Available online at www.sciencedirect.com

Surgical Neurology 69 (2008) 5–15www.surgicalneurology-online.com

Endoscopy

Success and complication rates of endoscopic third ventriculostomy foradult hydrocephalus: a series of 108 patients

Joshua R. Dusick, MDa, David L. McArthur, PhD, MPHa, Marvin Bergsneider, MDa,b,⁎aDivision of Neurosurgery, University of California at Los Angeles David Geffen School of Medicine, UCLA Medical Center,

Los Angeles, CA 90095-6901, USAbDivision of Neurosurgery, Harbor-UCLA Medical Center, Torrance, CA, USA

Received 4 July 2007; accepted 16 August 2007

Abstract Background: The clinical response to ETVof adult patients with noncommunicating hydrocephalus

Abbreviations: Cinterference in steadythird ventriculostomyacquisition; ICP, intraNPH, normal pressureLos Angeles; VP, ven

⁎ CorrespondingMedical Center, CHSUSA. Tel.: +1 310 20

E-mail address: m

0090-3019/$ – see frodoi:10.1016/j.surneu.2

may differ from that of children because of such factors as chronicity of hydrocephalus, physiologicdifferences in CSF dynamics, and changes in brain viscoelastic properties. We sought to determinewhich factors might predict clinical success and failure.Methods: A retrospective single-surgeon case series analysis was performed. This was a consecutivecase series for which the goal of the ETV procedures was shunt independence. One hundred ten ETVprocedures were performed in 108 adult patients (mean, 48 years; range, 17-88 years). There were52 cases of idiopathic aqueductal stenosis, 47 cases of mass lesions causing noncommunicatinghydrocephalus, plus 9 other miscellaneous obstructive etiologies.Results: Long-term shunt independence was achieved in 77% of patients. Two additional patients,who initially failed, later achieved success after reoperation and remained shunt free for the durationof their follow-up. Therefore, after reoperation, shunt independence was achieved in 79% of patients.Of the patients who ultimately failed, 11 failed within 1 month. Therefore, 52% who ultimately failedhad more than 1 month of shunt-free existence (mean, 10 months). There were 6 surgicalcomplications, including 2 deaths related to intracranial hemorrhage from brain tumors (not directlyrelated to ETV per se), and 10 medical complications. The median hospital length-of-stay was3 days. The median follow-up was 8 months (range, 0-95 months).Conclusions: Endoscopic third ventriculostomy is an effective treatment option for adult patients withnoncommunicating hydrocephalus.Althoughmost procedures resulted in long-term shunt independence,more than half of the eventual failures were delayed, and therefore, appropriate follow-up is required.© 2008 Elsevier Inc. All rights reserved.

Keywords: Adult hydrocephalus; Endoscopic third ventriculostomy; Neuroendoscopy; Intraventricular; Endoscope

1. Introduction

The procedure of ETV has gained popularity over the last2 decades in large part because of improved endoscope

CD, charged-coupled device; CISS, constructivestate; CSF, cerebrospinal fluid; ETV, endoscopic

; FIESTA, fast imaging employing steady statecranial pressure; MRI, magnetic resonance imaging;hydrocephalus; UCLA, University of California at

triculoperitoneal.author. UCLA Division of Neurosurgery, UCLA74-134, Box 956901, Los Angeles, CA 90095-6901,6 4100; fax: +1 310 825 7245.bergsneider@mednet.ucla.edu (M. Bergsneider).

nt matter © 2008 Elsevier Inc. All rights reserved.007.08.024

optics, high intensity lighting systems, and miniaturizedCCD cameras. In addition, more cases of noncommunicatinghydrocephalus are being diagnosed because of the improvedability to visualize ventricular anatomy with high-resolutionMRI. The increased clinical interest in ETV procedures hasbeen driven mainly by pediatric neurosurgeons, andconsequently, most of the larger published studies havefocused primarily on pediatric populations [7,8,11,15,32,51].In some reports, the authors did not distinguish adult vspediatric cases in their analysis [14,21,23,47,53,60].

The management of hydrocephalus in adult patients,however, differs from that of children. For example, CSFproduction and absorption changes over time. At one end of

Table 1Patient characteristics and pathologies

Pathology No. of patients(success rate)

Mass lesions 47 (76.6%)Pineal region 14 (71.4%)Pineocytoma 5Glioma 3Germinoma 2Benign pineal cyst 1Mature teratoma 1Cavernous malformation 1Pineal tumor not otherwise specified 1Midbrain and/or diencephalon (glioma or cyst) 13 (76.9%)Arachnoid cysts (prepontine, third ventricle,post fossa, or suprasellar)

7 (85.7%)

Neurocysticercosis (intraventricular,aqueductal, and tectal), nonresectable

4 (100%)

Cerebellar metastases 3 (66.7%)Fourth ventricle (medulloblastoma,choroid plexus papilloma, and glioma)

3 (66.7%)

Optic glioma 1 (100%)Petroclival meningioma 1 (100%)Colloid cyst 1 (0%)

Idiopathic aqueductal stenosis 52 (86.5%) ⁎

Triventricular hydrocephalus withoutimaging evidence of obstruction

4 (50%)

6 J.R. Dusick et al. / Surgical Neurology 69 (2008) 5–15

the spectrum, infants with presumed immature CSF absorp-tion mechanisms have generally been considered to have ahigher ETV failure rate compared with children older than 6to 12 months [5,7,12,15,20,22,26,31,37,38,48,64,66]. Inade-quate CSF absorption may play a role with advancing age,especially among elderly patients, in whom an increase inCSF absorption resistance [2,17,39,62,63] might likewiselead to a lower ETV response rate.

Chronic “compensated hydrocephalus,” in which bothstructural and physiologic hydrocephalic brain changes havebeen present for decades, may be less likely to respondfavorably to an abrupt change in CSF circulation. Alter-natively, adult patients with long-term shunted congenitalhydrocephalus may have developed increased resistance toCSF outflow, thereby predisposing to ETV failure secondaryto inadequate CSF absorption [2,17,39].

From a methodological standpoint, judging the success orfailure of an ETV operation may be more problematic inadults. For example, patients with subtle symptoms andchronic ventriculomegaly may not experience dramaticclinical improvements, or improvements may be delayed.Here, we present a decade experience of consecutive ETVoperations as the treatment of adult hydrocephalus.

Other 5 (40%)Fourth ventricle outlet obstruction bywebs/adhesions

3

Fourth ventricle outlet obstruction byChiari's malformation

1

Third ventricle obstruction by largemassa intermedia

1

⁎ Including 2 patients with successful ETV after reoperation.

2. Materials and methods

2.1. Patient population

We retrospectively identified all consecutive patients whounderwent ETV procedures, with a primary goal of shuntavoidance, performed by the senior author (MB) at UCLAand Harbor-UCLA Medical Centers, Torrance, Calif, from1995 to 2006. Patients younger than 16 years were excluded.Nineteen patients who underwent a combined CSF shuntplacement and an ETV procedure in the same setting duringthis period were not included in the analysis because theintent of the ETV procedure in these cases was not shuntavoidance. The resultant analysis consisted of 110 ETV pro-cedures performed in 108 patients.

The mean age was 48 years (range, 17-88 years), andthere were 48% males. Median hospital length of stay was3 days (range, 1-59 days). Median postoperative follow-upwas 8 months (range, b1 to 101 months). This study wasapproved by our institutional review board.

Diagnoses included 47 cases (43.5%) of mass lesionscausing noncommunicating hydrocephalus, 52 cases (48%)of idiopathic aqueductal stenosis, 4 cases (3.7%) oftriventricular hydrocephalus with a narrowed but patentaqueduct, and 5 other cases (4.6%) of hydrocephalus due tovarious pathologies (one-third ventricular obstruction due tolarge massa intermedia and 4 fourth ventricular outletobstructions due to Chiari's malformation or webs/adhesionsof unknown etiology, Table 1). The cases of fourth ven-tricular outlet obstruction demonstrated a dilated aqueductand proximal fourth ventricle with an obliterated andadhered distal segment including the area of the outlet

foramina. Most of these patients had a suspected, butunconfirmable, history of neurocysticercosis. In 2 of thesecases, the fourth ventricle was explored at the time ofsurgery, confirming the abnormal anatomy with completefourth ventricular outlet obstruction. None of the casesdemonstrated a definitive mass or cystic lesion that could besafely resected or undergo a biopsy.

In 25 cases, one or more additional endoscopic procedureswere also performed at the same time as the ETV procedure.These included tumor biopsy or resection (12 cases), cystexcision or fenestration (13 cases), septum pellucidotomy(5 cases), and foramen of Monro-plasty (3 cases).

2.2. Indications for surgery and reoperation

The clinical indications for treatment varied, but ingeneral, all patients had symptoms and/or clinical signsreferable to hydrocephalus. The most common symptomsincluded headache, nausea and vomiting, visual dysfunction,memory or cognitive decline, ataxia, urinary symptoms,shunt infection, and decline in mental status. Twenty-sixpatients presented with clinical symptoms suggestive ofnormal pressure hydrocephalus (disturbance of gait, urinarybladder control, and/or cognition); however, most of thesehadMRI evidence of noncommunicating hydrocephalus and,

7J.R. Dusick et al. / Surgical Neurology 69 (2008) 5–15

therefore, were not true idiopathic normal pressure hydro-cephalus cases. Only 4 of these patients had a true diagnosisof idiopathic normal pressure hydrocephalus without acomplete obstructing lesion visible on neuroimaging.

Seven patients presented with an existing shunt, slitventricles, and chronic, episodic, debilitating headaches.The shunt reservoirs, if present, depressed and refilled veryslowly. The headache symptoms were variable but was notcharacterized by postural (“overdrainage”) headaches.Many of them had been previously treated with shuntvalve changes (either to higher-level antisiphon devices orto programmable valves) with no success. They all had along history of hydrocephalus and had been chronicallyshunted. The documentation of intracranial hypertensionwas not considered as an essential part of the diagnosis,and in fact, many did not have elevated ICP at presentation.The initial procedure consisted of externalization of theshunt and placement of an intraparenchymal ICP monitor.Cerebrospinal fluid flow was completely stopped using a3-way stopcock, and then the patient was closely monitoredfor changes in alertness and/or pathologic elevations ofICP. Deterioration of level of consciousness was easilyreversed by controlled CSF drainage. The subset of patients(included in this study) who developed ventriculomegaly(typically within 48 hours), and who had aqueductalobstruction documented by CISS MRI, underwent anETV and shunt removal.

Fig. 1. A: Preoperative sagittal CISS MRI in a patient with aqueductal webs. Notbasilar artery and brainstem. B: Postoperative sagittal CISS MRI in the same patienthird ventricle through the ostomy to the premesencephalic and prepontine cistern

In every case, the primary treatment goals were sympto-matic improvement and shunt independence. Patients werenot treated if they had no new or progressive preoperativesymptoms. It was explained to the patient that the ETVprocedure was being performed in lieu of placement (orreplacement) of a shunt. Each patient understood that ashunt procedure would be recommended in the advent ofclinical failure.

The determination of anatomical eligibility for ETV wasbased on MRI. In 58% of cases (78% in the latter half ofpatients in the series), this was based on high-resolution T2-weighted sagittal imaging (CISS or, alternatively, FIESTAsequences) (Fig. 1) [3]. In addition to assessing for theanatomical location and characteristics of the site ofventricular obstruction, careful attention was also placedon visualizing the premesencephalic cisternal anatomy. Thepatients in which the brainstem and distal basilar artery wasjudged to be too close to the clivus (in general, less than2-mm space) were not considered for ETV, and eitherunderwent endoscopic aqueductoplasty or shunt placement.

Failure of the ETV was judged based on any of thefollowing: (1) worsening of hydrocephalus-related neurolo-gic symptoms, (2) progressive ventricular enlargement, or(3) continued (preoperative) symptoms persisting greaterthan 6 weeks. Placement of a CSF shunt was recommendedfor all ETV failures (as noted above) with the exceptionof 2 patients for whom a repeat ETV procedure was

e the adequate space in the premesencephalic cistern (arrow) anterior to thet. The open third ventriculostomy is visible as is a flow void from the anterior(arrowheads).

8 J.R. Dusick et al. / Surgical Neurology 69 (2008) 5–15

recommended. Amongst ETV failures, no patient declinedCSF shunt placement if it was recommended.

2.3. Imaging analysis

The mean ventricular index, determined by the ratio ofmaximal bifrontal horn diameter to the maximal intracranialdiameter, was 0.36 ± 0.06. The mean degree of bulging of thethird ventricular floor [35], determined by drawing a linefrom the optic chiasm to the mammillary commissure andmeasuring a straight line down from this line to the deepestpoint of the third ventricular floor, was 3.3 ± 2.6 mm. Forstatistical analysis, 2 types of postoperative imaging studieswere studied and independently analyzed for predictors ofoutcome. The first consisted of the first postoperativeimaging study obtained after the ETV procedure. Imagesfor the second set were chosen in the following hierarchicalorder: first choice, an imaging study obtained between 1 and12months after surgery demonstrating stable ventricular size;or second choice, the last available imaging study beforeplacement of a shunt.

The grading system devised by Kehler et al [35] was usedto stratify patients on a 5-point scale. Grading is dependenton 3 factors: (1) presence of imaging evidence of obstruc-tion, (2) degree of bulging of the floor of the third ventricle,and (3) presence of progression in preoperative symptomswithin the 12 years preceding the ETV.

2.4. Surgical technique

In more than 95% of cases, a 3.7- or 4.0-mm-diametersteerable flexible endoscope was used for the ETVprocedure. All other cases were performed using the Gaab'srigid neuroendoscope system (Karl Storz, Tuttlingen,Germany). Early in the series, a mistrajectory resulted in athalamic injury, prompting a change in protocol in whichframeless stereotactic neuronavigation was used for cannula-tion in any patient without significant ventricular enlarge-ment. In the 7 patients with shunt-related slit ventricles,ventriculomegaly was introduced by externalizing the shunt48 to 72 hours before the ETV [6,10].

For the flexible endoscope cases, the following techniquewas used. To minimize intraoperative pneumocephalus, wesupported the patient's head on a Mayfield horseshoeapparatus and brought up the back of the bed so that thebur hole, placed approximately 3 cm lateral to midline at thecoronal suture, was at or near the vertex. The ventricle wascannulated using a 14F peel-away catheter with a blunt-tipped obturator. Under endoscopic visualization, the floor ofthe third ventricle was perforated at the tuber cinereum, justposterior to the infundibular recess, using the nonfloppy endof a 0.035-in diameter Bentson Glidewire (Boston Scientific,Natick, Mass) passed transendoscopically. This rounded-tipwire was used to minimize the risk of adventitial damage tothe basilar artery terminus. The fenestration was enlargedwith a 3F elliptical balloon catheter (Cook Medical Inc,Bloomington, Ind). The premesencephalic cisternal space

was explored with the endoscope to confirm that the openingwas in continuation with the premesencephalic subarachnoidspace. Any bleeding encountered was allowed to sponta-neous stop under direct visualization and continuousirrigation. The bur hole was covered using a titanium burhole cover placed over a collagen sponge (Helistat; IntegraLifesciences Corp, Plainsboro, NJ).

2.5. Statistical analysis

The time course and proportion of success and shuntindependence after ETV was evaluated by Kaplan-Meier andproportional hazards regression analysis. In addition, impactof clinical predictors on the ETV success curve was evaluatedby proportional hazards regression. Outcomes were evalu-ated separately for 2 definitions of success: (1) shuntindependence without need for subsequent mechanicalshunt, and (2) postoperative improvement in preoperativesymptoms. Predictors for success (both shunt independenceand improvement in symptoms) and predictors of complica-tions were evaluated by a multivariate general linear regres-sion model and/or odds ratio analysis and 2-tailed Fisherexact test. The temporal course in relation to outcomes wasevaluated using a general linear model, each case numbered1 through 108 chronologically by procedural date. Allstatistical analyses were performed using the R programminglanguage [50].

3. Results

3.1. Outcomes—shunt independence

Eighty-three patients (77%) continued to be shunt freeafter the primary ETVoperation (mean follow-up, 8 months).Two patients, who clinically failed the primary ETVprocedure, underwent a second ETV operation based onMRI documentation of lack of CSF pulsations across thefenestration. Both achieved success after reoperation andthen remained shunt free for the duration of their follow-up(3 and 36 months). Therefore, a total of 85 patients (79%)achieved shunt independence.

Of the patients who achieved shunt independence, 36 hadspace-occupying lesions. Of these, only 7 had definitivetreatment of their lesion at the time of ETV or later. Anadditional 6 patients had chemotherapy and/or radiationtreatment at a later date. The remainder either had priordefinitive treatment of their lesion with persistent hydro-cephalus or had no surgical treatment indicated or possible.

Kaplan-Meier survival curve analysis, where survival wasconsidered as duration of shunt-independent time after ETV,predicted that 62% of patients would remain shunt free afterstabilizing after 27 months postoperatively (Fig. 2).

Of the patients who ultimately failed ETV, requiring aCSF shunt procedure, 11 (48%) failed early within the firstmonth. Therefore, 52% of patients who ultimately failed hadmore than 1 month of shunt-free existence. Six of thesepatients (50%) failed much later, after 6 months, and of all

Fig. 2. Kaplan-Meier plot showing predicted survival curve of ETVprocedure. The curve represents the proportion of patients remainingshunt free over time postoperatively. Dotted lines represent the 95% con-fidence interval.

Table 2Variables analyzed as predictors of outcome

Clinical

Additional endoscopic procedures performed at the same operationAgeCase no. in seriesComplicationsMedicalSurgical

SexInfection, history ofShunt infectionMeningitis

Prior mechanical shunt

Preoperative imaging

Bulging of third ventricular floorCISS MRICisternal compressionGrade [35]Site of obstructing lesionSlit ventriclesThird ventricular widthTransependymal edemaPresent?Amount

Type of obstructionSpace occupying lesionWebs/adhesions

Ventricular index

First postoperative imaging

Cisterns, improved or worseThird ventricular width, change inTransependymal edema, improved or worseVentricular index, change in

Stabilized postoperative imaging

Bulging of third ventricular floor, change inCisterns, improved or worseFlow void through third ventriculostomyOpen third ventriculostomy visibleThird ventricular width, change inTransependymal edema, improved or worseVentricular index, change in

9J.R. Dusick et al. / Surgical Neurology 69 (2008) 5–15

those patients who failed late, the average time to failure was10 months after ETV (Fig. 3). Of these patients who failedafter 1 month, 75% had transient postoperative improvementin symptoms before ultimate failure (see “Outcomes—symptomatic relief” section).

3.2. Predictors of success—shunt independence

Multivariate general linear regression model analysis wasused to evaluate clinical predictors of outcome, includingclinical data, preoperative imaging, and postoperativeimaging (Table 2). This analysis only revealed 2 statisticallysignificant predictors of shunt independence: ETV casesperformed later in the series and symptomatic improvement.The rate of successful cases improved in the latter part of theseries (P = .028). Enduring shunt independence was alsopredicted by postoperative improvement in hydrocephalus

Fig. 3. Number of patients that failed at specific postoperative months.

symptoms (odds ratio, 13.0; 95% confidence interval, 3.78-46.77, P b .0001). On univariate analysis, there was a trendtoward the location of obstruction being predictive ofsuccess. The success rate for aqueductal stenosis was 84%vs 67% for all other locations combined (P = .066), andobstruction of the third ventricle had the lowest success rate,55% (P = .05 compared with all others and P = .039compared with aqueductal stenosis).

The low success rate in patients with obstruction of thethird ventricle was in part because 2 of these patients diedfrom postoperative hemorrhage from malignant thalamicgliomas, one due to biopsy at the time of ETV and onewithout biopsy. The other failures in this group were in apatient with a nontumoral obstruction (large massa inter-media) who failed late (after 28 weeks) and a patient who

Table 3Complications

Surgical complication (6 patients) No. of patients

New or worsened hemiparesis (one mistrajectory) 2Thalamic injury with transient

akinetic mutism (mistrajectory)1

Transient seizures 1Operative mortality 2Intraventricular hemorrhage after thalamic biopsy 1Intracerebral hemorrhage into thalamic tumor(no biopsy)

1

Medical complications (10 patients) No. of patients

Pulmonary embolus and IVC filter placement 2Transient postoperative bradycardia 1Malignant hypertension-like syndrome related to

anesthesia medications1

Delayed cerebral salt wasting 1Meningitis 1Urinary retention 1Transient dystonic reaction to anesthesia medications 1Deep venous thrombosis requiring IVC filter

(no pulmonary embolus)1

Heparin-induced thrombocytopenia, PE, CHF,and pneumonia

1

IVC indicates inferior vena cava; CHF, congestive heart failure; PE,pulmonary embolism.

10 J.R. Dusick et al. / Surgical Neurology 69 (2008) 5–15

failed immediately after ETV with acute ventricular enlarge-ment. Only one failure was possibly related to tumor growthat 4 months postoperatively. Therefore, tumor growth and/orfurther tumor treatment contributed a very minor role in pooroutcomes of this patient group as a whole.

No other independent pre- or postoperative clinical orneuroimaging factors were found to be significant predictorsof outcome. Proportional hazards regression analysis of thesurvival curve for successful ETV revealed that the presenceof a complete visual obstruction on neuroimaging wasassociated with a predicted greater ETV survival. However,because the n for patients with no visual obstruction wassmall (n = 4), this was not statistically significant. Patientswith obstruction at the aqueduct also had greater long-termETV survival compared with obstruction at other sites.No other clinical factors significantly altered the survivalcurve profile.

We applied the grading system devised by Kehler et al[35] to 78 patients in which we had adequate imagingstudies. The grading scheme was not significant in predictingoutcome (P = 0.3). The greatest correlation with outcomeappeared to apply to those 16 patients with a score of5 (patients with bulging of floor of third ventricle greaterthan 5 mm, imaging evidence of obstruction, and progres-sion of hydrocephalus symptoms within 12 months beforeETV), 94% of whom remained shunt free.

Of the 4 patients with symptoms of normal pressurehydrocephalus without a complete obstruction apparent onpreoperative neuroimaging, 2 failed, requiring a VP shunt.

Of the 7 patients with slit ventricles, 6 (86%) were shuntindependent after shunt externalization, ventricular dilation,

and subsequent ETV. The one patient who ultimately failed,requiring reinsertion of a VP shunt, did so in a delayedfashion at close to 28 months postoperatively.

3.3. Outcomes—symptomatic relief

Of all patients, 88 (81.5%) had improvement of pre-operative symptoms. Eight (7.4%) were transient improve-ment. Of the patients who ultimately failed, requiringshunting, 48% had at least transient improvement insymptoms. Of the subset of these patients who failed early(within 1 month of the procedure), a likely explanation forthose with transient improvement relates to drainage of CSFat the time of the procedure. Of the subset of these patientswho failed after 1 month, 79% had transient improvement,and of the subset that failed after 6 months, 100% hadtransient improvement before ultimate failure. All 7 patients(100%) with slit ventricles had resolution of symptoms afterETV, one of which was transient in a patient who had a latefailure at 28 months.

3.4. Predictors of success—symptomatic relief

Multivariate general linear regression model analysis wasalso used to identify predictors of symptomatic improvement(Table 2). No pre- or postoperative clinical or neuroimagingfactors were found to be significant predictors of sympto-matic improvement.

3.5. Repeat ETV procedures

The clinical course of the 2 patients who underwent asecond successful ETV was unique among the series. In thefirst patient, he experienced a 1-month period of completeresolution of his headache symptoms after the initial ETV,followed by recurrence of headache. An MRI CSF flowstudy suggested obstruction of the stoma, and he wasreexplored endoscopically at 2 months. Intraoperatively, thestoma appeared open, so a VP shunt was placed with aprogrammable valve. His headache symptoms continued. Hedeveloped subdural hygromas and a shunt infection at 1 yearpostoperatively. The infected shunt was removed, and arepeat ETV performed at that time as a temporizing measure.The stoma had closed over this 1-year period. The patient'sheadache symptoms improved modestly, and he elected toremain shunt free.

The second patient was an elderly woman with aqueductalstenosis who underwent ETVand had an initially good result.However, over several months, her prior NPH-like symptomsgradually returned. She was followed with serial CISS MRIscans, and her stoma progressed to the point that it was apinpoint and her symptoms were worse than before. She wasalso offered a ventriculocisternal shunt or VP shunt. Shepreferred to reexplore the ETV, which was done about1½ years from the initial ETV. The stomawas confirmed to bevery small, and it was enlarged, along with arachnoidmembranes below the stoma. She did well after thisprocedure and continued to have symptomatic improvement

11J.R. Dusick et al. / Surgical Neurology 69 (2008) 5–15

and stable ventricular size. For all other patients with clinicalETV failure, MRI documented a patent stoma.

3.6. Complications

Six patients (5.5%) had postoperative surgical complica-tions (Table 3). This included 2 deaths (1.9%) secondary tointracranial hemorrhage. One death occurred in a patientwith a thalamic anaplastic astrocytoma, which did notundergo a biopsy during the procedure, yet had a massiveintracerebral hemorrhage from the tumor on postoperativeimaging. No direct trauma to the area was noted at surgeryand appears to be unrelated to direct surgical manipulation.The other death occurred in a patient who had a biopsy of athalamic anaplastic astrocytoma that resulted in largepostoperative intraventricular hemorrhage. Both deathsoccurred early in the series, and there have been no surgicalmortalities in the last 7 years.

Other complications included a mistrajectory of theendoscope into the ipsilateral thalamus in 2 patients.Symptomatically, this resulted in transient akinetic mutismin one patient and a new left hemiparesis in the other. Bothpatients with a mistrajectory had smaller ventricles than theaverage (ventricular index averaged 0.27 and 0.28 in patientswith mistrajectory vs average of 0.36 in all other patients).One other patient had transient postoperative seizures, andone patient had worsened hemiparesis after stereotacticpineal biopsy and ETV. Both patients with new or worsenedhemiparesis (one with mistrajectory discussed above and onewith stereotactic pineal biopsy) initially presented withadvanced aggressive malignant disease (pineal malignantglioma and small cell lung cancer metastases to the cerebellarvermis). They both died within a few months of surgerybecause of aggressive tumor progression.

Three other patients died in the follow-up period becauseof progression of underlying malignant disease or othercauses unrelated to hydrocephalus or the ETV procedure.

There were 10 (9%) patients with medical complications.Medical complications included one each of transientpostoperative bradycardia, a malignant hypertension-likesyndrome with anesthesia, delayed cerebral salt wasting,meningitis, urinary retention, a transient dystonic reaction toanesthetic medications requiring prolonged intubation, anddeep venous thrombosis without pulmonary embolusrequiring inferior vena cava filter placement. In addition,pulmonary embolus occurred in 2 patients with inferior venacava filter placement, and one patient had heparin-inducedthrombocytopenia, which lead to pulmonary embolus,congestive heart failure, and pneumonia.

3.7. Predictors of surgical complications

Analysis of predictors for complications revealed that siteof obstruction in the third or fourth ventricle was associatedwith an increased rate of complications (16% vs 2.5% for allother sites of obstruction, P = .03). Compared withobstruction of the aqueduct or basal cisterns, obstruction at

the third and fourth ventricle had 7.3 times greater odds ofsurgical complication (odds ratio, 7.3; 95% confidenceinterval, 1.05-62.7).

There were fewer surgical complications over time withETV performed latter in the series (complication rate of14% in the first 35 patients in the series and 1.4% in theremainder; odds ratio, 11.45; 95% confidence interval, 1.2-270.3; P = .015). Factors that were not significant inpredicting complications included preoperative size of thirdventricle (P = .7), preoperative ventricular index (P = 1.0),and whether the patient had CISS/FIESTA imaging studiespreoperatively (P = .2). However, the 2 patients withmistrajectories to the ventricles did have smaller ventricularsize compared with the average for the series (ventricularindex averaged 0.275 vs 0.36 for all others, P value notable to be calculated).

4. Discussion

In this clinical series of 108 adult hydrocephalic patientstreated by ETV, 79% achieved shunt independence. Thisrate, if considered an indicator of clinical success, is onpar with prior retrospective clinical ETV series reportingsuccess rates ranging from 50% to 94% [4,8,9,11,14,15,21,23,27,29,32,34,36,47,49,51-53,57,60,61]. Many of thelargest series of outcomes for ETV have been in primarilypediatric populations [7,8,15,47]. The largest purely adultseries of ETV, or series with separate analyses for the adultsubgroup [4,9,27,29,34,36,52,53,56,57], demonstrate success-ful outcomes ranging from 66% to 88% [9,27,29,34,52,53].In respect to both shunt independence and symptomaticimprovement, the outcomes in our series are comparable withthese prior studies.

Clinical predictors of outcome for ETV can be dividedinto 2 categories, preoperative characteristics that can beused to predict those patients who are likely to benefit for theprocedure and postoperative findings that indicate successand predict long-term shunt independence. The former canhelp to define indications for surgery and those patients mostlikely to benefit from the procedure. The latter may aid inmanaging the patient postoperatively and predicting whichpatients will remain shunt free and which will ultimately fail.This study analyzed both such types of factors. Most of theprior studies of risk factors for success and failure of ETVhave been in primarily pediatric series.

4.1. Preoperative predictors of outcome

One of the most commonly cited preoperative factor thatpredicts outcome is the etiology of the hydrocephalus. Inparticular, idiopathic aqueductal stenosis [14,23,24,27],arachnoid cysts [47], and external tumor compression[21,24,27,41,51] have all been found to be predictive ofsuccessful outcomes after ETV. In our analysis, obstructionof the cerebral aqueduct had the highest rate of success(84% vs 67% for all others, P = .066); however, on

12 J.R. Dusick et al. / Surgical Neurology 69 (2008) 5–15

multivariate analysis, the site of obstruction was notstatistically significant.

In our series, a history of shunt infection or meningitis orprior shunting was not predictive of outcome. In fact, of 7patients with a history shunt infection and/or meningitis, allhad functioning ETV during the follow-up period. Ourresults differ from those of Fukuhara et al [23] who reportedthat a history of shunt infection and/or meningitis waspredictive of failure of ETV. However, both Siomin et al [57]and Cinalli et al [16] have found that patients withhydrocephalus with a history of infection can have goodoutcomes after ETV.

Amini and Schmidt [4] reported that a history of priorshunting for congenital aqueductal stenosis is less oftensuccessful than patients without a prior shunt. Our data foundno apparent difference, and in fact, patients with prior shuntshad a slightly higher success rate (79%) compared with thosepatients with no prior shunting procedure before ETV (77%).

Kehler et al [35] applied a 5-point grading scheme scale,with the highest scores given to patients with imagingevidence of obstruction, progression of symptoms, andprofound bulging of the floor of the third ventricle, andfound that this combination of findings correlated withsuccess rates. In 78 patients in whom we could acquireadequate imaging to determine bulging of the floor of thethird ventricle, there was 94% success in the 16 patients witha score of 5 points compared with 79% success in all others.However, for lower scores, the success rates did not correlateas closely with the score (overall P = .3).

With regard to symptomatic improvement after ETV, nopreoperative factors were found to be significantly predictivein our analysis.

4.2. Postoperative findings predictive of outcome

Some studies have shown that postoperative reduction inventricular size is more pronounced in those patients whosucceed [7,24,55,59]. In particular, reduction in thirdventricular size is greater in successful cases [13,24]. Overallventricular size has been shown to reduce in both successesand failures but tend to do so to a greater extent in patientswho remain shunt free [40]. However, other studies havefailed to show a correlation of postoperative ventricularchanges with outcome, and it has been suggested that the sizeof the ventricles is not valid predictors of outcome, andreliance on imaging should be avoided [28].

A visible stoma or CSF flow void in the area of the thirdventriculostomy has also been shown to be an indicator ofsuccess [21,24,40,65]. We did not find postoperativeimaging, either change in ventricular size or presence of aflow void, to be predictive of shunt independence. There areseveral possible explanations for this seeming discrepancy.Most patients who failed had a patent stoma around the timeof failure. Likewise, a couple patients did not have a clearlyvisible stoma based on routine MRI protocols and/or CSFflow studies (not CISS MRI protocols); yet, they remainedshunt free. About midway through the study period, we

abandoned MRI CSF flow studies to assess stoma patencydue to a perceived high false-positive rate. Sagittal planeCISS imaging proved to be far superior to standard sagittalT1-weighted imaging for assessing anatomical patency inour experience. In addition, the statistical analysis was alsoaffected by the fact that so few ETV stomas weredetermined to be occluded using CISS imaging. An openstoma was endoscopically confirmed in many cases at thetime of shunt placement.

With regard to our current management of ETV patients,we obtain CISS imaging routinely at 3 months post-operatively, and then only as needed depending upon thesymptoms. For patients who are improved, no furtherimaging is typically obtained. We consider CISS imaging areliable indicator of stoma patency based on the reexplora-tion cases noted above. Using this protocol, we have rarelyfound occluded stomas. Of note, this may be in part due to anETV selection criteria that typically excludes patients with ahigh severe meningitis or chronic meningitis (ie, coccidio-mycosis, etc).

Prior studies reported a seemingly obvious correlationbetween clinical improvement and shunt-free existence[13,21]. This is consistent with our findings becausesymptomatic improvement was the only postoperative factorthat was found to be significantly predictive of long-termshunt independence. Of the patients who required implanta-tion of a mechanical shunt, however, 56% failed after1 month postoperatively (average, 10 months). Many ofthese patients had had transient postoperative improvementin symptoms. Although this statistic may seem obvious, itreinforces the fact that a patient's symptoms are the ultimatedeterminant of success of the procedure. A patient with arapid deterioration of hydrocephalus-related symptoms afterETV warrants a new procedure such as mechanicalshunting. Because we have not found postoperative imagingfindings to be predictive of outcome, the decision toreoperate is based almost entirely on symptom resolutionrather than neuroimaging.

4.3. The difficulty of defining success after ETV

In the existing literature, success for ETV has been mostcommonly defined as enduring shunt independence after theprocedure [4,8,21,23,28]. We chose shunt independence asour primary outcome measure for several reasons. First,shunt independence was one of the primary goals of the ETVprocedure and hence not an analytical afterthought. Second,from a retrospective data collection perspective, it was theleast ambiguous. A potential flaw, however, is that there maybe a hesitancy to recommend shunt placement in light ofsuspected ETV failure. Or, alternatively, one might imagine ascenario in which a minimally symptomatic patient, whoeither marginally improves or does not improve at all, mightchoose to postpone shunt placement until there is asignificant clinical deterioration. Studies including suchultralate ETV failures would likely require a much longerfollow-up period than ours.

13J.R. Dusick et al. / Surgical Neurology 69 (2008) 5–15

Some studies have considered symptomatic improvementas a criterion for success along with or independent of shuntindependence [14,18,27,34,61]. Of the adult series, Tisellet al reported at least partial improvement in hydrocephalussymptoms in 94% of 18 patients with primary aqueductalstenosis. Grunert et al reported excellent or improvedsymptoms in 81.6% of 120 adults with various causes ofhydrocephalus. Their best results (84.5%) were in patientswith benign aqueductal stenosis. In our series, symptomaticrelief was accomplished in 81.4% of our patients overall andin 86.5% of patients with benign aqueductal stenosis.

It has been suggested that simple shunt independence and/or symptomatic improvement is not the only measure of theworth and success of ETV for patients who have hydro-cephalus. For example, Beems and Grotenhuis [8] assertedthat it should not always be considered a complete failure if apatient requires a shunt after an ETV. They discuss examplesof patients with slit ventricle syndrome and multiloculatedhydrocephalus as well as CSF reabsorption problems. Themanagement of these patients and the requirement formultiple shunts may be decreased after ETV even if theyultimately require a shunt. In addition, in their experience aswell as others [14], patients who underwent ETV who laterrequired shunting were less prone to shunt dysfunction andhad less shunt operations. Others have suggested that a priorETV may reduce the symptoms of a malfunctioning shuntsystem [33].

4.4. Late ETV failure in adult hydrocephalus

Although many studies have shown that most ETVprocedures that fail do so early [23,47,52], there are severalseries that support our finding that a significant proportion ofultimate failures occur in a delayed fashion [4,8,18,19]. Ofthe 23 patients who ultimately failed and required a shunt inour series, 12 patients failed in a delayed fashion more than1 month after the ETV (average, 10 months; range, 2-28 months). Seven of these patients failed after 6 months. Ofthe 12 delayed failures, 9 had improved symptoms duringthe shunt-free interval. Seven of our patients failed after6 months. All had had temporary improvement in symptoms.Late rapid deterioration after ETV has been reported in theliterature. In our series, we did not see any late rapiddeterioration with the morbidity and mortality reported in theseries of Drake et al [19].

4.5. Complications

Our overall complication rate of 5.5% is comparable withthe range of 6% to 14% acknowledged in prior reports[4,8,14,21,28,29,34,46,52,53,57]. Most of our complicationsoccurred in patients with intracranial malignancies, includ-ing the 2 mortalities. This risk of significant morbidity ormortality among patients with brain tumor treated with ETV,particularly if a biopsy is performed, has been describedpreviously by Beems and Grotenhuis [8] in a large pediatricpopulation. In their series, although the overall rate of

complications was only 7.7%, they found a 19.5%complication rate in patients who had a biopsy at the timeof their ETV. Their only fatality was in such a patient.

The most troubling complication encountered in ourseries was injury related to ventricular cannulation ratherthan the actual perforation of the floor of the third ventricle.In our experience, this type of complication occurred as aresult of placing a 14F obturator and peel-away sheath inpatients with only mild to moderate ventriculomegaly. Thesecases tended to occur earlier in the series, before we usedstereotactic guidance. Whenever possible, we now routinelyused frameless stereotaxy for such cases, as has beenadvocated elsewhere for patients with normal-sized orsmaller ventricles [58]. Another option to improve the safetyof these procedures is to use a smaller “finding” catheter orneedle to initially tap the ventricle rather than use the 14Fcannula or endoscope sheath.

Injury to the basilar artery has been reported anecdotallyafter ETV [1,30,43,53,54]. From a methodological perspec-tive, we attempted to minimize this complication by acombination of careful patient selection combined with asurgical technique perceived as the lowest risk. Currently,we routinely perform sagittal CISS sequence MRI tovisualize the premesencephalic cistern anatomy and do notperform the procedure if the cisternal space is too small,potentially placing the basilar bifurcation in harms way.When performing the actual fenestration, we avoidedsurgical techniques considered by some to increase therisk of basilar injury, such as the use of electrical or lasercautery [30,43].

4.6. Indications

Although ETV is clearly an established treatment ofnoncommunicating hydrocephalus, including adult patients,the indications for the procedure remain to be clearlydefined. In 96% of the cases in our series, evidence of anobvious obstruction of ventricular CSF flow was presentbased on objective MRI findings. Can we expect similaroutcome results after ETV in patients without noncommu-nicating hydrocephalus? Among our 4 patients with a patentaqueduct and no other obvious site of ventricular outflowobstruction, 2 improved. This success rate (50%) was lessthan for all other patients (80%) but not statisticallysignificant because of the small sample size. The reportedclinical results of small series of patients with triventriculo-megaly idiopathic NPH have been inconsistent, reportingclinical improvement with ETV ranging from 21% to 72%[25,42,44,45].

5. Conclusions

Endoscopic third ventriculostomy for adult noncommu-nicating hydrocephalus is a well-tolerated procedure that hasa high likelihood of achieving shunt-free existence. Successrates in achieving shunt independence are generally more

14 J.R. Dusick et al. / Surgical Neurology 69 (2008) 5–15

than 70% for a variety of underlying pathologies. Adultpatients undergoing ETV should be followed for at least a1-year period in the advent of delayed failure.

References

[1] Abtin K, Thompson BG, Walker ML. Basilar artery perforation as acomplication of endoscopic third ventriculostomy. Pediatr Neurosurg1998;28(1):35-41.

[2] Albeck MJ, Skak C, Nielsen PR, Olsen KS, Borgesen SE, Gjerris F.Age dependency of resistance to cerebrospinal fluid outflow.J Neurosurg 1998;89(2):275-8.

[3] Aleman J, Jokura H, Higano S, Akabane A, Shirane R, Yoshimoto T.Value of constructive interference in steady-state three-dimensional,Fourier transformation magnetic resonance imaging for the neuroen-doscopic treatment of hydrocephalus and intracranial cysts. Neurosur-gery 2001;48(6):1291-5 [discussion 5-6].

[4] Amini A, Schmidt RH. Endoscopic third ventriculostomy in a series of36 adult patients. Neurosurg Focus 2005;19(6):E9.

[5] Balthasar AJ, Kort H, Cornips EM, Beuls EA, Weber JW, Vles JS.Analysis of the success and failure of endoscopic third ventriculostomyin infants less than 1 year of age. Childs Nerv Syst 2007;23(2):151-5.

[6] Baskin JJ, Manwaring KH, Rekate HL. Ventricular shunt removal:the ultimate treatment of the slit ventricle syndrome. J Neurosurg 1998;88(3):478-84.

[7] Beems T, Grotenhuis JA. Is the success rate of endoscopic thirdventriculostomy age-dependent? An analysis of the results of endo-scopic third ventriculostomy in young children. Childs Nerv Syst2002;18(11):605-8.

[8] Beems T, Grotenhuis JA. Long-term complications and definition offailure of neuroendoscopic procedures. Childs Nerv Syst 2004;20(11-12):868-77.

[9] Bognar L, Markia B, Novak L. Retrospective analysis of 400 neuro-endoscopic interventions: the Hungarian experience. Neurosurg Focus2005;19(6):E10.

[10] Boschert JM, Krauss JK. Endoscopic third ventriculostomy in thetreatment of shunt-related over-drainage: preliminary experience with anew approach how to render ventricles navigable. Clin NeurolNeurosurg 2006;108(2):143-9.

[11] Brockmeyer D, Abtin K, Carey L, Walker ML. Endoscopic thirdventriculostomy: an outcome analysis. Pediatr Neurosurg 1998;28(5):236-40.

[12] Buxton N, Macarthur D, Mallucci C, Punt J, Vloeberghs M.Neuroendoscopic third ventriculostomy in patients less than 1 yearold. Pediatr Neurosurg 1998;29(2):73-6.

[13] Buxton N, Turner B, Ramli N, Vloeberghs M. Changes in thirdventricular size with neuroendoscopic third ventriculostomy: a blindedstudy. J Neurol Neurosurg Psychiatry 2002;72(3):385-7.

[14] Choi JU, Kim DS, Kim SH. Endoscopic surgery for obstructivehydrocephalus. Yonsei Med J 1999;40(6):600-7.

[15] Cinalli G, Sainte-Rose C, Chumas P, Zerah M, Brunelle F, Lot G,Pierre-Kahn A, Renier D. Failure of third ventriculostomy inthe treatment of aqueductal stenosis in children. Neurosurg Focus1999;6(4):e3.

[16] Cinalli G, Salazar C, Mallucci C, Yada JZ, Zerah M, Sainte-Rose C.The role of endoscopic third ventriculostomy in the management ofshunt malfunction. Neurosurgery 1998;43(6):1323-7 [discussion 7-9].

[17] Czosnyka M, Czosnyka ZH, Whitfield PC, Donovan T, Pickard JD.Age dependence of cerebrospinal pressure-volume compensation inpatients with hydrocephalus. J Neurosurg 2001;94(3):482-6.

[18] Devaux BC, Joly LM, Page P, Nataf F, Turak B, Beuvon F, Trystram D,Roux FX. Laser-assisted endoscopic third ventriculostomy forobstructive hydrocephalus: technique and results in a series of 40consecutive cases. Lasers Surg Med 2004;34(5):368-78.

[19] Drake J, Chumas P, Kestle J, Pierre-Kahn A, Vinchon M, Brown J,Pollack IF, Arai H. Late rapid deterioration after endoscopic

third ventriculostomy: additional cases and review of the literature.J Neurosurg 2006;105(2 Suppl):118-26.

[20] Etus V, Ceylan S. Success of endoscopic third ventriculostomy inchildren less than 2 years of age. Neurosurg Rev 2005;28(4):284-8.

[21] Feng H, Huang G, Liao X, Fu K, Tan H, Pu H, Cheng Y, Liu W, ZhaoD. Endoscopic third ventriculostomy in the management of obstructivehydrocephalus: an outcome analysis. J Neurosurg 2004;100(4):626-33.

[22] Fritsch MJ, Kienke S, Ankermann T, Padoin M, Mehdorn HM.Endoscopic third ventriculostomy in infants. J Neurosurg 2005;103(1 Suppl):50-3.

[23] Fukuhara T, Vorster SJ, Luciano MG. Risk factors for failure ofendoscopic third ventriculostomy for obstructive hydrocephalus.Neurosurgery 2000;46(5):1100-9 [discussion 9-11].

[24] Gangemi M, Donati P, Maiuri F, Longatti P, Godano U, Mascari C.Endoscopic third ventriculostomy for hydrocephalus. Minim InvasiveNeurosurg 1999;42(3):128-32.

[25] Gangemi M, Maiuri F, Buonamassa S, Colella G, de Divitiis E.Endoscopic third ventriculostomy in idiopathic normal pressurehydrocephalus. Neurosurgery 2004;55(1):129-34 [discussion 34].

[26] Gorayeb RP, Cavalheiro S, Zymberg ST. Endoscopic third ventricu-lostomy in children younger than 1 year of age. J Neurosurg 2004;100(5 Suppl Pediatrics):427-9.

[27] Grunert P, Charalampaki P, Hopf N, Filippi R. The role of thirdventriculostomy in the management of obstructive hydrocephalus.Minim Invasive Neurosurg 2003;46(1):16-21.

[28] Hellwig D, Grotenhuis JA, Tirakotai W, Riegel T, Schulte DM, BauerBL, Bertalanffy H. Endoscopic third ventriculostomy for obstructivehydrocephalus. Neurosurg Rev 2005;28(1):1-34 [discussion 5-8].

[29] Hopf NJ, Grunert P, Fries G, Resch KD, Perneczky A. Endoscopicthird ventriculostomy: outcome analysis of 100 consecutive proce-dures. Neurosurgery 1999;44(4):795-804 [discussion -6].

[30] Horowitz M, Albright AL, Jungreis C, Levy EI, Stevenson K.Endovascular management of a basilar artery false aneurysm secon-dary to endoscopic third ventriculostomy: case report. Neurosurgery2001;49(6):1461-4 [discussion 4-5].

[31] Javadpour M, Mallucci C, Brodbelt A, Golash A, May P. The impact ofendoscopic third ventriculostomy on the management of newlydiagnosed hydrocephalus in infants. Pediatr Neurosurg 2001;35(3):131-5.

[32] Jones RF, Kwok BC, Stening WA, Vonau M. The current status ofendoscopic third ventriculostomy in the management of non-commu-nicating hydrocephalus. Minim Invasive Neurosurg 1994;37(1):28-36.

[33] Jones RF, Stening WA, Brydon M. Endoscopic third ventriculostomy.Neurosurgery 1990;26(1):86-91 [discussion -2].

[34] Kadrian D, van Gelder J, Florida D, Jones R, Vonau M, Teo C, SteningW, Kwok B. Long-term reliability of endoscopic third ventriculostomy.Neurosurgery 2005;56(6):1271-8 [discussion 8].

[35] Kehler U, Regelsberger J, Gliemroth J, Westphal M. Outcomeprediction of third ventriculostomy: a proposed hydrocephalus gradingsystem. Minim Invasive Neurosurg 2006;49(4):238-43.

[36] Kelly PJ. Stereotactic third ventriculostomy in patients with non-tumoral adolescent/adult onset aqueductal stenosis and symptomatichydrocephalus. J Neurosurg 1991;75(6):865-73.

[37] Koch D, Wagner W. Endoscopic third ventriculostomy in infants ofless than 1 year of age: which factors influence the outcome? ChildsNerv Syst 2004;20(6):405-11.

[38] Koch-Wiewrodt D, Wagner W. Success and failure of endoscopic thirdventriculostomy in young infants: are there different age distributions?Childs Nerv Syst 2006;22(12):1537-41.

[39] Konig K, Heissler HE, Zumkeller M, Rickels E. Age-dependence ofcerebrospinal parameters. Acta Neurochir Suppl 2005;95:315-8.

[40] Kulkarni AV, Drake JM, Armstrong DC, Dirks PB. Imagingcorrelates of successful endoscopic third ventriculostomy. J Neurosurg2000;92(6):915-9.

[41] Li KW, Roonprapunt C, Lawson HC, Abbott IR, Wisoff J, Epstein F,Jallo GI. Endoscopic third ventriculostomy for hydrocephalusassociated with tectal gliomas. Neurosurg Focus 2005;18(6A):E2.

15J.R. Dusick et al. / Surgical Neurology 69 (2008) 5–15

[42] Longatti PL, Fiorindi A, Martinuzzi A. Failure of endoscopic thirdventriculostomy in the treatment of idiopathic normal pressurehydrocephalus. Minim Invasive Neurosurg 2004;47(6):342-5.

[43] McLaughlin MR, Wahlig JB, Kaufmann AM, Albright AL. Traumaticbasilar aneurysm after endoscopic third ventriculostomy: case report.Neurosurgery 1997;41(6):1400-3 [discussion 3-4].

[44] Meier U. Shunt operation versus endoscopic ventriculostomy innormal pressure hydrocephalus: diagnostics and outcome. ZentralblNeurochir 2003;64(1):19-23.

[45] Meier U, Zeilinger FS, Schonherr B. Endoscopic ventriculostomyversus shunt operation in normal pressure hydrocephalus: diagnosticsand indication. Acta Neurochir Suppl 2000;76:563-6.

[46] Navarro R, Gil-Parra R, Reitman AJ, Olavarria G, Grant JA, Tomita T.Endoscopic third ventriculostomy in children: early and latecomplications and their avoidance. Childs Nerv Syst 2006;22(5):506-13.

[47] O'Brien DF, Javadpour M, Collins DR, Spennato P, Mallucci CL.Endoscopic third ventriculostomy: an outcome analysis of primarycases and procedures performed after ventriculoperitoneal shuntmalfunction. J Neurosurg 2005;103(5 Suppl):393-400.

[48] O'Brien DF, Seghedoni A, Collins DR, Hayhurst C, Mallucci CL. Isthere an indication for ETV in young infants in aetiologies other thanisolated aqueduct stenosis? Childs Nerv Syst 2006;22(12):1565-72.

[49] Oka K, Yamamoto M, Ikeda K, Tomonaga M. Flexible endoneur-osurgical therapy for aqueductal stenosis. Neurosurgery 1993;33(2):236-42 [discussion 42-3].

[50] R Development Core Team. R: a language and environment forstatistical computing. Version 2.3.1. Vienna (Austria): R Foundationfor Statistical Computing; 2006.

[51] Ray P, Jallo GI, Kim RY, Kim BS, Wilson S, Kothbauer K, Abbott R.Endoscopic third ventriculostomy for tumor-related hydrocephalus in apediatric population. Neurosurg Focus 2005;19(6):E8.

[52] Santamarta D, Diaz Alvarez A, Goncalves JM, Hernandez J. Outcomeof endoscopic third ventriculostomy. Results from an unselected serieswith noncommunicating hydrocephalus. Acta Neurochir (Wien)2005;147(4):377-82 [discussion 82].

[53] Schroeder HW, Niendorf WR, Gaab MR. Complications of endoscopicthird ventriculostomy. J Neurosurg 2002;96(6):1032-40.

[54] Schroeder HW, Warzok RW, Assaf JA, Gaab MR. Fatal subarachnoidhemorrhage after endoscopic third ventriculostomy. Case report.Neurosurg Focus 1999;6(4):e4.

[55] Schwartz TH, Ho B, Prestigiacomo CJ, Bruce JN, Feldstein NA,Goodman RR. Ventricular volume following third ventriculostomy.J Neurosurg 1999;91(1):20-5.

[56] Sgaramella E, Sotgiu S, Crotti FM. Neuroendoscopy: one year ofexperience-personal results, observations and limits. Minim InvasiveNeurosurg 2003;46(4):215-9.

[57] Siomin V, Cinalli G, Grotenhuis A, Golash A, Oi S, Kothbauer K,Weiner H, Roth J, Beni-Adani L, Pierre-Kahn A, Takahashi Y,Mallucci C, Abbott R, Wisoff J, Constantini S. Endoscopic thirdventriculostomy in patients with cerebrospinal fluid infection and/orhemorrhage. J Neurosurg 2002;97(3):519-24.

[58] SouweidaneMM. Endoscopic surgery for intraventricular brain tumorsin patients without hydrocephalus. Neurosurgery 2005;57(4 Suppl):312-8 [discussion -8].

[59] St George E, Natarajan K, Sgouros S. Changes in ventricular volume inhydrocephalic children following successful endoscopic third ventri-culostomy. Childs Nerv Syst 2004;20(11-12):834-8.

[60] Teo C. Third ventriculostomy in the treatment of hydrocephalus:experience with more than 120 cases. In: Hellwig D, Bauer BL, editors.Minimally invasive techniques for neurosurgery. Berlin: Springer;1998. p. 73-6.

[61] Tisell M, Almstrom O, Stephensen H, Tullberg M, Wikkelso C. Howeffective is endoscopic third ventriculostomy in treating adulthydrocephalus caused by primary aqueductal stenosis? Neurosurgery2000;46(1):104-10 [discussion 10-1].

[62] Tisell M, Edsbagge M, Stephensen H, Czosnyka M, Wikkelso C.Elastance correlates with outcome after endoscopic third ventriculost-omy in adults with hydrocephalus caused by primary aqueductalstenosis. Neurosurgery 2002;50(1):70-7.

[63] Trantakis C, Helm J, Keller M, Dietrich J, Meixensberger J. Thirdventriculostomy in communicating hydrocephalus in adult patients—the role of lumbar and cranial cerebrospinal fluid outflow measure-ment. Minim Invasive Neurosurg 2004;47(3):140-4.

[64] Wagner W, Koch D. Mechanisms of failure after endoscopic thirdventriculostomy in young infants. J Neurosurg 2005;103(1 Suppl):43-9.

[65] Wilcock DJ, Jaspan T, Worthington BS, Punt J. Neuro-endoscopicthird ventriculostomy: evaluation with magnetic resonance imaging.Clin Radiol 1997;52(1):50-4.

[66] Yadav YR, Jaiswal S, Adam N, Basoor A, Jain G. Endoscopic thirdventriculostomy in infants. Neurol India 2006;54(2):161-3.