breast calcification- a diagnostic manual

Breast CalcificationA Diagnostic Manual

00-EvansPre-cpp 19/6/02 12:49 pm Page ii

This page intentionally left blank

Breast CalcificationA Diagnostic Manual

Edited by

Andy EvansConsultant Radiologist

Breast ServicesInternational Breast Education Centre

City Hospital, Hucknall RoadNottingham, UK

Ian EllisReader and Consultant in Pathology

Department of HistopathologyCity Hospital, Hucknall Road

Nottingham, UK

Sarah PinderSenior Lecturer in Histopathology

Department of HistopathologyCity Hospital, Hucknall Road

Nottingham, UK

Robin WilsonConsultant Radiologist

International Breast Education CentreCity Hospital, Hucknall Road

Nottingham, UK

Greenwich Medical Media4th Floor, 137 Euston RoadLondonNW1 2AA

870 Market Street, Ste 720San Francisco CA 94109, USA

ISBN number 1841101117

First Published 2002

Apart from any fair dealing for the purposes of research or private study, or criticism or review, as permit-ted under the UK Copyright Designs and Patents Act 1988, this publication may not be reproduced, stored,or transmitted, in any form or by any means, without the prior permission in writing of the publishers, or inthe case of reprographic reproduction only in accordance with the terms of the licences issued by the appro-priate Reproduction Rights Organisations outside the UK. Enquiries concerning reproduction outside theterms stated here should be sent to the publishers at the London address printed above.

The rights of Andy Evans, Sarah Pinder, Robin Wilson and Ian Ellis to be identified as editors of this workhave been asserted by them in accordance with the Copyright Designs and Patents Act 1988.

The publisher makes no representation, express or implied, with regard to the accuracy of the informationcontained in this book and cannot accept any legal responsibility or liability for any errors or omissions thatmay be made.

A catalogue record for this book is available from the British Library.

Distributed worldwide by Plymbridge Distributors Ltd and in the USA by Jamco

Typeset by Phoenix Photosetting, Chatham, Kent

Printed by MPG Books Ltd, Bodmin, Cornwall

Visit our website atwww.greenwich-medical.co.uk

00-EvansPre-cpp 19/6/02 12:49 pm Page iv

Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiContributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixAcknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi

1 Benign Breast Calcification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1I Ellis and A Evans

2 Intraductal Epithelial Lesions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31S Pinder and A Evans

3 Invasive Carcinoma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53S Pinder and A Evans

4 Fine Needle Aspiration Cytology and Core Biopsy of . . . . . . . . . . . . . . . . . . . . . . 69Mammographic MicrocalcificationsS Pinder and A Evans

5 Large Bore Biopsy of Breast Calcification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83R M Wilson

6 A Practical Approach to the Reporting of Percutaneous . . . . . . . . . . . . . . . . . . . . 99Sampling of Breast CalcificationsI Ellis and S Pinder

7 A Practical Approach to Radiological Diagnosis of Breast Calcification . . . . . . . . 109R Wilson and A Evans

8 Localising Breast Calcification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115H Burrell

9 Clinical Aspects Diagnosing Microcalcification . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127D McMillan

10 High Frequency Ultrasound . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139W Teh

11 Computer-aided Detection of Mammographic Calcifications . . . . . . . . . . . . . . . . 149S Astley

12 MRI Detection of DCIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157F Gilbert

13 The Nature of Breast Tissue Calcifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171K Rodgers and R Lewis

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191

v

00-EvansPre-cpp 19/6/02 12:49 pm Page v

Preface

vii

Breast calcification seen on mammographyoften represents difficult and complex diag-nostic challenges. Calcifications are moredifficult to detect, analyse and biopsy thanother mammographic abnormalities. Therehas been a great deal of research and tech-nical advances in this area over the last 10years. Despite this, the value of detectingbreast calcification remains controversial andthe continued subject of debate amongstbreast radiologists, pathologists and clini-cians.

Current practice means that some womenare harmed by the detection of mammo-graphic calcification by being subjected to avariety of investigations ranging from recallfor magnification views to surgical localisa-tion biopsy for what proves to be benigndisease or normality.

Although current percutaneous biopsytechniques have reduced the number ofwomen undergoing surgical biopsy forbenign calcifications, the process of recall,further views, percutaneous biopsy andreturn for results is stressful. Most calcifica-tions recalled at screening are benign andtheir detection and investigation confers nobenefit.

Mammographic calcification is, however,an important feature of invasive and in situbreast cancer. Its detection enables thediagnosis of high-grade invasive cancer atsmall sizes when the prognosis is good. Thedetection of high-grade DCIS prevents thedevelopment of high-grade invasive breastcancer and so saves lives. The value ofdiagnosing low-grade DCIS is less clear as asignificant proportion of these lesions wouldnever have become invasive in the woman’slifetime.

This book brings together in one volume,current thinking on the detection and diag-nosis of breast calcification. Current issuessuch as computer-aided detection, choice ofbiopsy technique, assessing adequacy of per-cutaneous biopsy, predicting and diagnosinginvasion will be addressed as well as morenovel approaches and ideas, such as the useof synchrotron radiation for diagnosis andthe use of mammographic calcification as aprognostic factor.

We hope this book will help thoseinvolved in managing women with breastcalcification to re-evaluate their own meth-ods of detection and diagnosis, the value ofdetecting such calcification and where thefuture lies in this fascinating field.

Andy Evans May 2002

00-EvansPre-cpp 19/6/02 12:49 pm Page vii

Sue AstleyImaging Science and BiomedicalEngineeringUniversity of ManchesterStopford BuildingOxford RoadManchester, UK

Helen BurrellDepartment of RadiologyBreast Screening UnitCity HospitalHucknall RoadNottingham, UK

Fiona GilbertAcademic Department of RadiologyUniversity of AberdeenWest Block, Polwarth BuildingForesterhillAberdeen, UK

Robert A LewisSchool of Physics and Materials EngineeringMonash UniversityVictoriaAustralia

Douglas McMillanProfessorial Unit of SurgeryCity HospitalHucknall RoadNottingham, UK

Keith D RodgersDepartment of Materials and MedicalServicesCranfield UniversityShrivenhamSwindonWiltshire , UK

Will L TehDepartment of RadiologyNorthwick Park and St Marks HospitalWatford RoadHarrow, UK

ix

Contributors

00-EvansPre-cpp 19/6/02 12:49 pm Page ix

I would like to thank Joanne Cooper whodespite working tirelessly preparing themanuscript and organising the book hasremained helpful and cheerful.

xi

Acknowledgement

00-EvansPre-cpp 19/6/02 12:49 pm Page xi

Chapter

1Breast benign calcification

Andy Evans and Ian Ellis

Introduction 3

Fibrocystic change 4

Fibroadenoma and fibroadenomatoid hyperplasia 5

Stromal calcification 12

Calcification within atrophic lobules (involutional change) 12

Sclerosing adenosis 12

Duct ectasia 15

Blunt duct adenosis 19

Vascular calcification 19

Fat necrosis 20

Skin calcification 21

Suture calcification 24

Epidermal inclusion cysts of the breast 26

Metastatic calcification due to renal failure 26

Klippel–Trenaunay syndrome 26

Idiopathic granulomatous mastitis 26

1

01-EvansCh1-cpp 19/6/02 12:53 pm Page 1

Breast benign calcification

Introduction

Calcification is the mammographic featuremost commonly associated with benign,screening-provoked surgical biopsy1. Thewidespread introduction of image-guidedcore biopsy and digital stereotaxis has madethe non-operative diagnosis of indeterminatemammographic calcification easier and morereliable. A consecutive series of 174 benign,screen-detected calcification clusters suspi-cious enough to warrant biopsy indicatesthat 87% of such calcification clusters wereable to be confidently diagnosed on core

biopsy alone. As can be seen from Table 1.1,the commonest causes of benign indetermi-nate calcification are fibrocystic change,fibroadenoma, stromal calcification andfibroadenomatoid hyperplasia. Less com-mon causes include involutional change,sclerosing adenosis, duct ectasia, aprocrinechange, mucocele and blunt duct adenosis. It is interesting to note that 12% of benignindeterminate mammographic calcificationdiagnosed on core biopsy have associatedusual hyperplasia. This is most commonlyfound in association with fibrocystic change. Table 1.2 shows benign causes of

3

1

Pathology Number (%) Usual hyperplasia no. (%)

Fibrocystic change 50 (33) 11 (22)Fibroadenoma 27 (18) 0Stromal calcification 23 (15) 2 (9)Fibroadenomatoid hyperplasia 22 (15) 0Involutional change 17 (11) 1 (6)Sclerosing adenosis 11 (7) 2 (18)Duct ectasia 6 (4) 1 (17)Apocrine change 6 (4) 1 (25)Blunt duct adenosis 4 (3) 1 (25)Mucocele 3 (2) 0Vascular 2 (1) 0Fat necrosis 2 (1) 0LCIS 1 (0.6) 0Radiation change 1 (0.6) 0Foreign body reaction 1 (0.6) 0

Table 1.1 Causes of indeterminate calcification diagnosed by core biopsy alone (n = 151)

Pathology Number (%)

Fibrocystic change 10Atypical lobular hyperplasia 6Atypical ductal hyperplasia 4Papillary lesion 3Radial scar 1Duct ectasia 1Sclerosing 1Fibroadenoma 1

Table 1.2 Causes of indeterminate calcificationdiagnosed at surgery (n = 23)


Breast calcification

indeterminate calcification diagnosed at sur-gical biopsy. In this group almost 50% of suchlesions contain atypical hyperplasia of eitherductal or lobular type. Usual type hyperpla-sia is present in over 50%. This group alsocontains lesions deemed to be of uncertainmalignant potential such as papillary lesionsand radial scars on the diagnostic corebiopsy2. In the past, the commonest cause fordiagnostic surgery for calcifications was theinability of the radiologist to adequatelysample the lesion. With the advent of digitalstereotaxis, the commonest reason for surgi-cal biopsy of benign calcifications is ofuncertain unalignant potential such as atypi-cal hyperplasia, other forms of epithelialatypia, papillary lesions or radial scar diag-nosed on image-guided core biopsy.

Fibrocystic changeFibrocystic change describes a variety ofmorphological alterations believed to be an exaggerated response to physiologicalchanges in breast tissue. Symptomaticpresentation usually occurs before the

menopause as benign appearing masses dueto cysts. Symptoms can persist if women goon to hormone replacement therapy in thepostmenopausal period. Microscopically, thelining of the cysts found in fibrocystic changeare of two types, those lined by cuboidalluminal or attenuated epithelium and thoselined by apocrine type epithelium.Calcification when present usually occurswithin the cyst fluid (Fig. 1.1).

Calcifications due to fibrocystic change areextremely common and are the commonestcause of benign, indeterminate mammo-graphic calcifications. Many cases ofcalcification due to fibrocystic change arecharacteristic enough not to require recall.These cases often demonstrate bilateral dif-fuse calcification, although the extent ofcalcification is often asymmetric. Such calci-fications often show a multiple lobulardistribution and are often much easier to seeon the mediolateral oblique view than the CCview (Fig. 1.2). This is because the calcifica-tions lie within small microcysts and thecalcific fluid layers out giving a partial “teacup” appearance on the mediolateral oblique

1

4

Fig. 1.1Histology of calcification withincyst fluid.



view but amorphous, low-density, roundedcalcifications on the CC view. When a clusterof calcification is due to fibrocystic change,careful inspection of the opposite breast often reveals fainter but similar morphologycalcifications. This can often be helpful inpreventing recall of benign calcification.Fibrocystic change does, however, com-monly present with a single clustered area ofgranular microcalcifications which showvariation in size, density and shape and, assuch, are indistinguishable on routine mam-mographic views from ductal carcinoma insitu (DCIS, Fig. 1.3).

Magnification views are often helpful indemonstrating the tea cup appearance on thelateral magnification view (Fig. 1.4) butmany cases of fibrocystic change do notconvincingly show this appearance andtherefore require image-guided core biopsy.Magnification views of fibrocystic changealso commonly demonstrate similar calcifica-

tions elsewhere within the breast. This phe-nomenon does not, however, exclude thepresence of DCIS. Because fibrocystic changeis so common, it commonly co-exists with thepresence of DCIS. Therefore the presence ofone or two tea cups should not precludeimage-guided biopsy if some of the other cal-cifications within the cluster show featureshighly suspicious of DCIS.

Fibroadenoma andfibroadenomatoidhyperplasiaFibroadenomas are the commonest cause of abreast lump in the female population. Theymost commonly occur in women in their 20sand 30s. Calcification within fibroadenomasin women of this age is unusual. After themenopause, fibroadenomas can becomehyalanised and suspicious microcalcification

1

5

Fig. 1.2CC view showing low-density roundedcalcifications with ill-defined edges dueto fibrocystic change.



can occur which is of dystrophic type andoccurs within the stroma or the epithelialclefts (Fig. 1.5).

On mammography coarse, popcorn-likecalcifications are often seen in involuting

fibroadenomas (Fig. 1.6). Calcification withthis characteristic morphology is no cause forconcern and does not require recall. Finecalcification can, however, occur withinfibroadenomas and, if a dense background

1

6

Fig. 1.3Mammographic image showing anirregularly shaped cluster ofpleomorphic calcifications. Thisarea of fibrocystic change wasmammographicallyindistinguishable from DCIS.

Fig. 1.4Lateral magnification view ofmicrocalcifications demonstratingthe “tea cup” sign indicating fibro-cystic change.



pattern obscures the well-defined margins ofthe mass component of the fibroadenoma,such calcifications can cause diagnostic diffi-culty (Fig. 1.7) and confirmation of theirbenign nature with image-guided core biopsyis required.

Fibroadenomatoid hyperplasia is acommon cause of impalpable mammographiccalcification, although it has only beenrecently described as a cause of suspicious

microcalcification3. Fibroadenomatoid hyper-plasia histologically displays compositefeatures of fibroadenoma and fibrocysticchange. Like many benign conditions of thebreast it has been previously described undera number of different names, including scle-rosing lobular hyperplasia, fibroadenomato-sis or fibroadenomatoid mastopathy. It ischaracterised by a proliferation of fibrousstroma within which are hyperplastic epithe-

1

7

Fig. 1.5Histological image ofdystrophic calcification withinfibroadenoma.

Fig. 1.6Mammographic image showingcoarse popcorn-like calcificationwithin a fibroadenoma.



1

8

Fig. 1.7An indeterminate cluster ofcalcification with no apparentassociated mass. Histologically, thiswas due to calcification within afibroadenoma.

Fig. 1.8Histological image showingcalcification within the stromalcomponent of fibro-adenomatoid hyperplasia.



1

9

Fig. 1.9Mammographic image showingirregularly shaped cluster ofpleomorphic granularmicrocalcifications due tofibroadenomatoid hyperplasia.

Fig. 1.10Granular and punctate calcificationsvarying in size and density due tofibroadenomatoid hyperplasia.



1

10

Fig. 1.11Mammographic image showing acluster of microcalcification with noassociated density. The calcificationscontain rod-like forms as well asgranular and punctate calcifications.Histologically, this is fibro-adenomatoid hyperplasia.

Fig. 1.12Histological image showingcalcification of normal breaststroma.



1

11

Fig. 1.13An indeterminate elongated clusterof calcifications containing granularand punctate forms. Histologically,this was due to stromal calcification.

Fig. 1.14Mammographic image showing elongated linearcalcifications within a cluster. Histologically, thiswas stromal calcification.



lial elements. Unlike fibroadenoma, fibroade-nomatoid hyperplasia does not present as awell-circumscribed mass. Symptomatically,fibroadenomatoid hyperplasia presents inyoung women with palpable masses. In suchsymptomatic women, mammography oftenshows mass lesions and rarely shows calcifi-cation. Fibroadenomatoid hyperplasia does,however, cause suspicious microcalcificationin postmenopausal women. Calcification gen-erally occurs following hyaline degenerationof the stromal component (Fig. 1.8).

The radiological features of fibro-adenomatoid hyperplasia manifesting ascalcification are those of granular microcalci-fications that show variation in shape, sizeand density. It virtually always presents withcalcification in a localised, irregularly shapedcluster. Rod-shaped calcifications are com-mon. Branching calcifications and a ductaldistribution are seen in a minority of cases(Figs 1.9–1.11). Histologically, the calcifica-tions seen in fibroadenomatoid hyperplasialie within the stroma in the vast majority ofcases and calcification is occasionally seen inthe sub-epithelial region. Fibroadenomatoidhyperplasia does not appear to be associatedwith malignant lesions and it can usually beconfidently diagnosed on image-guided corebiopsy. Surgical excision is rarely requiredfor diagnostic purposes.

Stromal calcificationNormal breast stroma can occasionally calcify and cause indeterminate mammo-graphic microcalcifications (Fig. 1.12). The cal-cification clusters vary in size and occasion-ally can be quite large. In a proportion of casesthe calcifications are elongated towards thenipple and this raises a suspicion of DCIS. Thecalcifications are usually granular in shapeand often vary markedly in size and density.Occasionally, elongated rod-like forms arealso found. There is usually no associatedmammographic density (Figs 1.13 and 1.14).

Calcification within atrophiclobules (involutional change)Calcification within atrophic lobules canoccur (Fig. 1.15) and this can give rise to sus-picious microcalcification on mammography.Most cases consist of oval or round clustersof predominantly punctate calcifications butwhich vary in size and density (Fig. 1.16).Occasionally, however, more suspiciousfeatures such as a ductal distribution andelongated linear forms can occur (Figs 1.17and 1.18).

Sclerosing adenosisSclerosing adenosis is an entity composed ofa proliferation of epithelial myoepithelialand connective tissue structures within a ter-minal duct lobular unit. Microscopically, thisproliferation of epithelial and interlobularstromal elements results in distortion andexpansion of lobules within an overallnodular or diffuse appearance. The epithelialcomponent forms microacinar structures thathave small luminal spaces. These luminalspaces frequently contain microcalcification(Fig. 1.19).

Presentation of sclerosing adenosis is var-ied. Symptomatically, it can often present as apalpable mass that can be ill-defined andfixed, and give a clinical impression of carci-noma. More commonly, sclerosing adenosispresents as a mammographic abnormality.Mammographically, the most commoncorrelate is microcalcification but sclerosingadenosis can also present as areas ofparenchymal distortion or an ill-definedmass. Mammographically, the calcificationsseen in sclerosing adenosis are always fineand contain a mixture of granular and punc-tate elements. There is often quite markedvariation in size and density of the calcificflecks. The calcifications occur in eitherround or oval clusters and sometimes amultilobular distribution can be present (Figs

1

12



1

13

Fig. 1.15Histological image showingcalcification within atrophiclobules.

Fig. 1.16Mammographic image showingpredominantly punctate calcificationsdue to calcification in atrophiclobules.



1

14

Fig. 1.17Calcification cluster showing a ductal distribution andelongated rod-shaped forms due to calcification withinatrophic lobules.

Fig. 1.18Mammographic image showinggranular calcifications with a markedduct distribution due to calcificationof atrophic lobules.



1.20–1.22). Elongated rod and Y shapes and aductal distribution of calcifications are notusually found.

Duct ectasiaDuct ectasia consists of dilatation of predom-inantly the sub-areola ducts that are often

filled with pultatious material resemblingcomedo-type DCIS. The duct lining epithe-lium often contains interspersed inflam-matory cells and macrophages. The duct walland periductal stroma also contain an inflam-matory reaction. Duct ectasia is common insmokers, and is often complicated by recur-rent periductal abscesses and mammillary

1

15

Fig. 1.19Histological imagedemonstrating calcificationwithin luminal spaces insclerosing adenosis.

Fig. 1.20Mammographic image showing anelongated cluster of pleomorphiccalcifications due to sclerosing adenosis.



fistula. Duct ectasia is also a cause of nipple discharge and nipple retraction.Calcifications due to duct ectasia are oftencharacteristic. The features are those ofcoarse, rod and branching calcifications in aductal distribution (Fig. 1.23). These calcifica-

tions are formed by calcification of debriswithin dilated ducts (Fig. 1.24). These intra-ductal calcifications have been described ashaving a “broken needle appearance”.Unlike the ductal calcifications seen in DCIS,it is rare for there to be associated fine granu-

1

16

Fig. 1.21A diffuse cluster of granular and punctatecalcifications due to sclerosing adenosis.

Fig. 1.22 (A, B)A widespread microcalcification with a multilobular distribution due to sclerosing adenosis.

A B



lar calcifications. In DCIS, although rod andbranching calcifications are common, it isvery rare for the number of rods and branch-ing calcifications to be higher than thenumber of fine granular calcifications pre-sent. Duct ectasia is very commonly bilateraland this feature is quite useful in confirmingthe benign nature of small areas of duct ecta-sia (Fig. 1.25). It is commonly found that

the debris within the duct in duct ectasiaextrudes into the surrounding peri-ductal tissues. This causes an inflammatory reactionaround the duct. This inflammatory reactionoften calcifies, leading to “lead pipe” appear-ing calcification (Fig 1.26). Very occasionally,high-grade DCIS can produce very coarsecalcifications which can be confused withduct ectasia. Caution should therefore be

1

17

Fig. 1.23Coarse rod- and branching-shaped calcificationsin a ductal distribution due to duct ectasia.

Fig. 1.24Histological image showingcalcification of the wall of amuscular blood vessel.



1

18

Fig. 1.25Bilateral mammographyshowing bilateral widespreadrod-shaped calcifications dueto duct ectasia.

Fig. 1.26Mammographic image showing lead-pipecalcifications due to calcification of periductal fatnecrosis associated with duct ectasia.



exercised in diagnosing a focal, unilateralarea of duct ectasia.

Blunt duct adenosisBlunt duct adenosis is a common conditionwhich forms part of the spectrum of fibrocys-tic change. It may manifest as a dominantcondition but rarely produces a symptomaticlesion in isolation. Microcalcification of lumi-nal secretions can occur and this results in itsdetection by mammography (Fig. 1.27).Microscopically, blunt duct adenosis is char-acterised by the replacement of the normallobular luminal epithelium by a layer of tallcolumnar epithelial cells with basal nucleiand apical cytoplasmic snouts. No atypia isusually present but nuclei may be enlarged.Mild stromal proliferation can be seen inassociation with blunt duct adenosis. Thecolumnar cells may secrete mucinous mater-ial to form microcysts and calcification ofcyst contents can occur7,8.

Mammographically, blunt duct adenosis ischaracterised by a small oval or round clusteror granular microcalcification (Fig. 1.28).Linear and rod-shaped forms are usuallyabsent and there is normally no ductal distri-bution.

Vascular calcificationVascular calcification is common and isusually not a diagnostic problem. It is char-acterised by serpentine, tramline calcifica-tions (Fig. 1.29). Magnification views areoccasionally helpful in characterising thenature of vascular calcification. On occa-sion, early vascular calcification can causediagnostic problems. If only one side of onevessel calcifies within an area of densebreast tissue this can give a false appear-ance of ductal calcification. We have occa-sionally inadvertently confirmed vascularcalcification within image-guided corebiopsy or Mammotome™. The amount of

postprocedure compressions required insuch cases is longer than usual!

A large study from The Netherlands (theDOM project) has shown an associationbetween breast arterial calcification and anumber of disorders related to increased oraccelerated arterial sclerosis; these includehypertension, transient ischaemic attack andstroke and myocardial infarction. In olderwomen there is an increased risk of diabetes9.A similar study from the USA, looking atwomen who have had both mammographyand coronary arteriography, did find an asso-ciation between breast arterial calcificationand ischaemic heart disease but only whenthe women were aged 59 years or less10.

Fat necrosisFat necrosis is a benign non-suppurativeinflammatory process that most commonlyoccurs subsequent to accidental or iatro-genic breast trauma (Fig. 1.30). Clinically,fat necrosis has a multitude of featuresvarying from single or multiple smoothround nodules to fixed irregular masseswhich simulate malignancy. The mammo-graphic findings of fat necrosis includelipid cysts, microcalcifications (Figs 1.31and 1.32), coarse calcifications and spicu-lated masses. Fat necrosis uncommonlycauses focally clustered pleomorphic micro-calcifications which are indistinguishablefrom malignancy11.

Skin calcificationSkin calcification is commonly demonstratedon screening mammograms and they usuallyshow characteristic round calcifications witha lucent centre; these calcifications are oftenclustered but the individual calcifications arenormally of a similar size. Skin calcificationsare often bilateral and symmetrical (Figs1.33–1.35). Very rarely, skin calcification of adifferent morphology can be seen in systemic

1

19



disorders such as dermatomyositis.Dermatomyositis has been shown to causebizarre sheet-like branching calcifications12.Soft-tissue calcification can occur in this con-dition between 4 months and 12 years afterthe onset of the disease. There have been anumber of reports of resolution at the calci-fications associated with dermatomyositis

using low-dose warfarin treatment. Focalskin lesions commonly calcify and skinpapillomas are the commonest cause ofabnormal focal skin calcification. Calcifiedpapillomas have an obvious cauliflowermorphology and do not normally cause diag-nostic difficulties. Occasionally, however,focal skin lesions can cause indeterminate

1

20

Fig. 1.27Histological image showingcalcification in luminalsecretions due to blunt ductadenosis.

Fig. 1.28Mammographic image showing a roundedcluster of pleomorphic granularmicrocalcifications due to blunt duct adenosis.



calcification (Fig. 1.36). Their nature can usu-ally be confirmed with tangential views.Occasionally, core biopsy is required.

A number of skin creams, ointments andpowders that contain metallic salts canmimic breast calcification. These includedeodorants, talcum powder, zinc oxide andgold injections. Soap has also been describedas a cause of an artifact that can mimic intra-

mammary breast calcification13. Tattoo markshave also been described as simulating intra-mammary calcification.

Suture calcificationCalcification of surgical sutures is occasion-ally seen, especially in the irradiated breast.The coarse linear morphology of the

1

21

Fig. 1.29Mammographic image showingserpentine, tram-linecalcification characteristic ofvascular calcification.

Fig. 1.30Histological image showingcalcification within fat necrosis.


1

22

Fig. 1.31A mammographic image following previous excision of abenign abnormality. Widespread punctate calcificationsare demonstrated due to fat necrosis; in addition,calcified oil cysts are seen.

(A).

(B)



calcifications is usually very characteristicand virtually never causes diagnostic diffi-culties (Fig. 1.37).

Epidermal inclusion cysts ofthe breastAlthough the characteristic finding ofepidermal inclusion cysts are those of a well-

circumscribed low-density mass in a subcuta-neous location, approximately 30% are associ-ated with heterogeneous microcalcifications14.

Metastatic calcification due torenal failureWomen with a secondary hyperparathyroidinduced by chronic renal failure have

1

23

(C)

Fig. 1.33Mammographic image showingskin calcification symmetricallydistributed in the inferiorbreast.

Fig. 1.32A series of mammographic imagesshowing the development ofcalcification due to postoperative fatnecrosis. (A) Demonstrates elongatedlinear calcification which could beviewed as suspicious of malignancy. (B)Illustrates coarsening of the calcificationand the development of thecharacteristic curvilinear calcificationseen in fat necrosis. (C) Furthercoarsening of the calcificationsdemonstrated the calcification is nowobviously benign.



increased breast calcification. Such calcifica-tions include an increase in vascular andbreast parenchymal calcifications. The pat-tern of such calcification is, however, usuallyof a benign morphology. Suspicious micro-calcification needs to be treated in the sameway whether a woman has secondary hyper-parathyroidism or not.

Klippel–Trenaunay syndrome

Breast calcification has been reported in asso-ciation with Klippel–Trenaunay syndrome.The calcification occurred within the subcu-taneous adipose tissue due to capillary andsmall venial proliferation. These containedintramural calcium deposits15.

1

24

Fig. 1.34Bilateral mammography showingsymmetrically distributed skincalcification adjacent to thepectoral muscle.

Fig. 1.35Magnified mammographic imageof skin calcification. Thecharacteristic round calcificationswith lucent centres isdemonstrated.



1

25

(A)

(B)

(C)

Fig. 1.36(A) Mammographic image of an area ofindeterminate calcification for whichimage-guided core biopsy was planned.(B) The patient volunteered during theprebiopsy consultation that she had a skinlesion which had a similar appearance tothe calcification. (C) Mammographyfollowing placement of a lead shot overthe skin lesion confirmed that thecalcification was indeed within the skin.



Idiopathic granulomatousmastitisGranulomatous mastitis occurs in women ofreproductive age and usually presents as atender mass which may be associated withaxillary lymphadenopathy and is oftenbilateral. Microscopic features include agranulomatous inflammatory cell infiltratecentred on, and distorting, lobules.

Mammographic features include areas ofcoarse dystrophic calcification, which is oftenmultifocal and bilateral (Fig. 1.38). Biopsy israrely required as the calcification morphol-ogy is not usually suspicious of malignancy(Figs 1.39 and 1.40).

1

26

Fig. 1.37Mammographic image showing thecharacteristic appearance of suturecalcification.



1

27

Fig. 1.38(A) Bilateral mammographyshowing bilateral widespreadrelatively coarse calcifications.(B) A close-up view shows thecoarse dystrophic nature of thecalcifications with the absenceof fine granular calcifications.The appearances are ofgranulomatous mastitis.

(A)

(B)



1

28

Fig. 1.39Two cases showing a well-defined mass withperipheral calcification. Calcification with thisdistribution is normally found within calcifiedwalls of cysts, oil cysts or haematomas.

(A)

(B)



References1. Spencer NJB, Evans AJ, Galea M et al. Pathological–

radiological correlations in benign lesions excisedduring a breast screening programme. Clin Radiol1994; 49: 853–6.

2. Ellis IO, Humphries S, Michell M, Pinder SE, WellsCA, MDZ Guidelines for non-operative diagnosticprocedures and reporting in breast cancer screening:NHSBSP 2001, Report No. 50.

3. Kamal M, Evans AJ, Denley H, Pinder SE, Ellis IO.Fibroadenomatoid hyperplasia: a cause ofsuspicious microcalcification on screeningmammography. Am J Roentgenol 1998; 171: 1331–4.

4. Bundred NJ, Dover MS, Alunwihore N, FaragherEB, Morrison JM. Smoking and peri-ductal mastitis.Br Med J 1993; 307: 772–3.

5. Dixon JM, ed. ABC of Breast Disease. London: BMJBooks, 2001.

6. Hughes LE, Mansell RE, Webster DJT. BenignDisorders and Diseases of the Breast. London:Baillière Tindall, 1989.

7. Chinyama CN, J.D.D. Mammary mucinous lesions;congeners, prevalence and important pathologicalassociations. Histopathology 1996; 29: 533–9.

8. Fraser JL, Raza S, Chorny K, Connolly JL, Schnitt SJ.Columnar alteration with prominent apical snoutsand secretions. Am J Surg Pathol 1998; 22: 1521–7.

9. Van Noord PA, Beijerink D, Kemmeren JM, Van derGraaf Y. Mammograms may convey more thanbreast cancer risk: breast arterial calcification andarterio-sclerotic related disease in women of theDOM cohort. Eur J Cancer Prev 1996; 5: 483–7.

10. Moshyedi AC, Puthwala AH, Kurland RJ, O’LearyDH. Breast arterial calcification: association withcoronary artery disease. Radiology 1995; 194: 181–3.

11. Hogge JP, Robinson RE, Magnant CM, Zuurbier RA.The mammographic spectrum of fat necrosis of thebreast. Radiographics 1995; 15: 1347–56.

12. Nye PJ, Perrymore WD. Mammographicappearance of calcinosis in dermatomyositis. Am JRoentgenol 1995; 164: 765–6.

13. Thomas DR, Fisher MS, Caroline DF. Case report:soap-author artifact that can mimic intramammarycalcifications. Clin Radiol 1995; 50: 64–5.

14. Denison CM, Ward VL, Lester SC et al. Epidermalinclusion cysts of the breasts: three lesions withcalcifications. Radiology 1997; 204: 493–6.

15. Apesteguia L, Pina L, Inchusta M et al. Klippel–Trenaunay syndrome: a very infrequent cause ofmicrocalcifications in mammography. Eur Radiol1997; 7: 123–125.

1

29

Fig. 1.40Bilateral mammography showsdiffuse punctate calcificationswithin both pectoral muscles.This is due to previous infectionwith Trichinella spiralis.


Chapter

2Intraductal epithelial lesions

Andy Evans and Sarah Pinder

Introduction 33

Radiology of ductal carcinoma in situ 33

Atypical ductal hyperplasia 48

31

02-Evans-Ch2-cpp 19/6/02 12:58 pm Page 31

Intraductal epithelial lesions

IntroductionIntraductal epithelial proliferations in thebreast form a spectrum from usual epithelialhyperplasia (UEH) through atypical ductalhyperplasia (ADH) to DCIS. Whilst UEH iscommonly seen as a component of fibrocysticchange and is of limited clinical significance,identification of DCIS at breast screening hassignificant implications. DCIS is composed of a proliferation of cytologically malignantepithelial cells contained within breastparenchymal structures with no evidence ofinvasion across the duct basement membrane.

Radiology of ductalcarcinoma in situ

Frequency of abnormalmammography according toclinical presentationMicrocalcification is the commonest mam-mographic feature of DCIS and is seen in 80%

to 90% of those cases with a mammographicabnormality1. However, the proportion ofsymptomatic DCIS with a mammographicabnormality varies according to the clinicalpresentation.

It is our experience that virtually all casesof DCIS presenting with single nipple dis-charge have a mammographic abnormality.In contrast, only about half the women withDCIS presenting as Paget’s disease of thenipple have a mammographic lesion2. Withinthe literature there is a huge variation in thereported incidence of mammographic abnor-malities in patients with Paget’s disease ofthe nipple ranging from 0 to 100% (Fig. 2.1).The reason why patients with Paget’s diseaseshould have such a low incidence of mam-mographic abnormalities is not clear. Themajority of DCIS causing Paget’s disease ofthe nipple is high grade and of solid architec-ture. Although high grade DCIS is normallyassociated with necrosis and the presence ofcalcification this is not often seen in DCISassociated with Paget’s disease of the nipple.

33

2

Fig. 2.1Mammogram showing extensivepredominantly linear calcification in apatient with Paget’s disease of thenipple. It should be noted how coarsethe calcifications are and it is easy forlesions with this appearance to befalsely thought to have duct ectasia.



DCIS can also present symptomatically asa palpable mass. When it presents in thismanner it is more likely to also show a masslesion mammographically, either entirelysolid or a mixed cystic solid lesion whenvisualised on ultrasound.

DCIS detected by mammographicscreening is, by definition, mammographi-cally visible except for the very few lesionspicked up due to clinical features noted byradiographers during the screening exami-nation.

Cluster shape

Approximately 80% of cases of calcificDCIS have an irregular cluster shape (Fig.2.2) and about 10% of these irregular clus-ters are V-shaped. The irregular clustershape of DCIS is caused by the growth pat-tern of DCIS. DCIS has a tendency to growtowards and away from the nipple within asingle segment of the breast. About 15% ofDCIS clusters have an oval or round cluster

shape (Fig. 2.3). DCIS with a round or ovalcluster shape is more likely to be confusedwith benign process than DCIS presentingwith an irregular cluster shape. However,the shape of a cluster of calcifications canbe particularly helpful when the nature of the individual calcifications are non-specific. For instance, a round or ovalcluster of four or five granular micro-calcifications may well be viewed as havinga very low risk of being DCIS. A similarnumber of calcifications within an irregu-larly shaped cluster (especially if this clus-ter is elongated towards the nipple) shouldbe viewed with a higher degree of suspi-cion. Intermediate- and low grade DCIScan present with a multilobular distributionof calcifications where calcifications appearto lie within multiple round or oval clus-ters within one area of the breast (Fig. 2.4).This distribution which appears to repre-sent disease processes within individualacini of the breast is also commonly foundin fibrocystic change.

34

2

Fig. 2.2Mammographic view showing anirregularly shaped cluster which iselongated towards the nipple.Histologically, this lesion was high-grade DCIS.



Distribution of calcifications inDCIS

One of the commonest and most characteris-tic features of DCIS is that the calcificationsare aligned in a ductal distribution (Fig. 2.5).This distribution is common in both necrotic

and non-necrotic DCIS. If calcifications lackrod or branching shapes, a ductal distribu-tion can be extremely helpful in suggesting amalignant cause of the calcifications. Thedistribution of calcification is also helpful inother ways. Diffuse calcification involvingthe whole of the breast is unusual and DCIS

2

35

Fig. 2.3Mammogram showing a roundedcluster of calcifications due to DCIS.The pleomorphism of the granularcalcifications thankfully indicated itsmalignant nature. Pathologically, thelesion was high grade DCIS.

Fig. 2.4Mammographic view showing amultilobular distribution of calcifications.Such a distribution is commonly found infibrocystic change but on this occasion wasdue to intermediate grade DCIS.



of such a large extent will usually also havecharacteristic calcification features. DCIS israrely bilateral; diffuse bilateral calcificationsare always benign and do not warrant fur-ther investigation unless there is a focal areawhere there is different morphology to thecalcifications compared to those elsewherewithin the breasts.

The number of calcificationsApproximately 90% of calcification clustersat our unit shown to be DCIS have more than10 flecks of calcification. However, diagnos-ing DCIS is not uncommon in lesions withclusters of five or fewer flecks. The decisionwhether or not to recall three or four flecks ofcalcification should be made predominantlyon the morphology of the calcifications andtheir distribution and whether the calcifica-tions have changed over time. Recallingpatients with mammograms with three flecks

of calcification which are new and in a ductaldistribution often leads to a diagnosis ofDCIS.

Calcification morphology inDCISThe most common features of calcificationsdue to DCIS are granular calcifications withirregularity in density, shape and size com-pared with the other calcifications withinthe cluster. Although these features arepresent in over 90% of cases of DCIS, theirusefulness in benign versus malignant dif-ferentiation is limited, as these features arealso commonly found in benign causes ofcalcification. The more specific features ofDCIS such as a ductal distribution of calci-fications, rod and branching shapes aremuch less common (Fig. 2.6). In our serieswe found a ductal distribution and rod-

2

36

Fig. 2.5A mammographic imageshowing inumerable granularmicrocalcifications in a veryobvious ductal distribution. Thedensity of the granularmicrocalcifications within theduct give it an almost snakeskin-like appearance.



2

37

Fig. 2.6Mammographic image showinga combination of granular, rodand branching calcifications inan irregular cluster distribution.The appearances arepathognomonic of high-gradeDCIS.

Fig. 2.7Mammographic image showingquite coarse rod- and Y-shapedcalcifications with only a fewgranular elements. It would be easyto dismiss these calcifications asbeing due to duct ectasia, in factthey were due to high-grade DCISwith extensive necrosis.



shaped calcifications in about 70% of DCIScases. Branching calcifications are muchless common, being seen in only 40% ofcases. The commonest benign cause ofbranching- and rod-shaped calcifications isduct ectasia and one needs to be particu-

larly careful in making a radiological diag-nosis of duct ectasia if the calcifications areunilateral and focal. It is our experiencethat DCIS presenting as a duct ectasia look-alike is invariably DCIS of high histologicalgrade (Figs 2.7 and 2.8). 2

38

Fig. 2.8Mammographic image showing agenerally coarse cluster ofcalcifications with predominantlyrod and branching calcifications.Histologically, this lesion was high-grade DCIS.

Fig. 2.9Mammographic imageshowing a predominantlypunctate cluster ofmicrocalcifications in a roundcluster shape. It would be easyto dismiss this lesion as benignbut it in fact represented low-grade DCIS.



2

39

Fig. 2.10(A) Mammographic image showing coarsecalcifications showing definite “tea cupping”on the mediolateral oblique view. The presenceof tea cups falsely reassured us that this wasfibrocystic change. (B) Taken 3 years later, stillshows “tea cupping” but there is now a veryextensive ductal branching distribution of thecalcification. This was a case of intermediate-grade DCIS with mucin secretion. Calcificationwithin the mucin secretions layered and gavethis highly unusual tea cup appearance in thiscase of DCIS.

(A)

(B)



Punctate (round or oval) calcifications arealso commonly found in DCIS. In our seriesjust under 50% of calcification clusters con-tained punctate calcifications and about 15%of DCIS calcifications clusters were made upof calcifications which were predominantlypunctate in shape (Fig. 2.9). The take-homemessage is that the presence of relativelybenign looking punctate calcifications withina cluster does not exclude DCIS3.

Changes in the morphology of DCIScalcification over time

A recent study4 looking at the previous mam-mograms of women diagnosed as havingDCIS showed that in 22% of cases theprevious mammograms were, in retrospect,abnormal. This study showed that thefollowing features were commoner on theprevious films: predominantly punctate cal-cification (64% versus 12%, P = 0.001) andfewer than 10 calcifications in a cluster (54%versus 24%, P = 0.05). Features which wereless common on the previous mammogramthan the diagnostic mammograms were rod-shaped calcifications (27% versus 64%, P =

0.03) and a ductal distribution of calcifica-tions (45% versus 76%, P = 0.05). It can beseen from these results that the calcificationmorphology of the DCIS present on theprevious mammograms are much less char-acteristic of malignancy than those present atthe time of diagnosis. It is also of interest thatthese cases, which had such non-specific fea-tures at the time of previous mammography,were predominantly cases of high gradeDCIS. This indicates that the characteristiccalcification morphological features of highgrade DCIS, i.e. the presence of rods and aductal distribution may not be present whenthe lesions are small and that these character-istic features only occur when the lesiongrows to larger size (Fig. 2.10A,B). It wouldtherefore be wrong to assume that a small cluster of calcifications containinggranular and punctate calcifications repre-sents low grade disease when it is shown tobe malignant. Features that were presentboth on the diagnostic and the previousmammograms, which may have allowedearlier diagnosis, were granular calcifica-tions, which varied in size, density and shapeand irregular cluster shape (Fig. 2.11).

2

40

Fig. 2.11Mammographic image showinga rounded cluster of granularmicrocalcifications showingslight variation in size, densityand shape. This was due to low-grade DCIS.



Growth pattern of calcificationdue to DCISBy careful assessment and measurement ofmammographic calcification due to DCISwhich was missed on previous mammo-grams and measurement of the site and sizeof calcification on the later diagnostic imagesit is possible to gain information concerningDCIS growth rates and growth directions5.We have recently found that DCIS growstwice as fast in the nipple plane (i.e. in theplane towards and away from the nipple) asin the plane at 90° to this. However, DCISappears to grow at equal rates towards andaway from the nipple. Apparent DCISgrowth at 90° to this nipple plane may just bedue to passive expansion of the breast seg-ment. There appears to be a good correlationbetween both growth in the nipple plane andat 90° to the nipple with the cytonucleargrade of DCIS. This finding supports thevalidity of current grading systems for DCIS.

The pathological site ofmammographic calcificationsrepresenting DCISPunctate calcifications are commonly foundin non-necrotic DCIS. The calcifications areintraductal and occur in intercellular spaces.These spaces are often filled with secretionsand it is these that calcify to produce themammographically visible calcifications innon-necrotic DCIS. Calcifications in the sameintercellular spaces may also be granular cal-cifications. Granular calcifications can,however, be formed in necrotic DCIS as aresult of the necrotic debris within the ductsundergoing dystrophic calcification. Thesegranular calcifications can coalesce to formrod-shaped calcifications. If these rod-shaped calcifications occur where the breastduct is branching, then branching calcifica-tions result. Paul Stomper and co-workers

suggested that mammographic calcificationmay also represent high calcium concentra-tions within necrotic cells, but presented nodirect evidence to confirm this assertion6.

Appearance of DCIS according topathological sub-type

Methods for classification are at presentunder assessment with several groupsdescribing new methods for ascribinghistological grade of DCIS. The National Co-ordinating Group for Breast ScreeningPathology in the UK recommend a systembased on nuclear grade with categories ofhigh, low and intermediate nuclear grade7.Support for this classification is provided bythe finding that sub-type of DCIS is corre-lated with the histological grade of theassociated invasive tumour; low grade DCISprogresses more often into well-differenti-ated invasive cancer and high grade DCISinto grade 3 invasive tumour.

High nuclear grade DCIS is composed of pleomorphic large cells with abundantmitoses. The growth pattern is variable andcommonly the malignant cells distend theducts with central calcified necrosis (comedoDCIS) (Fig. 2.12). High-grade DCIS with a crib-riform or micropapillary growth pattern mayalso be seen. DCIS of low nuclear grade is com-posed of uniform cells with small nuclei.Tumour cell nuclei are hyperchromatic, cen-trally positioned and have indistinct nucleoli.This sub-type of DCIS most frequently has acribriform (with geometric “punched out”spaces) or micropapillary pattern (with bul-bous projections into the duct lumen). Veryoften both architectures co-exist, although thecribriform pattern usually predominates (Fig.2.13). If the neoplastic cells are less pleomor-phic than required for the diagnosis of highgrade disease but there is a lack of uniformityof low grade DCIS then the lesion is classifiedas being of intermediate nuclear grade (Fig.2.14). One or two nucleoli may be seen and are

2

41



more evident than in low nuclear grade DCISand the nuclear-to-cytoplasmic ratio of tumourcells is often high. As with other grades of DCIS,the architectural pattern may be solid, cribri-form or micropapillary. Rarely, variation in thecytonuclear grade of a single DCIS lesion maybe seen and then all the nuclear grades presentshould be recorded but the disease classifiedaccording to the highest grade present.

Several other systems for classifying DCIShave been proposed. Some authors have advo-cated a combined assessment of nuclear gradeand necrosis, with high grade, non high-gradewith necrosis; and non-high grade withoutnecrosis being recognised8.

The radiological appearances of DCISvary markedly according to the pathologicalsub-type. A variety of molecular markers

2

42

Fig. 2.12Histological image showing apleomorphic proliferation ofintraductal epithelial cells withcentral necrosis andcalcification. The appearancesare of high grade DCIS.

Fig. 2.13Histological image showing amonotonous intraductalproliferation of epithelial cells.A cribriform architecture ispresent and there iscalcification within thesecretions.



have also shown a correlation with the radio-logical features of the disease. The followingpathological variables have been shown tocorrelate with variations in the radiologicalappearance of DCIS:

● Architectural pattern● Cell size● Necrosis● C-erbB-2 expression● p53 expression and MIB1● Oestrogen receptor and progesterone

receptor expression

Architectural patternThe traditional classification of DCIS wasbased solely on architectural pattern;Holland and co-workers were the first todescribe variations in the radiologicalappearances of DCIS according to architec-tural pattern9. They recorded DCIS cases aseither predominantly comedo or predomi-nantly cribriform/micropapillary inarchitecture and noted that the pathogenesisof the calcification in these sub-types was dif-ferent. In comedo DCIS, calcification occurs

because of dystrophic calcification within thecentral necrotic debris, as described above. Incribriform/micropapillary DCIS, however,necrosis is not generally present and it is thesecretion in the intercellular spaces whichcalcifies. The morphology of the calcifica-tions in the two architectural sub-types isalso different. Holland et al.9 found that 80%of the cases of comedo DCIS had linear calci-fication but this finding was only present in16% of the cribriform/micropapillary group.This study found that only 53% of thecribriform DCIS group had mammographiccalcification compared with 94% of thecomedo group. This indicates that calcifica-tion in low grade DCIS is variable and oftendoes not occur. Indeed, if calcification doesoccur, it only occurs within part of the lesion.This paper was also the first to highlight thatmammographic estimation of DCIS lesionsize was more accurate in comedo DCIS thatin the cribriform DCIS. Eight per cent of thecomedo and 47% of the cribriform groupshowed greater than 2 cm discrepancybetween mammographic estimation of lesionsize and histological measurements9. A sub-

2

43

Fig. 2.14Histological image showingintraductal neoplastic cells lesspleomorphic than those ofhigh-grade disease but lackingthe uniformity of low-gradeDCIS. This is a case ofintermediate grade DCIS.Calcification of secretions isagain demonstrated.



sequent paper by the same group suggestedthat by the use of magnification views lesionsize estimation in low grade DCIS was asgood as lesion size estimation in high gradeDCIS10. This suggestion was, however, basedon a small number of cases and it is difficultto see how magnification views can delineatecalcification in areas of low grade DCIS thatdo not contain histological calcifications.Other authors have confirmed that linear cal-cifications are more common in the comedosub-type of DCIS and that granular calci-fications are more common in thecribriform/micropapillary types. It is, how-ever, impossible to predict reliably thearchitectural pattern of DCIS present whenonly granular calcifications are present.Granular calcifications can be present in bothcomedo and cribriform DCIS. If there isextensive linear and branching calcification itdoes, however, indicate a strong likelihood ofthe presence of high grade DCIS containingnecrosis. Modern pathological classificationsof DCIS are based on cytonuclear grade aloneor in combination with the absence orpresence of necrosis.

Cell sizeDCIS of large cell size is more likely to dis-play abnormal mammography (94% versus72%). Calcification is in particular more com-mon in large cell DCIS (95% versus 58%),although comparison of the calcificationmorphological features of small cell versuslarge cell DCIS does not show any statisti-cally significant differences. There is,however, a non-significant trend for large cellDCIS to more commonly display a ductaldistribution, rod-shaped calcification and anassociated asymmetric density. There is anon-significant trend for small cell DCIS cal-cifications to be predominantly punctate inmorphology1. The absence of a strong corre-lation between the cell size and calcificationmorphology is because not all small cellDCIS is free from necrosis and not all large

cell DCIS contains necrosis, and necrosis isthe major determinate of calcificationmorphology in DCIS.

NecrosisNecrosis within DCIS is an indicator ofaggressive biological activity. DCIS withnecrosis shows poorer disease-free survivaland a higher local recurrence rate comparedto DCIS without necrosis. Necrosis has there-fore been included as a major determinant insome of the more modern grading systemsdescribed for DCIS. As the two mechanismsof calcification formation postulated byHolland et al. are based on the presence orabsence of necrosis, it is not surprising thatthere are strong correlations between thepresence or absence of necrosis and themammographic features of DCIS. DCIS con-taining necrosis is more likely to showabnormal mammographic findings (95% ver-sus 73%), calcification (96% versus 61%),calcification with a ductal distribution (80%versus 45%) and rod-shaped calcifications(83% versus 45%). DCIS without necrosis ismore likely to show abnormal mammo-graphic features without calcification (39%versus 4%) and predominantly punctate cal-cification (36% versus 13%). The proportionof cases with normal mammograms orabnormal mammograms without calcifica-tion appears to show a relationship with boththe presence or absence of necrosis and alsothe degree of necrosis present (Fig. 2.15). Theabsence of mammographic calcificationalmost excludes the presence of DCIS withmarked necrosis1.

C-erbB-2 oncogene expressionC-erbB-2 (Her-2, Neu) is a member of thetype 1 tyrosine kinase growth factor receptorfamily. Amplification of c-erbB-2 oncogeneand expression of its protein is found inapproximately 60% of DCIS cases11. C-erbB-2expression in DCIS has been shown to corre-late with aggressive histological features

2

44



such as comedo sub-type, large cell size andnuclear pleomorphism. There has also beenan association demonstrated between c-erbB-2 expression and cellular proliferation12.C-erbB-2 expression has been shown to cor-relate strongly with the presence of necrosis.

It is therefore not surprising that the correla-tions of c-erbB-2 expression and theradiological features of DCIS are similar tothose seen between the presence of necrosisand the radiological features of DCIS. C-erbB-2-positive DCIS more commonly

2

45

Fig. 2.15Mammographic images showing amarked branching ductal distribution ofcalcifications which are, however, of verylow density. In this unusual case, thecalcification was within mucinoussecretions in a case of intermediate-grademicropapillary DCIS.

(A)

(B)



demonstrates calcification (92% versus 72%),a ductal distribution of calcification (78% ver-sus 57%), rod-shaped calcification (82%versus 54%) and granular calcification (97%versus 86%). C-erbB-2-negative DCIS morecommonly displays abnormal mammogra-phy without calcification (28% versus 8%)13.It can be seen from the above associationsthat c-erbB-2-positive DCIS more frequentlyshows calcification with morphological fea-tures characteristic of malignancy.

p53 and MIB1p53 is a tumour supressor gene, which hasbeen described as the “guardian of thegenome”. Mutated p53 may be assessedimmunohistologically and it has been foundto be expressed in about 25% of DCIS cases.p53 expression is associated with large cellsize and necrosis. Whilst one might thereforeexpect correlation to exist between p53expression and the mammographic featuresof DCIS, we could not find any such relation-ship14. Similarly, MIB1, which is a measure ofcellular proliferation and is a nuclear proteinexpressed by cells not in the G0 (restingphase) of the cell cycle, has not been shownto have statistically significant correlationswith the mammographic features of DCIS14.This is presumably because MIB1 activitycorrelates strongly with cellular proliferationbut less strongly with necrosis and it isnecrosis that is the major determinate of radi-ological features of DCIS.

Oestrogen and progesterone receptorsUp to 60% of DCIS cases are oestrogen recep-tor-positive. Oestrogen receptor positivity isfound more frequently in DCIS with smallcell size and in DCIS which does not containnecrosis. Oestrogen receptors are of particu-lar interest due to the investigation of the roleof tamoxifen in preventing local recurrenceafter wide local excision of DCIS. The morecharacteristic calcification morphologicalfeatures of DCIS such as granular and rod-

shaped calcifications are seen more fre-quently in oestrogen receptor-negative DCISthan in oestrogen receptor-positive cases.Similarly, a ductal distribution of calcifica-tion, rod shapes and branching calcificationsare seen more frequently in progesteronereceptor-negative cases14.

Diagnosing recurrent DCISPostconservation surveillance mammogra-phy is especially important in women whohave had DCIS treated by wide local exci-sion. Mammography is important for tworeasons within this group:

1. At least 50% of women with recurrentDCIS have invasive disease recurrence.Detection before metastasis has occurredis therefore of vital importance.

2. Mammography is the sole method ofdetecting recurrent DCIS in the vastmajority of cases.

A recent study of the mammographic fea-tures of locally recurrent DCIS demonstrated85% of local recurrences were detected solelyby mammography and that 95% of recurrentDCIS was visible mammographically15. Themammographic feature of locally recurrentDCIS was calcification in 90% of cases. Thisoccurred in the same quadrant as the origi-nally diagnosed disease in 90%. This studyalso showed that the calcification morphol-ogy of recurrent DCIS was the same as that ofthe original lesion in 82% of cases. Otherstudies have previously demonstrated thatlocally recurrent invasive breast cancer alsohas a tendency to recur with the radiologicalfeatures of the original tumour. One suchstudy showed that in five out of six patientswhere the original tumour appeared asmicrocalcification, the recurrence was alsodetected due to the presence of microcalcifi-cation16. It is self-evident from these findingsthat having the diagnostic mammogramsavailable for comparison will aid the report-

2

46



ing of follow-up mammography in womenwho have had wide local excision of DCIS.

Unfortunately, calcification is a commonfinding in the irradiated breast and, in some series, as many as 25% of patients whohave had wide local excision followed byradiotherapy develop benign-appearing cal-cifications. Although with time thesecalcifications appear coarse and curvilinearin nature, when they first appear they can befaint and pleomorphic and often raise a sus-picion of recurrent carcinoma. Calcificationsthat occur soon after wide local excision withradiotherapy (within the first 18 months) areless likely to be malignant than those thatoccur at a later time interval17.

What is the value of detecting DCIS atmammographic screening?

The introduction of mammographic screen-ing has led to a dramatic increase in thenumber of cases of pure DCIS diagnosed.Twenty to 25% of screen-detected breastcancers are DCIS compared with 5% ofsymptomatic breast cancer18,19. Screeningwomen under 50 years of age is associatedwith even higher proportions of pure DCISlesions than those seen when screeningwomen over 5020.

Ductal carcinoma in situ represents a spec-trum of disease. Low grade DCIS has only a25–50% chance of developing into (low-grade) invasive cancer at 30 years and haslow rates of local recurrence when treated bywide local excision21. Conversely, high gradeDCIS, which is often associated with necro-sis, in one small series had a 75% risk ofinvasive disease with a mean time to recur-rence of only 4 years22. A more recent studyhas shown that DCIS with poorly differenti-ated cytonuclear morphology has asignificantly higher risk for the developmentof invasive carcinoma than DCIS with welldifferentiated cytonuclear appearance23.

High grade DCIS is also associated with

high grade invasive cancer. High grade inva-sive cancer carries a poor prognosis unlessdetected when less than 10 mm in size. DCISthat is of high histological grade and, thepresence of necrosis are also known to pre-dict for higher rates of recurrence aftertreatment by wide local excision24–27.Approximately 50% of recurrent lesions fol-lowing wide local excision of DCIS haveassociated invasive carcinoma and there is astrong association between the grade of theoriginal lesion and the grade of the recurrentinvasive carcinoma28–30.

Critics of breast screening often claim thatthe high rates of DCIS seen represent over-diagnosis, many being lesions which wouldnever present clinically and threaten thewoman’s life31. This is compounded by thefact that such lesions may be extensive andtherefore frequently require mastectomy toobtain adequate excision. Such criticismwould only be valid if screen-detected DCISlesions were predominantly of low histo-logical grade.

Early studies have compared the histo-logical features of screen-detected andsymptomatic DCIS but have used DCIS clas-sifications based on the previously utilisedsystem of distinguishing sub-types based onarchitectural pattern. These studies foundthat comedo DCIS was commoner in screen-detected lesions than in lesions presentingsymptomatically32,33. Since these studies werepublished, a number of new DCIS classi-fications have been proposed based oncytological features and/or the presence ofnecrosis7,24,34,35. We have found that DCISdetected by mammographic screening is pre-dominantly of high nuclear grade and only13% is of low grade. Screen-detected DCIS isalso more likely to contain areas of necrosisthan symptomatic lesions. The most likelyexplanation for these findings is suggestedby a comparison of the radiological findingsof different DCIS sub-types. High gradeDCIS more frequently shows abnormal

2

47



mammographic features than low gradeDCIS, which is often mammographicallyoccult. The mammographic calcificationfound in high grade DCIS and DCIS withnecrosis is more characteristic of malignancy,often showing rod and branch shapes. Thegranular/punctate calcifications seen in lowgrade DCIS are non-specific and are morelikely to be confused with benign processesand possibly not recalled at mammographicscreening.

Thus screen-detected DCIS is predomi-nantly of high histological grade. Thedetection of high grade DCIS by screening islikely to prevent the development of poorprognosis, high grade invasive cancer withina few years and could be important in pro-ducing part of the mortality reduction seen inrandomised trials of mammographic screen-ing. The sub-type profile of screen-detectedDCIS suggests that most lesions wouldprogress to high grade invasive diseasewithin 5–10 years. Preventing such highgrade invasive disease is likely to have a sig-nificant impact on breast cancer mortality.The small proportion of low grade, non-necrotic DCIS lesions found atmammographic screening indicates thatdiagnosis of the more indolent forms of DCISis not common.

We therefore advocate the aggressiveinvestigation of suspicious microcalcificationseen at mammographic screening. Thisshould lead to the detection of predomi-nantly high grade DCIS and also enables thediagnosis of otherwise occult co-existing,small, grade 3 invasive carcinomas, whichhave already arisen within the DCIS lesions.The increased availability of stereotactic corebiopsy with digital imaging should meanthat an aggressive approach to mammo-graphic calcification should not give rise tohigh rates of surgical benign biopsy.

Atypical ductal hyperplasia ADH is a rare condition being seen in only4% of symptomatic benign biopsies. Theincidence increases in association withscreen-detected benign microcalcifications(31%) and ADH is seen most commonly as an incidental finding in association withanother lesion which has prompted biopsy(in up to 62% of cases36). The ability of mam-mography to detect microcalcification andthe use in screening programmes has thusresulted in an increase in the detection ofADH, although some groups have alsoreported that proliferative diseases of thebreast are increasing in prevalence37. Thesignificance of the diagnosis lies in the asso-ciated increased risk of invasive breastcarcinoma which is about four to five timesthat of the general population38. The risk isfurther increased if the patient has a first-degree relative with breast cancer to 10 timesthat of the female population38.

There is wide recognition of the imperfec-tions in the criteria used to diagnose ADH. Theoriginal definition of ADH38 was that of alesion not showing all the features of DCIS.This has been updated and, although the diag-nosis stills rests on an absence of all the fea-tures of DCIS, additional supporting criteriahave been described7. Page’s view that ADHshould be recognised if the cellular changes ofDCIS are present but occupy less than twoseparate duct spaces is widely accepted in theUK38, although other pathologists recommendthat the overall size of the lesion is more impor-tant and that a cut-off of 2 mm should beutilised39. These criteria recognise essentiallythe same lesions and there is widespreadagreement that ADH is small and focal, mea-suring less than 2–3 mm in size (Fig. 2.16). Itis clear that if such a proliferation is at the edgeof a biopsy it may represent the periphery of a more established in situ process andexcision of the adjacent tissue should beperformed. Similarly, if the appearances

2

48



equivalent to ADH are seen in a core biopsy,it is not possible for the histopathologist to dis-tinguish ADH from a more established, largerarea which would be classified as DCIS.

Thus ADH has, by definition, morpholog-ical similarities to low grade DCIS and theseentities are also alike in molecular phenotypeand DNA characteristics and are jointly dif-ferent from high-grade DCIS; 36% of ADHand 38% of cribriform DCIS show DNAaneuploidy compared to 93% of morphologi-cally high grade DCIS40. A proportion ofcases of ADH also fulfil one of the acceptedcriteria for a neoplasm, being monoclonal innature41. The atypical ductal proliferations ofthe breast epithelium are thus widelybelieved to form a spectrum of disease fromhigh grade DCIS at one end to low grade insitu disease and ADH at the other. At the lowgrade end of this spectrum, ADH and lowgrade DCIS share common histological andradiological features.

References1. Evans A, Pinder S, Wilson R et al. Ductal carcinoma

in situ of the breast: correlation between

mammographic and pathologic findings. Am JRoentgenol 1994; 162: 1307–11.

2. Ceccherini A, Evans AJ, Pinder SE, Wilson ARM,Ellis IO, Yeoman LJ. Is ipsilateral mammographyworthwhile in Paget’s disease of the breast? ClinRadiol 1996; 51: 35–8.

3. Evans AJ, Wilson ARM, Pinder SE, Ellis IO,Sibbering DM, Yeoman LJ. Ductal carcinoma in situ:imaging, pathology and treatment. Imaging 1994; 6:171–84.

4. Evans AJ, Wilson ARM, Burrell HC, Ellis IO, PinderSE. Mammographic features of ductal carcinoma insitu (DCIS) present on previous mammography.Clin Radiol 1999; 54: 644–9.

5. Thomson JZ, Evans AJ, Pinder SE, Burrell HC,Wilson ARM, Ellis IO. Growth pattern of ductalcarcinoma in situ (DCIS): a retrospective analysisbased on mammographic findings. Br J Cancer 2001;85: 225–227.

6. Stomper P, Connolly J. Ductal carcinoma in situ ofthe breast: correlation between mammographiccalcification and tumour sub-type. Am J Roentgenol1992; 159: 483–5.

7. National Coordinating Group for Breast ScreeningPathology. Pathology Reporting in Breast ScreeningPathology, 2nd ed: NHSBSP Publications no 3, 1997.

8. Silverstein MJ, Poller DN, Waisman JR et al.Prognostic classification of breast ductal carcinomain situ. Lancet 1995; 345: 1154–7.

9. Holland R, Hendriks JHCL, Vebeek ALM et al.Extent, distribution, and mammographic/histological correlations of breast ductal carcinomain situ. Lancet 1990; 335: 519–22.

2

49

Fig. 2.16Histological image of anintraductal small-cell epithelialproliferation lacking in thediagnostic features of DCIS.Calcification is noted. Theappearances are those of ADH.



10. Holland R, Hendricks J. Microcalcificationsassociated with ductal carcinoma in situ:mammographic–pathologic correlation. SeminDiagn Pathol 1994; 11: 181–92.

11. Ramachandra S, Machin L, Ashley S et al.Immunohistochemical distribution of c-erbB-2 in in situ breast carcinoma – a detailed morphologicalanalysis. J Pathol 1990; 161: 7–14.

12. Barnes DM, Meyer JS, Gonzalez JG et al.Relationship between c-erbB-2 immunoreactivityand thymidine labelling index in breast carcinomain situ. Breast Cancer Res Treat 1991; 18: 11–17.

13. Evans AJ, Pinder SE, Ellis IO et al. Correlationsbetween the mammographic features of ductalcarcinoma in situ (DCIS) and c-erbB-2 oncogeneexpression. Nottingham Breast Team. Clin Radiol1994; 49: 559–62.

14. Evans AJ. Mammographic and pathologiccorrelations. In: Silverstein, MJ, editor. DuctalCarcinoma in Situ of the Breast. Baltimore: Williamsand Wilkins, 1997, pp. 119–24.

15. Liberman L, Van Zee KJ, Dershaw DD, Morris EA,Abramson AF, Samli B. Mammographic features oflocal recurrence in women who have undergonebreast-conserving therapy for ductal carcinoma insitu. Am J Roentgenol 1997; 168: 489–93.

16. Burrell HC, Sibbering DM, Evans AJ. Domammographic features of locally recurrent breastcancer mimic those of the original tumour? Breast1996; 5: 233–6.

17. Dershaw DD, Giess CS, McCormick B et al. Patternsof mammographically detected calcifications afterbreast-conserving therapy associated with tumorrecurrence. Cancer 1997; 79: 1355–61.

18. Andersson I, Aspegren K, Janzon L et al.Mammographic screening and mortality from breastcancer: the Malmo mammographic screening trial.Br Med J 1988; 297: 943–8.

19. Smart C, Myers M, Gloecker L. Implications fromSEER data on breast cancer management. Cancer1978; 41: 787–9.

20. Evans WP, Starr AL, Bennos ES. Comparison of therelative incidence of impalpable invasive breastcarcinoma and ductal carcinoma in situ in cancersdetected in patients older and younger than 50 yearsof age. Radiology 1997; 204: 489–91.

21. Page DL, Dupont WD, Rogers LW, Jensen RA,Schuyler PA. Continued local recurrence ofcarcinoma 15–25 years after a diagnosis of low-grade ductal carcinoma in situ of the breast treatedonly by biopsy. Cancer 1995; 76: 1197–200.

22. Dean L, Geshchicter CF. Conedocarcinoma of thebreast. Arch Surg 1938; 36: 225–34.

23. Eusebi V, Feudale E, Foschini MP et al. Long-termfollow-up of in situ carcinoma of the breast. SeminDiag Pathol 1994; 11: 223–35.

24. Silverstein MJ, Poller DN, Waisman JR et al.

Prognostic classification of breast ductal carcinomain situ. Lancet 1995; 345: 1154–7.

25. Silverstein MJ, Lagios MD, Craig PH et al. Aprognostic index for ductal carcinoma in situ of thebreast. Cancer 1996; 77: 2267–74.

26. Fisher ER, Dignam J, Tan-Chiu E et al. Pathologicalfindings from the National Surgical Adjuvant BreastProject (NSABP) eight-year update of protocol B-17.Cancer 1999; 86: 429–38.

27. Solin LJ, I-Tien Y, Kurtz J. Ductal carcinoma in situ(intraductal carcinoma) of the breast treated withbreast conserving surgery and definitive irradiation.Cancer 1993; 71: 2532–42.

28. Douglas-Jones AG, Gupta SK, Attanoos RL et al. Acritical appraisal of six modern classifications ofductal carcinoma in situ of the breast (DCIS):correlation with grade of associated invasivecarcinoma. Histopathology 1996; 29: 397–409.

29. Cadman B, Ostrowski J, Quinn C. Invasive ductalcarcinoma accompanied by ductal carcinoma in situ(DCIS): comparison of DCIS grade with grade ofinvasive component. Breast 1997; 6: 132–7.

30. Lampejo OT, Barnes DM, Smith P, Millis RR.Evaluation of infiltrating ductal carcinoma with aDCIS component: correlation of the histologic typeof the in situ component with grade of theinfiltrating component. Semin Diag Pathol 1994; 11:215–22.

31. Baum M. Review of ABC of breast disease. J MedScreening 1995; 2: 233–4.

32. Evans AJ, Pinder SE, Ellis IO et al. Screeningdetected and symptomatic ductal carcinoma in situ.Mammographic features with pathologiccorrelation. Radiology 1994; 191: 237–40.

33. Bellamy CO, McDonald C, Salter DM et al. Non-invasive ductal carcinoma of the breast: therelevance of histologic categorization. Hum Pathol1993; 24: 16–23.

34. Poller DN, Silverstein MJ, Galea M et al. Ductalcarcinoma in situ of the breast. A proposal for a newsimplified histological classification. Associationbetween cellular proliferation and c-erbB-2 proteinexpression. Mod Pathol 1994; 7: 257–62.

35. Holland R, Peterse JL, Millis RR et al. Ductalcarcinoma in situ: a proposal for a newclassification. Semin Diagn Pathol 1994; 11: 167–80.

36. Stomper PC, Cholewinski SP, Penetrante RB, HarlosJP, Tsangaris TN. Atypical hyperplasia: frequencyand mammographic and pathologic relationships inexcisional biopsies guided with mammography andclinical examination. Radiology 1993; 189: 667–71.

37. Schnitt SJ, Jimi A, Kojiro M. The increasingprevalence of benign proliferative breast lesions inJapanese women. Cancer 1993; 71: 2528–31.

38. Page DL, Dupont WD, Rogers LW, Rados MS.Atypical hyperplastic lesions of the female breast. A

2

50



long-term follow-up study. Cancer 1985; 55:2698–708.

39. Tavassoli FA, Norris HJ. A comparison of the resultsof long-term follow-up for atypical intraductalhyperplasia and intraductal hyperplasia of thebreast. Cancer 1990; 65: 518–29.

40. Crissman JD, Visscher DW, Kubus J. Imagecytophotometric DNA analysis of atypical

hyperplasias and intraductal carcinomas of thebreast. Arch Pathol Lab Med 1990; 114: 1249–53.

41. Lakhani SR, Collins N, Stratton MR, Sloane JP.Atypical ductal hyperplasia of the breast: clonalproliferation with loss of heterozygosity onchromosomes 16q and 17p. J Clin Pathol 1995; 48:611–15.

2

51


Chapter

3Invasive carcinoma


Introduction 55

Extensive in situ component 61

Predicting invasion in mammographically detected microcalcification 62

Is calcification an independent prognostic feature? 65

Is spontaneous resolution of breast calcification a sign of malignancy? 66

53


Invasive carcinoma

Introduction

Calcification is a common mammographicfeature of invasive carcinoma. In particular,in a recent series from Milan, Italy, thepercentage of non-palpable invasive cancersdisplaying calcifications was shown tocorrelate strongly with age. In this series,calcification was seen mammographically in88% of the invasive cancers detected underthe age of 40 but only 22% of similar tumoursin women over the age of 701. One of thereasons for this trend was that an extensiveintraductal component was much morefrequently found in younger women; an EICwas found in 36% of women aged under 40but this dropped to 6.5% in women agedover 70 (see Table 3.1). This study alsoshowed that associated DCIS not to theextent of representing extensive intraductalcomponent, was also commoner in youngerwomen (see Table 3.2).

The mean invasive size of screen-detectedcarcinomas associated with both comedoand non-comedo suspicious calcifications inNottingham is 14 mm2. The only mammo-graphic sign associated with smaller invasivecancers is architectural distortion. This maybe difficult to identify and therefore therecognition of any associated calcification isa useful mammographic feature for thedetection of small invasive cancers.

Calcification also has strong correlationswith the histological grade of invasive breastcancers. Comedo calcification, in particular,

is strongly associated with grade 3 invasivecancer (Figs 3.1 and 3.2). Eighteen and a halfper cent of grade 3 invasive cancers displaycomedo calcification mammographicallycompared with less than 3% for grade 2 andgrade 1 invasive cancers. Similar, but lessstrong, associations are seen betweengranular microcalcifications (non-comedosuspicious calcification) and the grade ofinvasive cancer. In total, 40% of grade 3invasive cancers show some form of mam-mographic microcalcification, whereas only20% or fewer cases of grade 2 and grade 1invasive cancers display mammographicmicrocalcification2 (see Table 3.3). Otherstudies have confirmed that correlationbetween invasive tumour grade and calcifi-cation1. We have, however, been unable toshow any association between the presenceof mammographic calcification and lymphnode stage or vascular invasion status2.

Why is calcification associatedwith grade 3 invasive cancers?High-grade DCIS tends to give rise to high-grade invasive cancer, and low-grade DCISto low-grade invasive cancer3–5. This associ-ation between grade of invasive cancer andDCIS grade is present whatever gradingsystem is used. On average, 67% of inva-sive cancers associated with high-gradeDCIS using five different grading systemswere histologically grade 3. Using the Van

55

Age (years) No. of cases Cancers with microcalcifications (%)

> 40 33 87.9

40–49 167 68.3

50–59 283 54.4

60–69 156 40.4

≥ 70 46 21.7

Table 3.1 Correlations between age and mammographic calcification in invasive breast cancer1

3



Nuys system as many as 75% of invasivecancers associated with high-grade DCISare grade 3.

The vast majority (> 90%) of high-gradeDCIS cases display mammographic calcifica-tion, which is often of comedo morphology.Low-grade DCIS only display mammo-

graphic calcification in 50–60% of cases6,7. Asfew cases of low-grade DCIS are necrotic, themajority of the calcifications are granular orpunctate. These findings explain the associa-tion between calcification, particularlycalcification of comedo type and the grade ofassociated invasive cancer.56

3

Fig. 3.1Mammographic image showingan ill-defined mass withassociated suspiciousmicrocalcification. Histologically,this represented a grade 3invasive carcinoma.

<40 years 40–49 years 50–59 years 60–69 years ≥ 70 years

Ductal + DCIS (%) but not EIC 18.2 22.2 14.5 12.8 8.7

Ductal + EIC (%) 36.4 17.4 15.5 12.2 6.5

Table 3.2 Distribution of invasive tumour histology (%) according to patient age1

Tumour grade

Mammographic feature Grade 1 Grade 2 Grade 3n (%) n (%) n (%)

Comedo calcification 2 (2.6) 2 (2.4) 5 (18.5)

Non-comedo suspicious calcification 7 (9.0) 10 (12.1) 5 (18.5)

Benign calcifications 13 (16.7) 17 (20.5) 11 (40.7)

Table 3.3 Correlations between mammographic calcification and the grade of prevalent round screen-detectedinvasive breast carcinoma2


Invasive carcinoma

The grade of DCIS associated with inva-sive cancers has been shown to correlate withboth disease-free interval and survival. Thesedata indicate the biological importance ofDCIS and that the grade of associated DCIShas biological significance. The strong associ-ations that exist between the grade ofinvasive cancers and the grade of DCIS fromwhich they arose may, at least in part, explainthe above associations.

The importance of DCIS inenabling the detection of smallgrade 3 invasive tumoursData from the Swedish two-counties studyhas shown that grade 3 invasive cancers lessthan 10 mm in size have an excellent progno-sis8. This compares markedly with the poorprognosis of large grade 3 invasive cancers.In a recent study looking at screen-detected

3

57

Fig. 3.2(A) Calcification in the invasivecomponent of a grade 2/3carcinoma. (B) Calcification inhigh-grade DCIS in the centreof a grade 3 invasivecarcinoma.

A

B



grade 3 invasive cancers, we found that 54%had DCIS surrounding the invasive tumours(Fig. 3.3). Twenty-one per cent had no associ-ated DCIS and 25% had DCIS which wasconfined to within the invasive component(minimal DCIS)9.

As can be seen from Table 3.4, the presenceof surrounding DCIS in histological grade 3screen-detected invasive cancer is associatedwith smaller size of the invasive component

compared with similar invasive cancers witheither no DCIS or minimal DCIS. It can alsobe seen that there is a non-significant trendto node negativity in histological grade 3invasive cancers with surrounding DCIScompared with the other two sub-groups9.

Table 3.5 shows the correlations betweenthe mammographic appearances of screen-detected histological grade 3 invasive can-cer and histological size and nodal status.

3

58

Fig. 3.3(A) Histological image showing asmall grade 3 invasive carcinoma(B) associated with surroundinghigh-grade DCIS.

A

B


Invasive carcin

om

a

359

Nodal stage

DCIS status No. of grade 3 Size range Median size Mean size No. of grade 3 Stage 1 Stage 2 (%)tumours (%) (mm) (mm) (mm) tumours < 10 mm (%) or 3

No DCIS 13 13–35 20 20 0 7 5 (42)a

Minimal DCIS 15 (25) 10–30 21 20 2 (13) 8 6 (43)

Surrounding DCIS 33 (54) 1.5–33 14 15 10 (30) 23 7 (28)

aNumbers in parentheses are percentages

Table 3.4 Histological size and nodal status of grade 3 tumours by associated DCIS9


Breast calcifi

cation

360

Nodal stage

Mammographic No. of grade 3 Size range Median size Mean size No. of grade 3 Stage 1 Stage 2 Stage 3 Stage 2 or 3appearance tumours (%) (mm) (mm) (mm) tumours < 10 mm (%)

Granular/punctate 11 (18) 6–28 19 16 2 (18) 6 5 2 5 (43)calcification

Comedo calcification 14 (23) 1.5–33 12 15 6 (43) 10 3 1 4 (29)

All suspicious 25 (42) 1.5–33 19 15 8 (32) 16 8 1 9 (36)calcification

Calcification nomass 6 (10) 1.5–28 13 13 2 (33) 5 1 0 1 (17)

Calcification 19 (32) 6–33 19 19 5 (26) 11 7 1 8 (42)with mass

Mass without 35 (58) 5–35 18 18 4 (11) 22 7 4 11 (33)

calcification

Table 3.5 Histological size and nodal status of grade 3 tumours according to mammographic appearance9


Invasive carcinoma

It can be seen from Table 3.5 that calcifictumours were more likely to be less than orequal to 10 mm in size than non-calcifictumours (32% versus 11%, P < 0.05) (Figs3.4 and 3.5). This is predominantly becauseof the high frequency of tumours less thanor equal to 10 mm in size in the comedocalcification group. This group representsonly 23% of all the tumours but contained50% of all small tumours. Forty-three percent of tumours showing comedo calcifica-tion were less than or equal to 10 mm insize compared with only 13% of thosetumours without this feature9. This studyindicates that surrounding DCIS is com-

mon in histological grade 3 screen-detectedcancers and this enables detection mammo-graphically at a smaller size than grade 3tumours without surrounding DCIS. It istherefore self-evident that detection andaggressive investigation of mammographiccalcification is an important part of anymammographic screening programme.

Extensive in situ componentThe prognostic importance of an extensivein situ component (EIC) is controversial.Although it has been found that invasivecancers with an EIC have fewer nodal

3

61

Fig. 3.4Mammographic image showing a small clusterof pleomorphic calcification. Histologydemonstrated high-grade DCIS with a 4-mminvasive grade 3 carcinoma.



metastases and a significantly better 10-year survival, tumours with EIC were alsoof lower histological grade10. It may well bethat the presence of an EIC is not anindependent prognostic factor when histo-logical grade and type are taken intoaccount. This question would best beresolved in a large series including multi-variate analysis. The presence of an EIC ispractically at this time of greater impor-tance in patients being considered forbreast-conserving surgery. At a meeting ofthe European Organisation for Researchinto the Treatment of Cancer (EORTC) itwas concluded that the principle risk factorfor breast relapse after breast-conservingtreatment was a large residual burden andthe main source of this burden was EIC11.The presence of EIC may not be clinically

apparent and one of the important roles ofpreoperative mammography in patientsbeing considered for breast-conservingtherapy is to demonstrate mammographi-cally the presence of an EIC (Figs 3.6 and3.7).

Predicting invasion inmammographically detectedmicrocalcification

It is important to attempt to diagnose pre-operatively invasive disease associated withDCIS. This allows the patient to undergo asingle therapeutic operation with appropri-ate staging/treatment of the axilla. Becauseaxillary metastases are extremely rare inpatients with pure DCIS, routine axillary

3

62

Fig. 3.5Mammographic image showing a coarse clusterof pleomorphic calcifications with no obviousassociated mammographic mass. Histologicalexamination showed an area of high-grade DCISwith an associated grade 3 ductal carcinoma ofno specific type.


Invasive carcinoma

lymph node dissection is inappropriate insuch patients. However, multiple studieshave shown that about 20% of patients withpure DCIS diagnosed on core biopsy in facthave an invasive focus on surgical excision.

Lagios et al., in a paper from the 1980s, sug-gested that occult invasion was negligiblein association with mammographic clustersbelow 25 mm in size but showed a 44% riskof invasion in clusters larger than 25 mm12.

3

63

Fig. 3.6An ill-defined mass representing an invasivecarcinoma with extensive associated high-grade DCIS growing towards the nipple. Theextent of the DCIS made this lesion unsuitablefor breast-conserving surgery.

Fig. 3.7Mammographic image showing anill-defined mass which representedan invasive carcinoma with obviousductal calcification extendingtowards the nipple.



This has not been our personal experiencewith screen-detected DCIS as we have founda number of invasive foci in small calcifica-tion clusters. This prompted us to look inmore detail at predicting invasive foci withincalcification clusters. In a recent study13, welooked at 116 patients presenting with malig-nant mammographic calcification, withoutan associated mammographic or palpablemass. The final diagnosis was DCIS in 78patients and DCIS plus an invasive focus in38 cases (33%). In women where core biopsymade a malignant diagnosis, the sensitivityfor invasion was 55% and the negative pre-dictive value of core biopsy for invasion was79%. We have found that invasive foci wereequally common in patients with comedo orgranular-type calcification. However, in thefive patients with punctate calcification, wedid not find any invasive foci.

Table 3.6 shows the relationship betweencalcification cluster size and risk of invasion;it can be seen that there is not a significant cor-relation between cluster size and risk of inva-sion. The mean size of pure DCIS clusters was32 mm and the mean size of DCIS clusters withan invasive focus was not significantly differ-ent (35 mm). We have, however, found a sig-

nificant trend between increasing number ofcalcific flecks and the risk of invasive disease(Table 3.7). The risk of invasion in clusterscomprising fewer than 10 flecks, 10–40 flecksand over 40 flecks of calcification was 15%,24% and 43%, respectively (P < 0.05).

There was a significant correlationbetween the DCIS grade predicted on corebiopsy and the surgical DCIS grade. We alsofound that there was a correlation betweenthe DCIS core grade and the risk of invasion.If high-grade DCIS was found on corebiopsy, there was a 37% risk of invasivedisease, whereas none of the cases withintermediate- or low-grade DCIS on corebiopsy had an invasive focus at surgicalexcision.

Combining these features was found sig-nificantly to predict for invasion; we found a48% risk of invasion in those cases with high-grade DCIS diagnosed on core biopsy andmore than 40 calcifications. Conversely, therewas only a 15% risk of invasion in clustersdiagnosed preoperatively as high-grade DCIScontaining fewer than 40 flecks (see Table 3.8).

It is interesting to speculate as to why thenumber of calcifications is a better predictorof an invasive focus than the mammographic

3

64

Cluster size (mm) DCIS (n) Invasive

< 11 22 5 (18%)

11–30 24 13 (35%)

31–60 22 10 (31%)

> 60 13 7 (35%) P = 0.49

Table 3.6 Calcium cluster size and risk of invasion13

Number DCIS (n) Invasive

1–10 11 2 (15%)

11–40 45 14 (24%)

> 40 25 19 (43%) P < 0.05

Table 3.7 Number of cluster calcifications and risk of invasion13


Invasive carcinoma

cluster size. A possible explanation is that thenumber of calcifications is a more accuratepredictor of the total amount of breast tissueinvolved by DCIS. A focus of DCIS 60 mm inmaximum extent involving just one duct space where a focus 60 × 60 mm in area willhave a much higher total volume of breasttissue involved by DCIS than a 60 mm singleduct of DCIS. There will therefore be a higherrisk of an invasive focus in the former. Thepresence of calcifications is also likely to be apredictor of high-grade disease and thereforean increased risk of an invasive focus.

We conclude that if patients have high-grade DCIS on core and more than 40 flecksof calcification on their mammogram, the riskof invasion is almost 50% (Fig. 3.8). In thisgroup, an axillary staging procedure such as

a lymph node sample or a sentinel nodebiopsy may be appropriate at the same timeas primary excision of the lesion. An alterna-tive management strategy would be to try anddiagnose the invasive focus preoperatively bymore extensive sampling of the lesion with avacuum-assisted biopsy.

Is calcification an independentprognostic feature?A recent paper by Laslo Tabar et al. has sug-gested that the presence of casting calcificationis an independent prognostic feature for inva-sive screen-detected breast cancer that are lessthan 15 mm in size14. In this paper, 14% ofwomen with 1-mm to 9-mm tumours hadcasting-type calcification on mammography.

3

65

High grade Intermediate or low grade

Number DCIS Invasive DCIS Invasive

< 40 27 4 (15%) 11 0

< 40 11 10 (48%) 5 0

Total 38 14 16 0

Table 3.8 Core DCIS grade and cluster calcification number – prediction of risk of invasion13

Fig. 3.8Mammographic image showing anextensive area of DCIS with over 40calcific flecks demonstrated. Such acase with the presence of a corebiopsy showing high-grade DCIShas an almost 50% chance of aninvasive focus. Given this fact, alymph node sample or sentinelnode biopsy may be indicated inthese circumstances.



However, this group, surprisingly, accountedfor 73% of all breast cancer deaths (P < 0.001).In this study, the majority of breast cancerdeaths also occurred in women who werelymph node-negative. This paper suggestedthat histological grade was not a reliable pre-dictor of outcome; this is also an unusual find-ing in any series of invasive breast carcinoma.At our institution we have been unable to repli-cate these findings. In a similar study, we foundthat comedo calcification was not a prognosticfactor, even before adjusting for tumour grade.Lymph node status was the most powerfulpredictor of death in our series of patients. Wedid, however, confirm that histological gradeappears to be a less important prognostic fac-tor in small screen-detected invasive cancers.It is difficult to explain the disparity of the find-ings between these two studies. It may be thathistological sampling of areas of DCIS to lookfor occult areas of invasion and sampling oflymph nodes for small metastases was lessthorough in the 1970s compared to moremodern pathological examination.

Is spontaneous resolution ofbreast calcification a sign ofmalignancy?A recent study from Guildford and Londonhas shown that spontaneously resolvingbreast microcalcification is uncommon, beingseen in 0.03% of screening mammograms15.This incidence is in keeping with previousstudies. This study showed that women withclearly benign resolving microcalcificationwere not at increased risk of developinginvasive breast cancer. However, of the22 women with indeterminate resolvingmicrocalcification, eight (36%) developedcarcinoma. One of these lesions was a radialscar with a focus of lobular carcinoma in situ;the other seven cases were invasive carci-noma predominantly of ductal type. Theauthors suggest that resolution of indetermi-

nate microcalcification should prompt fullinvestigation and close follow-up.

References1. Ferranti C, Coopmans de Yoldi G, Biganzoli E et al.

Relationship between age, mammographic featuresand pathological tumour characteristics in non-palpable breast cancer. Br J Radiol 2000; 73: 698–705.

2. De Nunzio MC, Evans AJ, Pinder SE et al.Correlations between the mammographic featuresof screen-detected invasive breast cancer andpathological prognostic factors. Breast 1996; 5: 1–4.

3. Douglas-Jones AG, Gupta SK, Attomoce RL et al. Acritical appraisal of six modern classifications ofductal carcinoma in situ of the breast (DCIS):correlation with grade of associated invasivecarcinoma. Histopathology 1996; 29: 397–409.

4. Cadman B, Ostrowski B, Quinn C. Invasive ductalcarcinoma accompanied by ductal carcinoma in situ(DCIS): comparison of DCIS grade with grade ofinvasive component. Breast 1997; 6: 132–7.

5. Lampejo OT, Barnes DM, Smith P, Mills RR et al.Evaluation of infiltrating ductal carcinoma with aDCIS component: correlation of the histologic typeof the in situ component with grade of theinfiltrating component. Semin Diagn Pathol 1994;11: 215–22.

6. Holland R, Hendriks JH, Vebeek AL, Mravunac M,Shuurmans Stekhoven JH et al. Extent, distributionand mammographic/histological correlations ofbreast ductal carcinoma in situ. Lancet 1990; 335:519–22.

7. Evans A, Pinder SE, Wilson R et al. Ductalcarcinoma in situ of the breast: correlation betweenmammographic and pathologic findings. Am JRoentgenol 1994; 162: 1307–11.

8. Tabar L, Dufy SW, Vitak B. The natural history ofbreast carcinoma: what have we learned fromscreening? Cancer 1999; 86: 449–62.

9. Evans AJ, Pinder SE, Snead DR et al. The detectionof DCIS at mammographic screening enables thediagnosis of small, grade 3 invasive tumours. Br JCancer 1997; 75: 542–4.

10. Matsukuma A, Enjoji M, Toyoshima S. Ductalcarcinoma of the breast. An analysis of theproportion of intraductal and invasive components.Pathol Res Prac 1991; 187: 62–7.

11. Van Dongen JA, Fentiman IS, Harris JR et al. In situbreast cancer: the EORTC consensus meeting.Lancet 1989; ii: 25–7.

12. Lagios M, Westdahl PR, Margolin FR, Rose MR etal. Ductal carcinoma in situ. Relationship of extentof non-invasive disease to the frequency of occultinvasion, multicentricity, lymph node metastases

3

66


Invasive carcinoma

and short-term treatment failures. Cancer 1982; 50:1309–14.

13. Bagnall MJC, Evans AJ, Wilson ARM et al.Predicting invasion in mammographically detectedmicrocalcification. Clin Radiol 2001; 56: 828–32.

14. Tabar L, Chen HR, Duffy SW et al. A novel method

for prediction of long-term outcome of women withT1a, T1b and 10–14-mm invasive breast cancers: aprospective study. Lancet 2000; 355: 429–33.

15. Seymour HR, Cooke J, Given-Wilson RM. Thesignificance of spontaneous resolution of breastcalcification. BRJ 1999; 72: 3–8.

3

67


Chapter

4Fine-needle aspiration cytology

and core biopsy of mammographicmicrocalcification


Introduction 71

Core biopsy 72

69


Fine-needle aspiration cytology and core biopsy of mammographic microcalcification

Introduction

Fine-needle aspiration cytology (FNAC) hasbeen used to aid the diagnosis of mammo-graphic microcalcification for many years. Arecent review of the literature comparingFNAC with core biopsy has shown that theabsolute sensitivity of core biopsy is higherthan that of FNAC. This is particularly soif the procedures are performed stereo-tactically. Stereo FNAC has an absolutesensitivity of 62% compared with stereo corebiopsy at 91%. Ultrasound-guided corebiopsy is also superior to ultrasound FNACbut the difference is less marked, with theabsolute sensitivity of ultrasound FNACbeing 83% and that of ultrasound corebiopsy 97%1. This reduced sensitivity ofFNA compared with core seen in the litera-ture is also mirrored by practise in the UKbreast screening programme where FNAChas an absolute sensitivity of 54% comparedwith an absolute sensitivity of 75% for corebiopsy2.

Microcalcifications are particularly diffi-cult to biopsy compared with mass lesions.This is true both for core biopsy and forFNAC. The absolute sensitivity of FNACwhen biopsying microcalcification can behigh, for example in a particularly goodseries an absolute sensitivity of 71% wasobtained3. In general, however, the absolutesensitivity of FNAC is lower and, in particu-lar, in the diagnosis of DCIS is only in theregion of 53%4. Although the lower absolutesensitivity of FNAC in the diagnosis of DCISis of concern, the major issue when usingFNAC in the diagnosis of microcalcificationis the unreliability of FNAC to make a defin-itive diagnosis of benignity. For example, ina series from Guildford, even though theabsolute sensitivity was 71%, 36% of lesionswith C1 or C2 cytology were malignant3.Similarly, prior to 1994 when core biopsy wasintroduced in our unit, we found that 28% ofcalcification cases with a benign cytology

were malignant on surgical excision. This isdue largely to the fact that FNAC is unable toconfirm whether the sample was taken in theright place or whether there was a geograph-ical miss (sampling error). Whilst it ispossible to look for microcalcification onFNAC specimens, it is our experience thatthey are rarely seen. In addition, the resolu-tion of histology/cytology for calcification ismuch smaller than mammography and eventhe presence of calcification on a cytologyspecimen cannot confidently confirm accu-rate sampling of the lesion. This contrastswith core biopsy specimen radiography,which has been shown in many series to behighly accurate in confirming adequatesampling of the lesion.

Another factor to be taken into accountwhen deciding whether to perform FNAC orcore biopsy on an area of microcalcification ispatient comfort. A recent study from LondonUK, asked patients to rate the procedure painon a fixed-interval rating scale. This showedthat there was no difference in the painproduced by core biopsy, FNAC or cystaspiration. This study found a weak negativecorrelation between pain and the amount oflocal anaesthetic used5. Patient comfort istherefore not a reason to use FNAC ratherthan core biopsy.

Concern has been expressed regardingmalignant cell displacement at core biopsy.This may, however, be seen also withFNAC6,7. Malignant cell displacement doesoccur with both procedures, but the cellsappear to be non-viable; there is a verystrong relationship between the frequencyof malignant cell displacement and thetime between biopsy and surgery. If thebiopsy to surgery interval is less than 15days, tumour displacement is seen in 42%of cases. This frequency reduces to 15% at> 28 days8. In Steve Parker’s large multi-centre study, tumour track recurrence wasnot seen in almost 1000 cancers diagnosedby core biopsy9.

4

71



Core biopsyThe widespread introduction of automatedguns for image-guided core biopsy in themid 1990s was a significant advance in non-operative diagnosis of mammographicmicrocalcification. Stereotactically-guidedcore biopsy of indeterminate calcificationallows accurate diagnosis of the majority ofmicrocalcification clusters. The ability toperform specimen radiography to confirmthe presence of representative calcificationwithin the specimens is a significant advan-tage over FNA (Fig. 4.1). The more recentwidespread use of digital imaging has fur-ther enhanced the ability of stereotactic corebiopsy to accurately diagnose microcalcifi-cation. It has, however, also become clearthat there are a number of cases whereimage-guided core biopsy significantly

“understages” malignant microcalcification.The use of more invasive image-guided pro-cedures to address this problem will be dealtwith in a separate chapter.

Core biopsy equipmentStudies on the yields obtained with differentcore biopsy guns show that the Manan gun(Fig. 4.2) and the Bard gun (Fig. 4.3) give par-ticularly good yield of tissues10. The use ofdisposable Temno needles is associated withparticularly poor yields. Although whenusing upright stereotactic devices a 10-cmlength needle is usually adequate, in womenwith a large breast compressed thickness, it isoccasionally useful to use a 13-cm lengthneedle. Use of longer needles, such as a 16-cm needle, is often difficult using uprightstereotactic equipment because the tube headmay be unable to swing into position due tothe projection of the gun. In women with avery small breast compressed thickness,difficulties can arise when performing stereo-tactic core biopsy. To avoid hitting thetabletop, it is necessary to either use a c-armor to fire the gun from outside the breast.

The use of a short-throw gun is associatedwith very poor tissue yields and the long-throw (23-mm) gun should be used at alltimes (Figs 4.4 and 4.5). Fourteen-gaugeneedles should be used when performingstereotactic core biopsy of microcalcificationsas it has been shown that the preoperativediagnosis of malignancy is significantlypoorer if smaller gauge needles are used11.We have recently investigated the use of 12-gauge needles to diagnose mammographicmicrocalcifications. We have found that bothspecimen X-ray positivity and preoperativediagnosis of DCIS and invasive diseasewas identical when using 14- and 12-gaugeneedles12. We have therefore continued to use14-gauge needles when performing stereo-tactic core biopsy of microcalcifications.

Performing stereotactic core biopsy of

4

72

Fig. 4.1Specimen radiograph showing multiple flecks ofcalcifications.



microcalcification is difficult if only film-screen stereotaxis is available. The time lapsebetween taking a stereo pair and being able toview the image and adjust needle positionmeans that sampling errors are common. Inour experience, using film-screen uprightstereotaxis, a positive specimen radiographwas only obtained in just over 50% of cases.Indeed, the absolute sensitivities for diagnos-ing pure DCIS cases using film-screen stereo-taxis in our centre was only 34%, although theabsolute sensitivity when biopsying micro-calcification that contained an invasive focuswas slightly higher at 41%. When using film-

screen stereotaxis, only two or three checkpairs were obtained during the whole proce-dure. If more stereotactic pairs were taken, thepatient was often unable to tolerate the length-ening of the procedure. Complete sensitivitiesusing film-screen stereotaxis were also poor,being 52% for pure DCIS and 59% for DCISwith an invasive focus.

The introduction of digital stereotaxis hasenabled the use of many more check pairsduring a biopsy procedure. On average, ninecheck pairs are taken at our institution whenperforming stereotactic core biopsy. Thisallows very precise placement of the needle

4

73

Fig. 4.3The Bard gun previouslymanufactured by BIP.

Fig. 4.2The Manan long- and short-throwguns. We would recommend use ofthe long-throw gun at all times.



before firing and also shortens the intervalbetween obtaining an adequate position andfiring, thus the patient has less time to moveout of position. With the introduction of dig-ital stereotaxis our calcification retrieval rateimmediately rose from 55% to 85%. Ourabsolute sensitivity for the diagnosis ofpure DCIS rose from 34% to 69%, and thecomplete sensitivity from 52% to 94%. Thiscomplete sensitivity of 94% indicates that ageographical miss of calcifications usingdigital equipment is unusual. With the intro-

duction of digital stereotaxis, our absolutesensitivity for diagnosing clusters containingan invasive focus rose from 41% to 67%, andthe complete sensitivity from 59% to 86%13.Given further experience of the use of digitalstereotaxis, our calcification retrieval rate formicrocalcific lesions is now 96% and ourabsolute sensitivity for diagnosing pureDCIS is now 81%. These figures indicate thatthe results of upright digital stereotaxis (Fig.4.6) are similar to those achieved with pronetable stereotactic biopsies.

4

74

Fig. 4.4Photograph of a core biopsy performedusing the long-throw (23 mm) gun.

Fig. 4.5Photograph of short- and long-throw needles. We recommendthe use of the long-throwneedle at all times.



The use of the prone table (Fig. 4.7) never-theless has some advantages over the use ofupright digital stereotaxis. Patients in theprone position very rarely faint and patientmovement during the procedure is also lesscommon. When performing prone tablestereotactic biopsy the skin entry site is alsohidden from the patient. This is a particularadvantage if blood vessels are incidentallybiopsied during the procedure. Conversely,there are disadvantages to the use of theprone table, including the cost of the equip-ment, the fact that the prone table can only beused for stereotactic biopsies and the largesize of the equipment. Radiographic posi-tioning with the prone table is different fromradiographic positioning using routinemammographic equipment; positioningskills therefore take some time to acquire.Thus the advantages of upright digitalstereo-taxis are that the machine can be usedfor routine mammography when stereotacticprocedures are not being performed, radio-graphers are used to the positioning skillsrequired to perform the biopsy and that the

4

75

Fig. 4.6Photograph of a woman undergoing uprightstereotactic core biopsy.

Fig. 4.7Photograph of a womanundergoing a dedicated pronetable biopsy.



cost of a digital add-on is less than the cost ofpurchasing a dedicated prone table.

Acquiring specimen radiographs promptlyis important when performing stereotacticcore biopsies of microcalcifications14. If imme-diate specimen radiography is available itmeans that the number of cores taken can betailored to the amount of microcalcificationpresent on the specimen radiographs. The useof film-screen specimen radiographs intro-duces a significant delay between performingthe biopsy and knowing how successful thebiopsy has been. Therefore the use of digitalimaging to provide immediate specimen radi-ography is very helpful as there is no delaybetween performing the biopsy and knowingwhether the biopsy has been successful or not.

It also means that if the specimen radiographis negative, further cores can be taken with-out delay.

Performing stereotactic corebiopsyBefore image-guided biopsy is performed itis important that physical examination hasbeen conducted to assess palpability of thelesion. This information is important as itindicates whether wire localisation will berequired if the lesion is shown to be malig-nant. If a physical examination is onlyperformed after the biopsy it is often unclearwhether the lesion is palpable or just a post-procedure haematoma (Fig. 4.8A,B). It is also4

76

Fig. 4.8Mammogram before and after the freehand core biopsy showing a postprocedural haematoma with diffuse infiltrationof the fat in the upper breast.

A B



helpful for the patient to have previously metthe member of staff who will give them theirbiopsy result. The only contraindication tostereotactic core biopsy is anti-coagulation.Before the biopsy is performed it is importantto explain to the patient the nature of theirabnormality and how core biopsy will help.Whether the patient’s consent to the proce-dure is verbal or written is immaterial. Thepatient should always be aware before theprocedure that repeat biopsies do frequentlyneed to be performed and that core biopsy isnot always successful. Before the procedurebegins it is of value for the patient to hear thegun being fired as this will help them not tomove during the procedure. It is also impor-tant to discuss with the radiographer the bestplane of approach. The factors important indeciding this are ease of positioning both forthe radiographer and the patient, the needfor a centimetre or so of normal tissue deepto the lesion and which plane demonstratesthe abnormality most clearly. When usingupright stereotactic devices, lesions in theupper half of the breast can normally bebiopsied in the CC plane. Lesions in thelower medial breast can be biopsied in themediolateral oblique plane and lesions in thelower outer quadrant are usually biopsiedlateral to medial.

Once adequate check images have beenproduced by the radiographer it is importantthat the same fleck of calcification is identi-fied on both views. Failure to do this canmean quite large errors in needle positioning.Being able to check the point chosen on thestraight scout image can be quite helpful inidentifying those cases in which different cal-cifications have been picked on each checkimage. The skin is then cleansed and localanaesthetic applied to the skin and down tothe lesion, as long as the lesion is not sosubtle that it may be obscured by smallquantities of air within the local anaesthetic.In such cases, local anaesthetic should onlybe applied to the skin. The addition of adren-

aline is helpful in reducing procedure bleed-ing. A skin nick is normally performed withan 11-gauge blade. A small hook to move theskin to exactly where the needle tip is quitehelpful when performing the procedure.Obtaining optimal needle placement withthe use of multiple check images is veryworthwhile as the first few cores are often thebest. It is our normal practice to perform fivecores and then obtain a specimen radio-graph. It can be quite frustrating waiting fora specimen radiograph to go through a filmprocessor and we have found the wholeprocedure of stereotactic core biopsy to beenhanced by the use of digital imaging toprovide a specimen image.

The details of how many flecks of calcifi-cation are required on a specimen X-ray willbe addressed more fully later, but taking fiveto 10 more cores if required is appropriate. Ingeneral, patients much prefer to havetheir lesion adequately sampled on the firstoccasion than to come back for a repeat pro-cedure. It has been suggested that if corebiopsy specimens are kept in a water-basedsolution for 3 days, all the calcifications maydissolve. If samples are not going to be fixedand processed promptly, using an ethanol-based solution may prevent calcificationsfrom dissolving15.

After the procedure, compression shouldbe applied for approximately 5 minutes andthe patient is warned against physical exer-cise for the rest of the day. Patients shouldalso be advised concerning pain control andthe possible need for further compressionshould the wound ooze. Patients should notleave until there has been a clear arrange-ment made for the patient to receive theresult of their biopsy.

How many samples should be taken whenbiopsying calcification?

A number of studies have looked at the num-ber of core biopsy samples required to make

4

77



a definitive diagnosis of mammographicmicrocalcification. A study by Brenner in1996 compared the diagnostic yield of one tofive cores. Not surprisingly, they found atrend towards increasing accuracy withincreasing number of cores and that thistrend was especially marked in clusteredmicrocalcification16. A recent study fromLondon, UK also showed increasing absoluteand complete sensitivity with increasingnumber of cores; six or more cores gave abetter diagnostic yield than five cores17 (seeTable 4.1). These results highlight the fre-quent need to take multiple cores andcertainly 10 to 15 cores of microcalcificationare not excessive.

A recent study18 aimed to determine if thenumber of flecks of calcification retrievedwith stereotactic needle core, or the numbersof cores containing calcification, were relatedto biopsy sensitivity. This paper found that100% complete sensitivity was obtained oncethree individual calcific flecks were obtained,but for 100% absolute sensitivity five or moreflecks of calcification were required on speci-men radiography. This study also showedthat for 100% complete sensitivity it wasrequired that two of the cores showed at leastone fleck of calcification. For 100% absolutesensitivity, three separate cores each con-taining at least one fleck were required at

specimen radiography. The other importantfinding of this study was that three specimenX-rays which contained only one or twoflecks of calcification gave a benign result,even though the lesion was malignant onexcision. The lesson from this study is that ifon the initial specimen X-ray only one or twoflecks of calcification are obtained furthersamples should be taken. This will have thedual purpose of helping to exclude malig-nancy in benign lesions and improve theabsolute sensitivity if the lesion is malignant.A sub-analysis within this study showed thatthe amount of calcification needed to beretrieved at core biopsy was less for clusterscontaining less than 10 flecks of calcification;the retrieval of two flecks or two cores eachcontaining a fleck gave 100% absolute sensi-tivity. This finding is likely to be becausesmall calcification clusters are less likely tocontain both benign and malignant micro-calcifications18 (see Tables 4.2 and 4.3).

Results of core biopsy of microcalcification

The reported calcification retrieval rate ofstereotactic core biopsy is high, in the rangeof 86% to almost 100%. This high calcificationretrieval rate does not, however, lead to aparticularly high absolute sensitivity for the

4

78

Number of cores 2 3 4 5 6+

Microcalcifications: All (n = 125)

Absolute sensitivity 92 (73.6%) 93 (74.4%) 96 (76.8%) 98 (78.4%) 104 (83.2%)

Complete sensitivity 121 (96.8%) 97 (77.6%) 100 (80%) 105 (84%) 108 (86.4%)

Microcalcifications: Indeterminate (n = 67)


Complete sensitivity 47 (70.1%) 49 (73.1%) 51 (76.1%) 53 (79.1%) 68 (94%)

Microcalcifications: Malignant (n = 58)


Complete sensitivity 50 (86.2%) 51 (87.9%) 54 (93.1%) 55 (94.8%) 58 (100%)

Table 4.1 Relationship between number of cores and sensitivity17



diagnosis of malignancy in DCIS lesions andreported absolute sensitivities of stereotacticcore biopsy of DCIS lesions are in the order of44–67%. A non-definitive diagnosis is oftendue to partial sampling of the lesions. If alow-grade intraductal epithelial proliferationis seen in only two duct spaces, a histologicaldiagnosis of ADH is made; such lesion canonly be diagnosed as DCIS if more abnormalducts are demonstrated. Similarly, if a corebiopsy shows part of, or a single, duct spacecontaining a scanty population of more pleo-morphic epithelial cells, such as those seen inintermediate- or high-grade DCIS, this is

classified as suspicious rather than diagnos-tic of DCIS. These facts explain why there is alarge difference between the calcificationretrieval rate and absolute sensitivity forDCIS. Are there particular calcificationfeatures which lead to higher or lowersensitivity of core biopsy? Rich et al.17

demonstrated that in calcifications highlysuspicious of malignancy, the absolute sensi-tivity with six or more core specimens was ashigh as 97%. However, in the same series, forcalcifications graded as a lower risk of malig-nancy, core biopsy only achieved an absolutesensitivity of 72% with six or more core

4

79

Calcifications Core biopsy result Sensitivity (%)

(n) Normal or Uncertain Malignant Complete Absolutebenign malignant

potential or suspicious

0 21 1 3 16 12

1 2 2 7 81.8 63.6

2 1 3 6 90 60

3 1 5 100 83.3

4 2 1 100 33.3

> 5 17 100 100

Total 24 9 39

Table 4.2 Number of calcific elements on specimen radiology versus core histology18

Calcifications Core biopsy result Sensitivity (%)

(n) Normal or Uncertain Malignant Complete Absolutebenign malignant

potential or suspicious

0 21 1 3 16 12

1 3 6 12 85.7 57.1

2 2 8 100 80

3 8 100 100

4 6 100 100

> 5 2 100 100

Total 24 9 39

Table 4.3 Number of cores containing radiographic calcification versus core histology18



biopsy samples17. Bagnall et al., however,found no difference in the absolute sensitiv-ity of core biopsy when calcifications wereclassified as either granular or comedo18. It istherefore unclear as to whether the absolutesensitivity of core biopsy varies according toradiological appearance of malignant calcifi-cation.

How often is malignant calcification“understaged” by core biopsy?

It has been known for some time that a histo-logical diagnosis of ADH may be obtainedfrom a core biopsy of a lesion, which on exci-sion is classified as DCIS. Most series indicatethat approximately 50% of lesions with ADHon core show either DCIS or DCIS with inva-sive cancer at surgical excision19. ADH corebiopsy results from DCIS lesions are com-moner if the mammographic abnormality is amass rather than microcalcifications. Suchunderestimation of DCIS lesions by corebiopsy and vacuum-assisted biopsy are morecommon if fewer than 10 cores are taken20.Multiple studies have shown that approxi-mately 20% of lesions giving a core biopsyresult of DCIS have invasive disease atexcision biopsy21. Once again, such under-estimate of disease is more commonly foundif the mammographic abnormality is a massthan in microcalcific cases and underestima-tion is more common if fewer than 10 coresare taken20.

Interpreting benign results

Benign core biopsy results from calcificlesions should only be trusted if specimen X-ray shows unequivocal calcification. Abenign result from a lesion from which thespecimen X-ray only shows one or two flecksof calcification should be interpreted withcaution18. A repeat biopsy or diagnosticsurgical excision should be performed if ahistological benign core result is obtained

from a lesion that is highly suspicious ofmalignancy radiologically. It is also impor-tant to note that the presence of calcificationon histological examination is no substitutefor calcification on the specimen X-ray. Theresolution of histology for calcifications ismuch higher than radiography; thus calcifi-cations can often be found histologicallywhich are incidental and indeed are presentin radiologically non-calcific lesions. If noradiographically representative calcificationis retrieved and a benign histology result isobtained, the biopsy should either berepeated or the lesion excised. Conversely, ifa specimen X-ray shows unequivocal calcifi-cation and initial histological examinationdemonstrates only normal tissue withoutcalcification, further levels into the paraffin-embedded sample should be taken to searchfor the calcification deeper in the specimen.

References1. Britton PD. Fine needle aspiration or core biopsy.

Breast 1999; 8: 1–4.2. Britton PD, McCann J. Needle biopsy in the NHS

breast screening programme 1996/7: how much andhow accurate? Breast 1999; 8: 5–11.

3. Elliott AJ, Cooke JC, McKee G. A 4-yearretrospective analysis of screen-detected andstereotactically biopsied microcalcification withemphasis on ways to reduce the number of benignbiopsies. Breast 1996; 5: 410–14.

4. Venegas R, Rutgers JL, Cameron VL, Vargas H,Butler JA. Fine-needle aspiration cytology of breastDCIS. Acta Cytol 1994; 38: 16–143.

5. Denton ERE, Ryan S, Beaconfield T, Michell MJ.Image-guided breast biopsy: analysis of pain anddiscomfort related to technique. Breast 1999; 8:257–60.

6. Youngson BJ, Cranor M, Rosen PP. Epithelialdisplacement in surgical breast specimens followingneedling procedures. Am J Surg Pathol 1994; 18:896–903.

7. Youngson BJ, Liberman L, Rosen PP. Displacementof carcinomatous epithelium in surgical breastspecimens following stereotaxic core biopsy. Am JClin Pathol 1995; 103: 598–602.

8. Diaz LK, Wiley EL, Venta LA. Are malignant cellsdisplaced by large-gauge needle core biopsy of thebreast? Am J Roentgenol 1999; 173: 1303–13.

4

80



9. Parker SH, Burbank F, Jackman J et al. Percutaneouslarge-core breast biopsy: a multi-institutional study.Radiology 1994; 3: 359–63.

10. Krebs TL, Berg WE, Severson MJ et al. Large-corebiopsy guns: comparison for yield of breast tissue.Radiology 1996; 200: 365–8.

11. Nath ME, Robinson TM, Tobon H, Chough DM,Sumkin JH. Automated large-core needle biopsy ofsurgically removed breast lesions: comparison ofsamples obtained with 14-, 16- and 18-guageneedles. Radiology 1995; 197: 739–43.

12. Evans AJ, Whitlock JP, Burrell H, et al. Acomparison of 14- and 12-gauge needles for corebiopsy of suspicious mammographic calcification.Br J Radiol 1999; 72: 1152–4.

13. Whitlock JPL, Evans AJ, Burrell HC et al. Digitallyacquired imaging improves upright stereotactic corebiopsy of mammographic microcalcifications. ClinRadiol 2000; 55: 374–7.

14. Liberman L, Evans III WP, Dershaw DD et al.Radiography of microcalcifications in stereotaxicmammary core biopsy specimens. Radiology 1994;190: 223–5.

15. Moritz JD, Luftner-Nagel S, Westerhof JP, J.W. O,Grabbe E. Microcalcifications in breast core biopsyspecimens: disappearance at radiography afterstorage in formaldehyde. Radiology 1996; 200: 361–3.

16. Brenner RJ, Fajardo L, Fisher PR et al. Percutaneous core biopsy of the breast: effect ofoperator experience and number of samples ondiagnostic accuracy. Am J Roentgenol 1996; 166:341–6.

17. Rich PM, Michell MJ, Humphreys S, Howes GP,Nunnerley HB. Stereotactic 14-G core biopsy of non-palpable breast cancer: what is the relationshipbetween the number of core samples taken and thesensitivity for detection of malignancy? Clin Radiol1999; 54: 384–9.

18. Bagnall MJC, Evans AJ, Wilson ARM, Burrell HC,Pinder SE, Ellis IO. When have mammographiccalcifications been adequately sampled at needlecore biopsy? Clin Radiol 2000; 55: 548–53.

19. Liberman L, Cohen MA, Dershaw DD, et al.Atypical ductal hyperplasia diagnosed at stereotaxiccore biopsy of breast lesions: an indication forsurgical biopsy. Am J Roentgenol 1995; 164:1111–13.

20. Jackman RJ, Burbank F, Parker SH et al. Stereotacticbreast biopsy of non-palpable lesions: determinantsof ductal carcinoma in situ underestimation rates.Radiology 2001; 218: 497–502.

21. Liberman L, Dershaw D, Rosen PP et al. Stereotaxiccore biopsy of breast carcinoma: accuracy atpredicting invasion. Radiology 1995; 194: 379–81.

81


Chapter

5Large core biopsy for calcification

A. R. M. Wilson

Introduction 83

Very large core biopsy 83

Vacuum-assisted mammotomy 85

Ultrasound-guided mammotomy 90

X-ray stereotactic-guided mammotomy 90

Summary 91

83


Large core biopsy for calcification

IntroductionThe vast majority of microcalcifications canbe accurately and effectively sampled using14-gauge automated core biopsy but, in10–20% of cases, core biopsy fails to providea definitive diagnosis1–4. The reasons for thisinclude borderline pathological conditionsand abnormalities at sites in the breast thatare difficult to access using conventional corebiopsy1,2,5,6.

The traditional solution to these samplingproblems has been to obtain the tissuerequired for histological assessment by opensurgical biopsy after radiological localisa-tion. However, percutaneous biopsy devicesare now available that provide much largervolumes of tissue and these can be used toreduce the need for diagnostic open surgicalbiopsy for benign conditions and to pro-vide higher rates of preoperative diagnosisfor malignant disease7. Two differentapproaches have been developed. Oneapproach is to obtain a single very large coreof tissue up to 20 mm in diameter8,9. Theother approach retrieves multiple contiguous14-, 11- or 8-French gauge core samples bycombining core biopsy with a vacuumsystem for both acquiring and retrievingtissue samples (vacuum-assisted mammo-tomy, VAM). VAM is now a routineprocedure in many breast diagnostic cen-tres10,11.

Reasons for failure ofconventional core biopsyConventional core biopsy obtains separatenon-contiguous cores of tissue that are usu-ally sufficient in volume and architecturalinformation to allow for accurate pathologi-cal assessment. The most common reason forfailure to achieve accurate diagnosis withconventional automated core biopsy is aborderline pathological condition where the

pathologist requires a larger volume of tissuethan can be obtained by conventional corebiopsy to assess the true nature of the patho-logical process. These include conditionssuch as radial scar, papillary lesion, muco-cele-like lesion, differentiation of low-gradecarcinoma in situ from epithelial hyperplasiawith atypia (ADH) and the detection of invasive disease associated with in situ carcinoma12–18. All of these conditions can calcify and cause diagnostic difficulties forboth the radiologist and pathologist.

Another common reason for failure toretrieve any or sufficient representative cellular material from a cluster of micro-calcifications is difficulty in accurately target-ing the abnormality because of its small sizeor inaccessible site in the breast. Forsuccessful core biopsy, the needle must passdirectly through the tissue containing the cal-cifications at the correct depth. With bothupright and prone biopsy devices, successfulcore biopsy can prove to be difficult orimpossible in a proportion of cases becausethe cluster of calcifications is very small or isat a site difficult to access because of its posi-tion in the breast or the habitus of the patient.VAM is ideal in these circumstances as thistechnique only requires the sampling probeto be placed close to rather than through thearea to be sampled. The vacuum and abilityto sample tissue in a particular directionmeans that tissue in an otherwise inaccessi-ble site, for instance at the chest wall andimmediately behind the nipple, can be sam-pled. Using a lateral approach, VAM can alsobe used to obtain tissue from breasts that aretoo thin to sample when compressed usingthe conventional perpendicular approach.

Very large core biopsyThe technique developed by US Surgical Inc.was based on the principle of removing all ormost of the abnormality in a single core. Thissystem is called advanced breast biopsy 85

5



instrumentation (ABBI™) and involves theretrieval of a core of tissue from skin tobeyond the lesion with a choice 5-, 10- and20-mm diameter cores. The 20-mm diametercore is the most frequently used. The ABBI™system is only suitable for use with a pronebiopsy table. A variation on the very largecore principles similar to ABBI™ is known asSiteSelect™. This device differs in that it isdesigned to obtain a large core of tissue from

the area of the abnormality within the breastwithout removing the intervening of normaltissue between the skin and the lesion.

The ABBI™ and SiteSelect™ techniqueshave met with considerable criticism andboth are now not used. This is because com-pared to core biopsy and VAM, the techniqueis considerably more invasive, more costly,associated with much higher failure andcomplication rates and is not suitable to be

86

5

Fig. 5.1A&BThe Mammotome STTM driver andprobe shown for use with a pronebiopsy table.

A

B



used in a significant proportion of cases.Although ABBI was initially thought to offerthe opportunity to completely excise smalllesions, it has not proved to be an acceptablemethod for treating small malignancies19.There are very few cases where either ofthese very large core techniques offers anyadvantage over the other less invasive andless costly methods of percutaneous corebiopsy.

Vacuum-assistedmammotomyVAM devices have been developed by USSurgical Inc: Minimally invasive breast biopsy (MIBB™) and Breast Care Ethicon Endosurgery (Mammotome™). The MIBB device could only be used with aprone biopsy table; this device is no longer available. The Mammotome™ has the advantage that it can be used with both prone and upright stereotactic

5

87

Fig. 5.2A&BThe Mammotome HHTM probe usedfor hand held ultrasound guidedbiopsy.

A

B



devices (Mammotome ST™ – Fig. 5.1) andunder ultrasound guidance (MammotomeHH™ – Fig. 5.2)2,5,20–23. This device use theprinciple of a vacuum applied along a doublelumen needle to both obtain and retrievemultiple contiguous core samples withoutthe need to remove the needle from thebreast for each core specimen24. Core size canbe 14-, 11- or 8-gauge, delivering specimensof average weight of 35, 100 and 300 mgrespectively (Fig. 5.3). This compares to only17 mg for the average automated 14-gaugecore biopsy.

The principles of how the VAM techniqueworks are shown in Fig. 5.4. TheMammotome™ probe consists of three mainparts – an outer double lumen probe (1) witha lower section through which suction isapplied and an upper sampling chamber, a

hollow rotating motorised cutting trocar (2)and an inner specimen retrieval suction tro-car (3). The system is computer controlled forease of use with the sampling sequences pre-programmed by the user. Once placed in thebreast, tissue is sucked into the stationaryupper sample chamber and the motorisedhollow rotating inner cutting trocar separatesthe specimen. This is then retrieved from thesample site by withdrawing the trocar whileapplying suction through an inner secondtrocar. The biopsy probe remains in thebreast throughout the sampling process andmultiple radial contiguous core samples canbe obtained by rotating the whole probearound the biopsy site. Because there is noforward-throw action, sampling of lesionsthat are small, superficial or close to the chestwall can be easily achieved. VAM is therefore

5

88

Fig. 5.3Comparison of sample sizes with14 guage conventional core (a), 11 guage Mammotome (b) and 8 guage Mammotome (c) probes.

Fig. 5.4Schematic diagrams showing thevarious stages of the vacuum-assisted mammotomy process.

(a) (b) (c)



to be preferred where larger tissue samplesare required, when conventional core biopsyhas failed to provide a definitive diagnosis,for some very small lesions and for lesions atsites difficult to target with conventionalautomated core.

VAM is very well tolerated by patientsand, despite the needle size, many patientsprefer this procedure to automated corebiopsy, particularly for stereotactic X-ray-guided biopsy. With automated core biopsy,the noise of the gun firing and the shockwavethat this causes in the breast, and the need toremove and reinsert the device for eachsample, cause anxiety and distress despitereassurance and prior warning. Anxietycaused by the sampling technique itself issignificantly less with the mammotomytechnique; there is no loud noise when thesample is obtained and retrieved by the rela-tively slowly forward and backward rotatingcutting inner trocar and the device remains in the breast while multiple specimens areobtained.

VAM retrieves much larger volumes of tis-

sue using a rotating cutting trocar and forthis reason it is important and necessary touse more local anaesthesia than is usuallyneeded for automated core biopsy25. Localanaesthetic should be injected into the skinand deeply into the breast tissue around thetarget area. Local anaesthetic combined withadrenaline to promote vasoconstriction isrecommended for infiltration of the breasttissue around the biopsy site. The vacuumitself also appears to assist haemostasis andshould be used to aspirate any bleedingduring the sampling procedure. Volumes of10–15 ml of local anaesthetic can be safelyused. Local anaesthetic without adrenalinemay be preferred for the skin and in patientswho have contraindications to the use ofadrenaline. Longer-acting local anaestheticmay also be used around the biopsy site.However, there are no comparative studies to show that this more complicated localanaesthetic regimen provides any lesspatient morbidity than the use of a singlelocal anaesthetic preparation.

As with all biopsy techniques, the patient

5

89

Fig. 5.5The Mammotome STTM set up foruse with the GE Senovision™upright digital stereotactic systemusing the lateral approach.



should be fully informed verbally and withwritten instructions about why the proce-dure is taking place, what it involves, whatthey will experience during the biopsy, howlong it will take, what they will need to doafterwards and where they will receive theresult. Because of the cutting action of thetrocar anticoagulant, therapy is probably arelative contraindication to VAM comparedto automated core biopsy26.

In circumstances where it may be difficultat a later date to identify the site in the breastwhere the sample had been taken from, amarker should be placed in the biopsy cavitybefore the needle is removed. This can be asmall metallic clip (e.g. Micromark™ clip) orgel pellets (e.g. Gel Mark™)27. The most com-mon reason for marking the biopsy cavity iswhere it is likely that all the calcifications willhave been removed during the biopsy proce-dure and subsequent localisation for surgerymay be required. In these circumstances thegel pellets have the advantage of being easilyvisible with ultrasound, and localisation forsurgery can be performed under ultrasoundrather than X-ray guidance28,29. Displacementof calcifications at the time of biopsy has alsobeen described and for this reason it may bewise to always place some kind of marker atthe biopsy site where there is a high likeli-hood of subsequent surgical excision beingrequired30.

When the biopsy needle is removed fromthe breast, local compression should beapplied continuously directly over thebiopsy site as firmly as the patient can toler-ate for 10–15 minutes. When there is noevidence of any bleeding, the skin entry siteshould be closed with an adhesive strip orskin adhesive; a skin suture is not required.After-care is the same as for automated corebiopsy but it is advisable to also apply foldedswabs directly over the biopsy site with asecure skin dressing held in place by a tight-fitting brassiere or wrap-around bandage.The patient should keep this on for 48 hours.

If a large amount of tissue has been removed,a wrap-around pressure bandage should beused. All dressings can be removed after 48hours.

As with other biopsy procedures, patientsare advised not to undertake any vigorousexercise; particularly using the arm on theside of the breast biopsied, for at least 24hours. They should be given an instructionleaflet that explains the procedure that hasbeen performed, what to do if bleedingoccurs (apply manual pressure as applied atthe time of the biopsy), advice on the use ofanalgesia, which is not normally required,and a contact telephone number for adviceshould problems arise.

Indications for VAMMammotomy, with its ability to samplelarger volumes of breast tissue, has beenshown to be more reliable in confirming thatno frankly malignant change is present inassociation with conditions such as radialscar and ADH18,31–33. For the same reason re-biopsy rates are significantly less when VAMis used compared to conventional automatedcore biopsy32,34. However, despite the widersampling achievable by mammotomy, untilstudies show that this technique is com-pletely reliable in excluding associatedmalignancy, surgical excision is still recom-mended for definitive diagnosis in thesecircumstances19,35–39.

The scar tissue around the surgical site inthe conserved breast, particularly followingradiotherapy, can be extremely hard and dif-ficult to biopsy by conventional means whenrecurrence of malignancy is suspected. Themammotomy device with its motorisedcutting trocar can be used successfully toobtain sufficient tissue to achieve a reliablediagnosis.

For malignant lesions, where ascertainingexcision margins is fundamental to confirm-

5

90



ing adequate treatment, mammotomy mustnot be considered a therapeutic proce-dure19,38. Orientation of the piecemeal coresof tissue is difficult and it is impossible toascertain with any degree of certaintywhether adequate clearance of excision mar-gins has been obtained. However, for somebenign lesions, such as fibroadenoma, mam-motomy can be used for therapeutic excision.Ultrasound guidance is ideal for this procedure, which is significantly more cost-effective and associated with less morbiditythan surgical excision.

Indications for stereotactic VAM:

● very small cluster of microcalcificationsthat is likely to be difficult to sample withcore biopsy

● cluster of calcifications at a site difficult toaccess with core biopsy

● conventional core biopsy failed to providesufficient material for diagnosis

● indeterminate microcalcifications where itis likely that larger tissues volumes will berequired for diagnosis.

VAM will understage disease less thanhalf as often as will conventional corebiopsy12,14,16,40. The difference is particularlymarked in the understaging of ductal carci-noma in situ (DCIS). In a large review of coreand VAM carried out by Reynolds, DCIS wasfound at surgery following a biopsy result.ADH was reported in 41% of core biopsiesand only 15% of vacuum-assisted samples.Jackman and colleagues found that corebiopsy underestimated the presence of inva-sive malignancy associated with DCIS in justover 20% of cases, while this was found inonly 11% where VAM had been used forpreoperative sampling41. A similar studyreported by Rosenfield Darling found thatVAM underestimated the presence of inva-sive disease in patients with DCIS in half thenumber compared to core biopsy (10% com-pared to 21%) and understaged DCIS in 19%compared to 40%6,42,43.

VAM is ideal for sampling calcificationsassociated with papillary lesions as it allowsfor the whole lesion to be excised along witha rim of surrounding normal tissue44,45.Papillary lesions are often reported as inde-terminate by pathologists and removal of thewhole lesion and histological confirmationthat the lesion is entirely benign can beachieved without the need for surgicalbiopsy.

Indications for ultrasound-guidedMammotome™ biopsy include the follow-ing:

● mammographically detected architecturaldistortion visible on ultrasound (differen-tiation of radial scar from malignantdisease)

● focal and suspicious microcalcificationsvisible on ultrasound

● lesions too small for conventional corebiopsy

● lesions too superficial or deep in the breastfor conventional core biopsy

● previous failed conventional core biopsy● further evaluation of core or fine-needle

aspiration showing suspicious changes ofuncertain malignant potential (e.g. ADHor lobular carcinoma in situ or radial scar)

● diagnosis of recurrent disease in patientstreated by conservation surgery.

● abnormalities where wide sampling isconsidered important (e.g. mammo-graphic asymmetric density with anon-specific ultrasound correlate)

● removal of benign lesions such asfibroadenoma as an alternative to surgery

● removal of axillary lymph nodes for diag-nosis or as part of sentinel node biopsy.

The main advantages of VAM comparedto automated core biopsy are that it requiresonly a single pass of the probe into the breastto obtain multiple cores of tissue, the coresare contiguous and circumferential and thetissue volume removed is considerablylarger. VAM is also associated with signifi-

5

91



cantly less morbidity than core biopsy – mostwomen who have experienced both professto prefer mammotomy. The main compara-tive disadvantages are its increased cost (atleast 15 times more for consumables) andthat it takes significantly longer to perform.

Ultrasound-guidedmammotomyThe HH Mammotome device is light andeasy to use (Fig. 5.2)20,46,47. An initial ultra-sound scan is carried out to identify the bestdirection for access to the abnormality. Localanaesthetic is injected into the skin, the breasttissue down to the lesion and liberally infe-rior to and around the lesion itself. The localanaesthetic needle is used to identify the bestdirection and angle in which to insert theMammotome™ probe. The probe is placedthrough a 2-mm skin incision under directreal-time ultrasound vision so that the biopsynotch lies immediately behind, and notthrough, the lesion. The relationship betweenthe sampling notch and the lesion can easilybe identified on the scan by manually mov-ing the cutting trocar. Samples are obtainedby incremental rotating of the probe through

various angles up to 90° on either side of the12 o’clock horizontal position. The number ofsamples taken depends on the type and sizeof the abnormality. In a few cases there maybe difficulty in advancing the probe to therequired site through dense uncompressedbreast tissue. In these circumstances, forminga track for the probe with the local anaes-thetic injection is usually effective. Alter-natively, a radiofrequency outer sheath canbe placed for the mammotome probe to bepassed through to the biopsy site. As thepatient is lying supine, the procedure is welltolerated. Haematoma is kept to a minimumbecause the breast is not under compressionand intermittent suction is applied throughthe probe at the biopsy site.

X-ray stereotactic-guidedmammotomyThe techniques for prone table and uprightVAM are very similar to those for conven-tional automated core biopsy23,48. Specialattachments are needed for the probe guidesand all manufacturers can provide these. Thelocalisation software for the equipment mustalso be amended to allow for accurate place-

5

92

Fig. 5.6The Mammotome ST™ in use with theGE Senovision™ system.



ment of the probe. The depth of passage iscalculated as for core biopsy such that thecentre of the sampling chamber correspondsto the point target selected on the stereo-scopic images. VAM has been performedmore widely on prone table devices but it isalso easy to perform this technique onupright stereotactic devices using the BreastCare Ethicon Endosurgery MammotomeST™ device. This has been designed for bothprone and upright use (Figs 5.1 and 5.5). Inthe upright position, a lateral approach ispreferred for most procedures as the probe isinserted in the long axis of the compressedbreast (Figs 5.5 and 5.6). With this approachthe compressed thickness of the breast is nota factor – with a vertical approach the com-pressed breast cannot be less than 30 mm orthe sampling chamber will not be whollywithin the breast and the vacuum will notfunction correctly. The lateral approach alsoallows for easier access to lesions lyingsuperficially, close behind the nipple andclose to the chest wall. These are all sites thatcan be difficult to target with core biopsyeither with upright or prone biopsy tables48.

Some concern has been expressed aboutpossible long-term changes to the breaststructure shown on mammography as aresult of mammotomy but this has not beenshown to be a particular problem49. Libermanet al. reported that immediate sequelaeincluding air at the biopsy site and visiblehaematoma were common (72% and 60%,respectively) but that these changes resolvequickly leaving no long-term mammo-graphic interpretation problems50. Similarly,Lamm and colleagues retrospectivelyreviewed 744 stereotactic core biopsies andfound no abnormality at all in 96 of 225 vac-uum-assisted mammotome procedures andonly five mass lesions; in none of those witha residual abnormality did this cause any dif-ficulty with mammographic interpretation51.Similarly, displacement of viable malignantcells by the vacuum-assisted technique is not

thought to be a significant problem52,53.Significant acute complications, mainly

haematoma that requires intervention, areequally uncommon with VAM and corebiopsy occurring in approximately 0.15% ofcases54. Bruising following mammotomy iscommon for both procedures.

The contraindications to VAM are thesame as those for conventional automatedcore biopsy.

SummaryVAM is a useful adjunct to core biopsy in thequest to achieve as high as possible non-operative diagnosis of breast calcifications. Itis a flexible and accurate technique that canbe used under both ultrasound and X-rayguidance. It provides the much larger tissuespecimens needed by pathologists to makedefinitive diagnoses in borderline and otherdifficult cases and a method for the radiolo-gist to obtain material from lesions thatwould otherwise be inaccessible to percuta-neous biopsy. Despite the more accurate andreliable diagnoses achieved with VAM, theresults should always be discussed in a mul-tidisciplinary forum where the radiologicaland pathological concordance can be deter-mined and the appropriate managementdiscussed37,55. The main inhibition to themuch wider use of VAM is the cost per casecompared to conventional automated corebiopsy7.

References1. Teh WL, Evans AJ, Wilson ARM. Definitive non-

surgical breast diagnosis: the role of the radiologist.Clin Radiol 1998; 24: 11–9.

2. Parker SH, Burbank F, Jackman J et al. Percutaneouslarge-core breast biopsy: a multi-institutional study.Radiology 1994; 3: 359–63.

3. Vargas HI, Agbunag RV, Khaikhali I. State of the artof minimally invasive breast biopsy: principles andpractice. Breast Cancer 2000; 7: 370–9.

4. Russin LD. New directions in breast biopsy: review

5

93



of current minimally invasive methods andpresentation of a new coaxial technique. SeminUltrasound CT MR 2000; 21: 395–403.

5. Parker SH, Burbank F. A practical approach tominimally invasive breast biopsy. Radiology 1996;200: 11–20.

6. Darling ML, Smith DN, Lester SC et al. Atypicalductal hyperplasia and ductal carcinoma in situ asrevealed by large-core needle breast biopsy: resultsof surgical excision. Am J Roentgenol 2000; 175:1341–6.

7. Liberman L, Sama MP. Cost-effectiveness ofstereotactic 11-gauge directional vacuum-assistedbreast biopsy. Am J Roentgenol 2000; 175: 53–8.

8. Ferzli GS, Hurwitz JB, Puza T, Van Vorst-Bilotti S.Advanced breast biopsy instrumentation (ABBI): acritique. J Am Coll Surg 1997; 185: 145–51.

9. Ferzli GS, Hurwitz JB. Initial experience with breastbiopsy utilizing the advanced breast biopsyinstrumentation (ABBI). Surg Endosc 1997; 11:393–7.

10. Heywang-Kobrunner SH, Schaumloffel U, ViehwegP, Hofer H, Buchmann J, Lampe D. Minimallyinvasive stereotaxic vacuum core breast biopsy. EurRadiol 1998; 8: 377–85.

11. Liberman L, Dershaw DD, Rosen PP, Morris EA,Abramson AF, Borgen PI. Percutaneous removal ofmalignant mammographic lesions at stereotacticvacuum-assisted biopsy. Radiology 1998; 206:711–15.

12. Burbank F. Stereotactic breast biopsy of atypicalductal hyperplasia and ductal carcinoma in situlesions: improved accuracy with directional,vacuum-assisted biopsy. Radiology 1997; 202: 843–7.

13. Michell MJ, Andrews DA, Humphreys SEA. Resultsof 14-gauge biopsy of architectural distortionstellate lesions using a dedicated prone biopsysystem. Breast 1996; 5: 442.

14. Jackman RJ, Nowels KW, Shepard MJ, FinkelsteinSI, Marzoni F Jr. Stereotactic large-core needlebiopsy of 450 non-palpable breast lesions withsurgical correlation in lesions with cancer oratypical hyperplasia. Radiology 1994; 193: 91–5.

15. Liberman L, Cohen MA, Dershaw DD et al. Atypicalductal hyperplasia diagnosed at stereotaxic corebiopsy of breast lesions: an indication for surgicalbiopsy. Am J Roentgenol 1995; 164: 1111–13.

16. Jackman RJ, Burbank F, Parker SH et al. Atypicalductal hyperplasia diagnosed at stereotactic breastbiopsy: improved reliability with 14-guage,directional, vacuum-assisted biopsy. Radiology1997; 204: 485–8.

17. Reynolds HE. Core biopsy of challenging benignbreast conditions: a comprehensive literaturereview. Am J Roentgenol 2000; 174: 1245–50.

18. Brem RF, Schoonjans JM, Sanow L, Gatewood OM.Reliability of histologic diagnosis of breast cancer

with stereotactic vacuum-assisted biopsy. Am Surg2001; 67: 388–92.

19. Gajdos C, Levy M, Herman Z, Herman G, BleiweissIJ, Tartter PI. Complete removal of non-palpablebreast malignancies with a stereotactic percutaneousvacuum-assisted biopsy instrument. J Am Coll Surg1999; 189: 237–40.

20. Parker SH, Klaus AJ, McWey PJ et al.Sonographically guided directional vacuum-assisted breast biopsy using a handheld device. Am J Roentgenol 2001; 177: 405–8.

21. Parker SH, Dennis MA, Stavros AT, Johnson KK. Anew breast biopsy technique. J Diagn Med Sonogr1996; 12: 113–18.

22. Parker SH, Jobe WE, Dennis MA et al. US-guidedautomated large-core breast biopsy. Radiology 1997;187: 507–11.

23. Parker SH, Klaus AJ. Performing a breast biopsywith a directional, vacuum-assisted biopsyinstrument. Radiographics 1997; 17: 1233–52.

24. Brem RF, Schoonjans JM, Goodman SN, Nolten A,Askin FB, Gatewood OM. Non-palpable breastcancer: percutaneous diagnosis with 11-gauge and8-gauge stereotactic vacuum-assisted biopsydevices. Radiology 2001; 219: 793–6.

25. Brem RF, Schoonjans JM. Local anesthesia instereotactic, vacuum-assisted breast biopsy. Breast2001; 7: 72–3.

26. Melotti MK, Berg WA. Core needle breast biopsy inpatients undergoing anticoagulation therapy:preliminary results. Am J Roentgenol 2000; 174:245–9.

27. Liberman L, Dershaw DD, Morris EA, AbramsonAF, Thornton CM, Rosen PP. Clip placement afterstereotactic vacuum-assisted breast biopsy.Radiology 1997; 205: 417–22.

28. Burbank F, Forcier N. Tissue marking clip forstereotactic breast biopsy: initial placementaccuracy, long-term stability, and usefulness as aguide for wire localisation. Radiology 1997; 205:407–15.

29. Parker SH, Kaske TI, Gerharter JE, Dennis MA,Chavez JL. Placement accuracy andultrasonographic visualization of a newpercutaneous breast biopsy marker. Radiology 2001;221 (supplement): 431.

30. Lee SG, Piccoli CW, Hughes JS. Displacement ofmicrocalcifications during stereotactic 11-gaugedirectional vacuum-assisted biopsy with markingclip placement: case report. Radiology 2001; 219:495–7.

31. Philpotts LE, Shaheen NA, Carter D, Lange RC, LeeCH. Comparison of rebiopsy rates after stereotacticcore needle biopsy of the breast with 11-gaugevacuum suction probe versus 14-gauge needle andautomatic gun. Am J Roentgenol 1999; 172: 683–7.

32. Liberman L, Gougoutas CA, Zakowski MF et al.

5

94



Calcifications highly suggestive of malignancy:comparison of breast biopsy methods. Am JRoentgenol 2001; 177: 165–72.

33. Reynolds HE, Poon CM, Goulet RJ, Lazaridis CL.Biopsy of breast microcalcifications using an 11-gauge directional vacuum-assisted device. Am JRoentgenol 1998; 171: 611–13.

34. Liberman L, Smolkin JH, Dershaw DD, Morris EA,Abramson AF, Rosen PP. Calcification retrieval atstereotactic, 11-gauge, directional, vacuum-assistedbreast biopsy. Radiology 1998; 208: 251–60.

35. Philpotts LE, Lee CH, Horvath LJ, Lange RC, CarterD, Tocino I. Underestimation of breast cancer with11-gauge vacuum suction biopsy. Am J Roentgenol2000; 175: 1047–50.

36. Cangiarella J, Gross J, Symmans WF et al. Theincidence of positive margins with breastconserving therapy following mammotome biopsyfor microcalcification. J Surg Oncol 2000; 74: 263–6.

37. Liberman L. Clinical management issues inpercutaneous core breast biopsy. Radiol Clin NorthAm 2000; 38: 791–807.

38. Liberman L, Zakowski MF, Avery S et al. Completepercutaneous excision of infiltrating carcinoma atstereotactic breast biopsy: how can tumour size beassessed? Am J Roentgenol 1999; 173: 1315–22.

39. Won B, Reynolds HE, Lazaridis CL, P. JV.Stereotactic biopsy of ductal carcinoma in situ of thebreast using an 11-gauge vacuum-assisted device:persistent underestimation of disease. Am JRoentgenol 1999; 173: 227–9.

40. Lee CH, Carter D, Philpotts LE et al. Ductalcarcinoma in situ diagnosed with stereotactic coreneedle biopsy: can invasion be predicted? Radiology2000; 217: 466–70.

41. Jackman RJ, Burbank F, Parker SH et al. Stereotacticbreast biopsy of non-palpable lesions: determinantsof ductal carcinoma in situ underestimation rates.Radiology 2001; 218: 497–502.

42. Burak WEJ, Owens KE, Tighe MB et al. Vacuum-assisted stereotactic breast biopsy: histologicunderestimation of malignant lesions. Arch Surg2000; 135: 700–3.

43. Brem R, Berndt V, Sanow L, Gatewood D. Atypicalductal hyperplasia: histological underestimation ofcarcinoma in tissue harvested from impalpablebreast lesions using 11-guage stereotactically guideddirectional vacuum-assisted biopsy. Am JRoentgenol 1999; 172: 1405–7.

44. Guenin MA. Benign intraductal papilloma:diagnosis and removal at stereotactic vacuum-assisted directional biopsy guided bygalactography. Radiology 2001; 218: 576–9.

45. Dennis MA, Parker S, Kaske TI, Stavros AT, Camp J.Incidental treatment of nipple discharge caused bybenign intraductal papilloma through diagnosticmammotome biopsy. Am J Roentgenol 2000; 174:1263–8.

46. Wilson ARM, Teh W. Mini symposium: imaging ofthe breast. Ultrasound of the breast. Imaging 1998;9: 169–85.

47. Simon JR, Kalbhen CL, Cooper RA, Flisak ME.Accuracy and complication rates of US-guidedvacuum-assisted core breast biopsy: initial results.Radiology 2000; 215: 694–7.

48. Nisbet AP, Borthwick-Clarke A, Scott N. 11-Gaugevacuum-assisted directional biopsy of breastcalcifications, using upright stereotactic guidance.Eur J Radiol 2000; 36: 144–6.

49. Burbank F. Mammographic findings after 14-gaugeautomated needle and 14-gauge directional,vacuum-assisted stereotactic breast biopsies.Radiology 1997; 204: 153–6.

50. Liberman L, Hann LE, Dershaw DD, Morris EA,Abramson AF, Rosen PP. Mammographic findingsafter stereotactic 14-gauge vacuum biopsy.Radiology 1997; 203: 343–7.

51. Lamm RL, Jackman RJ. Mammographicabnormalities caused by percutaneous stereotacticbiopsy of histologically benign lesions evident onfollow-up mammograms. Am J Roentgenol 2000;174: 753–6.

52. Diaz LK, Wiley EL, Venta LA. Are malignant cellsdisplaced by large-gauge needle core biopsy of thebreast? Am J Roentgenol 1999; 173: 1303–13.

53. Liberman L, Vuolo M, Dershaw DD et al. Epithelialdisplacement after stereotactic 11-gauge directionalvacuum-assisted breast biopsy. Am J Roentgenol1999; 172: 677–81.

54. Lai JT, Burrowes P, MacGregor JH. Vacuum-assistedlarge-core breast biopsy: complications and theirincidence. Can Assoc Radiol J 2000; 51: 232–6.

55. Liberman L, Drotman M, Morris EA et al.Imaging–histologic discordance at percutaneousbreast biopsy. Cancer 2000; 89: 2538–46.

5

95


Chapter

6A practical approach to the reporting of percutaneous

sampling of breast calcificationsSarah E. Pinder and Ian O. Ellis

Introduction 99

Core biopsy 99

FNAC 102

97


A practical approach to the reporting of percutaneous sampling of breast calcifications

IntroductionNon-operative diagnosis is essential in breastscreening assessment to avoid surgical resec-tion of tissue for diagnosis, particularly ofbenign lesions. In the UK, up to the mid1990s, FNAC was the method of choice butthe more recent introduction of automatedcore biopsy guns has led to the increasing useof tissue biopsy. The role of either techniquefor obtaining a non-operative diagnosis inmalignancy is to provide a definitive diag-nosis with subsequent rapid referral fortreatment. Definitive non-operative diag-nosis of benign conditions is also useful,leading to prompt reassurance and dischargewith return to routine screening. It is wellrecognised that the greatest diagnosticaccuracy in the non-operative diagnosis ofbreast disease is achieved using a “tripleapproach”1,2. This utilises the results of imag-ing and clinical examination with FNACand/or core biopsy and, when all threemodalities agree, reaches a very high levelof diagnostic accuracy (> 99%)3. Similar levelof accuracy can be obtained when clinicalexamination is non-contributory for impal-pable lesions4. In the UK National HealthService Breast Screening Programme (NHSBSP), audit has identified improvedperformance for needle core biopsy5 andupdated guidelines recommend core biopsyrather than FNAC as more appropriate forthe assessment of calcific lesions6. Asdescribed in Chapter 4, the use of corebiopsy sampling has especial benefits in thediagnosis of mammographic calcific lesions.

Core biopsyHistological assessment of core biopsies can-not be performed reliably in isolation. Theclinical and mammographic findings arevital for full evaluation, including the natureof the lesion and the site of sampling. If thecore biopsy has been sampled from an area of

calcification, multidisciplinary discussionand specimen X-ray are essential.

Calcification in core biopsyAfter biopsies from mammographic micro-calcifications have been X-rayed todetermine whether calcium is present, theyshould be sent, ideally with a radiologicalcomment regarding the presence or absenceof representative microcalcification, to thehistopathology laboratory along with thecore sample. The specimen X-ray allows thepathologist to determine, not only the site ofthe calcification for which they are searching,but also the amount.

Examination of several levels (usuallythree) is performed initially, and if the calcifi-cation is not immediately apparent, furthersections can be undertaken. It may be helpfulfor the cores in which the microcalcificationis detected to be marked with a vital dye orsent to the laboratory in a separate specimenpot allowing “targeted” examination bydeeper levels etc. should calcification not bedetected in the initial series of sections. Thuscomparison with the core biopsy X-rayallows the pathologist to concentrate on theparticular core or portion of the sample thatbears the calcification. On occasion, althoughcalcification has been seen in the initiallevels, it is clear from specimen X-ray that theamount is not representative of that presentin the core and further levels may helpfullybe examined.

Once the core biopsy has been X-rayed itshould be placed immediately in fixativesolution and sent promptly to the laboratory.Optimal fixation is paramount no matterwhat the nature of the lesion and ideallybiopsies should be fixed for a minimum of6 hours. After processing, routine haema-toxylin and eosin (H&E)-stained sections areproduced in the laboratory. Calcium takesdifferent forms and has varying histological 99

6



appearances. For example, most calcifica-tions have a haematoxyphilic nature and canbe clearly seen with a deep purple colour onH&E sections. Calcium oxalate crystals, con-versely, do not take up H&E stain and areoften indistinct on routine sections but have acharacteristic birefringence and rhomboidalstructure when viewed with polarized light;further histochemical stains are therefore notin general required.

Diagnostic classification of corebiopsiesHistological examination of core biopsy pro-vides definitive diagnosis more often thanFNAC7 but it is important to recognise that itis not always possible to give an unequivocaldiagnosis in all cases, although this is possi-ble in the majority (> 90%). Most core biopsysamples can be classified as normal, benignor malignant but a small proportion of sam-ples cannot. Some difficult cases may requirefurther sectioning and, in problematic cases,histochemical and immunohistochemicalstudies may be helpful. Because of this andbecause of the requirements for fixation andprocessing, immediate reporting of core

biopsy is not possible, in contrast to FNACwhere a result may be available rapidly andwithin less than 1 hour.

Core categoriesB1 – normal

A core sample composed entirely of normaltissue is classified as B1, whether or not breastepithelial structures are present. Thus a sam-ple of normal breast ducts and lobules ormature adipose tissue or stroma alone is cat-egorised as B1. This may indicate that thelesion has not been sampled but this is not nec-essarily the case. Some benign lesions such ashamartomas and lipomas will provide normalhistological features on core biopsy.

Normal breast cores may contain micro-calcification, for example within stroma (Fig.6.1) or in involutional lobules. It is essentialthat these cases be discussed in a multidisci-plinary forum to confirm the appropriatenessof the microcalcification in the histologicalspecimen. Small foci of calcification withininvoluted lobules are common and may betoo small to be visible mammographicallyalthough they are evident microscopically(Fig. 6.2).

100

6

Fig. 6.1Histological image showingcalcification of normal breaststroma.



Microcalcification, either singly or in clus-ters less than 100 �m in diameter, is notvisible radiologically8. A histological reportthat merely notes the presence of this calcifi-cation without additional comment on itsnature, size and site is unhelpful, may bemisleading and can lead to false reassuranceand delay in diagnosis.

B2 – benign

A core biopsy sample is classified as B2 whenit contains a specific benign abnormality.Thus a range of benign lesions includingfibroadenomas, fibrocystic changes, scleros-ing adenosis and duct ectasia may containcalcification and will fall within the B2 cate-gory for core biopsy. This also extends toinclude other non-parenchymal lesions suchas abscesses and fat necrosis (Fig. 6.3).

Calcification which is deemed representa-tive of the mammographic lesionradiologically and which can be confirmed tobe in a specific benign lesion, such as ahyalinised fibroadenoma or fibroadenoma-toid hyperplasia9, can provide reassuranceand the patient can be advised that surgicalexcision is not necessary unless they wish to

have the lesion removed. A benign diagnosisfrom a calcific lesion in the absence of histo-logical calcification, however, may not beconsidered sufficient and repeat samplingshould be considered.

B3 – of uncertain malignant potential

This category is used for lesions which havebenign histological features in the corebiopsy sample, but the type of lesion identi-fied is known in a proportion of cases toshow heterogeneity with co-existing malig-nancy or to have an increased risk (albeitlow) of associated malignancy. Thusincluded under the B3 category are atypicalepithelial hyperplastic lesions where a uni-form population of cells involves one ductspace or only partially involves two or moreduct spaces. These appearances raise the pos-sibility of low grade DCIS but are insufficientin the tissue available to fulfil the diagnosticcriteria (see also Chapter 2, Intraductalepithelial lesions)10. There is a range ofdegree, from those which are insufficient fora definite diagnosis of DCIS but highly suspi-cious, to those which only show minordegrees of atypia which requires further

6

101

Fig. 6.2Histological image showingcalcification within atrophiclobules.



assessment. Appropriate categorisation intoB3 or B4 is undertaken by the pathologistbased on the degree of suspicion.

The definition of ADH relies on a combi-nation of histological and morphologicalfeatures and extent of disease and it is essen-tially an intraductal epithelial proliferationshowing the features of low grade DCIS butin less than two duct spaces or less than2 mm in diameter. For this reason, accuratediagnosis of ADH is not possible on corebiopsy. Series of subsequent surgical diag-noses in cases described as ADH innon-operative core biopsy show that in over50% the surgical excision biopsy has con-tained either in situ or invasive carcinoma11.This is not surprising as the limited tissuesampling that can be undertaken by corebiopsy guns (often by stereotactic methodsfor foci of microcalcification) often providesinsufficient material for definitive diagnosisof low grade DCIS. As described in Chapter2, a greater number of cores particularbearing increased number of calcifications12

or larger bore samples (such as theMammotome™ device, see Chapter 5) maybe helpful.

A proportion, usually of larger, radial

scars/complex sclerosing lesions are nowrecognised to harbour forms of breast carci-noma often in the form of low grade DCISor invasive tubular cancer. Papillomas,again usually the larger forms, can be het-erogeneous and have focal areas of in situcarcinoma. In addition, categorical distinc-tion between benign papillomas and papil-lary carcinomas in situ can be problematicin small tissue samples. Thus similar prin-ciples apply to lesions such as radialscars/complex sclerosing lesions and topapillomas (all of which may have associ-ated microcalcification) sampled by corebiopsy as apply to core biopsies bearingADH. Unless the lesion has been verywidely sampled by multiple core biopsies,or removed by mammotomy, currentguidelines recommend classification ofthese lesions as B3.

A wide variety of rare lesions, whichrarely contain calcification, such as phyllodestumour and spindle cell proliferation, canpresent mammographically and a confidentdiagnosis cannot be achieved on routine corebiopsy. In such circumstances we recom-mend that these are also classified as B3 andadvise diagnostic surgical excision.

6

102

Fig. 6.3A mammographic image following previous excision of abenign abnormality. Widespread punctate calcificationsare demonstrated due to fat necrosis; in addition,calcified oil cysts are seen.



B4 – suspicious

Rarely, apparently neoplastic cells are con-tained within blood clot or adherent to theouter aspect of the sample and these are clas-sified as B4 suspicious. Very small foci ofinvasive carcinoma, in which there is insuffi-cient material to allow immunocytochemicalstudies, may also on occasions be assigned tothis category.

A single complete duct space bearing anunequivocal high grade malignant epithelialproliferation is classified as B5 malignant andthe features are those of high grade DCIS.Care is, however, taken if one or only part ofa duct space is seen containing a highlyatypical epithelial process (particularly if nonecrosis is present); this may be regarded assuspicious rather than definitively malig-nant. Care is essential if the epithelial cellsshow apocrine features as this may representan atypical apocrine proliferation but notapocrine DCIS.

Another lesion, which may be derivedfrom mammographic calcification, andwhich must on occasion be classified assuspicious rather than malignant, is a non-high grade intraductal proliferation with asignificant degree of atypia which could rep-resent intermediate- or low grade DCIS.Particular care is taken when relatively fewinvolved duct spaces are represented in thebiopsy. If there is doubt in the mind of thepathologist, and the number of ducts pro-vided is sparse, they should take a pragmaticapproach and report an atypical intraductalproliferation, qualifying this according to thedegree of suspicion. On the basis of extent orseverity of atypia the core biopsy report isallocated to either the B3 or to B4 category.The management of cases classified as B3 orB4 will usually be either diagnostic excisionbiopsy of the area or repeat core or widerbore needle biopsy sampling to obtain defin-itive diagnosis. Multidisciplinary discussionis essential to ascertain what course of action

is most appropriate for each individual case.In some cases it may be felt by the patholo-gist that the limited sampling available evenwith wide-bore needle sampling, may notprovide sufficient material for reliablediagnosis and open biopsy may be moreappropriate. The UK NHSBSP Non-Operative Guidelines note that definitivetherapeutic surgery should not be under-taken as a result of a B3 or B4 core biopsydiagnosis16.

B5 – malignant

The B5 category is used for cases of unequiv-ocal malignancy on core biopsy. Furthercategorisation into in situ and invasivemalignancy is undertaken whenever possi-ble. One of the benefits of core biopsy is thatit can allow distinction between in situ andinvasive carcinoma, which is not possible onFNAC. Clearly, due to sampling error, pres-ence of DCIS alone in the core does notexclude the presence of an invasive focus inthe lesion, and in a proportion of cases sam-pled by standard methods, co-existinginvasive carcinoma will be identified in thesubsequent surgical excision specimen13,14.Conversely, invasive mammary carcinomacan be unequivocally identified in corebiopsy with a positive predictive value of98%14.

The nuclear grade, architecture and thepresence of necrosis of the DCIS can be indi-cated on the core biopsy report. The grade ofDCIS on core biopsy correlates with subse-quent grade of DCIS in the excision sampleand also provides some limited assistance inthe prediction of which cases may have asmall associated invasive focus13. Many ofthese cores will derive from screen-detectedmicrocalcifications and the presence of asso-ciated calcification should clearly be noted inthe histological report. The site of the calcifi-cation should also be recorded, whilst in themajority microcalcification will be present

6

103



within the necrotic debris of the DCIS (Fig.6.4); in some cases there may also be benigncalcification in a co-existing benign process.The mammographic features can bereassessed in view of this data and the size ofthe malignant process re-evaluated.

FNACIt is clear that FNAC is not the “gold stan-dard” diagnostic technique; false positivecytology will occur. In particular it is impor-tant to note that certain lesions classifiedhistologically as benign have malignantcytological features. These are generallyborderline hyperplastic lesions, such as atyp-ical hyperplasia. Histological diagnosis inthese cases relies on cytomorphology butalso extent and purity of the changes presentand these are clearly not demonstrable incytological preparations. Although ideally adefinitive diagnosis of malignancy or benig-nity can be made on the majority of FNACsamples, the proportion where this is possi-ble will increase with experience of both thepathologist and aspirator. If the sample ispaucicellular or if the preparation is sub-opti-mal a clear distinction may not be possible.

Calcification in FNAC FNAC from mammographically detectedfoci of calcification may be classified into anyof the five diagnostic cytological categoriesrecommended in the UK NHSBSP and it is ofvital importance that the significance ofeach is understood in the multidisciplinarycontext. It is essential that the pathologistcomments on the presence of calcificationwithin the sample. If calcification is present,the radiologist and the multidisciplinaryteam can be more certain that the lesion hasbeen sampled accurately and the likelihoodof a false negative due to an aspiration missis lower. It is important to note, however, thatthe presence of calcification in an FNAC froma calcific lesion does not discriminatebetween benign and malignant conditionsand that the background cytological featuresare paramount.

FNAC categoriesC1 – inadequate

The designation of an aspirate as “inade-quate” is somewhat subjective. It is generally6

104

Fig. 6.4Histological image showing apleomorphic proliferation ofintraductal epithelial cells withcentral necrosis andcalcification. The appearancesare of high grade DCIS.



based on the presence of sufficient numbersof epithelial cells to provide a sample ade-quate for confident assessment. There are anumber of reasons for categorising a smearas inadequate: the preparation may behypocellular, poorly spread or stained orobscured by blood. Conversely, aspiratesfrom certain lesions, such as cysts, abscesses,fat necrosis and nipple discharge specimensmay not contain epithelial cells but are notclassified as inadequate. It is extremelyunusual for a smear to contain microcalcifica-tion but insufficient epithelial cells forassessment and if this occurs, repeat sam-pling procedure should be performed,preferably core biopsy or diagnostic openbiopsy.

C2 – benign

A benign sample with no evidence of signifi-cant atypia or malignancy is categorised asbenign. The aspirate is usually poorly tomoderately cellular and consists largely ofregular epithelial cells. The background isusually composed of naked nuclei. A positivediagnosis of specific conditions, for examplefibroadenoma, fat necrosis, granulomatousmastitis, breast abscess or lymph node, maybe suggested if sufficient specific features arepresent to establish the diagnosis. Thepresence of microcalcification in a benignFNAC is supportive that the lesion has beensampled.

C3 – atypia probably benign

A C3 aspirate will have characteristics of abenign smear with, in addition, certain fea-tures not commonly seen in benign aspirates.These could be nuclear pleomorphism, someloss of cellular cohesiveness, nuclear or cyto-plasmic changes or increased cellularity. Inour unit the commonest benign lesion pro-ducing a C3 FNA is a fibroadenoma. Otherlow-grade and in situ carcinomas may also

give such a result15. Calcification may be seenin these lesions. The presence of calcificationin a C3 smear provides neither reassurancenor should it produce unease; whatever theradiological suspicion, repeat sampling ordiagnostic biopsy should be performed on aC3 result.

C4 – suspicious of malignancy

This category is used for aspirates wherethere are atypical features such that thepathologist is almost certain that they comefrom a malignant lesion. Confident diagnosiscannot be made because the specimen isscanty, poorly preserved or poorly preparedbut has features of malignancy or it mayshow some malignant features without overtmalignant cells present. Alternatively thesample may have an overall benign patternbut with occasional cells showing distinctmalignant features. As with C3 results, thecommonest malignant lesions in this cate-gory are low grade or special type lesionsand also DCIS15. Similarly, the presence ofcalcification is non-contributory to subse-quent management; further investigation isrequired. Definitive surgery should not beperformed on the basis of a C4 result16.

C5 – malignant

A malignant smear is usually cellular withcells showing discohesion, increase in cellsize and pleomorphism and thus is inter-preted as unequivocally malignant. It is wellrecognised that DCIS presenting as mammo-graphically detected microcalcification isoften high grade. FNAC of high grade DCIS,in general, if adequately sampled, bears largecharacteristically malignant cells along withnecrosis; in this situation confident diagnosiscan often be made. Low grade DCIS is moredifficult to diagnose on FNAC and may beimpossible to definitively diagnose; thesecases will not infrequently be categorised as

6

105



suspicious (C4) rather than malignant (C5)due to the paucicellular nature and the lackof large, overtly malignant cells. It shouldalso be remembered that DCIS and invasivecarcinoma cannot be distinguished bycytology alone.

The most common cause of false negativecytological diagnosis is an aspiration miss.There are, however, types of carcinomawhich, by their nature, may produce a falsenegative diagnosis. The most common ofthese are tubular and lobular type carcino-mas.

References1. Ellis IO, Galea MH, Locker A et al. Early experience

in breast cancer screening: emphasis ondevelopment of protocols for triple assessment.Breast 1993; 2: 148–53.

2. Lamb J, Anderson TJ, Dixon MJ, Levack P. Role offine-needle aspiration cytology in breast cancerscreening. J Clin Pathol 1987; 40: 705–9.

3. Zajdela A, Chossein NA, Pillerton JP. The value ofaspiration cytology in the diagnosis of breast cancer.Cancer 1975; 35: 499–506.

4. Azavedo E, Svane G, Auer G. Stereotactic fineneedle biopsy in 2594 mammographically detectednon-palpable lesions. Lancet 1989; 1 1033–5.

5. Britton PD, McCann J. Needle biopsy in the NHSbreast screening programme 1996/7: how much andhow accurate? Breast 1999; 8: 5–11.

6. Wilson R, Asbury D, Cooke J, Michell M, Patnick J,editors. Clinical Guidelines for Breast CancerScreening Assessment. Sheffield: NHS CancerScreening Programmes 2001.

7. Britton PD. Fine needle aspiration or core biopsy.Breast 1999; 8: 1–4.

8. Dahlstrom JE, Sutton S, Jain S. Histologic–radiologiccorrelation of mammographically detectedmicrocalcification in stereotactic core biopsies. Am JSurg Pathol 1998; 22: 256–9.

9. Kamal M, Evans AJ, Denley H, Pinder SE, Ellis IO.Fibroadenomatoid hyperplasia: a cause ofsuspicious microcalcification on mammographicscreening. Am J Roentgenol 1998; 171: 1331–4.

10. Coordinating Group for Breast Screening Pathology.Pathology Reporting in Breast Cancer Screening,2nd edn. NHS BSP Publications No 3, 1995.

11. Liberman L, Cohen MA, Dershaw DD et al. Atypicalductal hyperplasia diagnosed at stereotaxic corebiopsy of breast lesions: an indication for surgicalbiopsy. Am J Roentgenol 1995; 164: 1111–13.


13. Bagnall MJC, Evans AJ, Wilson ARM et al.Predicting invasion in mammographically-detectedmicrocalcification. Clin Radiol 2001; 10; 828–832.

14. Liberman L, Dershaw DD, Rosen PP et al.Stereotaxic core biopsy of breast carcinoma:accuracy at predicting invasion. Radiology 1995;194: 379–81.

15. Deb RA, Matthews P, Elston CW, Ellis IO, Pinder SE.An audit of “equivocal” (C3) and “suspicious”(C4)categories in fine-needle aspiration cytology of thebreast. Cytopathology 2001; 12; 219–26.

16. Coordinating Group for Breast Screening Pathology.Guidelines for Non-operative DiagnosticProcedures and Reporting in Breast CancerScreening. Sheffield: NHS Cancer ScreeningProgrammes, 2001.

6

106


Chapter

7A practical approach to the

radiological diagnosis of breast calcification

Andy Evans and Robin Wilson

Recall 109

Ultrasound 110

Physical examination 111

Magnetic resonance imaging 111

Scintimammography 111

Biopsy – which technique? 111

Repeat biopsies 112

107


A practical approach to the radiological diagnosis of breast calcification

RecallThe most important decision regarding thediagnosis of calcifications is whether thefeatures of the calcifications on the screen-ing mammogram warrant recall. If calcifica-tions are recalled to assessment it isunusual for the calcifications not to be cor-rectly diagnosed as benign or malignant.Factors that should be taken into accountwhen deciding whether to recall are theirmorphology, the distribution of the calcifi-cations and the cluster shape. Other impor-tant factors include period change andwhether similar calcifications are seen else-where in the same or opposite breast.Clinical history and physical findings areoccasionally important.

NumberThere is no magic number of calcific flecksabove which clusters should be recalled andbelow which clusters should not be recalled.Three granular calcifications in a ductal dis-tribution warrants recall, where four or fivepunctate calcifications in a round clusterwith similar calcific flecks scattered else-where within the ipsilateral and contralateralbreast probably do not warrant recall. Itshould always be remembered that thesmaller the cluster, the less characteristic themorphology is for DCIS.

MorphologyJudgement as to whether a calcificationcluster’s morphology warrants recall shouldbe based on the most suspicious of the mor-phological features present. The presence ofpunctate, rounded, oval calcifications withinthe cluster does not necessarily indicatebenignity if there are other calcifications withmore worrying morphological features, suchas granular, elongated rod or branching calci-

fications. Rounded calcifications with centrallucency are a reliable indicator of benignity.Variations in the size, shape and density ofthe calcifications increase the suspicion ofmalignancy although such features are found in many benign clusters. If elongatedrod-shaped calcifications are present, the dif-ferential diagnosis lies between DCIS andduct ectasia. Caution should be adoptedbefore diagnosing duct ectasia if the calcifica-tions are unilateral and especially if thecalcifications are not retro areolar. High-grade DCIS can give rise to quite coarsecalcification, which appears very similar toduct ectasia.

DistributionIf a ductal distribution is present, unlessthere are categorical features of duct ectasia,the calcifications should be recalled as theonly other common cause of a ductal distrib-ution of calcifications is DCIS.

Cluster shapeMost clusters of benign calcifications areround or oval, whereas the vast majority ofDCIS clusters have an irregular or V-shapedcluster shape. It is, however, not uncommonfor DCIS when the cluster is small to have a round or oval cluster shape and this isparticularly so in cases of low- or inter-mediate-grade DCIS. A multiple lobulardistribution of calcifications normally indi-cates fibrocystic change but occasionally amultilobular distribution of calcifications canbe found in intermediate- or low-gradeDCIS.

Period changeIncrease and decrease in the number ofcalcifications present over time is a com-mon feature of both benign and malignant 109

7



calcifications. New or increasing calcifica-tions are often benign and just becausecalcifications are new does not automati-cally mean that recall is required.Calcifications that are totally unchangedover a 3-year period are very unlikely torepresent DCIS and recall would only bewarranted if they were of particularlysuspicious morphology. It must also beappreciated, however, that decreasing cal-cifications do not necessarily imply thatthey are benign. We have had one case ofPaget’s disease of the nipple where obvi-ous comedo calcification was present onthe previous mammogram but which hadtotally disappeared by the time the patientpresented with Paget’s disease. A recentpaper has also demonstrated that a thirdof cases where indeterminate calcificationdisappear are associated with the develop-ment of invasive carcinoma.

Is calcification elsewhere?If a small cluster of calcification is found itis very important to critically review thewhole of the affected and the oppositebreast. If, on review, similar calcificationsare seen elsewhere within the affected andin the opposite breast and these calcifica-tions have a similar morphology, it isalmost certain that this represents fibro-cystic change and recall should only occurif the initial cluster identified has moreworrying morphological features than theother calcifications demonstrated.

HistoryThe threshold for recalling calcificationsshould be lower in patients whose clinicalhistory or findings raise the possibility ofDCIS. Low-risk calcifications in patients withsingle duct nipple discharge warrant furtherinvestigation.

Assessment of microcalcificationsMagnification views

High-quality magnification views shouldalways be obtained in the CC and lateralplanes. We routinely use a magnification fac-tor of 1.5. The use of magnification factorshigher than this can lead to blurring and cantherefore degrade the quality of the image.One of the most important reasons for doingmagnification views is to look for sedimenta-tion to confirm the presence of fibrocysticchange. The “tea cup” appearance should bevisible on the lateral view and the calcifica-tions should be more difficult to see and havean ill-defined rounded morphology on thecraniocaudal view. In cases of fibrocysticchange, magnification views will oftenidentify smaller but similar calcificationselsewhere within the breast. Magnificationviews in cases of DCIS will often revealsmaller additional calcifications within thecluster when compared with the standardmammographic views. The magnification ofviews are also helpful for confirming the sus-picious morphological features and a ductaldistribution in such calcifications. In cases ofmalignant calcification, the magnificationviews should also be used to assess lesionsize. Accurate assessment of lesion size isobviously important in counselling patientsas to whether they are suitable for breast-conserving surgery or not.

UltrasoundUltrasound is often useful in the furtherassessment of cases of microcalcification.This is especially true in cases where there isa dense mammographic background pattern.In this situation there will often be a mam-mographically occult, but ultrasound-visible,mass. Often this mass is smaller than the cal-cification cluster and may indicate aninvasive focus. Ultrasound-guided biopsy of110

7


A practical approach to the radiological diagnosis of breast calcification

invasive focus obviously benefits a patient asthis will alert the surgeon to the need for anaxillary staging procedure. Power Dopplerexamination of the area of microcalcificationcan sometimes show a focus of increasedvascularity and this can also be helpful inguiding ultrasound-guided biopsy of aninvasive focus. Ultrasound can also be help-ful even in the absence of a mass as thecalcifications can often be seen if high fre-quency 13 MHz probes are used1. In theabsence of a mass, it is debatable whetherultrasound-guided core or stereotactic corebiopsy should be performed. The results of ultrasound-guided core biopsy in theabsence of a mammographic mass in thisclinical setting tend to be slightly poorer thanthe results of stereotactic core if digital imag-ing is available.

Physical examinationBefore image-guided biopsy is performed, itis important that the patient has a physicalexamination. A physical examination is help-ful in identifying physical features associatedwith DCIS such as nipple discharge orPaget’s disease. It is also important to knowprior to biopsy whether the lesion is palpableand, if the lesion is shown to be malignant orsuspicious on biopsy, one needs to knowwhether the lesion requires localisation ornot. It is also helpful for the patient to meetthe clinician who will give them the result oftheir biopsy prior to the receipt of the biopsyresults.

Magnetic resonance imagingThree studies of dynamic, contrast-enhancedMRI have shown that this imaging modalityis both insensitive and non-specific in differ-entiating benign from malignant calcificationclusters2–4. It is unclear whether DCIS assess-ment with MRI is superior to high-qualitymagnification views.

ScintimammographyTwo studies have demonstrated that scinti-mammography in the clinical setting ofmammographic microcalcification is rela-tively insensitive to in situ malignancy4,5. Itsroutine use in this clinical setting is thereforenot advocated.

Biopsy – which technique?The use of FNAC to biopsy mammographicmicrocalcification is not recommended as itis not possible to confirm representativesampling. The prefered choice of techniqueis whether to use core biopsy or a vacuum-assisted device. The main advantages ofcore biopsy are low cost and speed. Thedisadvantages are a lower calcificationyield than mammotomy, difficulty in sam-pling calcifications behind the nipple andwidespread diffuse clusters. Core biopsy isalso more likely to understage DCIS andinvasive cancer, yielding ADH results incases of DCIS and DCIS results in casesthat have an invasive focus. Repeat biop-sies are more common following core thanfollowing mammotomy. The advantages ofmammotomy are the increased calcificationretrieval and less understaging of DCIS andinvasive carcinoma6. The disadvantages ofmammotomy are the cost of the dispos-ables and lengthened procedure when com-pared with core biopsy. Core biopsy is,however, able to accurately diagnose amajority of calcification clusters and, forcalcification clusters with 10 or more flecksin a tight cluster, the chances of diagnosticsuccess are high. We would advocate theuse of the mammotomy for small clusters,in cases of diffuse calcification and in calci-fication in the retro areola or inferior breast.Repeat biopsies should normally be per-formed by mammotomy. The major compli-cations for mammotomy and core biopsyare identical at 0.1%.

7

111



Repeat biopsiesRepeat stereotactic biopsies are often helpfulin patients with microcalcification7. If abenign result is obtained but no calcificationis present on the specimen X-ray, a repeatbiopsy is indicated. If a benign result isobtained and only one or two flecks of calci-fication were seen on specimen X-ray, arepeat biopsy is also likely to be required asfalse negative biopsies can occur in thesecircumstances8. A benign core biopsy and at least three flecks of calcification on thespecimen X-ray indicate the patient can bedischarged to routine follow-up unless thecalcifications demonstrated are highly suspi-cious of malignancy, in which case a repeatpercutaneous biopsy or surgical biopsyshould be performed. In patients where theresult of the needle biopsy is ADH, lobularcarcinoma in situ or suspicious of DCIS, arepeat biopsy should normally be per-formed. Such decisions should be made at amultidisciplinary meeting where the pathol-ogists and surgeon are present and thismeeting should occur before the patient hasbeen given the results of the biopsy. Werarely perform more than two percutaneousbiopsies. If two biopsies are non-diagnostic, asurgical diagnostic biopsy should be per-formed. If a calcification cluster has beenadequately sampled and a benign resultobtained, there is no need for short-term fol-low-up. It is our experience that womenplaced on short-term follow-up have ahigher incidence of anxiety and depressioncompared with women who are dischargedback to routine follow-up.

References1. Teh WL, Wilson ARM, Evans AJ, Burrell HC, Pinder

SE, Ellis IO. Ultrasound guided core biopsy ofsuspicous mammographic calcifications using highfrequency and power Doppler ultrasound. ClinRadiol 2000; 55: 390–4.

2. Westerhof JP, Fische U, Moritz JD, Oestmann JW.MR imaging of mammographically detectedclustered microcalcifications: is there any value?Radiology 1998; 207: 675–81.

3. Gilles R, Meunier M, Lcuidarme O et al. Clusteredbreast microcalcifications: evaluation by dynamiccontrast-enhanced subtraction MRI. J Comp AssistTomogr 1996; 20: 9–14.

4. Tiling R, Khalkhali I, Sommer H et al. Limited valueof scintimammography and contrast-enhanced MRIin the evaluation of microcalcification detected bymammography. Nucl Med Commun 1998; 19:55–62.

5. Vanoli C, Anronaco R, Giovanella L, Ceriani L,Sessa F, Fugazzola C. 99mTc-MIBI characterization ofbreast microcalcifications. Correlations withscintigraphic and histopathologic findings. RadiolMed 1999; 98: 19–25.

6. Jackman RJ, Burbank F, Parker SH et al. Stereotacticbreast biopsy of non-palpable lesions: determinantsof DCIS underestimation rates. Radiology 2001; 218:497–502.

7. Dershaw DD, Morris EA, Liberman L, AbramsonAF. Non-diagnostic stereotaxic core breast biopsy:results of rebiopsy. Radiology 1996; 198: 323–5.


7

112


Chapter

8Localising breast calcification

H. Burrell

Introduction 115

Hookwire localisation 115

Conclusion 122

113


Localising breast calcification

IntroductionThe majority of cases of DCIS regarded assuitable for breast-conserving surgery areimpalpable and therefore require image-guided localisation. Diagnostic surgicalbiopsy is also needed for mammographicallyindeterminate microcalcifications whereimage-guided biopsy has failed to make adefinitive diagnosis. This includes caseswhere it has not been possible to obtain arepresentative sample of microcalcificationat image-guided core biopsy or where thepathology of the tissue obtained is equivocalor suspicious for malignancy.

Some clusters of microcalcification may besuperficially located within a small breast. Inthese cases it may be sufficient to place amarker on the skin directly over the lesion.This can be done by placing a piece of leadshot on the skin directly over the lesionand securing it in position with a piece ofadhesive tape. Lateral and craniocaudalmammograms are then performed to checkthe position of the lead shot relative to thecluster of microcalcifications. If the positionis satisfactory, the site of the lead shot on theskin is marked with an “X” using indeliblemarker pen. The majority of clusters ofmicrocalcifications do not, however, liesuperficially within the breast and it is there-fore necessary to localise the cluster using amarker device placed under image guidance.

The most commonly used technique forimage-guided localisation involves the inser-tion of a hookwire into the area requiringexcision. Other techniques include radio-isotope localisation of clinically occult breastlesions; the use of carbon granules will alsobe discussed. The imaging modalities avail-able to guide insertion of the marker deviceare either ultrasound or mammography,usually using stereotaxis

Hookwire localisationThe ideal localisation device should beacceptable to the patient, easy to place, besecure in position from the time of place-ment until the time of surgery and facilitateaccurate surgical excision while allowing agood cosmetic result. A number of differenthookwires have been manufactured forlocalising clinically occult breast lesions(Fig. 8.1). A wire is inserted into the breastinside an introducing needle followinglocal anaesthetic infiltration of the skin.When the tip of the needle has been accu-rately placed within the lesion, the needleis withdrawn, leaving the tip of the wire inposition and the excess wire protrudingthrough the skin. The tip of the wire thenprovides a three-dimensionally stable guidefor the surgeon. The disadvantage of a flex-ible wire is that the surgeon has to followthe wire from the skin down to the lesionand therefore the surgical approach to thelesion is governed by the direction of wireinsertion. A flexible wire may also be inad-vertently cut at the time of surgery1,2. Acurved-end wire has the advantage that itcan be pulled back into the introducingneedle and repositioned within the breast ifnecessary and, if the localisation needle isleft in situ, this acts as a palpable, anchoredguide facilitating surgery3,4. The ability tostraighten out the curve and pull the wireback into the needle does, however,weaken the stability of the wire5.Hookwires that protrude from the side ofthe needle provide a rigid guide which canbe palpated by the surgeon and havegreater anchoring strength than the flexiblespringhook wire and the curved-endwire6–8. The Reidy breast localisation needleis flexible and has an X-shaped tip, whichis palpable and stable within the breast9.Once deployed within the breast, the Reidywire cannot be repositioned and somepathologists have had problems with its 115

8


use as it can be difficult to remove from thesurgical specimen without cutting into thetissue and interfering with specimen prepa-ration10. The main disadvantage of leavingthe introducing needle as well as the wirewithin the breast is the presence of the seg-ment protruding through the skin. As thisis not flexible, accidental trauma couldresult in tissue damage or displacement ofthe needle. This problem can be overcomeby removing the localisation needle follow-ing insertion of the wire. At the time ofsurgery the surgeon can then place a bluntcannula over the wire until the tip of thecannula reaches the tip of the wire. The tipof the cannula is palpable and indicates thesite of the lesion. The surgeon can thenchoose the optimal incision and route tothe lesion allowing removal of the lesionwith a good cosmetic result10,11.

Ultrasound guidance

The proportion of clusters of microcalcifica-tion visible on high-frequency ultrasoundvaries from 52 to 93%12–14. This depends on anumber of factors including the size of thecluster, depth of the cluster within the breast,presence of associated sonographic soft-tis-sue abnormalities and operator experience.For areas of microcalcification visible onhigh-frequency breast ultrasound, this is themethod of choice for insertion of the markerwire. When using ultrasound, the techniqueis similar to that used for ultrasound-guidedbreast biopsy. The woman is in a supineoblique position with the ipsilateral arm ele-vated behind the head. The skin is infiltratedwith local anaesthetic, and the introducingneedle containing the wire is placed throughthe area of microcalcification using ultra-116

8

Fig. 8.1Examples of various hookwiresused to localise non-palpablebreast lesions.



sound guidance. The ideal position is for thewire to pass through the cluster with the tiplying just deep to it. The wire is inserted at anangle approximately 30° to the skin to avoidthe potential complication of pneumothorax.

Stereotactic guidanceWhilst some have had success in recognisingmicrocalcifications using very high-fre-quency probes (10–15 MHz14), many clustersare difficult to detect unless there is an asso-ciated soft-tissue abnormality. Localisation ofmicrocalcification is therefore usually per-formed using stereotactic guidance. Thedirection of insertion of the localisationdevice depends on the location of the lesionwithin the breast and is chosen such that thewire traverses the minimum amount ofbreast tissue. When localising microcalcifica-tions using stereotaxis, the accuracy of theprocedure is dependent on choosing thesame part of the cluster as the target on bothstereotactic images. For lesions in the upperhalf of the breast, the wire should be insertedfrom above with the breast compressed in thecraniocaudal position. For lesions in thelower outer quadrant, the wire is insertedfrom the lateral side, and for lesions in thelower inner quadrant, the wire is insertedfrom the medial side. The skin is infiltratedwith local anaesthetic and the introducingneedle containing the wire inserted throughthe cluster of microcalcifications. As withultrasound localisation the ideal position isthe wire traversing the microcalcificationswith the tip of the wire immediately deepto the cluster. The potential difficulties ofperforming the procedure with uprightstereotaxis such as patient movement andvasovagal attacks should be reduced by theuse of digital technology as this significantlyreduces the time taken to carry out the proce-dure.

Following both stereotactic and ultra-sound localisation, the marker wire left

protruding through the skin is covered ingauze and the gauze taped to the skin. Thetip of the wire is fixed within the breast. Theprotruding wire is not taped to the skin inorder that it is free to move in and out of theskin as the shape of the breast alters accord-ing to the woman’s position. Cranio-caudaland lateral mammograms are then per-formed to check the position of the wire withrespect to the lesion. The ideal position of thewire is such that it passes through the clusterof microcalcifications with the tip lyingimmediately deep to the lesion (Fig. 8.2). Theposition is adequate provided that the tip ofthe wire is within 10 mm of the lesion. Goodcommunication between the radiologist andsurgeon is vital. It is useful for the radiologistto provide a written description of the direc-tion of insertion of the wire and its position inrelation to the cluster of microcalcifications.The check mammograms demonstrating thewire position should be made available to thesurgeon. If the tip of the wire is short ofthe lesion, the position is not adequate toguarantee successful surgical excision andthe procedure will need to be repeated. Someof the hookwires such as the Nottinghamwire can be removed by a firm tug. Otherssuch as the Reidy wire cannot be removedexcept by surgical excision under anaesthesiaand therefore, if the initial wire position isunsatisfactory, another wire should beinserted and the correct wire subsequentlyidentified to the surgeon. The wire shouldremain stable within the breast, allowingsurgery to be carried out later the same day.

Magnetic resonance imagingAlthough there have been some reports stat-ing that high-resolution magnetic resonance(MR) is able to visualise DCIS-inducedmicrocalcification, most researchers cannotconfirm this and MR has therefore no currentrole in the localisation of microcalcifica-tions15.

8

117


8

118

A

B



Outcome and complications ofhookwire localisationThe aim of the localisation procedure fornon-palpable breast lesions is the removal ofthe suspicious lesion at the first surgicaloperation. The failure of excision rates pub-lished in the literature varies between 1.5%and 10%10,16–21. Some studies have shown thatfailure of localisation is more likely withmicrocalcifications than other mammo-graphic lesions22. There are a number ofpossible reasons for failure to excise themammographic lesion. These include theaccuracy of the hookwire placement, experi-ence and skill of the individual surgeon and

migration of the wire within or out of thebreast following placement and beforesurgery22. Migration of the wire is most likelyto occur into the subcutaneous tissues withsubsequent extrusion through the skin. Wireshave, however, been reported to migratesome distance into the neck, axilla or even tothe subcutaneous tissue of the buttock23.

Other techniquesUse of dye and carbon

Preoperative marking of non-palpable breastlesions has been attempted using dyes suchas methylene blue. Unfortunately the dye

8

119

Fig. 8.2(A) Magnification view of cluster ofmicrocalcification. Stereotactic corebiopsy demonstrated malignantcalcification within an invasivecarcinoma with mucinous features.(B) Lateral and (C) craniocaudalmammograms demonstrating the tipof the hookwire within the cluster ofmicrocalcification.

C


tends to diffuse rapidly into the surroundingbreast tissue and hence there is inaccuracy ofmarking even if the surgery is carried outwithin 4 hours of injection24. Carbon suspen-sion as a marking medium was introduced in1978 and has been widely used in somecentres25. The non-palpable breast lesion ismarked with a 4% aqueous suspension ofcarbon injected through a needle from thelesion out to the skin while the needle isbeing withdrawn. A black track of carbonparticles is left within the breast and a tinyblack point of carbon acts as a tattoo markingthe site of injection in the skin. The carbontract remains inert within the breast and themarking may therefore be performed daysbefore surgery. At surgery, the surgeon dis-sects down the track to the lesion. Mullen etal. describe a 100% successful excision rate in

132 patients undergoing surgery followingmarking with carbon suspension26. In theirseries, the mammographic lesion underwentlarge-core needle biopsy using either a 14-gauge automated or 11-gauge vacuum-assisted device and, at the end of the biopsyprocedure, the needle track was marked withcarbon suspension injected through an 18-gauge needle. In the patients with apreoperative diagnosis of cancer there wasno significant difference in the size of thesurgical specimens with carbon markingcompared with conventional hookwire local-isation. No details were given regarding theadequacy of excision in these malignantcases. For the patients where localisation wasperformed for diagnostic purposes the meansize of the surgical specimen was 17 ml largerin the patients who had carbon marking

8

120

Fig. 8.3(A) Magnification view ofmicrocalcification due tointermediate grade ductalcarcinoma in situ.

A



8

121

(B),(C) Lateral and craniocaudalmammograms followingradiolabelled colloid localisation.The cluster of microcalcification isobscured by radio-opaque contrastmedium indicating accuratelocalisation.

B

C


compared with hookwire localisation. Theauthors attributed this to the fact that themajority of these cases were performed dur-ing the learning curve phase early in thestudy. No difference was seen in the postsur-gical cosmetic result. Carbon marking waswell tolerated by the patients, did not createany mammographic abnormality and there-fore did not interfere with subsequentmammographic follow-up. The main advan-tage of carbon marking is that it can be doneat the time of the image-guided large-corebiopsy and this avoids the need for a secondprocedure if the results of biopsy necessitatesurgery. Needle seeding of cancer along theneedle biopsy track has been described27. Asthe biopsy needle track is excised at the timeof surgery following carbon marking, this isa potential advantage of this procedurecompared with conventional hookwirelocalisation.

Radiolabelled colloid localisation

More recently, technetium-99m (99mTc)-labelled colloid albumin has been used forpreoperative localisation of non-palpablebreast lesions. Gennari et al. describe thetechnique in 647 patients28. In their series,3.7 MBq (0.1 mCi) of 99mTc-labelled colloidparticles of human serum albumin wereinjected into the non-palpable breast lesionusing either stereotactic or ultrasoundguidance. Frontal and lateral view planarscintigraphy images of the breast wereobtained with a gamma camera immediatelyand 5 hours after administration of the 99mTc-labelled colloid particles. Superimposingthese images on an appropriately enlargedmammogram checked the correct localisa-tion of the tracer in the lesion. Surgicalexcision was carried out within 24 hoursusing a gamma-detecting probe to identifythe area of maximum radioactivity corre-sponding to the site of the lesion. Afterexcising the specimen, the probe was used to

check for residual radioactivity at the exci-sion site and if this was present the excisionwas enlarged. Subsequent radiography ofthe specimen verified the presence and cen-tricity of the lesion in all but three patients(99.5%). In the remaining three patients, thelesion was successfully removed following awider excision of tissue from around the ‘hotspot’ of radioactivity. The mean absorbeddose of radioactivity to the surgeons’ handsfrom 100 such surgical procedures was esti-mated at 1% and 10% of the recommendedannual dose limits for the general populationand exposed workers, respectively. A com-parison of radio-guided excision with wirelocalisation of occult breast lesions showedthis new technique allowed a reduced exci-sion volume with a better cosmetic result29. Ifa small volume (0.1–0.2 ml) of radio-opaquecontrast medium is mixed with the radio-labelled colloid prior to injection, lateral andcraniocaudal mammograms done immedi-ately after the injection can be used toconfirm accurate positioning of the radio-isotope in the mammographic lesion (Fig.8.3). This obviates the need for scintigraphy30.

ConclusionIn conclusion, microcalcifications can besuccessfully localised using a variety of tech-niques. The most commonly used methodinvolves image-guided insertion of a hook-wire. The use of carbon marking at the timeof large-core needle biopsy avoids the needfor a second procedure if surgery is required.An alternative technique using an injectionof radiolabelled colloid has been shown to beas accurate and this may be more widelyadopted in the future.

References1. Hall FM, Frank HA. Preoperative localisation of

non-palpable breast lesions. Am J Roentgenol 1979;132: 101–5.

8

122



2. Kopans DB, DeLuca S. A modified needle-hookwiretechnique to simplify preoperative localization ofoccult breast lesions. Radiology 1980; 134: 781.

3. Homer MJ. Non-palpable breast lesion localizationusing a curved-end retractable wire. Radiology1985; 157: 259–60.

4. Homer MJ. Localization of non-palpable breastlesions with the curved-end, retractable wire:leaving the needle in vivo. Am J Roentgenol 1988;151: 919–20.

5. Kopans DB, Swann CA. Preoperative imaging-guided needle placement and localization ofclinically occult breast lesions. Am J Roentgenol1989; 152: 1–9.

6. Urratia EJ, Hawkins MC, Steinbach BG et al.Retractable-barb needle for breast lesionlocalization: use in 60 cases. Radiology 1988; 169:845–7.

7. Walker TM, Ross HB. Localization of impalpablebreast lesions using the Hawkins needle. Br J Radiol1992; 65: 929–31.

8. Czarnecki DJ, Berridge DL, Splittgerber GF, GoellWS. Comparison of the anchoring strengths of theKopans and Hawkins II needle-hookwire systems.Radiology 1992; 183: 573–4.

9. Chaudary MA, Reidy JF, Chaudhuri P, Millis RR,Hayward JL, Fentiman IS. Localisation ofimpalpable breast lesions: a new device. Br J Surg1990; 77: 1191–2.

10. Burrell HC, Murphy CA, Wilson ARM et al. Wirelocalization biopsies of non-palpable breast lesions:the use of the Nottingham localization device.Breast 1997; 6: 79–83.

11. Kopans DB, Meyer JE. Versatile spring hookwirebreast lesion localiser. Am J Roentgenol 1982; 138:586–7.

12. Cilotti A, Bagnolesi P, Moretti M et al. Comparisonof the diagnostic performance of high-frequencyultrasound as a first- or second-line diagnostic toolin non-palpable lesions of the breast. Eur Radiol1997; 7: 1240–4.

13. Cleverley JR, Jackson AR, Bateman AC.Preoperative localisation of breast microcalcificationusing high-frequency ultrasound. Clin Radiol 1997;52: 924–6.

14. Teh WL, Wilson ARM, Evans AJ, Burrell H, PinderS, Ellis IO. Ultrasound-guided core biopsy ofsuspicious mammographic calcifications using highfrequency and power Doppler ultrasound. ClinRadiol 2000; 55: 390–4.

15. Kuhl CK. MRI of breast tumors. Eur Radiol 2000; 10:46–58.

16. Homer MJ. Localization of non-palpable breastlesions: technical aspects and analysis of 80 cases.Am J Roentgenol 1983; 140: 807–11.

17. Homer MJ, Pile-Spellmann ER. Needle localizationof occult breast lesions with a curved-endretractable wire: technique and pitfalls. Radiology1986; 161: 547–8.

18. Bigelow R, Smith R, Goodman PA, Wilson GS.Needle localization of non-palpable breast masses.Arch Surg 1985; 120: 565–9.

19. Gisvold JJ, Martin JK. Prebiopsy localization ofnonpalpable breast lesions. Am J Roentgenol 1984;143: 477–81.

20. Hall FM, Storella JM, Silverstone DZ, Wyshak G.Non-palpable breast lesions: recommendations forbiopsy based on suspicion of carcinoma atmammography. Radiology 1988; 167: 353–8.

21. Tresadern JC, Ashbury D, Hartley G, Sellwood RA,Borg-Grech A, Watson RJ. Fine-wire localizationand biopsy of non-palpable breast lesions. Br J Surg1990; 77: 320–2.

22. Jackman RJ, Marzoni FA. Needle-localized breastbiopsy: why do we fail? Radiology 1997; 204:677–84.

23. Owen AWMC, Nanda Kumar E. Migration oflocalising wires used in guided biopsy of the breast.Clin Radiol 1991; 43: 251.

24. Svane G. A stereotaxic technique for preoperativemarking of non-palpable breast lesions. Acta Radiol1983; 24: 145–51.

25. Svane G. Stereotaxic needle biopsy of non-palpablebreast lesions. A clinic–radiologic follow-up. ActaRadiol 1983; 24: 385–90.

26. Mullen DJ, Eisen RN, Newman RD, Perrone PM,Wilsey JC. The use of carbon marking afterstereotactic large-core-needle breast biopsy.Radiology 2001; 218: 255–60.

27. Harter LP, Curtis JS, Ponto G et al. Malignantseeding of the needle tract during stereotaxic coreneedle breast biopsy. Radiology 1992; 185: 713–14.

28. Gennari RG, Galimberti V, De Cicco C et al. Use oftechnetium-99m-labeled colloid albumin forpreoperative and intraoperative localization of non-palpable breast lesions. J Am Coll Surg 2000; 190:692–9.

29. Luini A, Zurrida S, Paganelli G et al. Comparison ofradioguided excision with wire localization of occultbreast lesions. Br J Surg 1999; 86: 522–5.

30. Bagnall MJC, Rampaul R, Evans AJ, Wilson ARM,Burrell HC, Geraghty JG. Radio-guided occult lesionlocalization (ROLL) – a new method for locatingimpalpable breast lesions at surgery. Radiology2000; 3 (IOS Supplement): 49.

8

123

Chapter

9Clinical aspects of diagnosing

microcalcificationR. D. Macmillan

Introduction 127

Calcification and DCIS 127

Calcification and invasive cancer 129

Calcification and atypical hyperplasias 130

Conclusions 133

125

Clinical aspects of diagnosing microcalcification

IntroductionThe large majority of breast lesions present-ing as microcalcification are impalpable.Their clinical management depends on his-tology and extent and is greatly facilitated,and possibly more effective, if a preoperativediagnosis can be made. Extensive areas ofmicrocalcification rarely present a problem inthis regard and, as almost all of those that areclinically significant will be extensive DCIS,management decisions are not difficult.Small, localised clusters of microcalcificationmay be less amenable to radiological biopsyand an open diagnostic biopsy may berequired. When a preoperative diagnosis ofmalignancy has been made, therapeutic widelocal excision may be appropriate. Excisionof non-palpable lesions is now a relativelyroutine procedure for breast surgeons. Theextent of surgery may be influenced by mam-mographic findings both for calcificationassociated with pure DCIS and also formicrocalcification present within and aroundinvasive lesions. Microcalcification may alsobe the presenting feature of non-malignantlesions that predict increased breast cancerrisk. This chapter will discuss the clinicalaspects of these issues.

Calcification and DCISThe incidence of DCIS has increased sixfoldover the past 15 years. The mode of presenta-tion has also changed from a palpable mass,nipple discharge or Paget’s disease to thecurrent situation in which approximately90% of all DCIS cases present as clini-cally occult lesions on mammography.Approximately 80% of these are present asmicrocalcification. This is of course due tomammographic screening in which, forwomen aged 50–65, 20% of all cancersdetected are DCIS. If younger women arescreened, this percentage is reportedly ashigh as 40%.

The optimum management of DCIS is stilldebated. Uncertainty remains about whichwomen should be treated by mastectomyand which by wide local excision. No ran-domised trial has ever compared wide localexcision to mastectomy for the treatment ofDCIS and, whilst mastectomy is virtuallycurative, a small but significant risk of localrecurrence is reported by all series of widelocal excision. Factors used to predict thisrisk include close pathological margin status,high-grade and comedo histological subtype.Scoring systems have been developed whichweight these factors to predict risk of recur-rence and that proposed by the Van NuysGroup is perhaps the most widely used1.There is general agreement that marginstatus is of paramount importance. This is, ofcourse, not available until an attempt at ther-apeutic surgery has been performed. In threerandomised trials in which wide local exci-sion alone has been compared with adjuvantradiotherapy, local recurrence rates aresignificantly reduced by radiotherapy.However, it is clear that not all womenrequire radiotherapy (indeed the largemajority do not) and a few women will getlocal recurrence even after receiving it.

Several reports have assessed mammo-graphic appearances for predicting the likelysuccess of breast-conserving surgery forDCIS. Only one study has used actual localrecurrence rate as its primary outcomemeasure. This reported that proximity ofmicrocalcifications to within 40 mm of thenipple was associated with increased risk oflocal recurrence2. The local recurrence rate at5 years was 36.6% (11 of 46) in the closegroup compared to 12.8% (four of 83) in thedistant group. In another study of 37 cases,risk of local recurrence (as predicted by theVan Nuys prognostic index) was lowest withfine granular microcalcifications, moderatewith coarse granular microcalcifications andhighest with linear branching microcalcifica-tions3. This was principally due to the 127

9


association between linear branching micro-calcifications, and to a lesser extent coarsegranular calcification, with high nucleargrade. Linear pattern calcification, particu-larly if branching, is strongly associated withboth nuclear grade and presence of comedonecrosis4–8. Of 198 cases of DCIS diagnosed inKopparberg County, Sweden between 1977and 1994, 151 (76%) presented as microcalci-fication8. The distribution of grade accordingto type of calcification is shown in Table 9.1.

What is the clinical relevance of being ableto predict grade of DCIS by mammography?In both the NSABP and EORTC randomisedtrials of wide local excision for DCIS in whichwomen were randomised to adjuvant radio-therapy or not, high-grade/comedo necrosiswas a significant independent predictor oflocal recurrence risk9,10. In addition, invasivelocal recurrences in women with high-gradeDCIS tend to be grade 3 cancers11. Hencemortality from invasive recurrences of DCIS,although small, is greatest in those who hadhigh-grade DCIS. In the EORTC trial, 11 ofthe 14 women (79%) who developed metas-tases following an invasive local recurrencehad originally been treated for high-gradeDCIS10. Thus, if women with high-gradeDCIS are to be treated by breast-conservingsurgery, it is important that local treatment isadequate. Young women are more likely tohave high-grade DCIS and it is this groupwho are perhaps most likely to chose breast-conserving surgery. Knowing preoperativelythat a woman is likely to have high-grade

DCIS may influence the extent of surgicalexcision.

However, the DCIS collaborative groupsounded a cautionary note regarding the sig-nificance of grade12. Although it significantlycorrelated with local recurrence at 5 years(12% versus 3%), at 10 years this correlationwas all but lost (18% versus 15%). This wouldsuggest that grade of DCIS might be a pre-dictor of time to local recurrence as well asgrade of invasive local recurrence.

Currently, mammographic extent ofmicrocalcification rather than type of calcifi-cation is the main preoperative determinantof suitability for breast-conserving surgery.Discrepancy between this and pathologicalextent measured at microscopy is obviouslyof relevance. This was assessed by Holland etal.13; in a series of 82 mastectomy specimens,47% of the micropapillary/cribriform DCIStype showed a discrepancy of greater than20 mm between radiological and patho-logical extent compared to only 16% ofcomedo-type DCIS. However, in a later pub-lication analysing a series of 35 cases, therelationship between discrepancy in size andtype of DCIS was lost if magnification viewsof the calcification were performed4. This isnow standard practice. However, despitemagnification the pathological extent of 17%of all cases in this series was still underesti-mated by more that 2 cm.

Margin analysis of wide local excisionspecimens for DCIS is not, unfortunately, anexact science. Hence despite a widely held

128

9

Linear Coarse granular Fine granular

No. of cases (%) 46 (30.5%) 78 (51.7%) 27 (17.9%)

High grade 37 (80.4%) 35 (44%) 3 (11.1%)

Low/Intermediate grade 9 (19.6%) 43 (55.1%) 24 (88.9%)

From Tabar et al.8

Table 9.1 Distribution of grade according to type of calcification



belief that DCIS is a contiguous diseaseprocess, all series of breast-conservingsurgery report a rate of local recurrencedespite apparently clear margins. This rate ishigher if radiotherapy is omitted and affectsthe site of previous excision in approximately80% of cases. The chance of achieving clearsurgical margins is greatest at initial surgery,hence the importance of preoperative diag-nosis. It can be difficult at re-operation toidentify the original site of the lesion andhistological interpretation of scarred anddiathermised tissue can be problematic. In arecent analysis of the Nottingham DCISseries, 24/28 local recurrences (86%)occurred in women who had undergone re-excision to (apparently) clear margins aftereither diagnostic or initial therapeuticsurgery14. The chance of clear margins is alsogreater, unsurprisingly, with increasingwidth of excision15. There is an argument per-haps for recommending a wider (2–3 cm)margin for high-grade DCIS (linear calcifica-tion) than for low-grade DCIS (fine granularcalcification) and it may be that fewer casesof high-grade DCIS are suitable for breast-conserving surgery on the basis of thiscriteria alone.

Calcification and invasivecancerAssociated microcalcification also has signi-ficance for invasive lesions. It correlates withtumour grade, one study showing that itoccurred with 31% of high-grade cancerscompared to only 6% of low-grade cancers16.Six studies have shown that it correlates withan increased risk of involved surgical exci-sion margins after wide local excision17–22.Two studies have reported that two-thirds ofwomen who have cancers with associatedcalcifications have involved cavity wall shav-ings following wide local excision17,18. This isparticularly likely with linear calcifications17.

Beron et al. showed in a series of 190 patientsthat mammographic calcification of any sortwas associated with an increased chance offinding residual disease at re-excision (35%versus 11% with no calcifications)22. In astudy of 381 women, a stellate lesion withassociated microcalcification was associatedwith a 3.8-fold (95% CI 1.1–13.0) increase inlocal recurrence21.

Extensive in situ disease surrounding aninvasive cancer is also recognised as a riskfactor for local recurrence after wide localexcision. Five-year local recurrence rateshave been reported to be 3.5-fold higher inthose with an EIC23. In 105 cases, Healey et al.showed that cancers with EIC were morelikely to be associated with mammographiccalcification than those without EIC (83%versus 27%)24.

Absence of mammographic calcificationsurrounding an invasive ductal carcinomamay be the best predictor of unifocality21.Of 135 mastectomy specimens containingtumours < 4 cm (44% less than 2 cm), 90 hadno associated calcification. Of these, 39% hadmultifocal disease beyond 1 cm from thedominant mass histologically compared to62% in those with associated calcification.Surgeons should certainly take note of thepresence and type of calcification withinand around invasive cancers. Patients withmicrocalcifications may be selected for widerexcision or in some cases may be deemedunsuitable for breast-conserving surgery andprospective studies assessing this, perhapsassessing the significance of different pat-terns of microcalcification, are required.

Certain types of microcalcification haveeven been proposed as having indepen-dently significant prognostic value25. In aretrospective study of 343 patients withbreast cancers < 15 mm, casting-type calcifi-cations appeared to select a group of womenwith a poor prognosis. In view of the asso-ciation between this feature and tumourgrade, this is almost certainly a secondary

9

129


prognostic variable and it is difficult to envis-age its clinical utility.

Calcification and atypicalhyperplasiasA relatively common reason for a diagnosticexcision of microcalcification is ADH. Of alldiagnostic excisions performed for a pre-operative diagnosis of ADH, 25–47% willbe DCIS or invasive cancer26–29. One studysuggested that mild ADH found on mammo-tome biopsy may not need surgical excisionif all calcification was excised30. However, themain clinical difficulty encountered withADH is the variability in pathological inter-pretation of this lesion between differentpathologists. Nevertheless, surgical biopsy isnecessary and, as ADH is commonly associ-ated with malignant lesions, it may beappropriate to discuss therapeutic excision ofthe lesion as the first operative procedure(see below).

Wire-guided biopsyFor impalpable lesions the most commonmethod of performing either diagnostic ortherapeutic surgery is wire-guided biopsy.Other approaches are under investigation.

The surgical technique for both diagnosticand therapeutic wire-guided biopsy issimilar, with the exception that a diagnosticoperation is often incisional, the prioritybeing to remove the minimal amount oftissue sufficient for diagnosis with maximalconsideration for cosmesis. Accepting thisprinciple it is still sometimes prudent to con-sent the patient for and intentionally performa therapeutic excision for small lesions wherethe degree of radiological suspicion ishigh and the resultant additional effect oncosmesis is negligible.

In performing wire-guided surgery, theinitial step is to make a spatial appraisal of the

lesion from the mammograms in relation toits position within the breast, relationship tothe nipple, skin and chest wall. The next stepis to ascertain the full extent of the lesion, andinking of its perimeter on the mammogramsby the radiologist may aid this. To this end, amagnification view of the microcalcification isessential. The final preoperative step is todetermine the position of the lesion in relationto the tip of the wire. Ideally the wire tipshould be within the lesion or just beyond it(Figs 9.1 and 9.2). Two-view mammographyis required for this process and craniocaudaland true lateral films may be the ideal combi-nation. Attention is then turned to the breastand placement of the surgical scar. Thisshould be directly over the lesion. Minoradjustments to enhance cosmesis are accept-able but re-excision rates are high after bothdiagnostic and therapeutic procedures formicrocalcification and second excisions areeasier when the wound lies directly over thecavity. Most wire-guided localisation systemsinvolve a rigid cannula being fed over a flex-ible guide wire. This should be performedwith the wire straight. It thus needs to be heldunder slight tension and an assistant shouldmanipulate the breast to approximate theposition in which the wire was inserted(recorded by the radiologist), so straighteningit. Once the rigid cannula is advanced to thetip of the wire, the site of the lesion is ofteneasily palpable by balloting the tissuebetween the tip of the cannula and the over-lying skin. This in turn is made easier if thereis a relatively short length of cannula withinthe breast. At operation the position of thewire tip is repeatedly checked by balloting thecannula. When surgery is for an invasive can-cer or when there is a mass lesion on mam-mography, there is often a palpable noduleintraoperatively, which can guide excision.This is not the case for DCIS presenting asmicrocalcification only, for which there is usu-ally no palpable abnormality even intraoper-atively.

9

130



9

131

Fig. 9.1Mammography following stereotactic localisingwire insertion. This check image shows the tip ofthe wire just through the cluster ofmicrocalcifications. This is the ideal wire position.

Fig. 9.2Specimen X-ray on a diagnostic wirelocalisation showing adequatesampling of the calcification cluster.


The superficial plane of dissection overthe lesion is initially dissected. In fattybreasts, which can “crumble” easily duringsurgery, great care needs to be taken to avoiddisplacement of the guide wire. For a thera-peutic procedure, I then normally developthe planes of dissection parallel to the tip ofthe cannula, then dissect the margins distaland deep to the cannula tip before complet-ing the excision by removing the cannula anddividing the wire at the desired margin ofexcision. The specimen is marked, X-rayedand margins of excision assessed (Fig. 9.3).Two-dimensional X-ray is appropriate ifexcision margins have not been taken toskin superficially and pectoral fascia deeply.Surgical clips left in the cavity may help toidentify the site of the lesion if a re-excision isrequired.

With relatively large areas of microcalci-fication, or where it forms a more lineardistribution, bracketing by two wires may behelpful. This can allow large lesions to bewidely excised with cosmetic reshaping,reduction or volume replacement proce-dures.

Radio-guided biopsy

An alternative technique to wire-guidedsurgery is radio-occult lesion localisation(ROLL). This utilises 99mTc-labelled macro-molecules to localise the lesion, which canthen be detected using a gamma probe. Themacromolecules do not migrate within thebreast and are excised as part of the excisionspecimen. A check film is required to confirmcorrect placement of the marker. This can beachieved either by a scintigram or by inject-ing a small amount of radio-opaque dye withthe macromolecules and confirming correctplacement with a mammogram (Fig. 9.4).This technique has potential advantages overwire-guided biopsy. One advantage for sur-geons is that the site of the lesion is readilyidentified throughout the operation andabsence of radioactivity within the cavityconfirms that the lesion lies within the speci-men. It can be performed on the day or eventhe day before surgery and it is also notsubject to displacement, unlike guide wires.One disadvantage encountered in theNottingham series was that, for two patients,

9

132

Fig. 9.3Therapeutic wide local excisionspecimen X-ray showing the smallcluster of microcalcificationsrepresenting DCIS centrally within theexcision. Approximately 1–2 cm excisionmargin is seen radiologically.



the radio-opaque dye was inadvertentlyinjected into a duct and the check mammo-gram revealed a galactogram appearancewith diffuse radioactive uptake. Wire-guidedbiopsy was possible after allowing approxi-mately an hour for the dye to clear (Fig. 9.5).In a comparison with wire-localisation, onestudy found that ROLL excision specimenswere smaller and the lesion was bettercentred31. In a randomised trial comparingthese procedures, ROLL was quicker andeasier to perform. Accuracy of the two tech-niques was similar but patient satisfactionwas higher with ROLL32.

Conclusions

Clinically significant breast microcalcifica-tion has several implications for themanagement of premalignant and invasivebreast lesions. It can give important clues asto the biology of the disease process and thisinformation may currently be under-utilisedin the preoperative decision-making process.For both DCIS and invasive breast cancer,associated microcalcification can indicate theneed for wider excision and is strongly asso-ciated with grade. It should therefore be aconsideration in the decision to recommend

9

133

Fig. 9.4Mammographic check film of ROLL. Itcan be seen that the injection of water-soluble ionic contrast media overlies asmall cluster of microcalcifications.


breast-conserving surgery. Such surgeryrequires careful appreciation of the mammo-graphic findings and ROLL may be a usefulnew technique to facilitate it.

References1. Silverstein MJ, Lagios MD, Craig PH, Waisman JR,

Lewinsky BS, Colburn WJ, Poller DN. A prognosticindex for ductal carcinoma in situ of the breast.Cancer 1996; 77: 2267–74.

2. Stallard S, Hole DA, Purushotham AD et al. Ductalcarcinoma in situ of the breast – among factorspredicting for recurrence, distance from the nippleis important. Eur J Surg Oncol 2001; 27: 373–7.

3. Lee KS, Han BH, Chun YK, Kim HS, Kim EE.Correlation between mammographic manifestationsand averaged histopathologic nuclear grade usingprognosis–predict scoring system for the prognosisof ductal carcinoma in situ. Clin Imaging 1999; 23:339–46.

4. Holland R, Hendriks JH. Microcalcificationsassociated with ductal carcinoma in situ:mammographic–pathologic correlation. SeminDiagn Pathol 1994; 11: 181–92.

5. Dinkel HP, Gassel AM, Tshammler A. Is theappearance of microcalcifications on mammographyuseful in predicting histological grade ofmalignancy in ductal cancer in situ. Br J Radiol 2000;73: 938–44.

6. Hermann G, Keller RJ, Drossman S, Caravella BA,Tartter P, Panetta RA, Bleiweiss IJ. Mammographicpattern of microcalcifications in the preoperativediagnosis of comedo ductal carcinoma in situ:histopathologic correlation. Can Assoc Radiol J1999; 50: 235–40.

7. Tan PH, Ho JTS, Ng EN, Chiang GSC, Low SC, NgFC, Bay BH. Pathologic–radiologic correlations inscreen-detected ductal carcinoma in situ of thebreast: findings of the Singapore breast screeningproject. Int J Cancer 2000; 90: 231–6.

8. Tabar L, Gad A, Parsons WC, Neeland DB.Mammographic appearances of in situ carcino-mas. In: Silverstein MJ, Ed. Ductal Carcinoma In

9

134

Fig. 9.5Specimen X-ray following aROLL therapeutic excision.Microcalcifications can be seenadjacent to residual ioniccontrast.



Situ of the Breast. Williams & Wilkins 1997, pp.95–118.

9. Fisher ER, Dignam J, Tan-Chiu E, Costantino J,Fisher B, Paik S, Wolmark N. Pathologic findingsfrom the National Surgical Adjuvant Breast Project(NSABP) 8-year update of Protocol B-17: intraductalcarcinoma. Cancer 1999; 86: 429–38.

10. Bijker N, Peterse JL, Duchateau L et al. Risk factorsfor recurrence and metastasis after breast-conserving therapy for ductal carcinoma in situ:analysis of European Organization for Research andTreatment of Cancer Trial 10853. J Clin Oncol 2001;19: 2263–71.

11. Bijker N, Peterse JL, Duchateau L et al. Histologicaltype and marker expression of the primary tumourcompared with its local recurrence after breast-conserving therapy for ductal carcinoma in situ. Br J Cancer 2001; 84: 539–44.

12. Solin LJ, Haffty B, Fourquet A et al. An internationalcollaborative study: a 15-year experience. In:Silverstein MJ, Ed. Ductal Carcinoma In Situ of theBreast. Williams & Wilkins 1997, pp 385–90.

13. Holland R, Hendriks JH, Vebeek AL, Mravunac M,Schuurmans Stekhoven JH. Extent, distribution, andmammographic/histological correlations of breastductal carcinoma in situ. Lancet 1990; 335: 519–22.

14. Rampaul RS, Valasiadou P, Pinder SE, Evans AJ,Wahedna Y, Wilson ARM, Ellis IO, Macmillan RD,Blamey RW. Wide local excision with 10-mmclearance without radiotherapy for DCIS. Eur JCancer 2001: 37: 15.

15. Silverstein MJ, Lagios MD, Groshen S et al. Theinfluence of margin width on local control of ductalcarcinoma in situ of the breast. N Engl J Med 1999;340: 1455–61.

16. Lamb PM, Perry NM,Vinnicombe SJ, Wells CA.Correlation between ultrasound characteristics,mammographic findings and histological grade inpatients with invasive ductal carcinoma of thebreast. Clin Radiol 2000; 55: 40–4.

17. Macmillan RD, Purushotham AD, Cordiner C, DobsonH, Mallon E, George WD. Predicting local recurrenceby correlating preoperative mammographic findingswith pathological risk factors in patients with breastcancer. Br J Radiol 1995; 68: 445–9.

18. Walls J, Knox F, Baildam AD, Asbury DL, ManselRE, Bundred NJ. Can preoperative factors predictfor residual malignancy after breast biopsy forinvasive breast cancer? Ann R Coll Surg Engl 1996:77: 248–51.

19. Kollias J, Gill PG, Beamond B, Rossi H, Langlois S,Vernon-Roberts E. Clinical and radiologicalpredictors of complete excision in breast-conservingsurgery for primary breast cancer. Aust NZ J Surg1998; 68: 702–6.

20. Liljegren G, Lindgren A, Bergh J, Nordgren H,Tabar L, Holmberg L. Risk factors for local

recurrence after conservative treatment in stage Ibreast cancer. Definition of a group not requiringradiotherapy. Ann Oncol 1997; 8: 235–41.

21. Faverly DR, Hendriks JH, Holland R. Breastcarcinomas of limited extent: frequency,radiologic–pathologic characteristics, and surgicalmargin requirements. Cancer 2001; 91: 647–59.

22. Beron PJ, Horwitz EM, Martinez AA et al. Pathologicand mammographic findings predicting theadequacy of tumor excision before breast-conservingtherapy. Am J Roentgenol 1996; 167: 1409–14.

23. Vicini FA, Recht A, Abner A et al. Recurrence in thebreast following conservative surgery and radiationtherapy for early-stage breast cancer. J Natl CancerInst 1992; 11: 33–39.

24. Healey EA, Osteen RT, Schnitt SJ, Gelman R,Stomper PC, Connolly JL, Harris JR. Can the clinicaland mammographic findings at presentation predictthe presence of an extensive intraductal componentin early stage breast cancer? Intl J Radiat Oncol BiolPhys 1989; 17: 1217–21.

25. Tabar L, Chen HH, Duffy SW, Yen MF, Chiang CF,Dean PB, Smith RA. A novel method for predictionof long-term outcome of women with T1a, T1b, and10–14 mm invasive breast cancers: a prospectivestudy. Lancet 2000; 355: 429–33.

26. Brem RF, Behrndt VS, Sanow L, Gatewood OM.Atypical ductal hyperplasia: histologicunderestimation of carcinoma in tissue harvestedfrom impalpable breast lesions using 11-gaugestereotactically guided directional vacuum-assistedbiopsy. Am J Roentgenol 1999; 172: 1405–7.

27. O’Hea BJ, Tornos C. Mild ductal atypia after large-core needle biopsy of the breast: is surgical excisionalways necessary? Surgery 2000; 128: 738–43.

28. Gadzala DE, Cederbom GJ, Bolton JS et al.Appropriate management of atypical ductalhyperplasia diagnosed by stereotactic core needlebreast biopsy. Ann Surg Oncol 1997; 4: 283–6.

29. Moore MM, Hargett CW III, Hanks JB, Fajardo LL,Harvey JA, Frierson HF Jr, Slingluff CL Jr.Association of breast cancer with the finding ofatypical ductal hyperplasia at core breast biopsy.Ann Surg 1997; 225: 726–31.

30. Adrales G, Turk P, Wallace T, Bird R, Norton HJ,Greene F. Is surgical excision necessary for atypicalductal hyperplasia of the breast diagnosed byMammotome? Am J Surg 2000; 180: 313–15.

31. Luini A, Zurrida S, Paganelli G et al. Comparison ofradio-guided excision with wire localisation ofoccult breast lesions. Br J Surg 1999; 86: 522–5.

32. Rampaul RS, Bagnall M, Burrell H, Wilson ARM,Vuyaic P, Pinder SE, Geraghty JG, Macmillan RD,Evans AJ. Prospective randomised study comparingradio-guided surgery (ROLL) to wire-guidance foroccult breast lesion localisation. Eur J Cancer 2001:37: 2–3.

9

135

Chapter

10High-frequency ultrasound

W. L. Teh

Introduction 139

Background 139

Technique and application 140

Summary 143

137

High-frequency ultrasound

IntroductionThe mainstay of detection of microcalcifica-tions is mammography. As the typical size ofmicrocalcifications ranges from 50 to 500microns in size, mammography is well suitedto the task of demonstrating the presence ofbreast microcalcifications. Using modernscreen-film combinations, microcalcificationsof 50 microns in size can be detected. Themorphology of microcalcifications can befurther characterised using magnificationviews. Magnification views enable themicrocalcifications to be further analysedaccording to the degree of clustering, density,morphology and distribution as well aswhether any other associated features suchas the presence of any masses or distortions.Indeterminate and suspicious clusters ofmicrocalcifications can then be localised andbiopsied under stereotactic guidance in orderto achieve a histological diagnosis. For ultra-sound to be of clinical use in the diagnosis ofmicrocalcifications, the efficacy in the detec-tion, analysis, classification and guidance ofbiopsy procedures must be measured againstthat of conventional screen-film mammo-graphy. The ability to detect suspiciousclustered microcalcification in breast screen-ing is particularly important as it enables thedetection of DCIS, which may be associatedwith small high histological grade invasivetumours1.

BackgroundThe early literature using automated wholebreast water-path scanners in the early 1980sshows that palpable or mammographicmasses containing calcifications appear asmasses containing internal echoes or areas ofacoustic attenuation.

Microcalcifications could not be positivelyidentified. The limiting factor may be the lowfrequency (3.7–4 MHz) that is responsible forthe poor lateral resolution of 2 mm2,3.

However, microcalcifications larger than 0.5mm could be differentiated as strongechogenic foci in hypoechoic masses. The mainvalue was the ability to demonstrate massesthat were associated with invasive carcinoma.

The introduction of higher frequency7.5 MHz transducers with automatic scan-ners improved visualisation of microcalcifi-cations; Jackson et al.4 were able to detectmicrocalcifications sonographically in 57% ofcases but only when associated with masses.The use of real-time 7.5 MHz transducersfurther improved detectability of micro-calcifications as actual echogenic foci inhypoechoic areas5. These tend to appearlarger than the actual pathological size anddo not attenuate. Using 7.5–10 MHz real-time ultrasound equipment, ultrasoundabnormalities corresponding to clusteredmicrocalcifications can be identified in59.6–76% cases with a specificity for malig-nancy of 82–93%6–8. There is, however, loweraccuracy in the positive identification ofbenign microcalcifications. Not all malignantmicrocalcifications could be positively iden-tified on ultrasound; this is usually due to theabsence of a definite sonographic mass.There is also no clear distinction in whetherthe malignant lesions identified sonographi-cally were invasive carcinoma, which wereusually associated with a sonographic massor pure in situ disease. Other investigatorsusing similar equipment to localise impalpa-ble lesions presenting solely as micro-calcifications have not been able to achievethis level of detection9–11.

The use of higher frequency ultrasoundprobes (HFUS) with operating frequenciesabove 7.5 MHz and claimed axial and lateralresolution of 0.1–0.5 mm improved the abil-ity of detecting microcalcifications, and thisappeared to be substantiated by investiga-tors using 10 and 13 MHz transducers in themid-1990s12,13. For instance, a 5–10 MHzbroadband transducer with axial and lateral resolutions of 0.9 mm and 1.1 mm, 139

10


respectively, is capable of detecting micro-calcifications greater than 0.6 mm in size14.This compares to the detection of microcalci-fications of 0.15 mm in size, which can beidentified using a 13 MHz transducer with anaxial resolution of 0.118 mm. However, theresults are mixed with sensitivities rangingfrom 52% (31/52 patients) to 88% (15/17patients)8,13,15. The ability to visualise micro-calcifications is likely to be multifactorial,depending not only on the presence of anyassociated sonographic abnormalities butalso on operator experience.

Technique and applicationThe use of a broadband width linear trans-ducer with a mid-frequency above 7.5 MHzimproves the axial (defined by the pulsewidth) and lateral resolution (defined byecho beam width). Identification of the areaof interest is best guided by the knowledge ofthe mammographic location; this is easiest ifthe true lateral and cranio-caudal projectionsare used. The conventional orthogonal scan-ning planes can be used initially. Scanning inthe radial and anti-radial planes is particu-

larly useful when there is a high suspicion ofa ductal pathology where the mammogramssuggest a ductal or segmental distributionwith pleomorphic clustering of microcalcifi-cations.

More recently, there have also beenattempts to improve detection of malignantdisease associated with microcalcifications byusing colour or power Doppler ultrasound.Malignant breast disease is associated withneoangiogenesis in invasive breast cancer andincreased vascularity in high-grade DCIS16–18.In a recent study involving 44 patients, powerDoppler vascularity associated with clusteredmicrocalcifications was identified in 4/14(28.6%) benign lesions and 12/30 (40%)malignant lesions19. The presence of focalpower Doppler vascularity was instrumentalin detecting focal small masses or distortionsin eight cases, which in turn aided ultrasoundcore biopsy. In total, the combination of powerDoppler and 13 MHz transducer enabled thevisualisation of isolated clustered microcalci-fication in 93% of cases.

Where an ultrasound abnormality hasbeen identified, it is essential that it be corre-lated with the mammographic appearance.

140

10

Fig. 10.1 Benign microcalcifications instromal fibrosis manifest asechogenic foci in relativelyhypoechoic breast tissue.



In the presence of suspicious clustered micro-calcifications, the presence of a mass orecho-poor attenuating area should be consid-ered suspicious for malignant disease. Apositive ultrasound correlate is amenable topercutaneous needle biopsy under ultra-sound guidance. In the absence of asonographic mass, the sonographic lesionmay still be amenable to ultrasound-guided

needle biopsy. In this situation, specimenradiography should be undertaken to ensurethat representative microcalcification isobtained. The use of a large volume per-cutaneous sampling device such as theMammotome™ is likely to improve thediagnostic yield where the sonographicappearance is subtle or manifest as hyper-echoic foci in the absence of a mass.

10

141

Fig. 10. 2Lateral magnification viewshowing typical “tea-cupping” ofmilk of calcium in microcysts.

Fig. 10.3Ultrasound image depictingmicrocysts with milk of calciumlayering within the cysts.


Ultrasound appearances ofbenign microcalcifications

Benign calcifications can be difficult toidentify as many of these occur in fibro-glandular tissue where there is lack of con-trast required to distinguish the echogenic

foci within the echogenic tissue. Thepresence of stromal flecks of hyperechoicfoci representing stromal calcifications cansometimes be clearly identified (Fig. 10.1).Small microcysts containing “milk of cal-cium” as well as areas of focal fibrocysticchange containing microcalcifications are

10

142

Authors Transducer Benign Malignant Malignant frequency in situ invasive

Yang et al 199714 5–10 MHz linear – 0 (0/6) Invasive onlybroadband 100% (36/36)

DCIS andinvasive 52.4%(22/42)

Ranieri et al 19978 10 MHz annular 33.3% 33.3% 95.4%(7/21) (3/9) (21/22)

Cleverley et al 199713 10/13 MHz linear 87.5% 87.8% –(7/8) (7/8)

Gufler et al 200024 7.5 MHz linear 59.2% 100% 100%(16/27) (9/9) (11/11)

8/11 invasivewith in situ

Moon et al 200021 5–10 and 5–12 MHz 22.6% 76.7% 100% linear (14/62) (23/30) (8/8)

Teh 200019 13 MHz linear and 85.7% 85.7% 100% (10/10power Doppler (12/14) (6/7) invasive

with in situ)

Table 10.1 Detectability of clustered microcalcifications in prospective studies with histological correlation

Fig. 10.4A 13 MHz annular array imageshowing a dilated ductcontaining echogenic foci.There is associated parenchymalhypoechogenicity. The surgicaldiagnosis is high-grade DCIS.



other positive benign findings (Figs 10.2and 10.3). Fibroadenomas containing flecksof attenuating coarser calcifications are alsousually visible due to the presence of anassociated hypoechoic mass. Neverthelessthe definitive identification of benignmicrocalcification is comparatively lowerfrom that of malignant disease and rangesfrom 33.5% to 85.7% in prospective studieswhere histological correlation is available(Table 10.1).

Ultrasound appearances ofmalignant-typemicrocalcificationsMalignant lesions are usually more readilyidentified even in the absence of a mammo-graphic mass. The detectability variesdepending on whether the microcalcifica-tions are associated with DCIS or whetherthere is any associated invasive carcinoma.Comparative studies generally show detec-

10

143

Fig. 10.5(A) Mammogram showingsegmental distribution ofclustered malignantmicrocalcifications. (B) Ultrasoundimages confirm the presence ofstrongly echogenic foci. Anultrasound-guided core biopsyand subsequent surgical excisionconfirms comedo DCIS.

A

B


tion of pure DCIS to be superior to that ofbenign disease but inferior to that in whichinvasive disease is also present (Table 10.1).Morphological features described in DCISinclude the presence of dilated ducts contain-ing flecks of microcalcifications (Fig. 10.4)20.There may be associated sonographic par-enchymal changes or hypoechoic lesions.

Adjacent strongly echogenic foci represent-ing microcalcifications are also generallyseen (Fig. 10.5). Masses or irregular attenuat-ing areas may also be present particularlywith high-grade or comedo DCIS (Fig.10.6)20–22. These have the appearances of thetypical spiculated or irregular masses or non-strongly attenuating lesions or distortions. In

10

144

Fig. 10.6(A) Magnification view of clusteredcasting microcalcifications.(B) Ultrasound of the areademonstrates the presence ofseveral irregular masses withassociated parenchymal changes. Afleck of strongly echogenic focus isseen within the irregular mass.Ultrasound core biopsy confirmshigh-grade DCIS.

A

B



some cases, masses are not seen but sec-ondary signs of surrounding parenchymalabnormality (such as increased echogenicityof the intramammary fat) may be present.Where invasive carcinoma is present, thepositive identification of a sonographicabnormality approaches that of 100%. Thesetend be irregular, ill-defined masses. Theability to visualise a sonographic abnormal-ity is particularly high where themammogram shows a suspicious (e.g. pleo-morphic or typically casting or comedopattern) appearance or where there is cluster-ing of more than 10 mm in extent. Theincreased detection rate of malignant calcifi-cations using ultrasound has beensuccessfully exploited by investigators as ameans of performing ultrasound-guidedneedle biopsy or localisations8,13,19,21–23.

SummaryThe use of high-frequency ultrasound can beused to detect mammographic microcalcifica-tions. The ability to reliably detect benignmicrocalcifications remains low. Never-theless, clustered suspicious microcalcifica-tions demonstrated on mammography can bevisualised using ultrasound, particularlywhere associated with malignancy. Malignantclustered microcalcifications of suspiciousappearances, particularly if extensive, can bedetected as either masses or focally dilatedducts. The use of power Doppler may alsoimprove the detection of invasive foci of dis-ease.

References1. Evans AJ, Pinder SE, Snead DR, Wilson AR, Ellis IO,

Elston CW. The detection of ductal carcinoma in situat mammographic screening enables the diagnosisof small, grade 3 invasive tumours. Br J Cancer 1997;75: 542–4.

2. Kopans DB, Meyer JE, Steinbock RT. Breast cancer:the appearance as delineated by whole breast water-path ultrasound scanning. J Clin Ultrasound 1982;10: 313–22.

3. Lambie RW, Hodgden D, Herman EM, KoppermanM. Sonomammographic manifestations ofmammographically detectable breastmicrocalcifications. J Ultrasound Med 1983; 2: 509–14.

4. Jackson VP, Kelly-Fry E, Rothschild PA, HoldenRW, Clark SA. Automated breast sonography usinga 7.5-MHz PVDF transducer: preliminary clinicalevaluation. Work in progress. Radiology 1986; 159:679–84.

5. Kasumi F. Can microcalcifications located withinbreast carcinomas be detected by ultrasoundimaging? Ultrasound Med Biol 1988; 14 (Suppl 1):175–82.

6. Leucht WJ, Leucht D, Kiesel L. Sonographicdemonstration and evaluation of microcalcificationsin the breast. Breast Dis 1992; 5: 105–23.

7. Kasumi F, Sakuma H. Identification ofmicrocalcifications in breast cancers by ultrasound.In: Madjar H, Teubner J, Hackeloer B-J, Eds. BreastUltrasound Update. Basel: Karger, 1994, pp. 154–67.

8. Ranieri E, D’Andrea MR, D’Alessio A et al.Ultrasound in the detection of breast cancerassociated with isolated clusteredmicrocalcifications, mammographically identified.Anticancer Res 1997; 17: 2831–5.

9. Pamilo M, Soiva M, Anttinen I, Roiha M, Suramo I.Ultrasonography of breast lesions detected inmammography screening. Acta Radiol 1991; 32:220–5.

10. Potterton AJ, Peakman DJ, Young JR. Ultrasounddemonstration of small breast cancers detected bymammographic screening. Clin Radiol 1994; 49:808–13.

11. Rissanen TJ, Makarainen HP, Apaja-Sarkkinen MA,Lindholm EL. Mammography and ultrasound in thediagnosis of contralateral breast cancer. Acta Radiol1995; 36: 358–66.

12. Kenzel PP, Hadijuana J, Schoenegg W, MinguillonC, Hosten N, Felix R. Can the high-resolution 13-MHz ultrasonography facilitate the preoperativelocalization and marking of microcalcificationclusters? Rofo 1996; 6: 551–6.

13. Cleverley JR, Jackson AR, Bateman AC.Preoperative localization of breast microcalcificationusing high-frequency ultrasound. Clin Radiol 1997;52: 924–6.

14. Yang WT, Suen M, Ahuja A, Metreweli C. In vivodemonstration of microcalcification in breast cancerusing high resolution ultrasound. Br J Radiol 1997;70: 685–690.

15. Cilotti A, Bagnolesi P, Moretti M et al. Comparisonof the diagnostic performance of high-frequencyultrasound as a first- or second-line diagnostic toolin non-palpable lesions of the breast. Eur Radiol1997; 7: 1240–4.

16. Engels K, Fox SB, Whitehouse RM, Gatter KC,Harris AL. Distinct angiogenic patterns are

10

145



associated with high-grade in situ ductalcarcinomas of the breast. J Pathol 1997; 181: 207–12.

17. Lee AH, Happerfield LC, Bobrow LG, Millis RR.Angiogenesis and inflammation in ductal carcinomain situ of the breast. J Pathol 1997; 181: 200–6.

18. Guidi AJ, Fischer L, Harris JR, Schnitt SJ.Microvessel density and distribution in ductalcarcinoma in situ of the breast. J Natl Cancer Inst1994; 86: 614–19.

19. Teh WL, Wilson AR, Evans AJ, Burrell H, Pinder SE,Ellis IO. Ultrasound-guided core biopsy ofsuspicious mammographic calcifications using highfrequency and power Doppler ultrasound. ClinRadiol 2000; 55: 390–4.

20. Hashimoto BE, Kramer DJ, Picozzi VJ. High detectionrate of breast ductal carcinoma in situ calcificationson mammographically directed high-resolutionsonography. J Ultrasound Med 2001; 20: 501–8.

21. Moon WK, Im JG, Koh YH, Noh DY, Park IA. US of mammographically detected clusteredmicrocalcifications. Radiology 2000; 217: 849–54.

22. Schoonjans JM, Brem RF. Sonographic appearanceof ductal carcinoma in situ diagnosed withultrasonographically guided large-core needlebiopsy: correlation with mammographic andpathologic findings. J Ultrasound Med 2000; 19:449–57.

23. Rickard MT. Ultrasound of malignant breastmicrocalcifications: role in evaluation and guidedprocedures. Australas Radiol 1996; 40: 26–31.

24. Gufler H, Buitrago-Tellez CH, Madjar H et al.Ultrasound demonstration of mammographicallydetected microcalcifications. Acta Radiol 2000; 41:217–21.

10

146


Chapter

11Computer-aided detection of mammographic calcifications

Sue Astley

Introduction 149

Computer-aided mammography 150

Conclusions 153

147

Computer-aided detection of mammographic calcifications

Introduction

Most mammograms are acquired on X-rayfilm. However, in recent years, digital X-rayacquisition systems have been developed, in which the information in the image isrepresented directly as a matrix of numbers(pixels). Although digital images may beprinted on film for conventional viewing,they are more usually displayed and viewedon screen. Advances in digital acquisitionand display technology, coupled with theflexibility offered by digital imaging, havemade the prospect of routine digitalmammography more realistic, althoughresolution is still an issue, particularly forscreening applications where the early detec-tion of microcalcifications is important.

The use of digital mammograms offers anumber of significant advantages. First, theimages are readily amenable to manipulationby performing mathematical operations onthe matrix of numbers. Such processing canfacilitate viewing; for example, by adjustingthe contrast or brightness, or by using knowl-edge of the imaging physics to compensatefor degradation1. Specific image features canbe made more – or less – conspicuous to aidthe detection of abnormalities; for instance,linear structures might be enhanced toimprove visualisation of architectural distor-tion, or suppressed to aid detection ofmasses. Quantitative information can also beextracted to enable classification of detectedabnormalities for diagnostic purposes, ormeasurement for monitoring response totreatment. Images of the left and right breast,or films taken on different occasions, may beregistered to facilitate comparison. A furtheradvantage of using digital images is that theycan be rapidly transmitted to other sites, andmultiple copies may be made available with-out loss of image quality. Digital images canbe easily annotated without detriment to theoriginal data; multiple annotations can thusbe acquired for teaching and research pur-

poses. Intelligent software has been devel-oped to provide access to digital imagedatabases, not only on the basis of findingseveral examples of a given pathology, but to enable searches for images which shareparticular properties. This is useful both for teaching and for diagnostic purposes.

One of the most exciting prospects is thatof using a computer to automatically detectgroups of pixels corresponding to clinicallysignificant abnormalities2. Research in thisarea has been active for the last 25 years, withan increase in interest and particularly rapidprogress during the last 10 years. Most of thealgorithms to date have been developed andtested using digitised film images. Detectionof mammographic abnormalities is a chal-lenging problem both for human andcomputer, due to the sometimes ill-definedand subtle nature of abnormal signs, thecomplexity and variability of the underlyingmammographic structure and the similarityof normal and abnormal image features. Thisis compounded, in the screening context, bythe infrequency of clinically significantabnormalities and by the requirement forefficiency. The most successful results to datehave been for the detection of microcalcifica-tion clusters; although these may pose aproblem of perception for the humanobserver, they have a well-defined range ofappearance and are dissimilar to theparenchymal background. Current algo-rithms are capable of detecting a very highpercentage of microcalcification clusters witha low false positive rate. There has also beengood progress on the detection of spiculatedmasses, by virtue of the fact that a combina-tion of radiating linear stucture and a brightcentral region is not typical of the underlyingbackground, but the results so far are lessimpressive than those for microcalcificationdetection. Some signs of abnormality, in par-ticular asymmetry and distortion, are evenmore difficult to detect. For these signs, thereis less agreement between human observers 149

11


about what constitutes a significant degree ofabnormality, or merely a variation of normalappearance. Further algorithms have beendeveloped to analyse detected abnormalitieswith a view to improving the benign biopsyrate and providing additional diagnosticinformation. Again, the most successfulattempts at this are directed at the classifica-tion of calcification.

Computer-aidedmammography

The aim of much of the early research intoabnormality detection was to replace thehuman film reader by a fully automatic com-puterised system. However, in order to dothis, the computer must be able reliably todetect all manifestations of mammographicabnormality with both a very high sensitivityand an acceptably low false positive rate.Clearly, at present, it is not feasible to use amachine even as an automatic second reader;the only type of abnormality for whichdetection performance approaches the levelsattained by experienced human film readersis microcalcification.

Another potential model for computer-aided mammography is prescreening, inwhich the screening films are sorted by thecomputer into two groups: those that areunequivocally normal and those that maycontain an abnormality. The radiologistwould then look only at those films classifiedas potentially abnormal, reviewing just asmall number of the normal group for qual-ity control purposes. Provided a sufficientproportion of the films are correctly identi-fied as being unequivocally normal, thismodel would focus the radiologist on themore difficult and abnormal films, makingbetter use of their skills. Technically, thisrequires algorithms that can either reliablydetect normal images or that can detect all

manifestations of abnormality with greatsensitivity – but not necessarily with highspecificity. The detection of normality is anarea in which research is active; a startingpoint is the classification of mammogramsaccording to their glandular background, asthis has been shown to be related to risk3,4.Detection of abnormalities is easier in fattybreasts for both human and machine, and asignificant proportion of women in thescreening age group have predominantlyfatty breasts, so it may be that a combinedapproach could enable prescreening. Currentcomputer-aided detection systems are notdesigned for prescreening; the full range of abnormalities is not usually targeted, andtoo large a proportion of mammograms areflagged with suspicious regions to makeprescreening cost-effective.

There is, however, a way in which com-puter-based detection algorithms can beused to aid the process of detecting mammo-graphic abnormalities without having a fullsuite of algorithms, and without the require-ment that the algorithms must be almostperfectly sensitive. There is now evidencethat computer-based prompting can improvehuman detection performance. The idea ofprompting is to attract the reader’s attentionto regions of the original image which maybe abnormal. First, the computer uses algo-rithms to detect potential abnormalities, andtheir locations are presented to the humanfilm reader as prompts, which are usuallysmall symbols or outlines of suspiciousregions superimposed on a low resolutionversion of the mammogram. The humanreader, having first viewed the originalimage unaided, consults the prompt imageand reviews the original image accordingly.Some of the prompts may correspond to gen-uine abnormalities which the human readereither failed to see in their initial search of themammogram, or to abnormalities whichthey saw but dismissed as being insignifi-cant.150

11



This approach is known as computer-aided detection (CAD) and, with the adventof commercial CAD systems, we are begin-ning to understand the parameters whichmake prompting effective. Clearly, the algo-rithms need to be sensitive, and to detect notonly the more obvious abnormalities but alsosubtle signs that would otherwise be missedor dismissed by human readers. However,there is also evidence that too many falseprompts reduce the benefits offered by thetechnology5. This could happen by a numberof different mechanisms. For example, falseprompts could distract the human reader,drawing attention away from regions con-taining genuine abnormalities. If there are alarge number of false prompts, the radiolo-gist’s confidence in the significance ofprompts might also be reduced, causing himto routinely ignore prompting information.Most commercial CAD systems are centredaround a highly sensitive microcalcificationdetection algorithm, along with at least oneother algorithm (usually to detect masses).As all the prompts are presented on a singleprompt image, the response of the humanreader is complex, even though microcalcifi-cation and mass prompts are distinct. Thereis published evidence that CAD system algo-rithms are capable of detecting very earlycancers; researchers have looked at the previ-ous screening films of women with intervalcancers, and found that a number of thesehad prompts in the right place6. Despite this,there is as yet no evidence that a commercialCAD system can improve the performance ofhuman readers in the context of the NationalHealth Service Breast Screening Programme.

Microcalcification detectionMicrocalcifications are sometimes difficultfor the human film reader to detect becauseof their small size and low contrast, particu-larly if there are only a few particles, and if

these are superimposed on dense glandulartissue. However, of all the signs ofabnormality found on mammograms, micro-calcifications are the easiest for computers todeal with. Unlike small ill-defined masses,which may superficially resemble normalglandular tissue, microcalcifications haveproperties that differ significantly from thoseof normal background structures. Their smallsize is, in this respect, an advantage, as istheir relatively high attenuation coefficient.The computer can be trained to detect small,bright, regions with well-defined edges.Most algorithms make an initial attempt atdetection, followed up by a more specificanalysis to reduce the number of falsepositive responses. The small size of micro-calcification particles necessitates the use ofvery high-resolution digital images in whichmany pixels are used to represent eachsquare millimetre of the mammogram. Themore pixels in the image, the longer it takesto process, so the initial detection stage isused to reduce the number of pixels at whichdetailed further analysis will be applied.

False signals may be generated by over-lapping narrow linear structures, or artifactssuch as screen-film “shot” noise, but thesecan generally be excluded by further analy-sis7. The analysis phase focuses both on theproperties of individual candidate particles –shape, size and brightness – and also on thedistribution of candidate particles within theimage. One of the most powerful tools forreducing the false positive rate of detectionalgorithms is the application of clustering cri-teria; generally, a detection is only registered(and prompted) if a number of individualbright regions are found within a limitedarea. In other words, prompts are only placedwhere clusters of calcifications are found.

A more difficult problem is distinguishingbetween calcifications that would be dis-missed as insignificant by experiencedradiologists and those it is imperative todetect and prompt. Unless highly accurate

11

151


characterisation algorithms are available, it isprobably safer to prompt all clusters andleave the decision as to whether any furtheraction should be taken to the radiologist. A loss of confidence in microcalcificationprompts could result if some clusters areprompted and some are not.

There have been many published papersdescribing methods for detecting microcalci-fications in digital mammograms, but it isdifficult to compare their efficacy becausefew have been applied to the same sets ofdata. The majority of researchers work withlocal clinicians and have relied on theseclinicians to provide images considered to berepresentative examples. This is a highly sub-jective approach, as there are considerabledifferences in opinion as to what constitutesa representative data set. Some examples are much less conspicuous than others, and whilst a number of researchers haveattempted to compensate for this by classify-ing lesions according to subtlety, it is difficultto avoid bias. In addition, differences in dataarise because some data sets are based onscreening mammograms and others onsymptomatic cases, and because of differ-ences in screening practice.

The problem of subjectivity can be avoidedby using large samples (either randomlyselected or consecutive) taken from screeningdata. In this way, examples with multipleabnormality types and common artifacts willbe naturally included. The performance onvery early cancers can be measured using suit-ably reviewed previous screening films frominterval cancer cases, and normal data shouldbe taken from the previous screening films ofwomen who have had a subsequent normalscreening examination.

Further difficulties in evaluating algo-rithms arise because of the nature of theground truth against which the computer’sperformance is compared. As the majority ofcomputer-based detection methods areapplied to screening and symptomatic mam-

mograms, it is extremely difficult to get accu-rate pathological evidence of the location ofabnormalities – especially for microcalcifica-tions. We must rely on annotations providedby experts, and once again the problem ofsubjectivity arises. In the main, only theboundaries of calcification clusters aremarked, rather than individual particles, asannotation of individual particles isextremely time-consuming and difficult, gen-erally requiring access to both the digitalimage and the original film. To facilitatecomparison of methods, public databases ofdigitised mammograms have been madeavailable7.

The majority of attempts to automaticallydetect microcalcification clusters in digitalmammograms have directly exploited themost obvious properties of the particles: their small size, their increased brightnesscompared to their backgrounds and theirrelatively sharp edges. There exist manygeneric computer vision techniques whichcan be used to detect small bright blobs oredges in images, and researchers haveapplied several of these to mammograms.One class of techniques which has beeninvestigated for the detection of microcalci-fications by a number of researchers ismathematical morphology9–11. The mammo-gram may be considered as a surface, raisedaccording to brightness, with small peakscorresponding to microcalcifications, andlarger ones corresponding to other struc-tures. Morphological operations such aserosion and dilation modify the surfaceaccording to a predefined structuring ele-ment, and combinations of modified surfacescan be used to enhance structures of particu-lar sizes and shapes. Unfortunately, suchsimple techniques also detect any noisepeaks of similar size and shape present in theimages, and poor classification of image pix-els into calcification versus non-calcificationresults. These results improve when cluster-ing rules are applied, and sensitivities of

11

152



approximately 95% with about 0.5 false clusters per image have been reported11.Morphology may also be helpful in identify-ing suspicious regions, based on anexamination of the distribution of detectedpeak sizes in normal and abnormal tissue12.

There are similarities between the mor-phological approaches and that developedby Chan and colleagues13. Their methodinvolves subtracting an image in which themicrocalcifications have been suppressedfrom one in which they have been enhancedby means of a matched filter. Followingsubtraction, there are several stages of post-processing involving extraction of potentialmicrocalcifications and subsequent featureanalysis. More recently, methods based onneural networks have been used to reducethe number of false positive clusters. A sensi-tivity of 93.3% (measured on a ‘per case’basis) with 0.7 false clusters per image hasbeen obtained14.

An alternative approach is to attempt todetect the edges of calcification particles by amplifying gradient information in theimages. However, there are a large number ofedges naturally present in mammograms, allof which will be enhanced, and the subse-quent task of determining which edgescorrespond to microcalcifications is complex.Various combinations of edge and peakdetection have been investigated, but nonehas achieved sufficient sensitivity and speci-ficity to be clinically useful.

A more sophisticated method was devisedby Karssemeijer, who devised a statisticalpreprocessing algorithm for noise estimationand equalisation which was found to signifi-cantly improve results15. Each pixel isassigned one of four labels: background,microcalcification, line/edge, or film emul-sion error. The detection method, based onthe use of Bayesian techniques and a Markovrandom field model to describe spatial rela-tionships between pixel labels, involvesiteratively updating the labels to maximise

their probability. The method uses both con-trast and edge information. A cluster wasdefined as at least two detections in an areaenclosed by an empty region half a centi-metre wide. Results, plotted on FROCcurves, showed a sensitivity of more than90% with slightly less than one false detec-tion per image on a set of 40 mammograms.

Finally, commercial CAD systems are nowunder evaluation both in the UK and abroad.The detection performance of some of thealgorithms is impressive; the ImageCheckersystem from R2 Technology can currentlydetect over 86% of all abnormalities inNHSBSP films. However, the microcalcifica-tion detection algorithm far outshines thesystem’s mass detection capabilities, andwhilst the ImageChecker produces very fewfalse microcalcification prompts (approxi-mately one in every four cases), on averagethere is at least one false mass prompt percase. Despite this, the system is still the mostspecific commercial system available.

ConclusionsJust a few examples of computer-basedmethods for detecting microcalcifications are discussed here; many more have beendescribed in the literature, and still more areunder development. The most successfulmethods can detect 98% of microcalcificationclusters, but in most cases, their false positiverates are still too high to enable effectiveprompting. A rate of 0.5 false prompts perimage may sound good, but for a four-filmmammogram this rate corresponds to anaverage of two false prompts per case, andthe vast majority of cases will have at leastone false prompt.

Many of the false prompts for microcalci-fication detection algorithms correspond toclusters or image features an experiencedfilm reader would simply dismiss, and it maybe that work on characterisation andclassification of microcalcification clusters

11

153



will pave the way for more detailed analysisof candidate clusters. Characterisationresearch has, to date, focused both on theproperties of individual particles (for exam-ple, size, shape and density) and on theproperties of clusters (overall cluster shapeand distribution of particles). Some of thesefeatures – especially cluster shape – are verydifficult to deal with because of the nature ofthe imaging process. Three-dimensionalcluster shapes are projected into a two-dimensional image; even if the effects ofbreast compression are small, robust charac-terisation of the projected shapes is difficult,especially for clusters of only a few particles.

In summary, the detection of microcalcifi-cation clusters is the most successful applica-tion of computer-based detection inmammography. This is due to the differencesbetween microcalcifications – notably size andattenuation coefficient – and their back-ground. Agreat deal of research effort has beendirected at their detection, with the result thatmost clusters can be detected automatically,although false positive detections due to arti-facts and other background structures mayoccur. For automated detection to be a usefultool, the false positive rate must be reducedeven further and, if they are to be incorporatedinto a CAD system, the false prompt rates ofalgorithms for other abnormalities must alsobe reduced significantly.

References1. Highnam R, Brady JM. Mammographic Image

Analysis. Dordrecht: Kluwer Academic, 1998. 2. Boggis CRM, Astley SM. Computer-assisted

mammographic imaging. Breast Cancer Res 2000; 2:392–4.

3. Byng JW, Critten JP, Boyd NF et al. Analysis ofdigitized mammograms for the prediction of breastcancer risk. In: Doi K, Giger ML, Nishikawa RM,Schmidt RA, Eds. Proceedings of the ThirdInternational Workshop on Digital Mammography.Amsterdam: Elsevier, 1996, pp. 185–90.

4. Wolfe JN. Risk for breast cancer developmentdetermined by mammographic parenchymalpattern. Cancer 1976; 37: 2486–92.

5. Astley SM, Zwiggelaar R, Wolstenholme C, DaviesK, Parr TC, Taylor CJ. Prompting in mammography:how good must prompt generators be? In:Karssemeijer N, Thijssen M, Hendriks J, van ErningL, Eds. Digital Mammography. Dordrecht: KluwerAcademic Publishers, 1998, pp. 347–54.

6. Warren Burhenne LJ, Wood SA, D’Orsi CJ et al.Potential contribution of computer-aided detectionto the sensitivity of screening mammography.Radiology 2000; 215: 554–62.

7. Poissonier M, Brady M. Noise equalization, film-screen artifacts, and density representation. In: YaffeMJ, Ed. Proceedings of the Fifth InternationalWorkshop on Digital Mammography. Wisconsin:Medical Physics Publishing, 2001, pp. 605–11.

8. Heath M, Bowyer K, Kopans D, Moore R,Kegelmeyer P. The digital database for screeningmammography. In: Yaffe MJ, Ed. Proceedings of theFifth International Workshop on DigitalMammography. Wisconsin: Medical PhysicsPublishing, 2001, pp. 212–18.

9. Astley S, Taylor C. Detection of microcalcificationsin mammograms. In: Proceedings of the First BritishMachine Vision Conference, 1990.

10. Rick A, Muller S, Bothorol S, Grimaud M.Quantitative modeling of microcalcificationdetection in digital mammography. In: Proceedingsof the Medical Image Computing and Computer-Assisted Intervention, 1999, pp. 32–41.

11. Hagihara Y, Kobatake H, Nawano S, Takeo H.Accurate detection of microcalcifications onmammograms by improvement of morphologicalprocessing. In: Yaffe MJ, Ed. Proceedings of the FifthInternational Workshop on Digital Mammography.Wisconsin: Medical Physics Publishing, 2001, pp. 193–7.

12. Bruynooghe M. High resolution granulometricanalysis for early detection of smallmicrocalcification clusters in X-ray mammograms.In: Yaffe MJ, Ed. Proceedings of the FifthInternational Workshop on Digital Mammography.Wisconsin: Medical Physics Publishing, 2001, pp.154–60.

13. Chan H-P, Doi K, Galhotra S, Vyborny C,MacMahon H, Jokich P. Image feature analysis andcomputer-aided diagnosis in digital radiography. (I) Automated detection of microcalcifications inmammography. Med Physics 1987; 14: 538–48.

14. Gurcan MN, Chan H-P, Sahiner B, Hadjiiski L,Petrick N, Helvie MA 2002 Optimal NeuralNetwork Architecture Selection: Improvement inthe Computerized Detection of Microcalcifications.Academic Radiology 9: 420–29.

15. Karssemeijer N. Adaptive noise equalisation andrecognition of microcalcification clusters inmammograms. Intl J Pattern Recognition ArtifIntelligence 1993; 7: 1357–76.

11

154


Chapter

12MRI detection of DCIS

Fiona J. Gilbert

Introduction 157

MRI sensitivity and specificity in DCIS 158

Morphology of DCIS 159

Conclusion 162

155

12

MRI detection of DCIS

Introduction

Initial experience with MRI suggested thatbreast examination was not very useful asthere is overlap of T1 relaxation times ofbreast parenchymal tissue, benign lesionsand malignant disease. Following the intro-duction of intravenous gadolinium-basedcontrast medium it was found that dynamiccontrast-enhanced breast MRI was very sen-sitive to the detection of malignancy1. Therehave been a number of publications indicat-ing sensitivities of between 88 and 100% butwith variable lower specificity (Table 12.1).

The rapid leakage of contrast into theextravascular, extracellular space in cancersis responsible for the change in signal. Thelocal paramagnetic effects on T1-weightedimages shortening the relaxation time, result-ing in increased signal, allows detection ofabnormalities, particularly if the high signalfrom the adjacent fat is suppressed by usingeither a fat-suppression technique or imagesubtraction.

Lesion morphology, enhancement pat-terns and signal change over time are used todifferentiate benign disease from malignanttissue. Features suggesting malignancy arean enhancing mass with irregular or spicu-

lated margins2. In a series of 56 malignantlesions, 50 had these features, giving a posi-tive predictive value for malignancy of 89%3.Benign masses have smooth or lobulatedborders and have non-enhancing internalseptations2. Lesions with these characteristicsand non-enhancing lobulated masses werefound to be benign in 52 cases, with a negative predictive value for malignancy of100%4. Smooth masses most often repre-sented fibrocystic change, and lobulatedmasses with non-enhancing internal septa-tions were most likely to be fibroadenomas.DCIS is the most common histological corre-late of ductal enhancement, with a positivepredictive value of 40%. Enhancement in oneregion of the breast (regional enhancement)is usually due to fibrocystic change or malig-nancy (PPV 42%). Of these malignancies, halfare DCIS and half are DCIS with an invasivecomponent4.

Enhancement patterns are also thought tobe important in improving benign/malig-nant differentiation. In order to evaluatelesion enhancement, images need to beacquired every minute sequentially beforeand after injection of contrast for up to 5–8minutes. Large data sets are produced andthe limitation on the number of sequences

157

Author Year No. of No. of Benign Sensitivity Specificity patients cancers lesions (%) (%)

Heywang et al.26 1989 150 71 27 98 65

Kaiser & Zeitler27 1989 191 58 31 100 97

Harms et al.24 1993 30 47 27 94 37

Oellinger et al.28 1993 33 25 16 88 80

Gilles et al.29 1994 143 64 79 95 53

Boetes et al.30 1994 83 65 22 95 86

Orel et al.18 1995 176 72 112 92 88

Bone et al.9 1996 231 155 95 93 73

Nunes et al.31 1997 94 46 48 96 79

Kuhl et al.32 2001 192 15 100

Table 12.1 Overall sensitivity and specificity of MRI

12


acquired dynamically is the capacity of thesoftware to process the data. The T1 signalchange can be plotted over time to demon-strate the initial rate of contrastenhancement, the peak signal and subse-quent signal change.

Strong, early enhancement with a relativesignal increase of over 140% and a peak ofenhancement before 3 minutes is highly sug-gestive of malignancy together with earlywashout, i.e. a signal decrease of more than10% following maximum enhancement.Focal, irregular and non-homogeneousenhancement or enhancement that follows aductal pattern is suspicious. Rim or centri-petal enhancement, i.e. enhancement thatstarts at the periphery and progressestowards the centre of an abnormality, ishighly suggestive of malignancy. However, it is becoming increasingly clear that bothenhancement characteristics and lesion mor-phology are required to separate benign andmalignant lesions2.

MRI was initially not thought to be usefulin the detection of DCIS. With increasingasymptomatic mammographic screening, theproportion of DCIS cases had increased from5% of all breast cancers to 15–20% of alldetected breast cancers and 25–30% of allclinically occult cancers detected by mam-mography5. Due to this increased detectionof DCIS, recent breast MRI series haveincluded larger numbers of DCIS and radiol-ogists have also examined the utility of breastMRI to improve the diagnosis in mammo-graphically detected microcalcification.

MRI sensitivity and specificityin DCISMost of the published breast MRI series havebeen consecutive patients with either clinicalor mammographic abnormalities. DCIS hasbeen found together with invasive ductalcarcinoma and the sensitivity and specificity

for MRI for DCIS have been calculated as a sub-analysis4,6–11. MRI is performed for avariety of reasons, usually suspiciousmammographic features, palpable lump onclinical examination or newly diagnosedbreast cancer. DCIS tends to represent only asmall proportion of the patients in the series.The cases are identified retrospectively andthen the characteristic features of DCIS areanalysed. In almost all of the series, the MRIexaminations are read with the knowledgeand availability of the mammograms, withfew series reporting the MRI examinationblind. Sensitivity of MRI for the detection ofDCIS in these series varies between 77% and100%, the specificity varying between 28%and 100% (Table 12.2).

The value of MRI as a problem-solvingtool in patients with mammographic micro-calcification has been examined in two series.Westerhoff et al.12 examined 63 consecutivepatients with suspicious isolated clusteredmicrocalcification to assess the additionalvalue of MRI in relation to surgical manage-ment. In this cohort, 33 patients had DCIS,five patients invasive ductal cancer and 25patients had benign disease. The overallaccuracy of MRI was 56%, with 45% sensitiv-ity and 72% specificity. The sensitivity fordetection of DCIS was 67%. In this series, theMRI did not detect any additional diseasethat had not been identified on mammogra-phy and did not alter surgical management12.

A large, prospective two-centre studyincluded 172 women with isolated, clustered,suspicious microcalcification. All were des-tined for excision biopsy and examined withdynamic contrast-enhanced MRI. Eightmalignant lesions were found, of which 58were in situ cancers and 22 invasive carci-noma. The remaining 92 women had benignlesions. The overall sensitivity was 95% withMRI detecting early enhancement in 56 of 58cases of in situ disease6. However, while ahigh sensitivity was achieved, the authorsconcluded that specificity was poor, limiting158



the value of MRI in distinguishing benignand malignant mammographic microcalcifi-cation.

Overall, the sensitivity of breast MRI indetection of DCIS is lower than for invasivedisease. The reasons for this are complex andare discussed below. Sensitivity is improvedif the mammogram is available to directattention to areas of suspicious microcalcifi-cation. The specificity is variable and similarto invasive disease.

Morphology of DCISDCIS typically displays clumped or linearenhancement with a ductal distribution. Lesscommonly, DCIS can show a spiculatedappearance or even ring enhancement,although infiltrating invasive disease is morelikely to cause a spiculated appearance8.Ill-defined or diffuse enhancement in a seg-mental distribution tends to be the mostcommon finding. Satake, in a study designedto compare ultrasound detection of intra-ductal spread with mammography and MRI,found three morphological patterns – linear,regional and segmental enhancement, with

segmental and linear being the most com-mon correlate of DCIS13. Nunes, inattempting to correlate lesion appearancewith histological findings for a breast MRIinterpretation model, found 11 cases of DCISin a series of 192 patients. MRI detected 10 ofthese 11 cases, four had ductal enhancement(ductal is linear and branching), fourregional enhancement (where regional isdefined as diffuse, ill-defined pattern), oneirregular mass and one spiculated mass4.Linear, spotty enhancement, an area of linearenhancement, an enhancing area or masswithout distortion of the surrounding tissueand a well-circumscribed mass have all beenfound on MRI and reported in a series of 10cases of DCIS14. The spectrum of DCIS mor-phology is well demonstrated in Viehweg’sseries of 50 patients. Of the 48 lesions show-ing enhancement, the commonest appear-ance was of an ill-defined area of enhance-ment followed by a well-defined area, theremaining lesions demonstrating a ductalpattern (eight cases) and a diffuse area ofenhancement in five cases15. Segmentallyextended enhancement defined as vague,faint and diffuse enhancement forming a

12

159

Author Year No. of No. of patients Sensitivity Specificitypatients with DCIS

Gilles et al.6 1996 172 58 95% 51%

Bone et al.9 1996 231 17 82% 73%

Soderstrom et al.8 1996 22 22 100% –

Orel et al.7 1997 330 13 pure 77% –

Westerhof et al.12 1998 63 33 67% 72%

Kuhl et al.16 1998 33 33 100% –

Nunes et al.4 1999 192 11 90% –

Hiramatsu et al.33 1999 21 17 15/17 100%

Amano et al.11 2000 58 26 19/26 –

Satake et al.13 2000 46 15 91% –

Viehweg et al.15 2000 71 50 96% 28%

Table 12.2 MRI detection of DCIS

12


segmental “cone” shape was found in fourpure DCIS lesions, and 10 predominantlyDCIS cases with multiple focal masses con-fined to one quadrant were found in threeDCIS cases. No enhancement was found inseven cases of DCIS associated with invasivecancer11.

Fibrocystic enhancement patterns aremost likely to be confused with DCIS. Benignproliferative changes are seen as fine stippledenhancement8. Fibrocystic and proliferativechange appears as ill-defined irregular seg-ment or region of enhancement which is acommon appearance of DCIS. In summary,160

Fig. 12.1 APostcontrast T1-weightedimage with 3-cm invasive ductalcarcinoma showingheterogeneous enhancementand an irregular margin in thelower outer quadrant of theleft breast.

Fig. 12.1 BSubtraction of pre- andpostcontrast T1-weightedimage improves conspicuity ofabnormal enhancing areacaused by invasive ductalcarcinoma. Regions of interestdrawn round the enhancingtumour and also roundadjacent fat. The subtractedimage results in fatsuppression.



12

161

Fig. 12.1 CTime signal enhancement curvedemonstrates rapid rise inenhancement peak reached within 2minutes and more than 10% wash outimmediately following peakenhancement. The enhancement curveis typical of invasive carcinoma (type IIIcurve). The background fat signalshown by the dotted line shows a slowsteady rise in signal by approximately10%.

Fig. 12.2 AWedge-shaped area ofenhancement seen at the 2 o’clock position of the right breast on postcontrastT1-weighted images.

12


any focal area with diffuse ill-definedenhancement should be considered suspi-cious and investigated further. Enhancementfollowing a linear, branching or ductal pat-tern should be considered suspicious ofDCIS. Rarely, a focal or spiculated mass can be due to DCIS and this should beremembered when reporting this type ofabnormality.

Enhancement rate

DCIS enhancement rates can be very variablewith some lesions enhancing rapidly andsome slowly. Classically, invasive malig-nancy enhances early and reaches a peakwithin 3 minutes of injection time. Contrastinjection in the antecubital vein to breastenhancement is thought to take approxi-162

Fig. 12.2 BWedge-shaped segmental areaof enhancement on fat-suppressed images created bya subtraction of precontrastT1-weighted images frompostcontrast T1-weightedimage.

Fig. 12.2 CAdjacent slice with a ratherlinear ductal pattern at the 2 o’clock position of the rightbreast. On the left breast thereis an irregular lobulated lesionat the 12 o’clock positionwhich was a further 1.2 cmarea of invasive ductal cancer.The segmental area ofenhancement is due to DCIS.



12

163

Fig. 12.2 DRegion of interestdrawn aroundsegmental area ofenhancement andadjacent area ofbackground fat.

Fig. 12.2 EGraph demonstrates signal changeover time. Images acquired at 1-minute intervals. Continuous lineshows slow continuousenhancement over time with nowash out. This type II curve iscommonly found with DCIS as in thiscase. Again, background fat showslittle change in signal over time.

12


mately 46 seconds, although this will varywith heart rate and rapidity of the bolusinjection. Hiramatsu et al. found that DCIStended to be slowly enhancing in 10 cases ofpure DCIS14. In a larger series of 50 cases ofpure DCIS, enhancement was delayed in 65%of lesions15. The majority of cases showedstrong enhancement, with 16 showingindeterminate enhancement and five lowenhancement, and in only two cases wasenhancement absent15. Kuhl et al.’s16 series of33 patients showed fast enhancement in 49%, intermediate enhancement in 30% anddelayed in 21% with all patients with DCISshowing some enhancement. The time–signal intensity curves showed type I,steadily increasing, in 73%, with type II, earlyrise and plateau, in 18% and type III, peakwith washout, in 9%, whereas invasive dis-ease typically shows the reverse, i.e. 8%, 27%and 65%, respectively16. Similarly, inViehweg’s series, only 4% of DCIS showedthe malignant washout pattern15.

In the differentiation of DCIS from benigndisease it is unhelpful to use enhancementrates, as the rapid enhancement characteris-tic of invasive disease is often not present inDCIS. The enhancement patterns of benignproliferative change tend to overlap those ofDCIS with slow continuous enhancement.This can result in DCIS being overlooked. Onfinding slow diffuse enhancement, it isimportant to review the mammograms tosearch for evidence of DCIS. Most authorsadvocate interpretation of breast MRI only inconjunction with the mammograms. If themorphological changes are consistent withDCIS then a biopsy should be considered. Itis important to retain a high index of suspi-cion when diffuse enhancement patterns areeven. Timing of the examination is importantas during the second half of the menstrualcycle the normal breast parenchyma canshow marked enhancement. Imaging isrecommended between days 6 and 16 ofmenstrual cycle to reduce over-investigation

of parenchymal changes17. HRT can causesimilar effects and it is suggested stoppingHRT 4 weeks prior to examination to reducethe number of enhancing normal areas ofbreast tissue.

Relationship between DCIStumour grading and MRenhancementThere does not appear to be a relationshipbetween the grading of DCIS and the rate ofcontrast enhancement, or even whether theDCIS displays any signal change14. There isno significant difference between the rate ofenhancement according to grade, the pres-ence or absence of necrosis or microinvasion.Although 76% of comedo DCIS showed focalearly enhancement, compared to only 50% ofnon-comedo DCIS, no relationship betweencontrast enhancement and tumour grade hasbeen found12. Again, in Orel’s series of 19patients with DCIS (13 pure DCIS and sixDCIS with separate foci of invasive disease),there was no correlation between grade oftumour and enhancement. MRI detectedseven of nine cases with comedo high grade,three out of six non-comedo intermediategrade and three out of three non-comedo lowgrade18. Viehweg found ductal enhancementmore common in comedo-type DCIS (29%)compared to non-comedo (12%). ComedoDCIS was more likely to have early enhance-ment (50%) compared to non-comedo (29%),although none of these were shown to bestatistically significant15.

Lack of enhancement is the reason formost false negative cases in the literature.False negative cases of both comedo andnon-comedo type DCIS have been reported18.Stomper et al. reported three false negativecases, including a 4-cm area of comedo DCISwith 1 mm of microinvasion, and a second 6-cm area of comedo19. The one false negativecase in Sataki’s series was comedo DCIS in164



the sub-areolar area13. The retro-areolar areais notoriously difficult to assess on MRI aswell as mammographically. A 0.6-cm focus ofDCIS did not enhance and was the cause ofthe one negative case in a series of 11patients4.

It is important to be aware that not allcases of DCIS enhance. As MRI is sometimesused as a problem-solving tool to excludemalignancy and, clearly, DCIS cannot beexcluded using dynamic contrast-enhancedMRI as the negative predictive value is not ashigh for DCIS as it is for invasive disease.

Can MRI accurately demonstrateextent of DCIS?The classical mammographic features ofDCIS are irregular, linear or branching micro-calcifications with or without an associatedill-defined soft-tissue mass with, rarely, anill-defined soft-tissue mass being the onlyfinding. Mammography can underestimatethe extent of disease, and this is a particularproblem if DCIS is part of multifocal ormulticentric pathology. The extent of DCISdemonstrated on MRI has been shown tocorrelate reasonably well with histologicalassessment3,8,12. Westerhof, in his series of 63patients, found no additional disease on MRIor at histological examination compared tomammography8. Soderstrom et al., using thefat-suppressed RODEO technique, found theMRI estimation of disease extent concordantwith pathology in 22 cases, the only excep-tion being one case of pure DCIS where MRIoverestimated disease extent – the clumpingpattern was found which correlated withDCIS, fibrocystic change and sclerosingadenosis. In the same series, mammographyenabled accurate prediction of disease extentin 14 of 19 (74%) patients8.

While Gilles et al. found a significantcorrelation between the size of DCIS andhistopathological analysis (r = 0.85, P =

0.0085), they found that there were some dif-ferences in the correlation depending on thetype of enhancement pattern. When focalenhancement was seen, there was a good cor-relation with size at histological examination,in 20 out of 26 patients, but in the remainingsix MRI underestimated the extent of disease.Where there was early diffuse enhancement,in only two out of eight patients was thereaccurate histological correlation and, in theremaining patients, MRI overestimated theextent of DCIS. This was thought to be due toadjacent proliferative disease (four patients)or non-proliferative fibrocystic change3.

MRI has demonstrated mammographi-cally occult DCIS. In one series of 15 cases ofDCIS found on MRI, five of the cases werenot detected by mammography20. Similarly,in the series reported by Orel of 19 DCIScases detected on MRI, seven were mammo-graphically occult and six of these hadmultifocal disease (three in the same quad-rant and three in a separate quadrant)7. In aseries of 33 consecutive DCIS cases, 20 weremammographically occult21.

While MRI can demonstrate additionaldisease, the false negative cases mentioned in the previous section show that it is not asaccurate as histological examination. On balance, MRI is probably as good asmammography in demonstrating the extentof DCIS.

Does type of sequence usedinfluence MRI detection of DCISMost breast MRI studies have been per-formed on high field strength systems(1.0–1.5 T). These produce good signal-to-noise ratios, allow spectrally selected fatsuppression and increase the T1 relaxationtimes of tissues increasing the signal differ-ence between native and contrast-enhancedtissue.

Dedicated phased array breast coils allow

12

165

12


excellent spatial resolution and, with modernhardware and software, allow good temporalresolution. These requirements are necessaryfor the detection of small foci of DCIS. Fatsuppression is necessary to increase lesionconspicuity as subtraction techniques arehampered by patient movement which canobscure small abnormalities. Bilateral breastimaging is recommended due to high inci-dence of bilateral disease (3–6%)22, either inthe axial or coronal plane.

A variety of strategies have beenemployed to improve both detection of smallcancers and DCIS. This has mainly been byimproving spatial resolution and signal-to-noise ratio. Use of contrast is necessary todetect malignant lesions. The dose used isnormally 0.1 ml/kg, although some proto-cols have recommended a double dose toimprove lesion conspicuity23. Ideally, voxelsless than 3 mm should be achieved by usingslice thickness < 3 mm, with as small a rec-tangular field of view as possible to reducevolume averaging. T1-weighted sequenceprecontrast with postcontrast sequencesrepeated up to 5–8 times are recommendedwith each sequence lasting fewer than 60–90seconds.

The detection of small foci of DCISappears to be improved by the use of fat-suppressed three-dimensional volumespoiled gradient echo sequences (SPGR)7.Both Harms and Soderstrom have claimedthat three-dimensional rotating delivery ofexcitation off resonance (3D RODEO)sequence, which is fat-suppressed, hasimproved contrast and spatial resolution. In22 cases of DCIS, the true extent of diseasewas demonstrated in 21 of 22 cases8. Harmset al. demonstrated the value of this sequenceinitially and demonstrated DCIS in all sevencases24. It is thought that the magnetisationtransfer contrast inherent in the RODEOsequence improves the contrast-to-noiseratio between cancer and ductal tissue,which reduces the effects of volume averag-

ing between DCIS and normal parenchyma.Also it is claimed that the microcalcificationcan be detected directly on MRI as itproduces magnetic susceptibility artefacts.However, the susceptibility effects does nothelp distinguish between benign and malig-nant disease8.

New techniques have been introduced inan attempt to improve the temporal resolu-tion of breast imaging. Dynamic spiral MRIachieves improved temporal resolution butat a cost of lowering spatial resolution. The7–10 mm section thickness leads to greatervolume averaging, resulting in decrease inlesion detection. The technique thereforeneeds to be combined with high spatial reso-lution conventional fat-suppressed images.With spiral MRI, the susceptibility artefactsare greater due to the longer readout timeand the lack of magnetisation transfer sup-pression of muscle and normal fibro-glandular tissue. In six cases of DCIS therewas variable contrast enhancement. Thetechnique was very useful for utilising theexchange rate constant to differentiatebenign from invasive disease but it was not helpful in separating DCIS from benigndisease10.

Polarity-altered spectral and spatial selec-tive acquisition (PASTA) technique for fatsuppression has been used with three-dimensional fast spin echo sequences.PASTA uses narrow band spectral selective90° pulse to eliminate fat, and a reversepolarity 180° pulse to avoid contamination offat signals from other sections. Andoreported a series of 49 women imaged usingthis technique, in which there were sevenDCIS cases25. The method detects tumourvascularity directly instead of evaluatingpattern of enhancement and this could beused as a surrogate marker of malignancy. Inthe initial patients in the series, two-dimen-sional fast spin echo sequences were usedwith three to six slices 6-mm thick, but lat-terly a three-dimensional fast spin echo166



sequence was used in the sagittal plane.Maximum intensity projections are createdand peritumoural, intratumoural and mar-ginal vascularity is documented. Intra-tumour and marginal vascularity weresignificantly more common in invasive dis-ease than in DCIS but DCIS could not bereliably differentiated from benign disease.

ConclusionThe sensitivity of MRI to detect DCIS is lowerthan for invasive cancer, likely due to thevariable angiogenesis found with DCIS. Themost specific patterns are linear, clumped,regional and segmental enhancement butmore commonly a diffuse, ill-defined patternis seen. Signal change following contrasttends to be less than invasive disease butgreater than benign, proliferative changes.The enhancement rate can be rapid but morecommonly is slow, continuing to rise orplateau, with washout rarely seen. There isno relationship to grade of in situ disease.High resolution imaging is required to detectDCIS together with a high index of suspicionwhen dealing with ill-defined, diffuselyenhancing areas.

References1. Heywang-Kobrunner S, Hahn D, Schmid H,

Krischke I, Eiermann W, Bassermann R, Lissner J.MR imaging of the breast using gadolinium DTPA. J Comput Assist Tomogr 1986; 10: 199–204.

2. Orel S. MR imaging of the breast. Radiol Clin N Am2000; 38: 899–913.

3. Gilles R, Zafrani B, Guinebretiere J et al. Ductalcarcinoma in situ: MR imaging – histopathologicalcorrelation. Radiology 1995; 196: 415–19.

4. Nunes L, Schnall M, Orel S, Hochman M, LanglotzC, Reynolds C, Torosian M. Correlation of lesionappearance and histologic findings for the nodes ofa breast MR imaging interpretation model.Radiographics 1999; 19: 79–92.

5. Stomper P, Margolin F. Ductal carcinoma in situ: themammographer’s perspective. Am J Roentgenol1994; 162: 585–91.

6. Gilles R, Meunier M, Lucidarme O et al. Clustered

breast microcalcification: evaluation by dynamiccontrast-enhanced subtraction MRI. J Comput AssistTomogr 1996; 20: 9–14.

7. Orel S, Mendonca M, Reynolds C, Schnall M, SolinL, Sullivan D. MR imaging of ductal carcinoma insitu. Radiology 1997; 202: 413–20.

8. Soderstrom C, Harms S, Copit D et al. Three-dimensional RODEO breast MR imaging of lesionscontaining ductal carcinoma in situ. Radiology 1996;201: 427–32.

9. Bone B, Aspelin P, Bronge L, Isberg B, Perbeck L,Veress B. Sensitivity and specificity of MRmammography with histopathological correlation in250 breasts. Acta Radiol 1996; 37: 208–13.

10. Daniel B, Yen Y, Glover G et al. Breast disease:dynamic spiral MR imaging. Radiology 1998; 209:499–509.

11. Amano G, Ohuci N, Ishibashi T, Ishida T, Amari M,Satomi S. Correlation of three-dimensional magneticresonance imaging with precise histopathologicalmap concerning carcinoma extension in the breast.Breast Cancer Res Treat 2000; 60: 43–55.

12. Westerhof J, Fischer U, Moritz J, Oestmann J. MRimaging of mammographically detected clusteredmicrocalcifications: is there any value? Radiology1998; 207: 675–81.

13. Satake H, Shimamoto K, Sawaki A et al. Role ofultrasonography in the detection of intraductalspread of breast cancer: correlation with pathologicfindings, mammography and MR imaging. EurRadiol 2000; 10: 1726–32.

14. Hiramatsu H, Ikeda T, Mukai M, Masamura S,Kikuchi K, Hiramatsu K. MRI of ductal carcinomain situ of the breast: patterns of findings andevaluation of disease extent. Nippon IgakuHoshasen Gakkai Sasshi 2000; 60: 205–9.

15. Viehweg P, Lampe D, Buchmann J, Heywang-Kobrunner S. In situ and minimally invasive breastcancer: morphologic and kinetic features oncontrast-enhanced MR imaging. Magn ResonMaterials Physics Biol Med 2000; 11: 129–37.

16. Kuhl C, Mielcarek P, Leutner C, Schild H.Diagnostic criteria of ductal carcinoma in-situ(DCIS) in dynamic contrast-enhanced breast MRI:comparison with invasive breast cancer (IBC) andbenign lesions. Proc Intl Soc Mag Reson Med 1998;2: 931

17. Brown J, Buckley D, Coulthard A et al. Magneticresonance imaging screening in women at geneticrisk of breast cancer: imaging and analysis protocolfor the UK multicentre study. UK MRI BreastScreening Study Advisory Group. Magn ResonImag 2000; 18: 765–76.

18. Orel S, Schnall M, Powell C, Hochman M, Solin L,Fowble B, Torosian M, Rosato E. Staging ofsuspected breast cancer: effect of MR imaging andMR-guided biopsy. Radiology 1995; 196: 115–22.

12

167

12


19. Stomper P, Herman S, Klippenstein D et al. Suspectbreast lesions: findings at dynamic gadolinium-enhanced MR imaging correlated withmammographic and pathologic features. Radiology1995; 197: 387–95.

20. Heywang-Kobrunner S. Contrast-enhanced MRI ofthe breast. Invest Radiol 1994; 29: 94–104.

21. Abraham D, Jones R, Jones S et al. Evaluation ofneoadjuvant chemotherapeutic response of locallyadvanced breast cancer by magnetic resonanceimaging. Cancer 1996; 78: 91–100.

22. Fischer U, Kopka L, Grabbe E. Breast carcinoma:effect of preoperative contrast-enhanced MRimaging before re-excisional biopsy. Radiology1999; 213: 881–8.

23. Brown J, Coulthard A, Dixon A et al. Protocol for anational multi-centre study of magnetic resonanceimaging screening in women at genetic risk ofbreast cancer. Breast 2000; 9: 78–82.

24. Harms S, Flamig D, Hesley K et al. MR imaging ofthe breast with rotating delivery of excitation offresonance: clinical experience with pathologiccorrelation. Radiology 1993; 187: 493–501.

25. Ando Y, Fukatsu H, Ishiguchi T, Ishigaki T, Endo T,Miyazaki M. Diagnostic utility of tumor vascularityon magnetic resonance imaging of the breast. MagnReson Imag 2000; 18: 807–13.

26. Heywang S, Wolf A, Pruss E, Hilbertz T, EiermannW, Permanetter W. MR imaging of the breast with

Gd-DTPA: use and limitations. Radiology 1989; 171:95–103.

27. Kaiser W, Zeitler E. MR imaging of the breast: fastimaging sequences with and without Gd-DTPA.Radiology 1989; 170: 681–6.

28. Oellinger H, Heins S, Sander B. Gd-DTPA-enhancedMR breast imaging: the most sensitive method formulticentric carcinomas of the female breast. EurRadiol 1993; 3: 223–6.

29. Gilles R, Guinebietier J, Lucidarme O et al. Non-palpable breast tumours: diagnosis with contrast-enhanced subtraction dynamic MR imaging.Radiology 1994; 191: 625–31.

30. Boetes C, Barentsz J, Mus R et al. MRcharacterisation of suspicious breast lesions with agadolinium-enhanced TurboFLASH subtractiontechnique. Radiology 1994; 193: 777–81.

31. Nunes L, Schnall M, Orel SG, Hochman M, LanglotzC, Reynolds C, Torosian M. Breast MR imaging:interpretation model. Radiology 1997; 202: 833–41.

32. Kuhl C, Schmutzler R, Leutner C et al. Breast MRimaging screening in 192 women proved orsuspected to be carriers of a breast cancersusceptibility gene: preliminary results. Radiology2000; 215: 267–79.

33. Hiramatsu H, Enomoto K, Ikeda T et al. Three-dimensional helical CT for treatment planning ofbreast cancer. Radiat Med 1999; 17: 35–40.

168


Chapter

13The nature of breast tissue

calcificationsK. D. Rogers and R. A. Lewis

A crystallographic perspective 171

Crystallographic structures of biogenic calcifications 172

Characterisation of calcifications 175

X-ray diffraction 177

Synchrotron radiation studies 179

Breast tissue collagen organisation 182

Crystalline materials within breast tissues 183

Formation mechanisms 184

Work in progress 186

169

The nature of breast tissue calcifications

A crystallographic perspective

The formation of crystalline and semi-crystalline materials occurs extensivelywithin biological tissues due to natural bio-logical processes, disease, drug therapies andimplants. However, detailed crystallographicstudy of these organic and inorganic materi-als has generally been very limited, withperhaps the exception of bone. Traditionally,crystallographers determine and examinedetails of the regular atomic distributionsassociated with crystalline solids by exploit-ing diffraction phenomena. Measurementsof intensity distributions from diffractionexperiments are utilised to “solve” crystalstructures. Further information associatedwith structural disorder (e.g. crystallite sizeand “strain” and atomic substitutions), isalso available from diffraction data. Suchultrastructural characteristics have markedeffects upon macroscopic properties andbehaviour of materials such as reactivity andhardness. Crystallographic descriptions ofmaterials’ microstructures can provideinformation about formation mechanisms,formation environments and an indication ofhow the crystals will interact with theirenvironment.

Frequently, crystalline materials associ-ated with biological tissues are calcificminerals. These have been exploited diag-nostically when located in pathologicaltissues such as the brain1, thorax2, liver3,arteries4, joints5 and retina6. The most com-mon biological mineral, found in a widerange of crystalline forms and body loca-tions, is based upon the prototype structureof calcium hydroxyapatite [Ca10(PO4)6(OH)2].In vivo it occurs with extensive ionic sub-stitution, e.g. carbonate ions replacing theindigenous hydroxy or phosphate ions and,hence, it shall be referred to as b-HAP withinthis chapter. This material forms the bulk ofbone and teeth. As expected after many yearsof in vivo development, b-HAP has ideal

properties to behave as both an ion store andmechanical composite matrix. However, thisapatite is also the most common form ofdystrophic and metastatic mineral.

Other than bone and teeth, the most wellknown and studied human mineralisation isassociated with the urinary system (crystal-luria and urolithiasis), although thecrystalline constituents of uroliths are notexclusively mineral in nature7. Uroliths havebeen extensively structurally characterisedand found to consist of a wide range of crys-talline materials that most often combine toform elegant, laminar architectural macro-structures.

Currently, an important part of initialbreast cancer diagnosis is based upon themammographic appearance of radio-opaquedeposits commonly thought to be calcifica-tion. Such mammographically recognisedcalcification has been used, for some time,to indicate high-risk areas within breasttissues8,9 and increased specificity may begained from magnification of the images10.Indeed it has even been suggested thatmedial calcific sclerosis of breast arteries mayalso be diagnostic11 and that the shape of thecrystallites on mammograms can be instruc-tive12. The prognostic value of calcificationhas also been examined and a significantlygreater number (twice as many) of lymphnodes appear to be involved with tumours ofpatients with calcification13. This has sug-gested the possibility of calcium depositionbeing related to tumour cell metastasis.

Despite extensive diagnostic use made ofthese deposits (albeit with a relatively lowspecificity14), the detail of their chemical andstructural composition has not been fullydetermined. Thus, the clinical significance ofbreast tissue calcifications as diagnostic indi-cators of tumour type and stage is far frombeing fully explored. Further, there is littleevidence correlating the type of calcificationto its mammographic appearance, e.g. thelinear and granular distributions that have

13

171


been correlated to tumour subtype15. Alsoany relationship of calcific nature to histo-logical classifications, such as that byMaria Foschini16, has not been established.However, several studies have suggestedthat the type of calcification formed may actas a marker for malignancy and that simplythe presence of calcification may be of biolog-ical significance; 5- and 20-year survival ofpatients with calcifications has been found tobe significantly less than those without17,18.

Currently, there are several contradictoryand limited studies examining the diagnosticability of the chemical nature of calcifica-tions. Undoubtedly there is significantexperimental confounding within andbetween these studies due to differences insampling methods, specimen preparation/storage and analysis techniques. Crystallo-graphic characterisation of these calcifi-cations would certainly go some way toresolve many of these issues.

The aim of this chapter is to review thecrystallographic data related to breast tissuecalcifications and examine the implicationsand value of this information for (a) breastcancer diagnosis and (b) understanding thenatural history of the disease. This review is,for the most part, observational but placedwithin an appropriate context. The processesand language associated with crystallogra-phy and diffraction are also introduced.Much of the initial background is presentedin relation to b-HAP as this undoubtedlyforms the majority mineral of breast tissuedeposits.

Crystallographic structures ofbiogenic calcificationsThe term “crystalline” is used to indicate thepresence of long-range atomic order within amaterial. Within highly crystalline materials,such as silicon wafers used by the semicon-ductor industry, the regular arrangement of

atomic units can persist over several centi-metres and crystallographic investigationscan be performed using single crystals. Amore common material form, “polycrys-talline”, occurs when the crystal size(typically 0.1–100 µm) is many times smallerthan the interrogating X-ray probe and many,randomly oriented, crystallites are presentwithin the sample. Thus, although a highdegree of atomic order exists within eachcrystallite, intercrystallite orientational orderis absent. However, in some solid materials,the atomic order persists only over a singlemolecule (e.g. glasses) and such materials arereferred to as “amorphous”. This should notbe confused with the frequent use of thisterm to describe the visual and mammo-graphic appearance of breast tissue deposits;crystalline precipitates often form with an ill-defined visual morphology. Between thesingle crystal and amorphous extremes areintermediate states including “nanocrys-talline” materials in which the crystallite sizeis only a few nanometers. In fact, bone min-eral may be considered such a material andits extensive ultrastructural disorder (typicalof all b-HAP) described by a “paracrys-talline” model19.

Biological calcifications are typicallytreated as polycrystalline, although it hasbeen asserted that in vivo precipitation ofamorphous calcium phosphate is a precursorstage in the formation of b-HAP.

Diffraction data (upon which crystallo-graphic characterisation is based) from singlecrystal and polycrystalline forms are sig-nificantly different in nature and in theinformation that they can provide. For exam-ple, single crystal studies tend to providerelatively accurate and precise atomic posi-tion and vibrational data, whereas data frompolycrystalline experiments, although possi-bly not as precise, may also providecrystallite morphology and other microstruc-tural information. In particular, the “phase”(see following paragraph) of a material is

13

172



routinely identified from polycrystallinedata.

Conventionally, three-dimensional, atomicdistributions in crystalline materials arereduced to a description of (a) themorphology of the molecular repeating unitor motif, known as the “unit cell” anddefined by “lattice parameters”, and (b) theatomic species and atom positions (or“sites”) within a single unit cell. These con-cepts are illustrated, with reference tocalcium hydroxyapatite, in Figure 13.1.Different crystalline materials are distin-guished by differences in unit cell morphol-ogy and/or cell contents. Unique atomicdistributions and cell morphologies areidentified as “phases”. Thus calciumhydroxyapatite, calcium oxalate monohy-drate (COM) and calcium oxalate dihydrate(COD) are unique crystalline phases.However, within each crystalline phase,subtle differences in stoichiometry (e.g. thefraction of atomic sites occupied and atomicsubstitutions) and lattice parameters fre-quently exist. This is certainly the case for b-HAP that is almost exclusively found withhexagonal unit cell morphology – a phase

form of hydroxyapatite that is only sta-bilised by the inclusion of impurity atoms20.

Usually, changes to unit cell contents effectthe lattice parameters.* For example, in therange of biological interest, carbonate substi-tution for the PO4 ions of calcium hydroxya-patite causes a contraction of the a-axis by 6.2 ×10–4 nm/wt% and an expansion of the c-axisby 2.3 × 10–4 nm/wt%21. Other ionic substi-tution effects in hydroxyapatite are indicatedin Table 13.1. Similar effects are also observedwhen the hydration state of a material ischanged, for example in calcium oxalates.Thus, accurate determination of lattice para-meters can provide information on the precisenature of the material. However, an inherentdifficulty of employing lattice parameters forinterpretation of b-HAP stoichiometry, is theconfounding effect of multiple heteroionicsubstitutions that often must occur for chargebalance, e.g. when CO3

2– is substituted forPO4

3– in calcium hydroxyapatite, then a sub-stitution of Na+ for Ca2+ or loss of Ca is oftenobserved20. Heteroionic substitutions in b-HAP are extensive. For example, a detailedstudy22 of an enamel apatite determined itsstoichiometry to be,

13

173

Na+ substituting for Ca2+

unit cell

0.1 nm

latticeparameters

ac

b

CO32– substituting for PO4

3–

Ca2+ vacancies

Fig. 13.1Perspective view of HAP crystallographicstructure as seen along the c-axis (along theO-H channels).

∗ It should be realised that much of the crystallographic data concerning hydroxyapatite is derived from thesynthetic preparations of the mineral. Interpretation of structural data corresponding to b-HAP is derivedfrom these synthetic studies.


(Ca3.89Na0.09Mg0.02)I(Ca5.59Na0.13Mg0.03)II(PO4)5.5(CO3)0.5(OH)1.78

(the I & II subscripts referring to different Casites within the unit cell).

Small changes to stoichiometry and crys-tallite microstructure often have a significantimpact upon crystallite–environment inter-actions. For example, performance criticalparameters of b-HAP crystallites withinbone include their morphology and amountof carbonate substitution. Morphology isimportant for enzyme interactions, as theseare specific to particular growth surfaces ofhydroxyapatite23. Increased carbonate sub-stitution has been shown to enhancesignificantly the reactivity and dissolution ofsynthetic hydroxyapatite in vitro24. Further,the inflammatory properties of crystalliteswithin tissues have been shown to dependcritically upon the chemical and microstruc-tural characteristics (e.g. specific surfacearea) of the crystals25. The ability of HAPcrystals to lyse cell membranes has a similardependence26.

The formation conditions associated withcalcifications within breast tissues are impor-tant to assess; the extracellular environment

in which crystallites form significantly affectsboth the unit cell contents and crystallitemorphology. Thus, crystallographic detail ofcalcifications may be exploited to indicatelocal, dynamic changes in tissue physiologyand/or metabolism, i.e. disease onset. Invitro studies have shown that changes to themorphology of growing crystallites occur inresponse to changes in environmental pH27.An early study28 indicated a correlationbetween the nature of the calcificationsformed and breast tissue pH. The pH is sig-nificantly higher within breast ducts thansurrounding tissues and b-HAP direct pre-cipitation requires an alkaline environment.Further, calcification morphogenesis mayindicate if direct precipitation or phase trans-formation (for example from amorphouscalcium phosphate) is occurring. Alsocrystallite morphology is influenced by pref-erential adsorption of proteins. In particular,HAP has a high affinity for protein adsorp-tion but this is often associated withparticular crystal faces. For example, boneacidic proteins have been shown to adsorbpreferentially upon faces normal to the crys-tallite c-axis where calcium ions present a

13

174

Ion Substitution Lattice parameter changes

CO32– PO4

3– a1 c2

CO32– OH– a2 c1

F– OH– a1 c3

Cl– OH– a2 c1

Sr2+ Ca2+ a2 c2

Mg2+ Ca2+ a2 c1

Mn2+ Ca2+ a1 c1

Cd2+ Ca2+ a1 c1

Na+ Ca2+ a3 c3

Pb2+ Ca2+ a2 c2

1,2 indicates an increase and decrease respectively, 3 indicates no measurable change.

Table 13.1 Qualitative description of lattice parameter changes in HAPresulting from ionic substitutions



regular arrangement23. A consequence of thisis preferential growth along the c-axis.

It is highly likely that the histopatho-logical appearance of breast calcifications isdirectly influenced by the precise nature ofmineral. For example, the granular and lami-nar calcifications of DCIS characterised byFoschini et al.16 are clearly indicative ofdifferent crystallite growth environmentsand growth processes. Thus the crystallites’crystallographic features will inevitably bedifferent. Although not specific to breast tis-sues, a comprehensive review of biologicalcrystal formation in pathological tissues isprovided by Daculsi et al.29.

Characterisation ofcalcificationsWhen considering the microscopic and ultra-structural characterisation of biologicaltissues, it is important to appreciate theprecise nature of information generated byvarious analytical methods. Traditionally,histopathological examination of tissuesdoes not warrant the use of techniques thatprovide high specificity for crystalline phase.Stains such as Von Kossa and alizarin red,frequently used in the examination of breasttissues, are able to indicate high calciumconcentrations and even provide a degreeof phase differentiation through the use offurther stains such as silver nitrate/rubeanicacid and H&E safranin. However, specificityis low and it has been suggested that tradi-tional histopathological techniques can alterthe hydration states of materials. Further,specimen storage methods may have a sig-nificant influence upon the inorganic tissuedeposits30. The optical appearance of calcifi-cations is certainly not phase-specific and,even using polarising microscopy, it can bemisleading, for example when birefringencyis thought specific for COD12. A critical eval-uation of the use of polarising microscopy for

breast tissue calcification has recently beenprovided by the mineralogist Jill Pasteris,who concluded that there was significantmisapplication of technique and misleadingmineral identification within pathologyliterature31.

Scanning electron microscopy (SEM) is anincreasingly common and versatile analysistool that can be applied to histopathologicalslides with little specimen preparation.However, in its usual imaging mode it suffersfrom the same specificity limitations as opti-cal microscopy although it has significantlygreater sensitivity. It has been recentlyshown32 that SEM may detect calcificationsmissed by histopathology and, thus, it ishighly likely that the frequency of calcifica-tion within breast and other tissues issignificantly greater than currently reported.SEM studies tend to report crystallinedeposits as a range of forms, from largeaggregates to very small, fine needles. Thisclearly indicates differences in formationmechanisms and/or environments.

Most electron microscopes are modified toaccommodate detectors that measure X-raysproduced from the interactions betweenprobe electrons and atoms within the sample.The energies of these X-rays are characteristicof the atoms within the specimens and thusthis technique (known as electron beammicroprobe, EBM, or X-ray microanalysis)provides information concerning the atomicspecies within tissues. It is occasionally,and unfortunately, confused with X-raydiffraction33. Although EBM is not easilyquantifiable, it has been previously appliedto breast tissue calcifications34 where Ca/Pratios ranged from 1.5 to 1.7. Often it ismistakenly suggested that the technique isspecific for phase; for example, calciumphosphates are distinguished from calciumoxalates simply by the presence or absence ofP. However, the technique operates inde-pendently of the nature of the material;crystalline and amorphous solids are not

13

175


distinguished. A notable, extensive study ofbreast tissues was undertaken by BenjaminGalkin35 who examined discrete tissuedeposits from mastectomy specimens. Thestudy surprisingly concluded that many ofthe smaller deposits did not contain any cal-cium but a wide range of elements such assilicon and iron. The significance of this workremains unclear and the role of such depositsin calcification development is unknown.However, as a note of caution, due to thehigh sensitivity of this technique the poten-tial for contamination from specimenprocessing is high.

A third type of electron microscope exam-ination is the use of electron diffraction.Although specimen preparation is difficult,this method is specific for crystalline phase.However, its application to breast tissuedeposits has been very limited. Tornos et al.36

applied a combination of techniques thatincluded electron diffraction (to a singlespecimen) from which calcium oxalatemonohydrate crystals were identified. Asimilar, earlier study37 of a single histo-pathological section containing crystallitesindicated a good crystallographic match to b-HAP.

A technique often used by mineralogistsand, in particular, crystallographers study-ing bone mineral, is that of infrared (IR)spectroscopy. Through the determination ofpreferential IR absorption bands by bondedatoms, specific molecular species can beidentified. This has been previously appliedto breast tissue calcifications to demonstrate,in three cases, the presence of calcium oxalatedihydrate38. Recently, and increasingly, therehas been a more extensive use of IR spec-troscopy for examining biological systems, asthe analysis probe can be formed within anoptical microscope system39. Recent work40

has demonstrated significant differencesbetween the organic components of normaland malignant breast tissues. However, dueto the complex nature of the specimens, it is

difficult to assign differences to specificchanges in the tissue chemistry.

The greatest phase specificity undoubt-edly derives from diffraction methods. Themost common diffraction probes are X-rayswith a wavelength of approximately 0.1 nm,although neutron diffraction is generally alsowell utilised. However, X-ray diffractionmeasurements have rarely been performedon breast tissue materials, although they routinely provide definitive phase informa-tion within a wide range of fields such asmineralogy, forensic and materials science.Further details are provided in the followingsection.

In summary, and from a crystallographicperspective, despite the probable signifi-cance of phase and microstructuralcharacteristics of deposits forming withinbiological tissues, there have been very fewdefinitive studies, using appropriate (phase-specific) analytical techniques, of breastcalcifications. This has resulted in a commonbelief, propagated through contemporaryliterature, that the phases of breast calcifica-tions are well known. The principal sourcesof evidence upon which this belief is basedare summarised in Table 13.2 and comprise92 patients from a narrow target population.None of the experimental data were collectedwith sensitive detectors and no analysisbeyond phase identification has ever beenperformed. Further, most of the specimensexamined were only representative of“mature” calcification.

X-ray diffractionX-ray diffraction is a coherent scatteringprocess that occurs when X-ray photons passthrough any material. In fact, these scatteredphotons and the information containedtherein are produced during all routine radio-logical examinations. However, as they donot directly contribute to image formation,they are disregarded. Crystallographically,

13

176



all the structural information is provided bythese photons.

For examination of biological mineralsthere are several key characteristics that canbe derived from diffraction data. As is typicalof many analytical techniques, X-ray diffrac-tion data from polycrystalline materials maybe reduced to a pattern (“diffractogram”)containing intensity maxima at specificangular positions, θ, around the specimen(see inset of Fig. 13.2). The angular positionof each diffraction maxima is related simplyto the interrogating X-ray wavelength, λ, andthe perpendicular distance between atomicplanes within the sample material, d,through Braggs’ Law,

λ = 2dsinθ.

The integrated intensities and peak positionsare unique for each crystalline phase andthus may be exploited for phase identifica-tion. Even when the chemistry is similar, e.g.COD and COM, the diffraction patterns arequite distinct (see Fig. 13.2). Accurate deter-minations of lattice parameters can beobtained from the peak positions and thussmall changes in stoichiometry revealed and

quantified. Modern digital technologies (e.g.area or “moving point” detectors) for record-ing diffraction data have all but replaced filmas a recording medium and thus errors fromfilm processing, insensitivity etc. have beensignificantly reduced.

Microstructural information is inherentwithin the shape of each diffraction maxima.If crystallite dimensions are less than about0.2 µm (which is the case for most b-HAP)then the diffraction peaks become broadenedin a systematic manner. At these dimensions,optical microscopy and even SEM becomesvery limited. Anisotropic broadeningbetween different maxima may be employedto determine average crystallite morphology.A confounding effect, however, is that ofsmall, local variations in lattice parameters(referred to as “microstrain”) caused by dis-ordering effects such as stoichiometricgradients, which also broaden the diffractionmaxima. A diffraction peak may thus bebroadened by finite crystallite size or micro-strain, or more likely for biologicalcalcifications a combination of both. Forexample examine the bone diffraction datapresented in Figure 13.2 compared to that of

13

177

Experimental probe Recording medium Specimens Reference

X-ray diffraction Film Three cases, all carcinoma 28

X-ray diffraction Film 11 patients, malignant and 60benign

X-ray diffraction Film 49 patients, range of 43pathologies

X-ray diffraction Film 23 patients, range of 44pathologies

Electron diffraction Film One specimen (?), carcinoma 37

Electron diffraction Film One specimen, preselected 36 birefringent

Infrared spectroscopy Digital Three cases (sections), single 38crystals extracted

Table 13.2 Investigations where phase specific methodologies have been employed to examine breasttissue calcifications



the synthetic calcium hydroxyapatite. Thepeaks within the bone data are markedlybroader than those of the structurally more“perfect”, stoichiometric synthetic material.Fortunately, there are several analyticalmethods that may be applied to diffractiondata to separate these effects and provideestimates of crystallite size and microstrain41.

Diffraction data are also sensitive to theorientation distributions of crystallites withspecimens. This is important, for example, inthe mechanical functionality of bone thatis dependent upon the preferred growthdirections of the mineral crystallites.

Finally, the integrated intensities of dif-fraction peaks contain information con-cerning the contents of the average unit cell.Thus, by using these data, a complete struc-tural model of the unit cell may be derivedand refined. In the case of calcifications it isthe model refinement that is of most value, asthe structural prototype for b-HAP is wellknown. The refinement process for polycrys-talline materials is commonly known as theRietveld method and information about pre-cise atomic positions (and thereforedistortions) and site occupancies (e.g. Cadeficiencies) may, in principle, be obtained.

13

178

Fig. 13.2Diffraction data from relevant crystalinephases. Inset shows the conventional X-ray diffraction geometry.



Although diffraction analyses are utilisedextensively outside the field of medicine,there are limitations in their use applied tobiological tissues. Although diffraction ishighly phase-specific, with conventional,laboratory X-ray sources, the technique is nothighly sensitive and only provides spatialaverages. This can lead to ambiguity whenanalysing homogeneous materials such as b-HAP. With optimum laboratory conditions aphase determination of calcific materialsrequires at least ~10 µg of material. In thecontext of breast tissue calcifications, how-ever, the total mass of mineral in, forexample, a typical core-cut biopsy specimen,is relatively small and may be dispersedthrough a relatively large mass of organic tis-sue. Further, with a small crystallite size andlow structural symmetry (resulting in largenumbers of diffraction maxima), unambigu-ous phase identification and microstructuralcharacterisation is challenging, due to sig-nificant numbers of overlapping diffractionpeaks. Undertaking diffraction experimentsusing pathological slides is not feasible, dueto the limited mass of crystalline materialand because the embedding wax (even whenapparently removed using xylene) producesa strong diffraction signal masking that of thedeposits. However, early work using X-raydiffraction42,43 examined mammary calcifica-tions removed by microdissection. COD wasidentified from the diffraction data, althoughactual specimen numbers and any crystallineHAP identification were not made clear ineither of the studies. Further early diffractiondata were also collected from dissecteddeposits and were determined to consist ofCOD, HAP and amorphous calcium phos-phate. However, the distinction betweensmall quantities of truly amorphous andpoorly crystalline materials is very difficult.Perhaps the most comprehensive and sensi-tive X-ray diffraction work to date was thatof Fandos-Morera et al.44. Using a specialiseddiffraction camera, Fandos-Morera et al.

examined crystals that had been chemicallyisolated (aggressively?) from breast tissue.Phases identified were b-HAP, calciumoxalate (hydration state not provided), oxalicacid, calcite and aragonite.

Thus although, in principle, diffractioncan provide a large number of mineral char-acteristics; in practice for biogenic deposits, agreat deal of care is required, both in data col-lection and interpretation. However, it is oneof very few, truly phase-specific methodolo-gies and thus one that must be adopted if thisinformation is required. Therefore, recentresearch has involved the use of specialistradiation sources such as synchrotron radia-tion described below.

Synchrotron radiation studiesThe use of synchrotron radiation (SR) is notwidespread in the field of medicine and, infact, few health care professionals have evenheard of it. SR sources provide multiple,extremely intense and tuneable beams ofphotons over a wide range of energies fromIR through to hard X-rays. Its advent hasrevolutionised many experimental tech-niques and SR is increasingly being appliedacross many fields from macroscopic imag-ing to molecular dynamics. It has spawnedseveral methods for studying live and wettissue samples yielding information on bothstructure and composition on all lengthscales down to atomic resolution. Such tech-niques have played a crucial role in thedevelopment of molecular biology and thesolution of protein structures. The applica-tion of SR in the field of radiology is nowexpanding and it is clear that very substantialimprovements in image quality and patientdose can be realised.

Although there are currently more than 50SR sources world-wide (for example, see Fig.13.3), the use of SR for medical research isstill for the most part, in its infancy. Theapplications of SR to medicine are many and

13

179


varied. The use of SR sources offers thepotential for considerable improvements insome of the techniques already routinelyused in medicine, and also makes possibleentirely new procedures. Some of these tech-niques can be applied in vivo, whilst othersare more suited to in vitro studies that aretherefore potential constituents of the pathol-ogist’s tool kit.

The properties of synchrotron radiationcan be summarised as follows:

● very high intensity● a broad and continuous spectrum from

infrared to X-rays● natural collimation● small source size● high polarisation● pulsed time structure.

It is the combination of these properties thatmakes SR a unique and vastly superior X-raysource than conventional sealed tubes for avery wide range of scientific and technicalapplications. In particular, the high-intensity,natural collimation and continuous spectrummakes possible the production of intense,tuneable, monochromatic beams of radiation

which simply cannot be produced in anyother way. The selection of a small wave-length band from the broad white radiationspectrum is achieved by using monochroma-tors fabricated from “perfect” single crystalsthat exploit Braggs law (see the section on X-ray diffraction).

The traditional and most frequent use ofSR has been for X-ray diffraction experi-ments. These have played a major role in thedevelopment of molecular and structuralbiology, areas that are set to expand further.The human genome project alone is likely toidentify in excess of 10 000 proteins, many ofwhich will have to be structurally analysedin order to determine their function. Thestructure of a protein has a major effect uponits functionality and knowledge of proteinsequence is not sufficient to predict itsfunction. Molecules pack in crystals in a non-random manner and are governed by thesame rules that apply when the moleculebinds to its macromolecular receptor. Theresults of high-resolution structure analysesbased upon X-ray diffraction studies of crys-tals, provide information that is invaluablefor modelling drug–receptor binding. Drug

13

180

Fig. 13.3The European Synchroton RadiationFacility sited at Grenoble, France



design on the basis of this information isin its infancy, but has accelerated in recentyears as a deeper understanding of therequired structural principles has begun toemerge.

SR X-ray diffraction is not restricted to thestudy of crystalline materials but can alsoyield a wealth of information on fibroustissues and even solutions, where small andweakly scattering or dilute samples can stillproduce useful diffracted intensities. Mostbiological systems have some degree of spa-tial ordering and SR X-ray diffraction can stillbe used to provide extremely useful struc-tural information. This type of diffraction isknown by the generic term of non-crystallinediffraction, and there are a large number ofmedically significant projects taking placethroughout the world, including studies ofprotein folding, tubulin assembly and trans-dermal drug delivery. Specific examples ofstudies of fibrous biological tissues includeexaminations of the cornea45 (which has astromal region of stacked lamellae, eachbeing composed of a parallel array of colla-gen fibrils) and muscle46. A further examplerelevant to breast structures is described ingreater detail below.

Radiological applications of SR haverecently included development of severalimaging modalities. These include coronaryangiography (including dichromography),bronchography, multiple energy computedtomography, and fluorescent computedtomography. All have been shown to pro-duce high quality images in vivo, with futureexploitation expected to provide newinsights into diagnosis and disease progres-sion. Radiotherapy applications, exploiting anovel approach using multiple SR micro-beam applicators, have shown remarkableresults in terms of cellular survivability whencompared to conventional methods. Both theimaging and therapeutic applications of SRhave been comprehensively reviewed byLewis47.

Mammography with SR

Using conventional mammography, lesionsof less than 2–3 mm are very difficult todetect. The dedicated mammography X-raysources are operated at 25–30 kVp. The spec-trum from such sealed tubes consists of twofluorescence lines at 17.4 and 19.6 keV super-imposed upon a Bremsstrahlung continuum.The characteristic lines themselves representonly 25% of the total flux. Whilst the opti-mum energy for mammography is known tobe in the region of 17–21 keV (it variesslightly with breast thickness and density),the continuum of radiation above 20 keVproduces a diffused background in theresulting radiograph that serves only todiminish contrast. It was reported in 199248

that tuneable monochromatic SR beamsoffered the potential to image the breast withhigher contrast and perhaps a lower dosethan is possible with a conventional mam-mography set. The basic principle isillustrated in Figure 13.4(a). The monochro-maticity of SR results in all the incidentphotons usefully contributing to the imageand the inherent collimation of the beamallows slits to be placed both before and afterthe breast thereby greatly reducing scatter.The image is constructed by scanning thepatient or possibly the beam.

The first tests on phantoms and excisedbreast tissue, performed at the Adone SRsource in Frascati49, demonstrated thatimages of higher contrast could be recordedwith a similar or lower dose than on a con-ventional mammography set. Subsequentwork with phantoms50 has shown that 20-keV SR images have comparable contrast tothose obtained in conventional mammo-grams but with a skin dose that is 10 timessmaller. At 17 keV, substantially higher con-trast is obtained with less than half the skindose for a conventional mammogram. Ourwork on the UK’s SR source at Daresbury hasrecently confirmed these results51. To fully

13

181


establish the potential of this technique, adedicated mammography beamline is beingbuilt on the “ELETTRA” SR machine atSincrotrone Trieste, Italy52.

A very exciting recent development is thelatest work at the National Synchrotron LightSource (USA). Termed diffraction-enhancedimaging (DEI), a second analyser crystal isplaced after the breast tissue which thentransmits only those photons having a spe-cific energy (see Fig. 13.4b). Not only doesthis system have excellent scatter rejection,it has been used to produce some quitespectacular contrast enhancements by usingthe analyser to select those photons whichhave been very slightly deviated from thebeam53.

It has been demonstrated that with SR,and particularly DEI, greatly improved spa-tial and contrast resolution can be achievedand thus there is the potential to detect andcharacterise calcifications during mammog-raphy. This may therefore result in highersensitivity and specificity for diagnosis,based upon appearance rather than simplydistribution of the calcifications.

Breast tissue collagenorganisationIn a current research programme at theDaresbury SR source, breast tissue samplesare being studied by X-ray diffraction. It ishoped that the work will determine whetherdifferences at the molecular level, in the tis-sue surrounding tumours, can be observedby diffraction techniques. Work is concen-trating on collagen structure, calcificationsand adipose tissue. Recently we have beenexamining collagen structures within breasttissues using small angle X-ray scattering(SAXS) to characterise collagen’s supramole-cular ordering. The rationale for this work isbased upon observations that breast tumoursdisplay aberrant collagen organisation in thetumour stroma54,55. Furthermore, extensivealteration of the extracellular matrix has beenobserved in invasive colorectal carcinomaswhere the collagen derangement is attrib-uted to both enzymatic degradation andaltered neosynthesis.

Our results thus far have demonstratedthat highly significant differences can be

13

182

Fig. 13.4Mammography using Synchrotonradiation. Both of the systemsillustrated utilise a tuneablemonochromatic beam and slits toreject scattered radiation. In (b) ananalyser crystal is used to enhancescatter rejection and improvecontrast.



detected in the collagen order and atomicspacings of normal, malignant and benignbreast tissues56.

Perhaps, more importantly, it appears thatthese differences are detectable several cen-timetres away from the lesion. An example ofthe SAXS data is presented in Figure 13.5,which illustrates the marked differencebetween normal and malignant breast tis-sues. We are currently pursuing further workinvolving larger patient numbers and awider range of tumour types.

Crystalline materials withinbreast tissuesThe difficulty in describing the crystallo-graphic details of breast tissue calcificationis that they are simply unknown. The realsignificance and formation mechanisms ofbreast calcifications are also unknownalthough it has been suggested that thedifference between lamellar and granulararchitectures is due to calcium precipitationin either proteinaceous or cellular environ-ments respectively16. Thus only a briefoutline of current evidence is providedbelow. There are probably a number of differ-ent formation mechanisms in breast tissues,some passive (related to necrosis) and someactive. These inevitably produce a rangeof phases and mineral stoichiometries.

Generally, formation processes and modelshave been inferred from the location of thecalcifications within tissues rather than tissuechemistry and calcification crystallography.Perhaps the most fundamental (and, on thewhole, currently lacking) crystallographicinformation is simply the phase of materialbeing formed.

Specific interest in the crystalline phasesforming within breast tissues has previouslyarisen, principally due to:

1. Understanding discrepancies betweenradio-opacities seen on mammogramsand those observed during histopathol-ogy. This has been ascribed partly due todifficulties observing what is assumed tobe COD in biopsy specimens (normalhistopathological stains are ineffective)and its apparent preferential loss duringsectioning57.

2. Assessment of the diagnostic value ofcrystalline phase. It is often assumed thatbreast calcification phase is either (i) cal-cium oxalate dihydrate, formed perhapsas a result of secretory activity of epithelialcells58 and/or neoplasia, or (ii) dystrophiccalcium hydroxyapatite59. However, care-ful examination of the definitive evidence(i.e. phase-specific) upon which thisassumption is founded (see Table 13.2)reveals several unsatisfactory features of

13

183

Fig. 13.52-dimensional, small angle X-rayscattering distributions from breast tissuecore-cut biopsy specimens of (a) normaltissues and, (b) malignant tumour tissue.Note the lack of well defined diffractionmaxima in (b) indicating the loss ofultrastructural order. The darkrectangular central region is a leadabsorber to stop the direct beamreaching the detector.


the data. These include very small speci-men numbers, confounding specimenpreparation methods and highly selectivetissue types (only very mature/largedeposits have thus far been studied).

The association of calcium oxalates withbenign disease or LCIS is in contrast to anassociation of b-HAP with both benignand infiltrating carcinomas.60 Thus somediagnostic value may be directly attribut-able to phase. However, the detaileddiffraction work of Fandos-Morera44 didnot coincide with this generally held view.Perhaps a further difficulty here is thecomplete lack of any genuine “control”data arising from non-diseased breasttissues.

3. Examining the effects of silicone pros-thetic devices to determine the likelihoodof silicate deposition and toxicity of siliconbased minerals31,61.

The b-HAP of breast tissues is undoubtedlynon-stoichiometric and it is highly likely thatthe “amorphous calcium phosphate” oftenreferred to within current literature is actu-ally crystalline b-HAP. It is probably the mostfrequently observed calcific phase withinbreast tissues, where it may appear asopaque, basophilic deposits with small crys-tallite size. It has also been referred to59 as a“type II” calcification and exclusively (proba-bly mistakenly) associated with infiltratingcarcinoma61. It has been suggested that b-HAP is the result of any pathologicalalteration of breast tissue, whereas theformation of other phases may suggest aspecific tissue change.

In contrast, calcium oxalate dihydrate(mineral name “weddellite”), probably tendsto form relatively large polyhedral and bire-fringent crystallites. It has been referred to as“type I” calcifications and its presence hasbeen associated with lobular carcinoma insitu or benign lesions (including apocrinemetaplasia). Although its most common for-

mation route is through oxalic acid, this israrely observed to crystallise in breast tis-sues. This, of course, may be a consequenceof the non-phase-specific analytical tech-niques employed.

Although calcium oxalate dihydrate is anend point of metabolism (there are no oxalatedegenerative enzymes in humans), oxalateresorption has been demonstrated in renaland postmortem thyroid tissues62. Any suchresorption would inevitably result in tissuechanges in response to the liberation of toxicoxalate ions. Calcium oxalate dihydrate ismore unstable thermodynamically than themonohydrate phase. In other biosystems,COD has been shown to convert to the mono-hydrate forms unless stabilised63. Perhapssurprisingly, calcium oxalate monohydrate israrely reported to be found within breast tis-sues. Within the thyroid it has beensuggested that examination of (what isassumed to be) COD, features may indicatethe functional state of lesions64. Within occu-lar tissues, dystrophic oxalosis has beenascribed to the presence of intraocular ascor-bic acid65 and activity of the retinal pigmentepithelium66.

Through a careful examination of previ-ous investigations, there would appear to besome evidence for a range of phases formedwithin breast tissues. These are listed in Table13.3. The majority of these are not regularlyreported, due to the low specificity of themethods routinely employed to examinebreast tissues and, perhaps, destructive spec-imen processing techniques. If indeed thereis a wide range of crystalline phases beingformed then undoubtedly specific crystalli-sation mechanisms are responsible.

Formation mechanismsFor any physiologic or pathologic crystaldeposition, a sequence of fundamentalevents involving tissue changes must occur.These include a local increase in ionic

13

184



species, a change in pH and the formation ofa deposition nucleator or promoter68. Thenature of these events in breast tissue has notbeen established. Pathologic b-HAP is oftencorrelated with the presence of membranedebris such as that found associated with cellnecrosis. However, in many studies thecause–effect direction has not been estab-lished; studies have been unable todetermine if crystals cause necrosis ornecrotic tissue facilitates crystal growth67.Further, within breast tissues, many of thecrystalline deposits, particularly formedfrom relatively small crystallites, are notassociated with areas of necrosis, although acharacteristic of all necrotic breast tissue isthat it absorbs calcium.

The apparent similarity between the b-HAP formed within breast and that of osteotissues, has prompted several studies exam-ining common features of the formationmechanisms. However, calcific phase deposi-tion in breast tissues is unlike “normal”ossification, as breast calcification is notuniquely associated with collagen and, inany case, the collagen of breast carcinomatissues does not possess the structuralorder required to support ossification68,69.

Nonetheless, several bone matrix proteinshave been identified at raised concentrationsin malignant lesions and calcified tissues.Macrophages rather than tumour cells, thathave been shown to express osteopontin70

(this has a high binding affinity for HA), havealso been associated with necrotic areas ofbreast tumours, although this is not uniqueto breast tissues71. Further, it has beenrecently demonstrated72 that breast cancercells express bone sialoprotein (normallyassociated with bone mineralisation andremodelling) and this may be used as apotential indicator of the ability of the cells tometastasise to bone. An early stage markerfor this is likely to be calcification.Intriguingly, it has also been suggested thatsuch bone matrix proteins, produced bybreast cells, may further be responsible forthe high osteotropism of breast cancer cellsand a molecular basis for this, based uponinteractions between metastatic cancer cellsand osteoclasts, has recently been pro-posed73.

Other indicators of specific formationmechanisms that involve active protein par-ticipation may be derived from elementalanalyses of specimens; many enzymes

13

185

Chemical name Chemical formula Mineral name Reference

Calcium hydroxyapatite Ca5(PO4)3(OH)1 Dahllite 60

Calcium oxalate dihydrate Ca(CO2)22H2O Weddellite 60

Calcium oxalate monohydrate Ca(CO2)22H2O Whewellite 36

Calcium oxalate Ca(CO2)2 – 44

Calcium carbonate CaCO3 Calcite 44

Calcium carbonate CaCO3 Aragonite 44

Oxalic acid C2H2O4 – 44

Tricalcium phosphate Ca3(PO4)2 Whitlockite 28

Amorphous CAP CaxPy – 60

1 Refers to b-HAP, dahllite being carbonate substituted HAP. Where multiple reference sources are available(e.g. in cases when b-HAP is reported), only one is provided.

Table 13.3 Phases reportedly identified within breast tissues


require metallorganic cofactors. Such studieshave shown that several elements, includingMg and Zn74, are significantly elevatedwithin cancerous tissues. It is well knownthat Mg has an inhibiting effect on the con-version of amorphous calcium phosphate tohydroxyapatite. Other elements such assilicon may also be implicated in specificformation mechanisms.75

Another facet to breast tissue–calcificationinteractions has been developed through thecareful examination of calcium phosphatecrystallites associated with osteoarthriticjoint disease. Here a clear link between thepresence of basic calcium phosphate crystals(these include b-HAP, octacalcium phos-phate and tricalcium phosphate) and thedegree of cartilage degeneration has beenestablished76. The causal nature of this linkhas been shown to be probably due to theapparent ability of b-HAP crystallites toinduce mitogenesis and secretion of matrixmetalloproteases77,78. It has also been demon-strated that human articular cartilage matrixvesicles can be stimulated to produce crys-talline calcium pyrophosphate dihydrate ifexposed to adenosine triphosphate (ATP)and b-HAP crystals without ATP79,80.Recently, a genetic basis for control of tissuecalcification through regulation of cellularpyrophosphate levels in a mouse model hasbeen postulated81. The consequences of theserecent findings for calcification of breast tis-sues are unclear. No pyrophosphate phasehas ever been reported observed withinbreast tissues.

In summary, breast tissue is a complex,dynamic chemical environment that, in prin-ciple, enables the formation and resorption ofa range of inorganic crystalline deposits. It isclear that currently there are no satisfactorymechanistic models for the formation ofbreast tissue calcifications and no definitivedata relating the deposition of specificinorganic phases to the aetiology and devel-opment of breast disease. There is, however,

a growing body of circumstantial evidencethat may form the basis of such a model andthere is certainly a central role to be playedby the calcification ultrastructure within anysuch model.

Work in progressThe crystallography of breast tissue calcifica-tions is important (a) to assess the diagnosticand prognostic potential of the calcific struc-tures and (b) to help map the natural historyof disease processes, through premalignant,cancerous and metastatic stages. In selectingappropriate data collection and analysistechniques, both specificity and sensitivityneed to be assessed. Ideally the calcificationsshould be examined within the tissues inorder to retain any spatial relationshipbetween them. Thus, correlation and corrob-oration with histopathological data would bepossible. We are unaware of any such studiesthat have been performed and even the detailthat arises from diffractometer experimentsis rarely, if ever, reported.

Our recent work has involved diffractioncharacterisation of a range of breast tissues.We have used small angle diffraction tostudy changes to collagen with malignancy(see previous section) and wide angle diffrac-tion to study the nature of calcifications.Although not yet attempted with breast tis-sues, in principle, these two measurementregimens may be combined into a singleexperiment.

Our diffraction studies of calcificationshave included the examination of both dis-sected crystalline masses and core-cut biopsyspecimens containing crystallite populations.The dissected samples were studied in thelaboratory using conventional powder dif-fraction measuring facilities. The b-HAP datafrom breast tissue deposits (see Fig. 13.6b)was subsequently compared to that fromother biogenic calcifications (bone, uretericcalculi and aortic medial calcification) using

13

186



the Rietveld refinement method. Althoughspecimen numbers were limited, it wasapparent that the breast calcification wassomething of an extreme mineral, possessingthe greatest lattice parameter (along the c-axis) and smallest crystallite size82. The fullsignificance of this will be realised as furtherdata become available. However, it may berelated to differences in the initial depositionmechanisms. Further preliminary work (asyet unreported) employing synchrotron radi-ation examined, for the first time, core-cutbiopsy specimens. Utilising a small (0.5 × 0.5mm) incident X-ray beam, each unfixedspecimen was oscillated through the beam,

ensuring complete illumination during 30-minute data collections. An image platerecorded the diffraction data (see Fig. 13.6a),which was subsequently reduced to one-dimension by radial integration. A phase-matching routine indicated the presence of b-HAP and other, as yet unidentified crys-talline materials that do not appear tocorrespond to those within Table 13.3.

We hope in future to undertake more exten-sive studies on similar specimens, and pro-duce a complete account of the calcific phasesformed and their crystallographic detail. Invivo characterisation may, to some extent, bedeveloped from emerging technologies such

13

187

Fig. 13.6Diffraction data from core-cutbiopsy specimens. (a) section of 2-D intensity

using synchrotronradiation,

(b) 1-D data from large,dissected deposits.


as diffraction-enhanced imaging (see previoussection) and/or absorption edge-enhancedimaging (the linear absorption coefficient ofHAP is almost three times that of COD at20 keV). Thus the full diagnostic potentialof breast calcifications would be exploited.Naturally, this is critically dependent upondemonstration of the clinical significance ofcalcification crystallography.

References1. Piattella L, Zamponi N, Cardinali C, Porfiri L,

Tavoni M. Endocranial calcifications, infantile celiacdisease, and epilepsy. Childs Nervous System 1993;9: 172–5.

2. Brown K, Mund DF, Aberle DR, Batra P, Young DA.Intrathoracic calcifications: radiographic featuresand differential diagnoses. Radiographics 1994; 14:1247–61.

3. Rieber A. Das fibrolamellare Karzinom. ZeitschriftGastroenterol 1994; 32: 651–3.

4 Kragel PJ, Aquino MO, Fiorella R, Chapman J.Clinical, radiological and pathological features ofmedial calcific sclerosis in the breast. Southern MedJ 1997; 90: 518–21.

5 McCarty DJ, Lehr JR, Halverson PB. Crystalpopulations in human synovial fluid; identificationof apatite, octacalcium phosphate and tricalciumphosphate. Arthritis Rheum 1983; 26: 1220–4.

6. Mustafina ZhG, D’iakova GA, Galimova RZ,Karzhaubaeva GG. Pathological mineralization inretinoblastoma. Vestnik Oftalmologii 1992; 108: 37–8.

7. Sperrin MW, Rogers KD. The architecture andcomposition of uroliths. Br J Urol 1998; 82: 781–4.

8. Price JL, Gibbs NM. The relationship betweenmicrocalcification and in situ carcinoma of thebreast. Clin Radiol 1978; 29: 447–52.

9. Egan RL, McSweeney MB, Sewell CW.Intramammary calcifications without an associatedmass in benign and malignant diseases. Radiology1980; 137: 1–7.

10. Galkin BM, Feig SA, Frasca P, Muir HD, Soriano RZ.Photomicrographs of breast calcifications:correlation with histopathologic diagnosis.Radiographics 1983; 3: 450–77.

11. Kragel PJ, Aquino MO, Fiorella R, Chapman J.Clinical, radiographic and pathologic features ofmedial calcific sclerosis in the breast. Southern MedJ 1997; 90: 518–21.

12. Frouge C, Meunier M, Guinebretiere J-M et al.Polyhedral microcalcifications at mammography:histologic correlation with calcium oxalate.Radiology 1993; 186; 681–4.

13. Holme TC, Reis MM, Thompson A, Robertson A etal. Is mammographic microcalcification of biologicalsignificance? Eur J Surg Oncol 1993; 19: 250–3.

14. Albertyn LE. Mammographically indeterminatemicrocalcifications: can we do any better?Australasia Radiol 1991; 35: 350–7.

15. Stomper PC, Connolly JL. Ductal carcinoma in situof the breast: correlation between mammographiccalcification and tumor subtype. Am J Roentgenol1992; 159: 483–5.

16. Foschini MP, Fornelli A, Peterse JL, Mignani S,Eusebi V. Microcalcifications in ductal carcinoma insitu of the breast: histochemical andimmunohistochemical study. Hum Pathol 1996; 27:178–83.

17. Tsuchiya A, Kanno M, Hara K, Kimijima I, Abe R.The evaluation of mammographic microcalcificationas biological malignancy in breast cancer.Fukushima J Med Sci 1996; 1: 17–22.

18. Tabar L, Chen H, Duffy SW, Yen MF, Chiang CF,Dean PB, Smith RA. A novel method for predictionof long-term outcome of women with T1a, T1b, and10–14 mm invasive cancers: a prospective study.Lancet 2000; 355: 429–33.

19. Wheeler EJ, Lewis D. An X-ray study of theparacrystalline nature of bone apatite. CalcifiedTissue Res 1977; 24: 243–8.

20. Morgan H, Wilson RM, Elliott JC, Dowker SEP,Abderson P. Preparation and characterisation ofmonoclinic hydroxyapatite and its precipitatedcarbonate apatite intermediate. Biomaterials 2000;21: 617–27.

21. Handschin RG, Stern WB. X-ray diffraction studieson the lattice perfection of human bone apatite(Crista Iliaca). Bone 1995; 16: 355S–63S.

22. Wilson RM, Elliott JC, Dowker SEP. Rietveldrefinement of the crystallographic structure ofhuman dental enamel apatites. Am Minerol 1999;84: 1406–14.

23. Fujisawa R, Kuboki Y. Preferential adsorption ofdentin and bone acidic proteins on the (100) face ofhydroxyapatite crystals. Biochim Biophys Acta 1991;1075: 56–60.

24. LeGeros RZ, Kijkowska R, Bautista C, LeGeros JP.Synergistic effects of magnesium and carbonate onproperties of biological and synthetic apatites.Connective Tissue Res 1995; 33: 203–9.

25. Prudhommeaux F, Schiltz C, Liote F et al. Variationin the inflammatory properties of basic calciumphosphate crystals according to crystal type.Arthritis Rheum 1996; 39: 1319–26.

26. Wiessner JH, Mandel GS, Halverson PB, MandelNS. The effect of hydroxyapatite crystallinity onhemolysis. Calcified Tissue Intl 1988; 42: 210–19.

27. Osborne CA, Davis LS, Sanne J, Unger LK, O’BrienTD, Clinton CW, Davenport MP. Identification andinterpretation of crystalluria in domestic animals: a

13

188



light and scanning electron microscopic study. VetMed 1990; 85: 18–37.

28. Hassler O. Microradiographic investigations ofcalcifications of the female breast. Cancer 1969; 23:1103–9.

29. Daculsi G, Bouler J-M, LeGeros RZ. Adaptive crystalformation in normal and pathological calcificationsin synthetic calcium phosphate and relatedbiomaterials. Intl Rev Cytol 1997; 172: 129–91.

30. Moritz JD, Luftner Nagel S, Westerhof JP, OestmannJW, Grabbe E. Microcalcifications in breast corebiopsy specimens: disappearance at radiographyafter storage in formaldehyde. Radiology 1996; 200:361–3.

31. Pasteris JD, Wopenka B, Freeman JJ, Leroy Young V,Brandon HJ. Medical mineralogy as a new challengeto the geologist: silicates in human mammarytissue? Am Mineral 1999; 84: 997–1008.

32. Terzakis JA. Detection of calcifications in breastbiopsies by scanning electron microscopy. UltrastrPathol 1998; 22: 181–4.

33. Mihaescu A, Burri G. Calcium oxalate crystals inbenign breast cyst fluid. Diagn Cytopathol 1995; 12:67–70.

34. Torell JA, Knight JP, Marcus PB. Intraluminalcalcium hydroxyapatite crystals in breast carcinoma:an ultrastructural study. Ultrastr Pathol 1984; 6:9–14.

35. Galkin BJ, Frasca P, Feig SA, Holderness KE. Non-calcified breast particles. A possible new marker ofbreast cancer. Invest Radiol 1982; 17: 119–28.

36. Tornos C, Silva E, El-Nagger A, Pritzker KPH.Calcium oxalate crystals in breast biopsies. Themissing calcifications. Am J Surg Pathol 1990; 14:961–8.

37. Ahmed A. Calcification in human breastcarcinomas: ultrastructural observations. J Pathol1975; 117: 247–51.

38. Going JJ, Anderson TJ, Crocker PR, Levison DA.Weddellite calcification in the breast: 18 cases withimplications for breast cancer screening.Histopathology 1990; 16: 119–24.

39. Estepa-Maurice L, Hennequin C, Marfisi C, Bader C,Lacour B, Daudon M. Fourier transform infraredmicroscopy identification of crystal deposits intissues. Clin Chem 1996; 105: 576–82.

40. Jackson M, Mansfield JR, Dolenko B, Somorjai RL,Mantsh HH, Watson PH. Classification of breasttumours by grade and steroid receptor status usingpattern recognition analysis of infrared spectra.Cancer Detect Prevent 1999; 23: 245–53.

41. Delhez R, deKeijser TH, Langford JI, Louer D,Mittemeijer EJ, Sonneveld EJ. Crystal imperfectionbroadening and peak shape in the Rietveld method.In: Young RA, Ed. The Rietveld Method. USA,IUCr–Oxford University Press, 1993.

42. Frappart L, Remy I, Lin HC, Bremond A, Raudrant

D, Grousson B, Vauzelle JL. Different types ofmicrocalcifications observed in breast pathology:correlations with histopathological diagnosis andradiological examination of operative specimens.Virchows Arch A 1986; 410: 179–87.

43. Frappart L, Boudeulle M, Boumendil J et al.Structure and composition of microcalcifications inbenign and malignant lesions of the breast: study bylight microscopy, transmission and scanningelectron microscopy, microprobe analysis and X-raydiffraction. Hum Pathol 1984; 15: 880–9.

44. Fandos-Morera A, Prats-Esteve M, Tura-Soteras JM,Traveria-Cros A. Breast tumours: composition ofmicrocalcifications. Radiology 1988; 169: 325–7.

45. Meek KM, Blamires T, Elliot GF, Gyi TJ, Nave C.The organisation of collagen fibrils in the humancorneal stroma: a synchrotron radiation x-raydiffraction study. Curr Eye Res 1987; 6: 841–6.

46. Lombardi V, Piazzesi G, Ferenczi M, Thirlwell H,Dobbie I, Irving M. Elastic distortion of myosinheads and repriming of the working stroke inmuscle. Nature 1995; 374: 553–5.

47. Lewis RA. Medical applications of synchrotronradiation X-rays. Physics Med Biol 1997; 42:1213–43.

48. Burattini E, Gambaccini M, Marziani M et al. X-raymammography with synchrotron radiation. Rev SciInstrum 1992; 63; 638–40.

49. Burattini E, Cossu E, Di Maggio C et al.Mammography with synchrotron radiation.Radiology 1995; 195: 239–44.

50. Johnston RE, Washburn D, Pisano E et al.Preliminary experience with monoenergetic photonmammography. SPIE 1995; 2432: 434–41.

51. Lewis RA, Hufton AP, Hall CJ, Helsby WI, Towns-Andrews E, Slawson S, Boggis CRM. Improvementsin image quality and radiation dose in breastimaging. Synchrotron Radiation News 1999; 12:7–14.

52. Arfelli F, Bravin A, Barbiellini G et al. Digitalmammography with synchrotron radiation. Rev SciInstrum 1995; 66: 1325–8.

53. Zhong Z, Thomlinson W, Chapman D, Sayers D.Implementation of diffraction-enhanced imagingexperiments: at the NSLS and APS. Nucl InstrumMethods Physics Res 2000; A450: 556–67.

54. Schonermark MP, Bock O, Buchner A, Steinmeier R,Benbow U, Lenarz T. Quantification of tumor cellinvasion using confocal laser scan microscopy. NatMed 1997; 3: 1167–71.

55. Kauppila S, Stenbäck F, Risteli J, Jukkola A, RisteliL. Aberrant type I and type III collagen geneexpression in human breast cancer in vivo. J Pathol1998; 186: 262–8.

56. Lewis RA, Rogers KD, Hall CJ et al. Breast cancerdiagnosis using scattered X-rays. J SynchrotronRadiat 2000; 7: 348–52.

13

189


57. D’Orsi CJ, Reale FR, Davis MA, Brown VJ. Iscalcium oxalate an adequate explanation fornonvisualization of breast specimen calcifications?Radiology 1992; 182: 801–3.

58. Edmundo J, Gonzalez G, Caldwell R, Valaitis J.Calcium oxalate crystals in the breast. Pathologyand significance. Am J Surg Pathol 1991; 15: 586–91.

59. Radi MJ. Calcium oxalate crystals in breast biopsies.An overlooked form of microcalcification associatedwith benign breast disease. Arch Pathol Lab Med1989; 113: 1367–9.

60. Busing CM, Keppler U, Menges V. Differences inmicrocalcification in breast tumors. VirchowsArchiv A 1981; 393: 307–13.

61. Bellahcene A, Castronovo V. Increased expression ofosteonectin and osteopontin, two bone matrixproteins, in human breast cancer. Am J Pathol 1995;146: 95–100.

62. Bergstrand A, Collste LG, Franksson C et al.Oxalosis in renal transplants followingmethoxyflurane anaesthesia. Br J Anaesth 1972; 44;569–74.

63. Rogers KD, Sperrin MW, MacLean EJ. Asynchrotron study of bladder urolith architecture.Powder Diffraction 2000; 15: 94–100.

64. Katoh R, Kawaoi A, Muramatsu A, Hemmi A,Suzuki K. Birefringent (calcium oxalate) crystals inthyroid diseases – a clinicopathological study withpossible implications for differential diagnosis. Am JSurg Pathol 1993; 17: 698–705.

65. Gardner A. Retinal oxalosis. Br J Ophthalmol 1974;58: 613–19.

66. Jensen OA. Calcium oxalate crystals localised in theeye. Acta Ophthalmol 1975; 53: 187–96.

67. Boskey AL, Bullough PG, Vigorita V, DiCarlo E.Calcium–acidic phospholipid–phosphate complexesin human hydroxyapatite-containing pathologicdeposits. Am J Pathol 1988; 133: 22–9.

68. Rogers KD, Lewis RA, Hall CJ et al. Preliminaryobservations of breast tumor collagen. SynchrotronRadiation News 1999; 12; 15–20.

69. Pucci-Minafra I, Luparello C, Andriolo M, BasiricoL, Aquino A, Minafra S. A new form of tumor andfetal collagen that binds laminin. Biochemistry 1993;32: 7421–7.

70. Hirota S, Ito A, Nagoshi J et al. Expression of bone

matrix protein messenger ribonucleic acids inhuman breast cancers. Lab Invest 1995; 72: 64–9.

71. Brown LF, Papadopoulos-Sergiou A, Berse B et al.Osteopontin expression and distribution in humancarcinomas. Am J Pathol 1994; 145: 610–23.

72. Bellahcene A, Castronovo V. Expression of bonematrix proteins in human breast cancer: potentialroles in microcalcification formation and in thegenesis of bone metastases. Bull Cancer 1997; 84:17–24.

73. Yoneda T. Cellular and molecular basis ofpreferential metastasis of breast cancer to bone.J Orthopaed Sci 2000; 5: 75–81.

74. Santoliquido PM, Southwick HW, Olwin JH. Tracemetal levels in cancer of the breast. Surg GynecolObstet 1976; 142: 65–70.

75. Carlisle EM. Silicon. Biochem Essent UltratraceElements 1984; 3: 257–91.

76. Halverson PB, McCarty DJ. Patterns of radiographicabnormalities associated with basic calciumphosphate and calcium pyrophosphate crystaldeposition in the knee. Annu Rheum Dis 1986; 45:603–5.

77. Cheung HS, Story MT, McCarty DJ. Mitogeniceffects of hydroxyapatite and calciumpyrophosphate dihydrate crystals on culturedmammalian cells. Arthritis Rheum 1984; 27: 668–74.

78. McCarthy GM, Mitchell PG, Struve JS, Cheung HS.Basic calcium phosphate crystals cause co-ordinateinduction and secretion of collagenase andstromelysin. J Cell Physiol 1992; 153: 140–6.

79. Derfus BA, Kurtin SM, Camacho NP, Kurup I, RyanLM. Comparison of matrix vesicles derived fromnormal and osteoarthritic human articular cartilage.Connective Tissue Res 1996; 35: 337–42.

80. Derfus BA, Kranendonk S, Camacho NP, Mandel N,Kushnaryov V, Lynch K, Ryan LM. Humanosteoarthritic cartilage matrix vesicles generate bothcalcium pyrophosphate dihydrate and apatite invitro. Calcified Tissue Intl 1998; 63: 258–62.

81. Ho AM, Johnson MD, Kingsley DM. Role of themouse ank gene in control of tissue calcification andarthritis. Science 2000; 289: 265–70.

82. Rogers KD. Initial Rietveld characterisation ofbiological calcifications. Powder Diffraction 1997;12: 175–9.

13

190


ADH see atypical ductal hyperplasiaadvanced breast biopsy instrumentation (ABBI),

85–87amorphous material, 172, 184aneuploidy, rates in DCIS and atypical ductal

hyperplasia, 49anticoagulant therapy, vacuum-assisted

mammotomy and, 90anxiety, follow-up and, 112artefacts, 24arthritis see osteoarthritisatrophic lobules, calcification, 12, 13 (Fig.), 14

(Fig.)atypical ductal hyperplasia, 48–49

core biopsy, 79, 80, 101, 102microcalcification, 130

atypical intraductal proliferation, 103axillary metastases, diagnosed ductal carcinoma

in situ, 62–63

b-HAP (ion-substituted calcium hydroxyapatite),171, 173–174

breast vs bone, 184osteoarthritis, 186

Bard gun, 72benign calcification, 3–29, 109

core biopsy, 80ultrasound, 142–143

benign lesionscore biopsy, 101fine-needle aspiration cytology, 105magnetic resonance imaging, 157, 164vacuum-assisted mammotomy for, 91

biopsysurgical, diseases diagnosed, 3–4see also core biopsy

blunt duct adenosis, 19bone matrix proteins, 185bone mineral

vs breast calcification, 185X-ray diffraction, 178see also b-HAP (ion-substituted calcium

hydroxyapatite)Bragg’s law, 178broken needle appearance, in duct ectasia, 16

C-erbB-2 oncogene, ductal carcinoma in situ,44–46

calcificationdisappearance of, 110mechanisms, 184–186

calcification retrieval rates, core biopsy, 77–78calcium hydroxyapatite, 171calcium oxalate crystals

formation, 184microscopy, 100phases, 173technical errors, 175, 183X-ray diffraction, 177 (Fig.), 178

calculi (uroliths), 171carbon suspension, calcification localisation,

120–122carbonate substitution, hydroxyapatite, 173, 174carcinoma see ductal carcinoma in situ; invasive

carcinomacartilage, pyrophosphate phase, calcification, 186casting-type calcification, prognosis, 65–66,

129–130cell size, ductal carcinoma in situ, 44characterisation research, computer-aided

mammography, 153–154classifications

core biopsies, 100–104ductal carcinoma in situ, 41–43fine-needle aspiration cytology specimens,

104–106clinical examination, for core biopsy, 76–77, 111cluster shape, calcification

ductal carcinoma in situ, 34recall of screening patients, 109

cluster size, ductal carcinoma in situ, invasionrisk, 64

collagen, synchrotron radiation studies, 182–183colour Doppler ultrasound, 140comedo calcification

disappearance, 110ductal carcinoma in situ, invasive foci, 64invasive carcinoma, 55, 56 (Table), 61prognosis and, 66

comedo ductal carcinoma in situ, 41magnetic resonance imaging, 164–165screen-detected, 47ultrasound, 144–145

comedo necrosis, ductal carcinoma in situ,recurrence risk, 128

computer-aided detection (CAD), 150–151computer-aided mammography, 149–154consent, core biopsy, 77contrast enhancement, magnetic resonance

imaging, 157–158morphology, 159–162timing, 162–164

core biopsy, 72–80benign results, 80diseases diagnosed, 3fine-needle aspiration cytology vs, 71, 99, 100large, 85–95number of samples, 77–78 191

Index

Index

core biopsy – continuedreasons for failure, 85specimen assessment, 99–104understaging, 80, 91vacuum-assisted mammotomy vs, 91–92, 111validity for invasive focus in DCIS, 64

cornea, synchrotron radiation studies, 181coronary artery disease, vascular calcification, 20cosmetics, radiodense, 24cribriform pattern, ductal carcinoma in situ, 41,

42 (Fig.), 43–44crystallography, 171–175cysts, 4, 28 (Fig.)

ultrasound, 142

databases, digital images, 149DCIS see ductal carcinoma in situdepression, follow-up and, 112dermatomyositis, 23diabetes mellitus, vascular calcification, 20diffraction see X-ray diffractiondiffraction-enhanced imaging, 182digital mammography, 149digital stereotaxis, 73–74duct ectasia, 15–19

vs ductal carcinoma in situ, 16–19, 33 (Fig.), 38ductal carcinoma in situ, 33–48, 127–129

calcificationscriteria for recall of screening patients,

109–110distribution, 35–36morphology, 36–40number, 36, 64–65

change over time, 40classification, 41–43core biopsy diagnosis, 103–104vs duct ectasia, 16–19, 33 (Fig.), 38extensive in situ component (EIC), invasive

carcinoma with, 55, 61–62, 129fine-needle aspiration cytology, 105–106growth patterns, 41invasive carcinoma and, 55–65, 103magnetic resonance imaging, 157–167missed diagnosis, 40, 41ultrasound, 142 (Fig.), 143–145see also recurrent ductal carcinoma in situ

dyes, calcification localisation, 119–120dynamic spiral magnetic resonance imaging, 166dystrophic calcification

eye, 184fibroadenoma, 6

edge detection, computer-aided mammography,153

electron beam microprobes (EBM), 175–176electron diffraction, 176electron microscopy, 175

enhancement see contrast enhancement, magneticresonance imaging

epidermal inclusion cysts, 26epithelial proliferations, intraductal, 31–51erect posture, digital stereotaxis, 74–76extensive in situ component (EIC), invasive

carcinoma with, 55, 61–62, 129eye, oxalosis, 184

fast-spin echo sequences, 166–167fat necrosis, 20–21, 22 (Fig.), 23 (Fig.)fibroadenomas, 5–7, 8 (Fig.)

magnetic resonance imaging, 157ultrasound, 143

fibroadenomatoid hyperplasia, 7–12fibrocystic change, 4–5, 6 (Fig.)

magnetic resonance imaging, 157, 160magnification views, 110ultrasound, 142–143

film-screen stereotaxis, core biopsy, 72–73fine-needle aspiration cytology, 71

vs core biopsy, 71, 99, 100false negative, 106inadequate samples, 104–105specimen assessment, 104–106

fixation, biopsy specimens, 99

gadolinium see contrast enhancement, magneticresonance imaging

gel pellets, biopsy site marking, 90grading of carcinoma, calcification and, 55–57,

128–130granulomatous mastitis, idiopathic, 26–29guns, core biopsy, 72, 89

haematomas, 28 (Fig.)haemostasis, vacuum-assisted mammotomy, 90heteroionic substitution, apatite, 173–174high-frequency ultrasound transducers, 139–141,

142 (Table)histology

calcification, 99–100vs mammography, microcalcification, 128stains for calcification, 175

hookwire localisation, 115–116see also wire-guided biopsy

idiopathic granulomatous mastitis, 26–29image processing, 149–154ImageChecker system (R2 Technology), 153inadequate samples, fine-needle aspiration

cytology, 104–105inclusion cysts, epidermal, 26infrared spectroscopy, 176intercellular calcification, ductal carcinoma in

situ, 41intraductal epithelial proliferations, 31–51192

14-Evans-Ind-cpp 19/6/02 1:14 pm Page 192

Index

invasive carcinoma, 55–66core biopsy diagnosis, 103disappearance of calcification, 110incidence, 47magnetic resonance imaging, contrast

enhancement rates, 164microcalcification and prognosis, 129–130ultrasound, 143–145

involutional change (atrophic lobules),calcification, 12, 13 (Fig.), 14 (Fig.)

ionic substitution, apatite, 173–174ischaemic heart disease, vascular calcification, 20

Klippel—Trenaunay syndrome, 26

large-cell ductal carcinoma in situ, 44large core biopsy, 85–95lateral approach, vacuum-assisted mammotomy, 93lattice parameters, crystalline materials, 173lead-pipe calcification, ductal ectasia, 17, 19 (Fig.)lobular type carcinoma, fine-needle aspiration

cytology, 106local anaesthesia

core biopsy, 77ultrasound-guided mammotomy, 92vacuum-assisted mammotomy, 89

localisation of breast calcification, 115–123lymph node metastases, tumour calcification and,

171

magnesium, 185magnetic resonance imaging, 111, 117

ductal carcinoma in situ, 157–167magnification views, 110, 120 (Fig.), 139malignant cell displacement, biopsy, 71mammography, synchrotron radiation, 181–182Mammotome HH™, 87 (Fig.), 88

ultrasound-guided procedures, 92Mammotome ST™ driver and probe, 86 (Fig.), 93Manan gun, 72margins, wide local excision, 129markers

biopsy sites, 90calcification localisation, 115, 119–122

see also hookwire localisationmenstrual cycle, magnetic resonance contrast

enhancement, 164metastases (malignant disease)

bone matrix proteins, 185tumour calcification and, 171

metastatic calcification, renal failure, 26methylene blue, calcification localisation, 119–120MIB1 (cell cycling marker), ductal carcinoma in

situ, 46microcalcification

assessment when found, 110–112biopsy, 71

blunt duct adenosis, 19clinical aspects, 127–135computer-aided detection, 151–154ductal carcinoma in situ, 33epidermal inclusion cysts, 26fibroadenomatoid hyperplasia, 12fibrocystic change, 5invasive carcinoma, 55normal biopsy cores, 100, 101normal stroma, 12, 100–101predicting carcinoma invasion, 62–65sclerosing adenosis, 12on screening, 48spontaneous resolution, 66ultrasound, 110–111, 139–144

micropapillary pattern, ductal carcinoma in situ,41, 43–44, 45 (Fig.)

microscopypolarising microscopy, 175scanning electron microscopy, 175see also histology

microstrain, 178migration, localisation wires, 119milk of calcium see teacup appearanceminimally invasive breast biopsy (MIBBT), 87multifocality, vs calcification, 129

nanocrystalline material, 172necrosis

calcification and, 184–185ductal carcinoma in situ, 37 (Fig.), 44

recurrence risk, 128screen-detected, 47, 48

needle size, core biopsy, 72nipple plane growth, ductal carcinoma in situ, 41non-crystalline diffraction, 180normal appearance, computer identification, 12normal biopsy cores, 100–101normal stroma, microcalcification, 12, 100–101number of calcifications

ductal carcinoma in situ, 36invasion risk, 64–65

recall of screening patients, 109

oestrogen receptors, ductal carcinoma in situ, 46oil cysts, 28 (Fig.)osteoarthritis, b-HAP (ion-substituted calcium

hydroxyapatite), 186osteopontin, 185oxalate metabolism, 184

p53 gene, ductal carcinoma in situ, 46Paget’s disease of nipple

disappearance of calcification, 110incidence of radiological abnormalities, 33

pain, fine-needle aspiration cytology vs corebiopsy, 71 193

Index

papillary lesions, vacuum-assisted mammotomy,91

papillomascore biopsy, 102skin, 23

paracrystalline material, 172PASTA (polarity-altered spectral and spatial

acquisition), 166pathology see histologypectoral muscles, calcification, 28 (Fig.)pH, calcification, 174phased array breast coils, magnetic resonance

imaging, 165–166phases, crystalline materials, 173

Bragg’s law, 178physical examination, for core biopsy, 76–77, 111polarising microscopy, 175polarity-altered spectral and spatial acquisition

(PASTA), 166polycrystalline material, 172popcorn-like calcification, 6, 7 (Fig.)power Doppler ultrasound, 140prescreening, computer-aided, 149progesterone receptors, ductal carcinoma in situ,

46prognosis

casting-type calcification, 65–66, 129–130comedo calcification, 66invasive carcinoma, microcalcification and,

129–130prompting, 150–151prone digital stereotaxis, 75–76proteins

absorption, crystallites, 174structural studies, 180

punctate calcification, ductal carcinoma in situ, 40pyrophosphate phase, calcification, 186

radiodense materialstissue deposits, 176topical medications, 24

radiography see specimen radiography;stereotactic procedures

radioisotope studies, 111radiolabelled colloid, calcification localisation,

121 (Fig.), 122biopsy, 132–133, 134 (Fig.)

radiotherapyadjuvant, 127calcification from, 47synchrotron radiation, 181

recall of patients, from screening, 109–110recurrent ductal carcinoma in situ

calcification vs grading, 127–128diagnosis, 46–47

Reidy localisation needle and wire, 115–116renal failure, metastatic calcification, 26

repeat biopsies, 111–112resolution, spontaneous, microcalcification, 66Rietveld method, X-ray diffraction, 178rod-shaped calcification, ductal carcinoma in situ,

40, 41ROLL see radiolabelled colloidrotating delivery of excitation off resonance

(RODEO), magnetic resonance imaging, 166

sampling errors, fine-needle aspiration cytology,71

scanning electron microscopy, 175scar tissue

carcinoma in, 102vacuum-assisted mammotomy for, 90

scintimammography, 111sclerosing adenosis, 12–15screening

ductal carcinoma in situ detection, 47–48recall of patients, 109–110

seeding see track recurrencesialoprotein, 185SiteSelect™ (large core biopsy), 86sitting posture, digital stereotaxis, 74–76skin calcification, 21–24, 25 (Fig.)small angle X-ray scattering, collagen, 182–183,

186small-cell ductal carcinoma in situ, 44soap, radiodense, 24specimen radiography, 76, 99

performance, 71wire-guided biopsy, 131 (Fig.), 132

specimen storage, 77spiral magnetic resonance imaging, 166spoiled gradient echo sequences (SPGR), 166spontaneous resolution, microcalcification, 66stains (histology), calcification, 175statistics

algorithms, computer-aided mammography,153

see also validitystereotactic procedures

core biopsy, 72–80vs fine-needle aspiration cytology, 71

hookwire localisation, 117, 118–119 (Fig.)vacuum-assisted mammotomy, 92–93

indications, 91stromal calcification, 10 (Fig.), 11 (Fig.), 12subjectivity, computer-aided detection and, 152subtraction technique, computer-aided detection,

153surfaces, mammographic images as, 152surgical biopsy, diseases diagnosed, 3–4suture calcification, 24–26synchrotron radiation studies, 179–182, 187

tattoos, 24194


Index

teacup appearancecysts, 4, 5, 6 (Fig.)ductal carcinoma in situ, 39magnification views, 110ultrasound, 141 (Fig.)

technetium-99m-labelled colloid see radiolabelledcolloid

thyroid, calcium oxalate, 184tissue studies, synchrotron radiation, 182–183topical medications, radiodense materials, 24track recurrence, biopsy, 71

carbon suspension localisation and, 122Trichinella spiralis, pectoral muscle calcification, 28

(Fig.)triple approach, non-operative diagnosis, 99tubular type carcinoma, fine-needle aspiration

cytology, 106

ultrasound, 139–145microcalcification, 110–111, 139–144

ultrasound-guided procedurescore biopsy

vs fine-needle aspiration cytology, 71vs stereotactic core biopsy, 111

hookwire localisation, 116–117Mammotome HH™, 92vacuum-assisted mammotomy, indications, 91

understagingcore biopsy, 80, 91vacuum-assisted mammotomy, 91

unifocality, vs calcification, 129unit cells, crystalline materials, 173upright digital stereotaxis, 74–76uroliths, 171usual type hyperplasia, 3, 4, 33

vacuum-assisted mammotomy, 85, 87–92vs core biopsy, 91–92, 111stereotactic procedures, 92–93

indications, 91validity

core biopsy for invasive focus in DCIS, 64magnetic resonance imaging, DCIS, 158–159

vascular calcification, 19–20, 21 (Fig.), 171vascularity, Doppler ultrasound, 140very large core biopsy, 85–87

weddellite, 184wide angle X-ray diffraction, 186wide local excision, ductal carcinoma in situ,

recurrence, 127wire-guided biopsy, 130–132wires see hookwire localisation

X-ray diffraction, 176–179synchrotron radiation studies, 179–182

X-ray microanalysis, 175–176

zinc, 185

195


breast calcification- a diagnostic manual

Documents