Potential risks of operation include death (-0.5%),wound infection (1%), ventriculitis/meningitis (23%),cerebrospinal fluid (CSF) leakage (1%) cerebellar signs(permanent 10%, transient 25%) and the pseudobulbarposterior fossa syndrome (3-5%).

PREOPERATIVE PREPARATIONPatients are usually treated for at least 24 hours preopera-tively with corticosteroids, which decrease peritumoraledema and may improve symptoms of hydrocephalus.Seizures are an infrequent problem in treating either cer-ebellar astrocytomas or their associated hydrocephalus,and prophylactic anticonvulsants are not needed. Preop-erative antibiotics are not needed; several studies havedemonstrated that intraoperative antibiotics effectivelydecrease the risk of infection.

In the past, hydrocephalus was often treated with a


Figure 1. Axial, coronal, and sagittal magnetic resonance images of a partially cystic juvenile pilocytic astrocytoma.


434


CSF shunt several days before the craniotomy. That is in-frequently appropriate now, because hydrocephalus istreated effectively by an external ventricular drain insertedintraoperatively just prior to tumor removal and postop-erative shunts are needed in only 25% of patients afterremoval of cerebellar astrocytomas.

SURGICAL TECHNIQUE

AnesthesiaPreoperative sedation is usually not given to patients withposterior fossa tumors, especially those with hydroceph-alus, because sedatives may depress respirations, causingarterial PCO2 and intracranial pressure to rise. After pa-tients are in the preoperative “holding area” and are undercontinuous close observation, intravenous sedation may begiven, often with midazolam. Two peripheral venous cath-eters provide the customary venous access. There is debateabout the role of central venous catheters: they provide in-formation about intravascular volume and allow aspirationof some air in case of venous air embolism, but their inser-tion via the internal jugular or subclavian vein risks a he-matoma that may compress the vein and increase venousand intracranial pressure, risks a pneumothorax, and theyprobably do not permit the aspiration of adequate volumesof air. Pulse oximeter monitoring is standard. An arterialcatheter is commonly inserted for continuous blood pres-sure monitoring and for withdrawal of blood samples.

Anesthesia is usually induced with thiopental andmaintained with a mixture of intravenous fentanyl plusnitrous oxide or with an inhalation agent such asisoflurane. The chest should be auscultated after the headis in its final position; the neck is commonly flexed ineither the prone or sitting position and the endotrachealtube may advance into the right mainstem bronchus afterthe neck is flexed. A precordial Doppler monitor is oftenapplied to detect air embolism, which causes a character-istic washing-machine sound. If air is detected, the op-erative field can be flooded with saline and the venouspressure can be increased, either by lowering the head ofthe operating table or by jugular venous compression.

Operative PositioningIf the patient has hydrocephalus or if hydrocephalus islikely to develop in the postoperative period, an externalventricular drain (EVD) should be inserted before the cran-iotomy is performed. That drain can be inserted into thefrontal horn of the lateral ventricle through a small holeat the junction of the coronal suture and the pupillary line,aiming in an anteroposterior trajectory 1 cm anterior tothe ear and a transverse trajectory so that the catheter endsat the midline. The distance from the skull to the foramen

of Monro can be estimated from the CT scan and is usu-ally 5-6 cm. The distal end of the catheter should be drawnseveral centimeters subcutaneously before it exits througha stab wound and is then connected to a closed drainagesystem. The system is closed during positioning of thepatient for craniotomy but is set to drain at approximately10-15 cm above the lateral ventricles; the EVD is openedduring the craniotomy if the dura is tense.

Surgeons who prefer the patient in a sitting positionmay insert the EVD catheter via the occipital approachonce the patient is in position, inserting it through a burrhole 6 cm above the inion and 2.5 cm lateral to the mid-line, aiming at the ipsilateral inner canthus. This tech-nique is usually successful if the ventricles are large, butother-wise there may be difficulty entering the ventricle.

The prone (angulated Concorde) position (Fig. 2) isthe position used most commonly to remove cerebellarastrocytomas. The risk of air embolism is low and there isless risk of tearing bridging veins between the cerebel-lum and tentorium than in the sitting position. The pa-tient orientation is direct and the surgeon can stand com-fortably. The sitting position (Fig. 3) is the second mostfrequently used position. For tumors in the lateral cer-ebellar hemisphere, operations are often done with thepatient in the lateral or “park bench” position (Fig. 4),with a roll under the axilla and the head fixed in theMayfield head holder.

For the prone position, patients are positioned at thesurgeon’s side of the operating table and are supported onchest rolls that extend from the shoulders to the pelviccrest, with no compression of the abdominal viscera. Headsupport varies with the age of the patient: less than 10-12months, the skull is usually too soft to permit reliable pinfixation and the head is therefore supported in the foam-padded pediatric horseshoe head holder; pediatric pinsare used in 1-2 year olds, and adult pins are used thereaf-ter. The head is fixed in the three-pin (Mayfield) headholder and angled about 30° away from the midline, turn-ing the posterior fossa toward the surgeon. The neck isflexed approximately 30°; a distance of one to two fingerbreadths must be left between the chin and the chest.

The sitting position is used by some neurosurgeonsfor removal of midline and paramidline posterior fossaastrocytomas. Its advantages include direct orientationof the patient/tumor position and the removal of bloodand CSF by gravity, but its disadvantages include therisk of air embolism and the fatigue of the surgeon’selevated arms. For the sitting position, patients are po-sitioned on the operating table with their shoulders atthe end of the main portion of the table. The head isfixed in the three-pin head holder and the crossbar isattached to side rails near the foot of the

435

Figure 2. Prone (angulated Concorde) position. The patient is at thesurgeon’s edge of the operating table. The head is turned 30° away fromthe midline and the neck flexed 30°.

Figure 3. Sitting position. Shoulders are at the end of the table and theneck is flexed but the chin is 1-2 finger breadths from the chest.

Figure 4. Lateral position. The vertical retromastoid incision is cen-tered over the maximal tumor volume.

ALBRIGHT : CEREBELLAR ASTROCYTOMAS




436


table. The table is maximally flexed; folded blankets maybe placed under the buttocks to keep the shoulders at thedesired height. The table back is flexed upward, the headholder is connected to the crossbar, and the table’s head-rest is folded downward to a position comfortable for thesurgeon’s elbows to rest on. The head is kept in the mid-line for midline and paramedian tumors.

Skin IncisionTo approach midline and paramidline tumors, a midlineincision is made from the inion to the C1-2 interspace.For lateral hemispheric tumors, a vertical incision is madebetween the midline and mastoid, centered over the maxi-mal volume of the tumor.

Operative ProcedureThe epidermis can be opened with a No. 15 blade and thesubcutaneous tissues opened in the midline with the needletip of the coagulating monopolar cautery. If the cautery isnot used, scalp clips (Raney, Leroy, or Children’s Hospi-tal clips) must be applied to the scalp margins. Tissuesare opened in the midline with the cautery down to theoccipital bone. A Weitlaner or posted Jannetta retractor isinserted and periosteum is dissected off the occipital bonewith the cautery or with a broad periosteal elevator later-ally for 3-4 cm from the midline; the bony and dural ex-posure must be sufficient to permit the hemispheres to beretracted away from the tumor. Venous emissary chan-nels must be immediately and meticulously waxed to mini-mize air embolism. Tissues need to be dissected away fromthe C1-foramen magnum space and off the C1 spinousprocess and lamina. An Adson periosteal elevator is in-serted to dissect dura away from the the posterior rim ofthe foramen magnum. The completed exposure is shownin Figure 5A.

The traditional method of occipital bone removal isby a craniectomy, removing bone with Kerrison and Leksellrongeurs from the foramen magnum upward until the infe-rior margin of the transverse sinus is seen. Equivalent ex-posure can be achieved with a craniotomy, which avoidsthe postoperative concavity so common after a craniectomy.Two burr holes are made in the occipital bone just belowthe expected level of the transverse sinus. Bone is oftenmuch thinner over the tumor than elsewhere. Dura is dis-sected away from the bon( in three directions: from theburr holes downward to the foramen on either side andacross the midline between the holes. Craniotomy cuts aremade first from the burr holes down to the foramen mag-num and then across the midline; the likelihood of a duraltear is decreased if dura is depressed with a Penfield dis-sector inserted from the opposite side while the bone is

being cut across the midline. Bone margins are waxed. AC1 laminectomy does not have to be performed in the ma-jority of these operations, but may be needed if the cer-ebellar tonsils have herniated to the C1-2 interspace. If thedura is tense, the EVD should be opened and CSF draineduntil the dura is slack before it is opened.

To approach midline and paramidline tumors, the duraIs opened in a Y-shaped manner (Fig. 5B), beginning withthe lateral limbs of the Y and incising toward the midline.As the opening nears the midline, the occipital sinus mustbe occluded before the dura is divided. That occlusioncan be accomplished by: (a) temporary occlusion withhemostats while the dura is cut (Fig. 5C), then coagulat-ing/ligating the sinus, or (b) permanent occlusion withWeck clips, which may distort the postoperative MRI scan.The inferior limb of the Y is then opened down to Cl.Dural tackup sutures, usually 4-0 Nurolon, are insertedand the dural leaves are retracted backward over 0.5 × 2-inch cottonoid strips, kept moist during the procedure toprevent dural drying and shrinkage. The arachnoid overthe cisterna magna is opened. Cottonoid patties are in-serted at the foramen magnum to lessen caudal migrationof blood during the operation and, if appropriate, are in-serted between the tonsils and onto the floor of the fourthventricle to protect the brain stem. Blades are attached toa retractor system (Jannetta, Greenberg, Yasargil).

If the outer surface of the astrocytoma is not visible,folia overlying the tumor are usually thinned and widened.At this point in the operation, neurosurgeons in the pastwould insert a blunt needle into the tumor cyst to decom-press the posterior fossa; that is infrequently needed now ifthe associated hydrocephalus has been treated by an EVD.Pia is coagulated with bipolar forceps (Fig. 5D) and openedwith a No. 15 blade and fine-tipped scissors. Underlyingwhite matter along that line is suctioned away down to thetumor surface and retractor blades are inserted to hold theparenchymal surfaces apart. Once the tumor is visible, manyneurosurgeons elect to biopsy the tumor for a frozen sec-tion diagnosis, although the subsequent resection is infre-quently altered by that information. If a diagnosis of cer-ebellar (juvenile pilocytic) astrocytoma (JPA) is made,meticulous attempts to achieve a complete removal are in-dicated because the tumor is generally surgically curable.JPAs have a fleshy appearance and frequently have a dis-tinct plane demarcating them from surrounding cerebel-lum. Small tumors can be excised en toto, dissecting inthat plane with Penfield or microdissectors to extract themfrom surrounding tissue. Larger tumors are internallydebulked with the ultrasonic aspirator (Fig. 5E), then theirexterior is dissected out (Fig. 5F). For JPAs that have a“mural nodule” and an associated cyst, removal

437


Figure 5. A, completed exposure of the occipital bone and C1 lamina.The sites for burr holes and for the craniotomy are depicted. A postedJannetta retractor and blades are demonstrated. B, completed cran-iotomy, exposing dura from the inferior aspect of the transverse sinusdown to the C1 lamina. A left hemispheric tumor and the plannedclural incisions are shown. If the tumor has a major cystic componentand if the dura is tense, the cyst can be aspirated via a 14-16-gaugeblunt (Cone) needle, inserted before the dura is opened. C, lateral du-ral incisions have been made. The midline occipital sinus is occluded

temporarily by hemostats before the midline is divided. The sinus issubsequently coagulated and ligated and the inferior limb of the Y isthen opened. D, dural leaflets are sutured backward over the moistcottonoid strips. A cottonoid strip is inserted into the cisterna magnato prevent blood from entering the cisterna magna/foramen ofMagendie. Cerebellar folia overlying the tumor can be opened trans-versely or vertically, depending on the maximal diameter of the tumorand thinness of the folia. Pia is coagulated with a bipolar cautery andopened with microscissors.


438


Figure 5. (Continued) E, cerebellum Is retracted away from the pialopening and tumor with blades attached to a posted Jannetta retractor orto a Greenberg retractor. The ultrasonic aspirator is used to decompressthe interior of predominantly solid astrocytomas. F, after the majorityof the interior of the tumor has been removed, the periphery of the solidtumor is dissected away from adjacent cerebellum, retracting with thesuction tip and coagulating small bridging vessels with the bipolar cau-

tery. Inset, for cystic astrocytomas with a mural nodule and a membra-nous cyst wall, the nodule can be dissected out with Penfield dissectorsor the bipolar cautery. Membranous cyst walls need not be removed;walls 1-2 mm thick because of tumor infiltration must be dissected out.G, primary dural closure is usually possible if the dura is kept moist. Ifprimary closure is not possible, a small graft of fascia or lyophilizeddura is inserted, The bone flap is held in place with 2-0 Vicryl sutures.


439


of the tumor is usually curative (Fig. 5, inset). Cyst wallsthat are membranous and translucent are typical; the cystmargin infrequently contains tumor cells. If there is sus-picion about the presence of tumor in the wall, a frozensection biopsy should be obtained. For astrocytomas with-out a clear peripheral dissection plane, the ultrasonic as-pirator is used to remove tumor, beginning in its core andproceeding toward the periphery until normal tissue ap-pears.

After the resection is complete, the cerebellar tonsilsshould be inspected and, if herniated, lifted above the fo-ramen magnum. Meticulous hemostasts is obtained withbipolar cautery in the tumor bed and hemostasis is checkedby Valsalva’s maneuver, performed for approximately 10seconds. The walls of the cavity are often lined withSurgicel to aid hemostasis; Gelfoam is less appropriatebecause it swells postoperatively and may be confusedwith residual tumor on the immediate postoperative CTscan.

Closure TechniquesThe dura. is closed by an initial single central suture toapproximate the three corners of the dural flaps together,followed by either a running suture or multiple interruptedsutures to close the dura (Fig. 5G). If the dura cannot beclosed primarily or if the primary closure is tight becauseof edema of the underlying cerebellum, a dural patch graftis indicated. Small grafts can be obtained from adjacentfascia; larger grafts are of lyophilized dura or fascia lata.

The bone flap is sewn back in place with 2-0 Vicrylor silk sutures rather than wire, which distorts postopera-tive scans. If there is no bone flap, some neurosurgeonslay a piece of Gelfoam over the dura in the hope of reduc-ing the risk of a CSF leak and some insert calvarial frag-ments removed during the opening. The various layers(deep fascia, superficial fascia, subcutaneous tissues, andskin) are closed individually with either running or inter-rupted sutures. Subcutaneous drains are not needed.

MONITORINGIntraoperative monitoring is probably not indicated if thetumor is located in a cerebellar hemisphere, becauseevoked potentials cannot evaluate cerebellar pathways. Fortumors that may invade the brain stem, monitoring soma-tosensory or auditory evoked potentials and direct poten-tials from the sixth and seventh nerves may add a mea-sure of safety. Intraoperative ultrasonography isinfrequently useful; tumor margins are usually evident andmany ultrasound probes are inappropriately large.

SPECIALIZED INSTRUMENTATIONHigh-speed drills such as the Midas Rex drill have facili-

tated posterior fossa craniotomies; they are more maneu-verable than traditional drills and cut through thick oc-cipital bone more rapidly and with less effort. The ultra-sonic aspirator is a more efficient tool than the laser forremoving most astrocytomas. Posterior fossa astrocyto-mas were traditionally removed without magnification orperhaps with loupes, but the operating microscope pro-vides illumination and clarity of vision which may in-crease both the safety and completeness of tumor resec-tion.

COMPLICATIONSPostoperative pseudomeningoceles and CSF leaks areusually due to inadequate drainage of CSF and are betterprevented than treated: EVDs should be kept 510 cm aboveventricular level for the first 2-3 days postoperatively, thenincreased by 5 cm/day until the drip chamber is 25-30 cmabove ventricular level—a process that takes 5-6 days—and then clamped for 12-24 hours and removed if the pa-tient has no symptoms of increased pressure.Pseudomeningoceles have been treated by repeated per-cutaneous aspirations which are sometimes effective butrisk introducing bacteria.

The likelihood of a postoperative hematoma—in theresection cavity or the epidural space—large enough toproduce signs of increased pressure and to warrant re-moval, is small, approximately 1-2%, if the techniquesdescribed above are used. The risk of postoperative woundinfection is similar in spite of intraoperative antibiotics.The likelihood of staphylococcal infection of the EVD is5-10%. CSF samples should be cultured daily from theEVD; the infection rate may be decreased by the dailyinstillation of vancomycin, 10-20 mg, into the EVD cath-eter.

The risk of postoperative cerebellar deficits dependson the tumor location and on the trauma of its removal,and ranges from 10 to 25%. Corticosteroids are usuallygiven in high doses for 2-3 days postoperatively, and thentapered over 3-7 days. Postoperative edema is not a con-spicuous feature of most posterior fossa astrocytomas.

The presence of residual tumor is important sinceJPAs are usually surgically curable tumors. A CT scanshould be obtained if possible within the first 48 hourspostoperatively. If the scan demonstrates residual tumorthat appears to be safely excisable, it is appropriate toreturn to the operating room and remove it.

The posterior fossa pseudobulbar syndrome occursin 3-5% of patients. It is characterized by the delayed onset(12-48 hours after operation) of confusion, agitation, apha-sia, and at times, ataxia, or hemiparesis. Its cause andtreatment are unknown. CT, MRI, and cerebral blood flowscans have been normal.

440

EXTREME LATERAL LUMBARDISC HERNIATION

ROBERT S. HOOD, M.D.

INTRODUCTIONThe development of high-resolution computed tomogra-phy (CT) scans allowed reliable identif ication ofextraforaminal or extreme lateral lumbar disc herniationsfor the first time. It is now apparent that extreme lateraldisc herniation occurs more frequently than once appre-ciated, perhaps representing up to 10% of clinically sig-nificant disc herniations. Surgical techniques were poorlyadapted to manage these herniations and surgeons wereless familiar with the operative anatomy in this region.Extreme lateral disc herniations, when inadequately re-moved, represent one cause of failed back surgery. Op-erative techniques needed to be modified so the surgeoncould approach this area more easily without the neces-sity of exploring within the spinal canal. The techniquedescribed herein allows easy access to the extreme lateraland neural foraminal. region under direct visualization,with the option of spinal canal exploration, while avoid-ing instability due to facet destruction.

The clinical syndromes associated with extreme lat-eral disc herniation differ from typical intraspinal hernia-tion, in that extreme lateral herniations seem to occur withequal frequency at L3-4, L4-5, and L5-S1. Because thepedicle of the vertebra below the disc prevents caudalmigration of fragments, cephalad migration of fragmentsusually occurs, making midlumbar radicular syndromesmuch more likely. Herniations occurring within the spi-nal canal typically produce a radicular syndrome affect-ing the nerve root exiting below the disc space, in con-trast to the extreme lateral herniation which affects thenerve root above (e.g., the L4 nerve root with extremelateral herniation at L4-5). The clinician should thus bealert that pain or sensory, motor, or reflex deficit in theanterior thigh may well be caused by an extreme lateraldisc herniation.

High-resolution CT or magnetic resonance imaging(MRI) scans are the diagnostic procedures of choice for

extreme lateral lumbar disc herniation. Myelography withpostmyelogram. CT is occasionally helpful to exclude con-current intraspinal or axillary herniation or stenosis.Myelography alone and discography have generally notbeen helpful or necessary.

Unless major neurologic deficit has occurred, allpatients are treated conservatively with an initial periodof rest, limited activity, use of anti-inflammatory agents(nonsteroidal or steroids), and, occasionally, physicaltherapy. Conservative treatment is undertaken for a mini-mum of approximately three weeks. If the patient is notsufficiently improved, such that there remains significantlimitation from radicular pain or neurologic deficit, thenmore aggressive treatment is warranted.

In addition to long-term conservative treatment, bothchemonucleolysis and automated percutaneous lumbardiscectomy may be alternatives. However, since extrudedmigrated fragments occur in 95% of extreme lateral discherniations, and the nerve is significantly displaced nearthe site of usual needle or cannula placement for each ofthese procedures, they seem less likely to give satisfac-tory relief. These alternatives may also have a higher riskof nerve injury than when they are used for herniationwithin the spinal canal.

Several surgical alternatives exist. Traditional lami-notomy risks failure, because the annular defect and thefragments cannot be visualized directly or reached easilywith standard instrumentation. Extension of the laminec-tomy to include facetectomy allows adequate exposure,but entails more bone removal than is necessary and risksinstability. The approach to the extreme lateral space byundermining the lamina with medial facetectomy froman intraspinal direction allows exposure to the medial neu-ral foramen, but less easily to the extreme lateral portionof the vertebral body and disc. It also has the disadvan-tage that the nerve the surgeon is trying to protect anddecompress is not visualized until after the disc hernia-tion is removed, risking nerve injury. The far lateral re-gion can be reached satisfactorily using a paramedianapproach through the paraspinal muscles, which allows


441

HOOD : EXTREME LATERAL DISC HERNIATION

good lateral access. However, the target is deep and theapproach is more disorienting. The paramedian approachalso does not allow exploration within the spinal canalfor caudally migrated fragments, if present, because thepedicle below limits lateral access.

The following surgical approach is used for extremelateral or foraminal herniations, which may include someherniations in the axilla above. It is also useful where ex-treme lateral foraminal and intraspinal herniation coex-ists with foraminal stenosis.

OPERATIVE PROCEDUREThe patient is operated upon under general inhalationalanesthesia. An intraoperative antibiotic irrigation solutionis used, but neither antibiotics nor steroids are given sys-temically during the pre- or intraoperative period. Thepatient is placed prone on the operating table, on chestrolls or a lumbar frame, with the table flexed and thepatient’s legs flexed at the knees. Standard draping tech-niques using paper drapes with an adherent plastic drapeon the skin are used.

The first portion of the procedure is performed us-ing operating loupes, and a high-intensity headlight, withan operating microscope brought into use later. The skinincision is in the midline, about 4-5 cm in length, and iscentered slightly more cephalad than the usual incisionfor a typical intraspinal disc herniation, because the neu-ral foramen lies at and above the disc space and extrudedfragments tend to migrate cephalad (Fig. 1). A small sub-cutaneous fat graft is excised and preserved in saline forlater use. An ipsilateral fascial incision is made about 1cm lateral to the midline. The paraspinal muscles arecleared with a periosteal elevator and curettes over thesuperior articular process, centered over the neural fora-men, which is then easily visualized. At this point a por-table, lateral x-ray film is made with a marker placed atthe foramen to localize the dissection accurately. Muscleexposure should be carried far enough lateral to includethe articular processes both above and below the foramenand should extend caudal to but not beyond the laminaunless intraspinal exploration is also necessary. A self-retaining retractor, usually a single-bladed long Williamsretractor, is inserted, centered over the neural foramen (Fig.2). Exposure does not need to include the transverse pro-cesses.

The edge of the foramen is identified and the liga-mentum flavum, which extends even lateral to the fora-men, is separated from the bony edge using a small angledcurette (Fig. 3). A high-speed drill is then used to unroofthe foramen from lateral to medial, extending from thepedicle above to the pedicle below such that the full ex-tent of the foramen and disc are visualized adequately(Fig. 4). Adequate bone removal for exposure is essential

for both visualization and adequate decompression. Theextent of medial bone removal is dependent on the exactlocation of the herniation as shown on the preoperativestudies and at surgery. Bone removal is completed withrongeurs and can be carried medially to visualize the mainthecal sac, which is necessary when dealing with axillaryor medial foraminal herniation. Bone removal must alsoinclude the cephalad one-quarter to one-third of the facetbelow, which always overhangs the far lateral portion ofthe disc. Preoperative studies occasionally suggest thepresence of disc fragments both far lateral and within thespinal canal below the disc space. If exploration in bothplaces is warranted, then partial hemilaminectomy canalso be performed, leaving at least a strut of laminar boneto the facet, in order to avoid potential instability.

The operating microscope is then brought into placeand is used for the remainder of the procedure. Use of themicroscope at an earlier stage is acceptable but can slowthe procedure and may lead the surgeon to the pitfall ofinadequate bone decompression, with a greater possibil-ity of overlooking residual pathologic changes, due to thelimited focus as a result of magnification.

The first step in dissection is to find and thus protectthe nerve, which is always easily located at the

Figure 1. A midline incision. oriented adjacent to the neural foramen,slightly cephalad to the disc space.


442


Figure 2. A paramedian approach with a self-retaining retractor exposing the neural foramen and facet joint.

Figure 3. An extreme lateral disc herniation is shown displacing the nerve and ganglion before boneremoval (the ligamentum flavum is not shown).



443

HOOD : EXTREME LATERAL DISC HERNIATION

Figure 4. Exposure following bone and ligamentum flavum removal, with the underlying disc herniationshown distorting the nerve and ganglion.

cephalad margin of the foramen beneath the ligamentumflavum where it courses next to the pedicle above. Theligamentum flavum is easily separated from the pedicle.This is the location, therefore, where one should elevatethe yellow ligament with a nerve hook or angled curetteand visualize the nerve, then excise the ligament withrongeurs or by sharp dissection, while protecting the nerveand ganglion throughout. The ligamentum flavum is re-moved laterally to the intertransverse ligament, allowingfull exposure of the nerve, its ganglion, and its more pe-ripheral extent to the edge of the vertebral body and disc.There is usually an abundance of epidural fat and veins inthe region. The disc herniation usually lies inferior andmedial to the nerve, which is distorted posteriorly andlaterally. The nerve is protected with cottonoids, and bi-polar cautery is utilized if needed.

Although sequestrated fragments may be seen,more commonly the fragments of extruded disc lie be-neath the posterior longitudinal ligament over the bodyof the vertebra cephalad to the disc space. More later-ally placed herniations may indeed lie directly over orlateral to the disc space. Thus, the posterior longitudi-nal ligament is opened and a thorough search made forfragments, which are removed, using nerve hooks andmicro-pituitary rongeurs. One must be certain to ex-plore to the lateral edge of the vertebral body and discas well as medially to the dural margin, in order to avoidretained fragments. The annular defect is then identi-fied and entered with a variety of pituitary rongeursfor enucleation of the lateral portion of the disc. While

curettes are occasionally used to assist with enucle-ation, vigorous curettement is not done. The nerve isthen inspected throughout its course to ensure thoroughdecompression (Fig. 5).

Following final inspection, the subcutaneous fat graftthat was previously removed and preserved is placed overthe nerve and far lateral space. Small pledgets of Gelfoamsoaked in saline are occasionally used for hemostasis orin the absence of useful fat. The wound is then closedwith absorbable interrupted fascial, subcutaneous, andsubcuticular sutures. Steristrips are applied to the skinfollowed by a light, nonadherent, breathable bandage.Drains are never used.

POSTOPERATIVE MANAGEMENTPostoperatively the patient is ambulated briefly on the dateof surgery, more frequently on the following day, and isusually discharged by the second postoperative day. In-jectable narcotic analgesics are used, if necessary, on theday of surgery and oral narcotics thereafter, briefly fol-lowing discharge. A mild muscle relaxant is used forincisional stiffness for about one week and a rapidly ta-pering prednisone regimen is used postoperatively for 5-6 days. The patient is activated progressively, includingthe initiation of an exercise program with aerobic exer-cise within three weeks following surgery.

COMPLICATIONSThe complications attendant to this procedure are com-mon to all procedures for disc surgery. Significant


444


Figure 5. The ganglion and nerve are well decompressed following excision of the disc herniation.

bleeding or infection has not occurred and should be rareif attention is paid to technique. Major vessel injury hasnot been encountered. Intraoperative nerve injury can beavoided with careful microdissection and with particularattention to avoid excessive manipulation of the highlysensitive ganglion. Facet pain and instability have notoccurred with this procedure. Recurrent disc herniationseems to occur at a rate similar to or less than that follow-ing intraspinal disc surgery.

RESULTS AND CONCLUSIONSForaminal and extreme lateral disc herniations are rela-tively common. This technique has been used for morethan six years by the author. More than 90% of patientshave experienced satisfactory relief of pain, with returnto usual activity. The surgeon should be prepared to dealproperly with herniations in this region, adapting the tech-nique to the patient’s needs. This technique provides suchan adaptation.


445

TENTORIAL MENINGIOMASLALIGAM N. SEKHAR, M.D.

ATUL GOEL, M. CH.

INTRODUCTIONThe tentorium is a tough double-layered dural wall sepa-rating the cerebral hemispheres from the cerebellum. Thetentorial notch or incisura is of the shape of a necklacearound the midbrain and is the only communication be-tween the supra- and infratentorial spaces. Its dimensionsare variable. The tentorium is shaped like a “tent,” slop-ing downward from its apex at the posterior edge of thetentorial notch to its attachments anteriorly to the petrousridge and posterolaterally to the inner surface of the oc-cipital and temporal bones. The anatomy of the tentoriumand its anterior extensions into the anterior and posteriorclinoid processes has been described in the various re-views on this subject.

The problems of surgically excising tentorial menin-giomas are related to the difficult access, especially withmedial tentorial lesions, and the relationships to the tem-poral lobe, brain stem, cranial nerves, blood vessels, andvenous sinuses. The relationships of the posterior cere-bral and superior cerebellar arteries, the perforating ar-teries, the vein of Labbé, and the veins draining the brainstem are particularly important. Thus the classificationof these meningiomas according to their location intomedial, lateral, posterior, and falcotentorial is useful inplanning surgery and comparing the results (Fig. 1). Thetumors can arise from either the inferior or the superiorsurface of the tentorium and can extend correspondinglyinto the posterior fossa or supratentorially. Tentorial tu-mors usually have extensions in both superior and infe-rior compartments. Similarly, these tumors may extendinto the petroclival or falcine areas.

Despite the advancements in microsurgical operativetechniques and instrumentation, neuroanesthesia, radiol-ogy, and the medical management of elevated intracra-nial pressure, meningiomas arising from the tentorium

pose a challenge for the surgeon due to the following char-acteristics:

1. The tentorial meningiomas are usually diagnosed when they arelarge in size.

2. They have a close relationship to the brain stem, the cranial nerves,important blood vessels, and the cavernous sinus.

3. The neurologic deficits are often minimal early in their presenta-tion in spite of the location and the large size of the tumor.

4. The tumors are located deep and require the retraction of vitalparts of the brain to approach them.

5. The blood supply of the tumor is derived from arteries which usu-ally arise deep to the large tumors and are relatively difficult toobliterate early in the operation.

6. Factors relating to the tumor, such as arterial encasement, com-promise of important venous structures, degree of brain stem com-pression, tumor vascularity, tumor consistency (soft as opposed tofirm), and the presence (or absence) of a subarachnoid plane be-tween the tumor and the brain stem are important variables whichultimately decide the course of the surgical resection and the re-sults. Presence of scar tissue from previous surgery and prior ra-

diation therapy are also important problems for the surgeon.

PRESENTATION AND INVESTIGATIONSSigns of elevated intracranial pressure or dementia (dueto the large size of the tumor or to hydrocephalus) are themost common forms of presentation, followed by cerebel-lar symptoms and pyramidal and cranial nerve involve-ment. Psychomotor seizures may occur if the tumor com-presses the temporal lobe.

Computed tomography (CT) with soft-tissue and bonealgorithms continues to remain the primary investigationfor tentorial meningiomas. Magnetic resonance imaging(MRI) before and after contrast administration demonstratesthe relationship of the tumor with important surroundingblood vessels better than any other investigation presentlyavailable. In addition, the presence or absence of an arach-noid plane can also be discerned from MRI. Angiographyis important to delineate the extent of the vascularity of thetumor, its source of blood supply, and the degree of in-volvement and shifts of the blood vessels. Therapeuticembolization of the feeding tentorial arteries can be per-formed whenever the feeders are large and can be© 1991 The American Association of Neurological Surgeons

446


Figure 1. The common sites of tentorial meningiomas are shown.

cannulated via the internal carotid artery (ICA). Whenthey cannot be cannulated, sometimes the ICA may betemporarily occluded with a balloon, and absolute alco-hol may be infused into the feeding meningohypophysealartery or other branches. The vertebral angiogram andMRI must be examined carefully to note the location ofthe basilar artery and its terminal branches in relation tothe tumor.

GENERAL PRINCIPLES

Electrophysiological MonitoringMonitoring of somatosensory evoked potentials (SSEP),brain stem evoked responses (BSER), and electroencepha-logram (EEG) keeps the surgeon aware of the changes invital functions. Cranial nerve (CN) monitoring may in-clude CNs VI to XII.

InstrumentationAs the operation may be prolonged, the surgeon should bein a comfortable operating position during the surgery. Acomfortable chair with armrests and a Zeiss Contraves mi-croscope are usually used by the senior author. A Malisirrigation bipolar forceps is useful for tumor dissection andfine work, whereas Aesculap bipolar equipment is usefulfor rapid tumor removal. CO

2 and neodymium-yttrium-alu-

minum-garnet (Nd-YAG) lasers and a Cavitron ultrasonic

aspirator are useful for debulking the tumor in selected situ-ations. In addition to standard microsurgical instrumenta-tion, the surgeon should also have available instrumentsfor the repair of small and large vessels.

Brain Retraction and Tumor ExposureBrain retraction should be minimized as much as pos-sible. Adequate basal bone removal and wide exposure,the opening of appropriate cisterns and fissures, the drain-age of cerebrospinal fluid (CSF), and sometimes the ju-dicious resection of noneloquent areas of brain are usefulin avoiding excessive retraction. The use of osmotic di-uretics, moderate hyperventilation, low-dose barbiturateinfusion, appropriate anesthetic agents, steroids, and thereplacement of volume loss with colloids rather than crys-talloids helps in maintaining optimum brain relaxation.

Tumor RemovalAll dissections must be carried out under direct vision.Use of a monopolar nerve stimulator can help in the iden-tification of cranial nerves, some of which may be verythin and splayed over the tumor. With lateral tentorialmeningiomas, the major technical problem is the involve-ment of the venous sinuses. Medial tentorial lesions aretechnically more difficult considering their relationships.

The general principle of tumor resection includesearly devascularization of the tumor, if possible, and re-duction of tumor bulk by coring out the center of the tu-mor before attempting dissection from the brain. Thearachnoidal planes are usually preserved around the tu-mor and dissection must be carried out in that plane. Incases of large tumors, however, the arachnoidal plane maybe missing and in such cases efforts are made to preservethe pial plane. When critical vessels and nerves are en-countered, dissection is done from the normal to the ab-normal area parallel to the direction of the encased struc-ture. Sharp dissection is preferable in the proximity ofblood vessels and nerves. Dissection of the small perfo-rating vessels, of encased and narrowed arteries, of com-pressed brain stem, and through scar tissue are most dif-ficult tasks.

Total tumor removal with excision of the involvedtentorium is the goal of the surgery. However, this maynot be feasible with media] tentorial lesions because oftumor extension onto the petrous ridge or the cavernoussinus or because of occasional dense adherence of thetumor to important blood vessels.

OPERATIVE APPROACHThe approach to a tentorial meningioma depends uponits location on the tentorium and its size. Generally,when the tumor is small in size or extends only intothe supra- or infratentorial spaces, the tumor


447

SEKHAR AND GOEL : TENTORIAL MENINGIOMAS

may be removed through an approach only on one side.Larger tumors, however, will require exposure on bothsides of the tentorium.

Lateral Tentorial MeningiomasThe craniotomy is performed in the occipital or temporalarea and in the posterior fossa, encircling the tumor. Thetumor is removed as described. Where there is adequatecollateral circulation, and when the sinus is nondominant,when the sinus is already occluded, and in younger pa-tients, the sinus may be resected to achieve complete tu-mor resection. When the sinus is dominant, a trial occlu-sion with temporary clips is performed prior to permanentocclusion even if adequate cross circulation is present. Inall patients, a reconstruction of the venous sinus is at-tempted, for potential long-term benefits.

Medial Tentorial MeningiomasSmall medial tentorial meningiomas may be removed bya subtemporal, transzygomatic, and transpetrous apexapproach. Larger medial tentorial meningiomas require asubtemporal, presigmoid, retrolabyrinthine, ortranslabyrinthine approach.

Falcotentorial MeningiomasLarge falcotentorial meningiomas are removed by a com-bined occipital-transtentorial and suboccipital approach,with the temporary division of the transverse sinus.

Posterior Tentorial MeningiomasLarger tumors in this area need a bilateral occipital andsuboccipital craniotomy with the exposure of both trans-verse sinuses and the superior sagittal Sinus. Unless thetorcular herophili is already occluded by the tumor, a to-tal resection of these meningiomas cannot be performed.Fortunately, most of the tumors in this area involve only acomer of a venous sinus such that the tumor can be re-moved and the sinus repaired.

SPECIFIC OPERATIVE APPROACHES:EXAMPLESPosterior Subtemporalwith Presigmoid ApproachA posterior subtemporal and presigmoid approach is usedto resect large or giant-sized medial tentorial meningio-mas. While in some patients one can proceed by aretrolabyrinthine approach, in others a translabyrinthineapproach with sacrifice of ipsilateral hearing is needed.This can be done more readily if hearing is very poor inthat ear. The removal of the zygomatic arch with the condy-lar fossa permits a further inferior mobilization of thetemporalis muscle and improves the exposure of the in-terpeduncular cistern area.

The advantages of this approach are:

1. There is wide and shallow exposure to difficultanteromedially placed tentorial meningiomas. Theoperating distance to the tumor is shortened by about3-4 cm.

2. The retraction of the cerebellum and the temporallobe is minimized.

3. The surgeon has reasonable access to the anterior andlateral aspects of the brain stem.

4. The continuity of the transverse venous sinus is notaffected.

5. Tumor extensions into the cavernous sinus and cli-vus can be dealt with during the same sitting.

Operative StepsA curvilinear scalp incision is made behind the ear ex-tending from the retromastoid to the temporal region (Fig.2). The ear canal is transected and reflected forward withthe flap. The incision in the ear canal is made through thepinna and the skin is resutured at the end of the operationif the exposure is retrolabyrinthine. If a translabyrinthineexposure is performed, the external ear canal is dividedand closed as a blind sac and covered additionally with afascial flap. A temporal craniotomy is first performed fromthe pterion anteriorly to about 2 cm behind the mastoidprocess posteriorly. The dimensions of this flap are var-ied to suit the tumor. The transverse sinus is separatedunder direct vision and then a retromastoid craniectomyis performed, usually 2-3 cm in size. Middle fossa dura isseparated from the floor.

The temporomandibular joint capsule is then openedand the meniscus is depressed from the glenoid fossa.Using a reciprocating saw, a zygomatic osteotomy is per-formed with its anterior limit being just behind the lateralorbit. Posteriorly, the cuts include the glenoid fossa. Oneshould be careful while making the posterior cut andshould not go beyond the confines of the glenoid fossabecause of the proximity to the petrous ICA. Posteriorlylie the middle ear and the external ear canal. When thezygomatic osteotomy is carefully reapproximated at theend of the operation, no dysfunction of the temporoman-dibular joint is observed.

A mastoidectomy is performed next, with completeskeletonization of the sigmoid sinus. In older people, avery thin shell of bone should be left over the sigmoidsinus because it may be easily torn as a result of adhe-sions during attempted separation. Next, a labyrinth-ectomy is performed, with removal of the posterior, su-perior, and lateral semicircular canals. The tympanicand mastoidal segments of the facial nerve are skel-etonized but not displaced. By an extradural middlefossa approach, the petrous ICA is exposed in

448


Figure 2. Drawing showing the exposure for a posterior subtemporalwith presigmoid approach. The ear canal has been transected and thescalp flap reflected anteriorly. Labyrinthectomy has been carried out

and the facial nerve skeletonized. The dashed line shows the site ofdural incision.

Figure 3. Drawing showing the wide exposure of the tumor obtained with a posterior subtemporal with presigmoid approach.



449


its horizontal segment, and a petrous apicectomy is per-formed if needed.

The dura is opened in the presigmoid area andalso in the temporal region (Figs. 2 and 3). After su-ture ligation of the superior petrosal sinus, the tento-rium is divided from a lateral to medial direction,posterior to the tumor. The temporal lobe and cerebel-lum are gently retracted, taking care to avoid Injuryto the vein of Labbé. Any middle fossa tumor is re-moved f irst. The lateral wall of the cavernous sinus ispeeled posteriorly, and Meckel’s cave is opened. Thesetumors often extend into Meckel’s cave and the pos-terior cavernous sinus. Removal of the tumor fromthe posterior cavernous sinus may allow the tumor tobe devascularized further by the division of themeningohypophyseal artery.

The tumor is then debulked centrally. Dissection fromcritical structures follows. These include the brain stem,basilar artery, posterior cerebral and superior cerebellararteries, and CNs III, IV, V, and VI. With giant-sized tu-mors, it is usually not possible to preserve CN IV. It maybe reconstructed with a graft if the proximal stump ispresent and by finding the distal end in the lateral wall of

the cavernous sinus. As much as possible, dissection ofthe tumor from the brain stem must be made in thesubarachnoidal plane.

Illustrative CasesJP: This 26-year-old man experienced a single grand malseizure. There was no neurologic deficit. Investigations(MRI and CT scan) showed a large tumor extending onboth sides of the tentorium (Fig. 4, A-C). An attempt toremove the tumor at another institution was unsuccessfulbecause the tumor was very vascular, and the subtemporaland retrosigmoid exposure was restricted by a dominantsigmoid sinus and a large vein of Labbé on the ipsilateralside. At our institution, the patient underwent a cerebralangiogram (Fig. 4D) and a balloon occlusion test of theleft internal carotid artery. Absolute alcohol emboliza-tion of the tentorial branch of the internal carotid arterywas then carried out. Following this procedure the pa-tient developed a partial third nerve and a complete sixthnerve paralysis on the left side, presumably due to devas-cularization of these cranial nerves. The operation wasperformed by using the approach described above,with labyrinthectomy. A total tumor resection

Figure 4. A-C, MRI scans of patient JP showing the large tentorial meningioma compressing the brain stem. D, a vertebral artery angiogramshowing the marked vascular tumor blush.


450


was achieved. The patient’s facial paresis and ocular nervepalsies resolved. The final pathological diagnosis washemangiopericytoma. Adjuvant radiotherapy and chemo-therapy were administered.

CG: This 44-year-old male patient presented withhistory of a single grand mal seizure. He was also notedto have progressive imbalance, staggering of gait, anddecreased hearing on the right side. Diplopia had beenpresent for two years and became worse in the last sixmonths before presentation. Cranial nerve examinationrevealed a mild right third nerve paresis with slight pto-sis, enlargement of the pupil, and restriction of upwardand medial gaze. Radiologic investigations showed a largetentorial tumor also involving the cavernous sinus (Fig.5). The tumor was excised, using a posterior subtemporalpresigmoid and retrolabyrinthine approach. Postoperativecerebrospinal fluid leakage and meningitis occurred be-cause of transient hydrocephalus. These were successfullytreated. The patient suffered an abducens palsy but other-wise recovered uneventfully.

Combined Supra- andInfratentorial ApproachA combined supra- and infratentorial approach can be usedfor a giant falcotentorial (pineal region) meningioma. Themajority of surgeons currently use either a supracerebellarinfratentorial approach or an occipital transtentorial ap-proach to the lesions situated at this site, depending on

the extensions of the tumor. However, for some giant sizedand extensive lesions, the two approaches can be com-bined by division of the tentorium and the nondominant(communicating) transverse sinus. This combined ap-proach makes use of the advantages of both approaches,while avoiding the disadvantages. The transverse sinus isresutured at the end of the procedure to re-establish theblood circulation.


1. The procedure is indicated for giant and extensivemeningiomas. It provides a direct and wide exposureof the lesion. The brain stem and the vital veins ofthis region are widely exposed so that they can bedissected free under vision early in the operation. Alarge additional exposure is obtained at the depth bythe section of the transverse sinus, because it is closeto the surgeon’s line of vision.

2. The retraction of both the cerebellum and the occipi-tal lobe is minimized.

3. The transverse sinus is resutured at the end of theoperation, especially in younger patients, to restoreits patency. This maintains the natural venous com-munication, in case the patient develops occlusionof the contralateral venous sinus later in life.

4. The surgeon operates in a comfortable working po-sition, without the disadvantages attendant to the pa-tient being in a sitting position.

Figure 5. A, B, MRI scans showing the large tumor of patient CG.


451


Operative StepsThe patient is placed in right lateral decubitus (Sugita) po-sition. A horseshoe-shaped incision is made starting in thesuboccipital region, curving well above the lambda, andthen curving back on the other side (Fig. 6). A large freebone flap is first cut in one occipital area. After separationof the sagittal sinus, a second occipital flap is elevated.After separating the transverse sinus under direct vision,the suboccipital flap is elevated. The dura covering the pos-terior aspects of both occipital lobes and cerebellar hemi-spheres and the three venous sinuses is thus exposed.

The occipital dura is opened on the side of thenondominant transverse sinus, bordering the superior sag-ittal sinus and the transverse sinus (Fig. 6). The dura isalso opened in the suboccipital area in a linear fashionjust below the transverse sinuses, and the occipital sinus

is ligated. The occipital lobe is gently retracted away fromthe tentorium up to the tentorial notch.

Two temporary clips are placed across the nondominant(usually the left) transverse sinus approximately 1 cm lateralto the torcular herophili. After an observation period to checkfor possible brain swelling, the transverse sinus is divided be-tween the two clips. The tentorium is then incised parallel toand about 1 cm from the straight sinus up to the tentorial notch(Fig. 7). There is usually significant bleeding from the venoussinuses in the tentorium, but it can be controlled with gentlepacking with oxidized cellulose and with bipolar coagulation.The left half of the left transverse sinus and the tentorium areretracted away to provide a large space between the cerebel-lum and the occipital lobe. Very often, the occipital dura onthe other side will also have to be opened to divide the tento-rium on the other side, lateral to the tumor. Be-

Figure 6. Drawing showing the exposure for a combined supra- and infratentorial approach. The dashed line shows the site of dural incision.


452


Figure 7. Drawing showing the large exposure of the tumor obtainedafter the combined approach. The transverse sinus has been sectioned

and the two edges are held in clips. The inset shows the resuturing ofthe transverse sinus.


453


cause the blood supply to these tumors is derived fromthe tentorial arteries, the tumor is completelydevascularized thus.

The tumor is then debulked with bipolar cautery andscissors or a Cavitron ultrasonic aspirator. Dissection ofthe tumor is performed from the cerebellum with furtherpiecemeal excision. The upper pole of the tumor is thendelivered inferiorly, and careful dissection is performed,separating the tumor from the vein of Galen and its tribu-taries. The tumor is then dissected away from the com-pressed brain stem. Finally, the dural attachment of the tu-mor is excised. However, a small area of the tumorattachment around the straight sinus and the entrance ofthe vein of Galen can only be cauterized. Reconstructionof the transverse sinus can be performed if preferred. Thisis done with 5-0 or 6-0 Prolene interrupted sutures (Fig. 7,inset).

Illustrative CasePG: A 45-year-old man presented with a one-year historyof progressive loss of balance while walking and bilateralhearing loss. Except for mild abnormality of cortical func-tion, and difficulty in tandem walking, he had no neuro-logic deficit. Magnetic resonance imaging showed a largemass in the pineal region, straddling the tentorium, sug-gestive of a meningioma (Fig. 8). There was associatedhydrocephalus. Angiography showed that the tumor wasfed mainly by a large tentorial branch of the rightmeningohypophyseal trunk. The right transverse sinus wasdominant, but there was good communication across thetorcular herophili.

A combined supra- and infratentorial approachwas used to resect this tumor completely. The patientmade a complete recovery after a short period of reha-

Figure 8. A-D, MRI scans of patient PG showing the large pineal region (falcotentorial) meningioma.


454


bilitation. Postoperative scans showed the absence of thetumor and the patency of the transverse sinus.

Subtemporal, Transcavernous, andTranspetrous Apex ApproachA subtemporal, transcavernous, and transpetrous apexapproach (Fig. 9) is ideal in situations where there is alarge anteromedial tentorial meningioma which involvesthe cavernous sinus. This approach is suitable for lesionsin the region of the upper clivus and petrous apex area.The approach is the modification of the transpetrous apexapproach described by Kawase.


1. This approach is useful when there is involvement ofthe cavernous sinus by a small or medium sized tento-rial meningioma. It may be emphasized that the duraof the superior and lateral walls of the cavernous sinusare extensions of the tentorium and hence it is involvedfrequently by medial tentorial meningiomas.

2. The removal of the glenoid fossa leaving the carti-lage of the joint intact does not affect the function ofthe joint significantly.

3. Because the approach is very basal, the retraction ofthe brain is greatly minimized and the surgical depthof the lesion is reduced.

4. A wide working space is available for the surgeon tooperate in the critical areas anterior to the brain stem.

Operative StepsA curvilinear scalp incision is made starting in the

temporal region, curving inferiorly above the pinna of theear and extending to the preauricular area. The skin flapis reflected deep to the superficial layer of the temporalfascia to preserve the frontal branches of the facial nerve.Below the level of the zygomatic arch the scalp is reflectedjust over the masseteric fascia to avoid injury to the pa-rotid gland. A temporal craniotomy is performed, with itsposterior extent just above the mastoid process. This isfollowed by zygomatic osteotomy. The glenoid fossa ofthe temporomandibular joint may be included in the os-teotomy if a more extensive exposure of the region is nec-essary.

Extradural middle fossa dissection is then performedfrom a lateral to medial and from a posterior to anteriordirection to identify the following landmarks: tegmentympani, arcuate eminence, the lesser and greater super-ficial petrosal nerves (LSPN and GSPN), the middlemeningeal artery, the mandibular nerve (V3), and thehorizontal segment of the petrous internal carotid artery(if it is uncovered by bone). The GSPN may be distin-guished from the LSPN by the fact that the LSPN joinsthe middle meningeal artery at the foramen spinosum,and by electrical stimulation of the GSPN, with observa-tion of resulting contraction of the facial musculature onelectromyography.

The dura is opened and the temporal lobe is gently

Figure 9. Drawing showing the subtemporal and transpetrous apex approach. The arrows indicate the directions in which the surgeon canapproach the tumor after the initial exposure.


455


retracted after splitting the sylvian fissure widely, work-ing with the aid of the surgical microscope. The arach-noid membrane of the perimesencephalic cistern isopened, taking care not to injure the posterior cerebraland the superior cerebellar arteries which lie immediatelyunder the arachnoid in this area. The fourth cranial nerveis identified and the tentorium is divided well posterior tothe tumor and posterior to the entrance point of the tro-chlear nerve. The division of the tentorium is extendedposterolaterally, just posterior to the superior petrosal si-nus. The tentorial division is then extended anterolaterallyinto the middle fossa, across the superior petrosal sinus,and opening Meckel’s cave. The superior petrosal sinus ispacked with oxidized cellulose. The trigeminal root isidentified.

The dissection in the cavernous sinus will depend onthe extent of the tumor involvement and the preoperativestate of the extraocular muscle function. The lateral wallof the cavernous sinus is peeled away from a posterior toan anterior direction, exposing the trigeminal ganglion,and cranial nerves IV and V1. If patent, some of the cav-ernous sinus is packed with Surgicel between cranialnerves IV and V1. The dura is also opened all the wayforward along the lower border of the trigeminal root,trigeminal ganglion, and V3, and the lateral wall ofMeckel’s cave is opened completely. It may be essentialto dissect the fourth cranial nerve in the lateral wall of thecavernous sinus and move it superiorly to allow the sur-geon to work between cranial nerves IV and V inParkinson’s triangle. The surgeon may also work in thecavernous sinus between cranial nerves III and IV andbetween rootlets of cranial nerve V in Meckel’s cave.

The tumor in the prepeduncular area is now exposedwidely and can be dissected. For additional exposure thepetrous apex bone is removed with a highspeed drill, lat-eral to the trigeminal root and ganglion, workingintradurally. The lateral border of the removal of the pe-trous apex is the horizontal segment of the petrous inter-nal carotid artery. It is often easier to do this intradurallyrather than extradurally since the dura restricts the expo-sure of the petrous apex during extradural bone removal.The sixth cranial nerve will be identifted just medial tothe trigeminal root and can be followed into the cavern-ous sinus at this time; care has to be taken not to damageit. The surgeon may have to remove all the bone of thedorsum sellae, both posterior clinoids, the floor of thesella turcica, and any of the involved sphenoid and pe-trous apex bone.

Illustrative CasePS: This woman had a five-year history of headache andleft retro-orbital pain which recently increased in sever-ity and a one-year history of progressive facial numb-ness. There was no other neurologic deficit. The investi-gations (CT scan and MRI, Fig. 10) revealed a petroclivalmeningioma extending into the ipsilateral cavernous si-nus. She underwent a subtemporal, transcavernous, andtranspetrous apex approach for this lesion which was to-tally excised. The patient had postoperative third and sixthnerve palsies which resolved completely by eight monthsafter the operation.

COMPLICATIONSComplications following surgery for tentorial meningio-mas maybe related to brain retraction, injury to the brainstem, injury to the basilar artery branches, injury to ma-jor veins or venous sinuses, CN injuries, CSF leakage,and infection. Fortunately, with adequate exposure andwith the use of microsurgical technique such complica-tions can be greatly minimized.

Figure 10. An MR1 scan of patient PS shows the medial tentorial tu-mor extending into the petroclival region and the cavernous sinus.


neurosurgical operative atlas volume 1

Documents