MRI in breast cancer diagnosis and treatment

Foreword There is little doubt that mammography has had an enormousimpact on the problem of breastcancerdetection.The procedurehas undergonecontinuous modification and improvement in the past decade,andthis progresscontinues in the exploration of new techniquesto improve the accuracyof detectionand diagnosis.MRI seemsan ideal medium for the study of breastdisease,but its role is asyet undefined.In this monographDr. Harms addressesthis issueand helps clarify the place of MRI in this problem. We believe this is vital information for any of us who are involved in mammography. E. Keats, MD

Theodore

Editor-in-Chief

Steven E. Harms, MD, received his medical degree and diagnostic radiology residency from the University of Arkansas. He is the past chairman of Radiologic Devices Panel of FDA, past president of SMRI, officer in ISMRM and RSNA. He serves on the editorial boards of the Journal of Magnetic Resonance Imaging, Quarterly Magazine for Magnetic Resonance, and Journal of Computed Aided Surgery. Dr. Harms has reveived awards from AMA, AADR, ARRS, SNM, RSNA, American Society of Ophthalmology, Plastic and Reconstructive Surgery, NCI, and the Susan G. Komen Foundation. His publications exceed 140 scientific papers, abstracts, 35 book chapters, and six patents. 192

Curr

Probl

Diagn

Radiol,

November/December

1996

MRI in Breast

Cancer Diagnosis

and Treatment

Breast cancer diagnosis and treatment are important health care issues in the Industrialized World. About 180,000 new breast cancers are discovered annually in the United States. Becausethis cancer often occurs in premenopausal women, breast cancer is a leading cause of potential life years lost. Breast magnetic resonanceimaging (MRI) is capable of producing detailed information concerning the extent and character of breast lesions. The technique and alternatives for generating high-resolution breast MR images are reviewed. Characteristic features of a pulse sequencefor breast imaging includes heavy T, weighting and magnetization transfer weighting for more effective gadolinium contrast, fat suppression,and rapid acquisition time. MRI is best employed for breast cancer diagnosis as a supplement to conventional breast imaging. Diagnostic groups particularly well suited to breast MRI include women with radiographically dense breasts, silicone augmentation, and postoperative scar.The capacity of breast MRI to show diseaseextent is employed to plan and localize for breastconservation therapy. Tumor size and multiple tumors can be characterized for more-effective surgical management. Ductal carcinoma in situ can be imaged and staged for tailored therapy. MRI-directed biopsy and localization can be used to optimize lumpectomy surgery and reduce the potential for histologically positive margins. MRI can define the effectivenessof radiation therapy and chemotherapy to provide improved information on nonsurgical treatment of breast cancer. The clinical implementation of breast MRI in the future depends on the careful coordination of quality MRI images and interpretations with skillful therapeutic management.

q

he evolution of mammographic screeningfor breastcancertransformedbreastimaging into primary care. Screeningmammograms often are the entry point for women with breastcancerinto the health caresystem.As a result, the knowledge and responsibilities of the breastimager are quite different from the environment to which most radiologists are accustomed.The breastimager must possessa knowledge of clinical managementin addition to the usual responsibility for the radiologic diagnosis. BreastMRI providesimprovedvisualizationof breast cancerbut is considerablymore expensivethan conventional breast-imagingmethods.The integrationof MRI into the diagnosticmanagementof breastcancerrequires a careful analysis in the decision processthat results from its use.Appropriate use of MRI can result in reducedoverall costsfor breastcancermanagement. hrr

Probl

Diagn

Radiol,

November/December

1996

This article reviews the current statusof breastMRI and describesthe potential integrationof MRI into the managementof breastcancer. General Features of Breast Cancer An estimatedonein nine American womenwill develop breastcancerat some time during their lifetime.’ The breast is the most frequent site of cancer incidence amongAmericanwomenandis the secondleadingcause of cancer mortality.2 These deathselevatethe importance of breastcancer as a leading causeof years of potential life lost because 15% of breastcancerdeaths occur amongpremenopausalwomen.3 Women who die of breastcancerhavedistant occult metastases,indicating the presenceof either diseaseat the time of local treatment or metastasesfrom inadequately treatedlocal or regional disease.To provide 193

thebestchanceof overall survival in women with breast cancer,treatmentgoals develop optimal local control (remove the tumor and, as much as possible, conserve normal breasttissue)andstagethe diseaseso that those with advancedregionalor distantdiseaseareidentified for a courseof systemictreatment.The implementation of this strategyis often impaired by inaccuraciesin the estimation of local diseaseextent. Treatment is often basedon the probableextent of diseaseestablishedby clinical trials rather than the actual demonstrationof diseasein a woman.As aresult, most patientsaregiven more treatment than would be necessaryfor elimination of the clinically apparentdiseaseto treat the clinically occult disease.This additionaltreatmentincreases thecostandmorbidity associatedwith breastcancercare. In some cases,however, the subclinical diseaseis so extensivethat the additional treatment remains inadequate,and the diseaserecurs.Most clinical-treatment trials are designedto addressthis problem.A more accurate method for determining the extent of disease within the breastshould havea profoundimpact on the clinical managementof breastcancer.

with state-of-the-artconventional imaging including mammography,galactography,sonographyiandpathologic examinationsthat include serial sectionanalysis on mastectomyspecimens.This article summarizesour experiencewith RODEO (Rotating Delivery of Excitation Off-resonance)breastMRI andits integrationinto clinical managementof breastcancer. RODEO MRI

General MRI Technical Features. Precontrastand postcontrastfat-suppressedimages areobtainedfor all studies.To visualize the early enhancementfeaturesof cancercontrastedwith breastparenchyma,which enhanceslater,postcontrastimages were obtainedimmediately aftera 0.1 mmol/kg (usually 8 to 16ml) intravenousbolus of gadopentetatedimeglumine.All imaging was performed on a current 1.5 T GeneralElectric Signa(Milwaukee)scanner.Imagereformationsandmaximum-intensity projection (MIP) ray tracingswere performed on a General Electric Independent Console (Milwaukee). A quadratureradiofrequency(i-f); transmit-receive breastcoil (MRI Devices,Milwaukee) was developedspecifically for breastMR imaging. For all Breast MRI examinations, the patients were scannedin the prone Breast applications of MRI were predicted in the late position without breastcompression. Three-dimensionalimaging wasusedto improve the 1970~.~,~ Early clinical systems failed to sustain this speculation.6-9By the late 1980s breastMRI was con- image resolution andfacilitate image-processingmethsidereda clinical failure by most experts.“-” However, ods.The display matrix of 128 x 256 x 256 produces the introduction of gadolinium contrast gaveusers of voxel resolution of -1.4 x 0.7 x 0.7 mm for an 18 cm breastMRI new hope. Europeanresearchersdemon- field of view (FOV). The use of the image-processing stratedconsistent enhancementof breast cancer after workstationallows nearreal-timereformationsto faciligadoliniumcontrastadministration.Theseworkersdem- tate the depiction of oblique image planes. The onstratedthe clinical application of thenegativepredic- hyperintensebreastnoduleswithin theentirevolumecan tive valueof breastMRI asa methodfor improving the be demonstratedwith MIP ray tracing. The MIP methspecificity of imaging diagnosis.The combination of ods provide a quick survey of diseaseextent and anaMRI after a positive mammogram could improve the tomic relations.‘9-22 potential for a positive biopsy.12-l4 In the United States, Pulse Sequence.A steady-statefat-suppressedsethis concepthasnot achievedwidespreadsuccess;how- quencewasdevelopedto achieveoptimal imagecontrast ever,the needlebiopsy is becoming a popular alterna- for breastimaging with a T,-weighted and magnetizative to surgical biopsy. Needle biopsy approachesthe tion-transfer weighted 3D image. RODEO usesa sine reliability of surgical biopsy and can be obtainedfor a excitation on fat resonancefollowed by a similar sine fraction of thecost of the surgicalbiopsy or the MRI.15-17 excitation 180 degreesphaseshifted (Fig. 1). The secIn 1990our group introduced a new pulse sequence ond excitation drives fat back longitudinally (aligned for the efficient acquisition of high-resolution, fat- with theappliedmagneticfield), resultingin suppressed suppressedthree-dimensional(3D) images that could fat signal in the image. Becausewateris off-resonance achieve the heavy T, weighting and magnetization- for both excitations,both rf pulsesareadditive for watransferweighting neededfor MRI breastcancerstag- ter resonanceto result in transversemagnetizationthat ing.‘s-20Since then, a seriesof more than 1020breasts producesMR signals.The off-resonance-shaped rf prohas beenimaged at Baylor University Medical Center vides robustmagnetization-transferweighting without with the sameMRI sequence.All casesarecorrelated the use of additional rf pulses. The magnetization194

Curr Probl

Diagn

Radiol,

November/December

1996

transfer weighting is usedto suppressthe intensity of RF -f-Lnormal ductal tissue. The RODEO method can be achieved with a very short repetition time (TR; 18.5 msec)/echotime (TE; 3.9 msec) and a single excitation for a very efficient 3D acquisition of about 5 minutes. Even onT,-weightedimagesproducedby steady-state fast-scanpulsesequences, longT, fluids arehyperintense. Radiofrequencyspoiling is a techniquethat randomizes FIG. 1. RODEO pulse sequence. Rotating Delivery of Excitation Offthe phaseof the digital rf to reducethe signal intensity resonance (RODEO) uses nonselective sine rf excitation fallowed by a dueto thelong T, steadystate.The rf-spoiling technique 180-degree reverse sine excitation. This sequence suppresses a narwasusedto distinguishlesionswith ahigh fluid content. row bandwidth of resonance about the center of the excitation bandIf the center is set on fat resonance, the fat will be excited by the Theprecontrastimageswereperformedwithoutrf spoil- width. first sine rf pulse but will be subsequently reversed by the 180-degree ing. The RODEO images without rf spoiling demon- phase reversed sine rf pulse. The water spins that are off-resonance stratedhyperintensefluid causedby thesteady-state build- will experience both sine rf pulses as additive. Bound water is also by this off-resonance excitation to provide magnetization-transfer up on long T, spins. The postcontrast images were excited contrast. This sequence provides an efficient mechanism for suppressperformedwith rf spoiling to reducethe signal intensity ing fat or silicone while providing heavy T, weighting and mognetizaof lesionswith a high free-fluid content. tion-transfer weighting. The contrast, signal-to-noise, and speed of this With the variableuseof rf spoiling, cystsandenhanc- sequence makes it ideal for breast imaging. ing massesbehaveexactly oppositein signal intensity on the precontrastand postcontrastimages. Cysts are method that is effective at all field strengths.Convenhyperintense,whereasmassesare hypointenseon the tional STIR sequencesaretime consumingand havea precontrastimages. The postcontrastimages demon- low SNR. IR-preparedfast scanssuffer from blurred stratehyperintensity from enhancingmasses,whereas imagesresultingfrom thefilling of k-spacewith samples fluid-filled lesionsarehypointenseasaresultof rf spoil- that areoff the fat nu11.23-25 ing (Fig. 2).19-22 Fat presaturationusesa frequency-selective 90-degree pulseon fat resonancein preparationfor the usualpulse Other Breasf MRI Methods sequence.Fat presaturationis available on most highFat Suppression. In many cases,detectionof a patho- field MR systems.The lengthenedTR associatedwith logic processis hinderedby the presenceof alarge sig- this methodmakesit most applicableto 2D methods.It nal from neighboring adiposetissue.In such cases,the is susceptibleto motion artifacts, andthe less effective suppressionof thefat signaloffersimprovedconspicuity fat suppressionoften results in fat signal that is nearly and sensitivity. Fat suppressionis commonly usedfor isointensewith muscle.*‘j contrast-enhancedimages in which the hyperintense Another chemical-shift method exploits the differsignalfrom enhancingabnormalitiesmay be maskedby encesin phaseof fat and water that can occur with adhyperintensefat on conventionalMR images.The vari- justment of the readouttiming. Phase-differencemethous choices for fat-suppressiontechniqueare summa- ods (Dixon) can be applied at any field strength but rized along with the relative merits and limitations in require multiple repetitions. The trade-off is usually lower resolution or a longer scan.z7-30 Table 1. Subtractionis the most popularmethod for eliminatImage datacan be gatheredwith phaseencoding in ing fat intensity in contrast-enhancedimaging (Fig. 3). all threespatialdirections,which preservesthefrequency Subtractionis simple andwidely availableon a variety information for spectroscopy of individual voxels. of platforms and field strengths.Subtraction has not Chemical-shift imaging hasnot achievedpopularity for becomeapopularmethodfor applicationsotherthanthe musculoskeletalapplicationsbecauseof theconsiderably breastbecauseof misregistration from movement be- longer scantimes associatedwith the method.31-34 tween scans and the low signal-to-noiseratio (SNR) In 3D imaging, selectiveexcitationof the signalfrom waterby using adiabatictechniquesis possible,because resulting from the subtraction.12-14 The inversion time of inversion-recoverypulse se- no slice-selection gradient is required. These fatquencescanbe selectedsothat fat recoveryis at or near suppressed3D images have high SNRs without the the null during the readout.Short tau-inversionrecov- errors associatedwith 2D methods.RODEO provides ery (STIR) has a long history as a fat-suppression excellentfat suppressionwith rapid 3D acquisition.The Curr

Probl

Diagn

Radiol,

November/December

1996

195

FIG. 2. lntracystic

papilloma.

Mediolateral

oblique

mammogram

sag&al RODEO MRI image demonstrates hyperintense spoiling (B, right) RODEO MRI image shows hypointense papilloma (straight arrow) enhances on the postcontrast

(A) demonstrates

off-resonance excitation of RODEO produces a very T,weighted image along with magnetization-transfer weighting. Unfortunately, the RODEO method is not widely available and performs best with a transmitreceive coil.21 Recently spatial and frequency-selective excitation has emerged on imaging systems. These spatial-spectral pulses can select water signal in a definable region. The pulses are inherently of long duration and favor long TR pulse sequences. The longer pulses also are more susceptible to motion artifacts.35 Contrast and Resolution. The intensity of a voxel depends on the average of all the components of that voxel. Detection of abnormal signal is not a problem when the mass is large and homogeneous but can be difficult when the tumor infiltrates surrounding structures. Breast tumors commonly extend like fingers into surrounding tissue. A voxel composed of tumor and fat may register a signal intensity below the threshold for detection. This situation is most severe in DCIS, in which the abnormal cells distend but do not invade the duct. The inability to visualize tumors because of volume averaging is worse when the voxels are large or the contrast between tumor and surroundings is low. For example, the high density of metal can be used to detect tiny amounts of metallic foreign body on computed to196

a subareolar

mass.

The

precontrast

without

spoiling

(B, left)

fluid (curved arrow) and hypointense papilloma (straight arrow). The postcontrast with fluid (curved arrow) caused by the disruption of the long T, steady state of water. The image. Thus with two sequences, tumor and fluid-containing masses are easily separated.

mography (CT) scans because the average intensity of the voxel is higher than that of the surroundings. For foreign bodies with which the contrast is less, however, the CT examination, even with high resolution, may be difficult. For this reason, we prefer to image with the highest practical resolution and with a pulse sequence that maximizes the contrast between tumor, parenchyma, and fat. The quality of fat suppression is important because a less robust suppression leads to lower contrast and a need for even smaller voxels. The contrast between tumor and parenchyma is also important. Heavy T, weighting improves this contrast. If less T, weighting is provided, then higher resolution and smaller voxels will be needed to avoid a more severe volume-averaging problem. Another mechanism for enhancing the tumor-to-parenchyma contrast is magnetization-transfer contrast. Magnetization transfer is often provided by an additional prepulse in the sequence. Magnetization transfer reduces the intensity of tissues with high amounts of protein-bound water. RODEO provides fat suppression, heavy T, weighting, and magnetization-transfer weighting in an efficient, high-SNR acquisition.21 MRI Interpretation Receiver-operating characteristic (ROC) curves are often used to define the image interpretation characteristic of a Curr

Probl

Diagn

Radiol,

November/December

1996

1. Fat-suppression

TABLE

methods

Method Phase

difference

Chemical

Chemical

shift

shift

(Dixon) presaturation

(time

domain)

RODEO

Spectral-spatial Inversion

selective

lntrensic

MR parameter

exploited

Chemical

shift

Requires acquisition;

Chemical

shift

Favors long TR pulse sequences; more motion artifacts due to long excitation incomplete fat suppression due to T, recovery before readout

Probl

best

suited

to 2D

pulses;

4DFf acquisition is time consuming and/or low resolution; therefore this method has not been extensively used except for spectroscopic evaluations

Chemical

shift

Narrow frequency range can be selectively suppressed; optimized for high-resolution 3D acquisitions: single repetition fast 3D imaging with good SNR; can be performed at mid-field; favors transmit-receive coils

Chemical

shift

Long pulse length favors available on few machines

T.

Radial,

multiple repetitions; usually spin echo

shift

recovery

Diagn

features

Chemical

November/December

20;

greater

motion

T, is adjusted to nuil the T, corresponding to fat: may be used as a preparation pulse spin echo, RARE, or turbo gradient echo sequence; when combined with fast imaging (turbo) sequence the images are blurred due to acquisition of data off the null; can be used at any field strength

diagnostic-imagingexamination.For common diseases with diagnosesthat result in relatively inexpensiveconsequences,the diagnostic-imaging criteria should be weighted toward the upper region of the ROC because sensitivity is the primary goal. Criteria may be adjusted to improve the specificity, resulting in a movementtoward the left sideof the ROC curve.However,the key is to develop diagnostic criteria that achievethe highest specificity without sacrificing sensitivity. If the disease is uncommonor the diagnosticconsequences areexpensive,thenthepriorities would shift towardtheleft side of the curveandadjustmentswould be madeto achievethe highestsensitivity without sacrificing specificity. Our initial breastMRI diagnosticcriteria weredevelopedto emphasizesensitivity.Any enhancementof any size andlocation was called positive. TheseMRI diagnostic criteria wereusedin our initial seriesof 88breast examinations.In 30 of these88 breasts,the MRI findings were compared with the gold standard, serialsectionpathologic study of mastectomy specimens.As expected,thesensitivityis very high (94%)but the specificity is low (37%) becausethe diagnostic criteria window is wide and the serial-sectionanalysis allows the diagnosis of very small lesions that otherwise would not be sampled with routine pathologic examination. Curr

Major

1996

artifacts;

for

The high specificity of conventional imaging results from the assignmentof false-positiveMRI findings as true negativesby conventionalimaging. Theseresults are summarizedin Table 2.21 When the remaining 58 breaststhat were subjected to biopsy or lumpectomy rather than serial sectioning wereanalyzed,therewereno falsenegatives.Thesefindings,however,arerelatedto lesion-selectionbiasinherentin the lumpectomylbiopsy gold standardandshould not be considereda true measurementof the ability of MRI to visualize clinically occult disease. The lower reporting of false negativesresults in an overestimationof sensitivity. An example of accuracy in the estimation of diagnostic imaging becauseof an inadequategold standardcan be seenin the nuclear medicineliteraturein which sulphurcolloid liver-spleen scansare reportedto havea sensitivity of 90% for the detectionof liver metastases.When more-sensitiveimagingmethodsweredevelopedthat could identify more disease,the liver-spleenscanswereshownto be far less sensitive. MRI Diagnostic

Criteria

Over the last 3-year period, our researchhas emphasized the developmentof refined criteria for the inter197

TABLE 2. Lesions identified in breasts mastectomy pathologic analysis

with serially

sectioned

Statistic

MRI

Conventional

True positives False positives True negatives False negatives Sensitivity Specificity Accuracy

44 17 10 3 94% (44/44+3) 37% (10/10+17) 72% (44+10/74)

26 3 24 21 55% (26/26+21) 89% (24/24+3) 67% (26+24/74)

imaging

eligible for breast-conservationsurgery.StageS1 could potentially be treatedwith lumpectomy alone. Stages S2 andS3 would probably be treatedwith lumpectomy with radiation therapy.If the lesion/breastsize ratio is large enoughto preventbreastconservation,stagesS1, S2, S3, and Ml could potentially undergo induction

FIG. 3. Subtraction breast images obtained on a current midfield scanner. T,-weighted spin-echo pulse sequences with 5 mm sections were used with a bilateral breast coil. Top, before contrast; middle, after contrast; and bottom, a subtraction. The area of slightly increased intensity in the lateral aspect of the left breast on the subtraction image was called tumor. However, even higher intensity can be seen in the heart and chest wall caused by misregistration artifact.

pretationof MRI images.A good analogyis the developmentof criteria for the diagnosisof meniscaltearson MRI. If all hyperintensesignals were called positive for a tear, then the sensitivity would be very high, but the large occurrence of asymptomatic globular hyperintensity would result in a low specificity. If we call positive only those linear areasof hyperintensity that contact a meniscal surface(gradeIII), then we can expecta specificity that approaches100%at theexpense of a somewhat lower sensitivity. Some linear hyperintensities that do not extend to a mensical surface (gradeII) will alsobe tears(falsenegatives).Many imaging developmentsduring the past 8 years since the original proposalof thesecriteria haveemphasizedimprovementsin resolution that aid in the discrimination betweengradeIII andgradeII signal intensities. MRI diagnostic criteria were developedfrom lesion distribution and structure.Lesion distribution was developedfrom previousclinical treatmenttrials andpathologic analysesthat correlateddiseaseextentwith treatment results. The lesion-distribution categoriesalong with the pathologic and MRI characteristicsare listed in Table 3. StagesSl, S2, and S3 could be potentially 198

therapywould result in downstagingto S1, S2, andS3. Thesestageswould potentially be treatablewith breastconservationsurgery.StagesMl, M2, and M3 would typically betreatedwith mastectomy.Completeresponse to chemotherapywould be classified as stage0. Induction chemotherapywould be used in some of the M2 andM3 breaststhat havelocally advancedcancer.Partial responseto chemotherapy would correspond to ~50% enhancementby volume comparedwith thepretreatment image. Partial responseto chemotherapyin thesegroupsusually resultsin a reclassificationto stages Ml, M2, M3, or M5. Morphologic categorizationis usedto bettercharacterize lesions and reducethe false-positivediagnoses. The morphologic patternsthat we have usedarelisted in Table 4. All publishedbreastMRI seriesindicate a high negative predictive value for breastMRI. Even when comparedwith serial-sectionpathologicanalysis,the malignantpotentialfor anonenhancinglesionis extremelylow. Of 74 lesions identified in our initial series,only one failed to enhance, a microscopic metastasis to an in&mammary lymph node.Thereforeanylesionthatdoes not enhanceis classifiedasbenign.Punctuateor stippled areasof enhancementare almost always benign(98%), particularly if they arewidely distributedthroughoutthe breastor within theupperouterquadrant.This enhancement pattern most commonly is associated with fibrocystic changebut canbe seenwith atypical hyperplasia,lobularcarcinomain situ,andoccasionallyDCIS. This patternis seenalso in premenopausalwomen and Curr Probl

Diagn

Radiol,

November/December

1996

TABLE

3. Extent

of disease

categories

with pathologic

Grade

Extent

0

No cancer

Sl

of disease

and MRI findings Pathologic No gross microscopic

Solitary cancer single quadrant

in a

cancer but to a single

findings or disease

Results

of RODEO

Solitary focus of abnormal enhancement single quadrant

Multiple confirmed quadrant

Multiple foci of abnormal enhancement in a single quadrant

in a

s2

Multiple confined quadrant

53

Larger (>l cm) single quadrant cancer with multi-quadrant small foci (4 cm) of disease

Solitary larger (>I cm) malignancy with EIC-positive or other small (4 cm) infiltrating lesions two or more quadrants

Solitary high probability larger focus (>l cm) with other foci of abnormal enhancement 4 cm in diameter two or more quadrants

Ml

Single large cancer involving more than one quadrant

Solitary malignancy extending over more than one quadrant

Solitary focus of abnormal extending over more than

M2

Multiple cancers involving more than one quadrant

Evidence of DCIS or infiltrating cancer (~1 cm) in more than one quadrant

Foci of abnormal enhancement more than one quadrant

(4

cm) in

M3

Extensive disease (>l cm) in more than one quadrant

Foci of gross disease (>I cm) in more than one quadrant

Foci of abnormal enhancement more than one quadrant

(>l

cm) in

can changeduring the menstrualcycle. Smoothly marginated enhancementis predictive of benign disease (95%). Smooth margination is often seenwith sclerosing adenosis,fibroadenoma,atypical hyperplasia,and fibrocystic change.Lobulation andseptationareusually associatedwith fibroadenomaandarehighly predictive of benign disease(95%). Mutinous carcinoma (2% of breastcancers)can havea similar appearance.The two mutinous carcinomasin our serieshad lobulatedmargins with internal septations.Theselesionsarealsohard to distinguishfrom fibroadenomaon physical examination, mammography,andsonography.Recentreportsindicate that thesemalignanciesalso havea dynamic enhancement pattern that is indistinguishable from fibroadenoma.We have found that coalescentglobules of abnormalenhancement(clumped) havebeenhighly associatedwith DCIS (83%). Another sign of DCIS is thearrangementof clumpedenhancement in alinearfashion along a duct extendinglike a ray towardthe nipple. This duct patternis highly predictive of DCIS involvement (100%) but was seenin only 15% of DCIS cases. Microcalcifications, seenas small foci of hypointensity amid the clumped enhancement,are associatedwith comedo-typeDCIS (90%). Spiculatedenhancementis highly predictiveof malignancy(85%),usually invasive Curr

Probl

Diagn

Radiol,

November/December

1996

of cancer in a single only

MRI

No significant enhancement

Solitary focus malignancy in a single quadrant foci

breast

enhancement one quadrant

cancer.Our invasivelobular carcinomasshowedspiculatedenhancementin 90% of cases.We hadtwo casesof microabscessesaftersurgicalbiopsythat showedspiculatedenhancement.Ring enhancementis evenmorepredictive of malignancy(90%)but wasmuch lesscommon (18%of malignancies)andalmostalwaysseenwith spiculation (15%). The useof morphologic andextentclassificationshas greatly improved the sensitivity, specificity, and accuracy of MRI diagnosiswhen comparedwith serial-section pathologic analysis.This classification also facilitatesthe practical implementation of MRI information andthe clinical managementdecisions.A true positive and a true negativearedefined asa match betweenimaging interpretation and pathologic studies(benign or malignant histology and extent determined by serial sectioning).When the imaging study overestimatesthe extentof malignancycomparedwith pathologicresults, it is called a false positive. When the imaging study underestimatesthe extentof disease,it is categorizedas a false negative.These criteria were appliedto a series of 82 serially sectionedbreasts;theresultsaretabulated in Table 5. Some of the mammograms were not available or exhibited suboptimal quality and were not consideredin the mammographystatistics. 199

TABLE MRI

4. MRI morphologic

finding

and usual

pathologic

finding

Usual

No enhancement Tiny stippled enhancement Smoothly marginated enhancement Lobulated enhancement Septated enhancement Clumped globular enhancement Clumped interspersed with tiny magnetic susceptibility effects Ductal pattern enhancement Ring enhancing Spiculated

TABLE 5. Breast MRI morphologic compared with mammography Statistic True positives False positives True negatives False negatives Sensitivity Specificity Accuracy

MRI 63 3 6 6 94% (68/68+6) 66% (6/6+3) 90% (68+6/82)

and extent

pathologic

finding

Benign Benign, usually fibrocystic, change Benign Benign, usuallyfibroadenoma Benign, usually fibroadenoma Malignant, usually DCIS Malignant, usually comedo-type DCIS Malignant, Malignant, Malignant,

classification

usually usually usually

DCIS infiltrating infiltrating

cancer cancer

MRI as a tool for the management therapy.

of breast cancer

Mammography 20 2 1 27 42% 33% 42%

Diagnostic

(20/20+27) (l/1+2) (20+1/50)

These data demonstrate a significant benefit in the diagnostic accuracy for MRI when the morphologic and extent criteria are employed. As compared with our original analysis, MRI still performs with twice the sensitivity of mammography yet with a considerable improvement in specificity. The specificity improved to 66% compared with 37% by lesion-enhancement criteria alone. Of concern is the increase in the false negatives for MRI. In fact, only one MRI case failed to show any lesion when the pathologic studies were positive. In this case, induction chemotherapy produced a significant response that was called complete by MRI but showed microscopic foci of disease by pathologic examination. The remainder of the MRI cases that were called false negative actually represented underestimation of disease extent. The major impact of the lesion classification is the limitation of the effect of false-positive MRI examinations. By considering disease extent, a more practical exploitation of breast MRI data can be implemented for medical decision analysis.36-39

Clinical Applications The clinical application of breast MRI for cancer diagnosis is generally divided into two groups of issues: (1) MRI as a supplement to conventional breast imaging for the diagnosis of breast cancer, and (2) 200

diagnosis.

Issues

Mammography is established as the imaging method of choice for breast cancer screening. Although mammography is known as a highly sensitive tool for breast cancer detection, it has a known false-negative rate. The BCDDP study revealed that about 9% of palpable cancers were not seen by means of mammography.40 Certain clinical and histologic subgroups have been established as difficult to evaluate by mammography. Dense Breasts. The limitations of mammography in the evaluation of dense breasts is well recognized in the breast-imaging field. 41.42Breast density, according to the American College of Radiology Breast Imaging Reporting and Data System (Lexicon), is categorized in four levels based on the potential sensitivity of mammography: (1) almost entirely fat, (2) scattered fibroglandular elements that could obscure a lesion, (3) heterogeneously dense that may lower sensitivity of mammography, and (4) extremely dense that lowers the sensitivity of mammography. 43 Breast density that significantly impairs mammographic interpretation is estimated to occur in 25% of the population. 44 Dense breasts are generally more common in younger, hormonally active women. Dense breasts can, however, be seen in postmenopausal women and are more common with hormonal supplementation.45,46 The limitations of dense breast tissue for mammography do not exist for MRI. An example of the improved demonstration of breast cancer by MRI in mammographically dense breasts is shown in Fig. 4. Silicone. The attenuation of x-rays by silicone limits the mammographic demonstration of breast cancer.46-48 This problem is most severe in breasts with silicone injections. Specialized push-back or Eklund mammographic views have improved detection in breasts with Curr Probl

Diagn

Radiol,

November/December

1996

FIG.

4. Mammographically

dense

cytokeratin-positive adenocarcinoma no masses or abnormal calcifications. all four classified

quadrants. as M3.

breasts

in a 36-year-old

of likely Sagii-tal

woman

with

breast primary. Mediolateral projection RODEO image

Probl

Diagn

mass.

Radiol,

November/December

1996

The

biopsy

oblique mammogram (6) demonstrates extensive

Most of the areas of abnormal enhancement show spiculation, and The patient was treated with induction chemotherapy before mastectomy.

silicone implants. 47Even with special views, however, a portion of the breast remains obscured by the implant. 46-48Silicone does not impair MRI for cancer detection.‘9-22,24-26,29,49 Special silicone sequences can be used to provide specific information on the composition of a suspected mass (Fig. 5). When these sequences are used in combination with contrast enhancement, masses in patients with silicone implants can be readily characterized (Fig. 6). PostoperativeScaP:Densities attributable to scarring are often difficult to evaluate by mammography. Scars are routinely seen after surgery, either lumpectomy or biopsy. The clinical question of surgical scar versus recurrent tumor is commonly encountered in a busy breast practice. Several studies have demonstrated a benefit for MRI in this application when the patient has a clinical Curr

an axillary

some

of these

axillary

lymph

(A) shows very dense breast areas of abnormal contrast are

ring

enhancing.

The

MRI

nodes

showed

parenchyma enhancement

with in

distribution

was

and mammographic problem.38s50-54 An example of a patient with suspected recurrent tumor versus scar by mammographic and clinical examinations is shown in Fig. 7. Therapeutic

Issues

Breast-ConservationSurgery.With advances in breast cancer detection, the potential is seen for breast-conserving surgery (i.e., lumpectomy, partial mastectomy, or segmental mastectomy with the motivation of producing less deformity than mastectomy). Several modem prospective, randomized trials have compared mastectomy and breast-conservation treatment as defined by primary tumors of ~4 or 5 cm with no distant metastases or fixed axillary nodes.55-61 Whole-breast irradiation to doses of 4500 to 6000 cGy was used in all of 201

FIG.

6. Silicone

implants

and

cancer.

This

woman

with

silicone

plants had a palpable mass thatwas suspected to be a silicone ultrasound. The fat-suppressed (leff) and silicone-suppressed sagittal RODEO of an intracapsular

(A) images rupture

(/eff) and postcontrast (B) show a spiculated

FIG. 5. Silicone leak. A, woman with a palpable silicone leak was imaged with fat-suppressed

mass (left)

suppressed (right) slightly hyperintense

images (arrow)

suppressed images silicone-suppressed chemically masses. RODEO [arrows)

RODEO. The sagittal silicone implant (left). The implant sagittal RODEO

RODEO and leak

and suspected and silicone-

demonstrates the implant.

and leak(arrow) suppress on the image (right). This information is

the larger

leak

and several

capsule,

(right) sagittal fat-suppressed RODEO enhancing mass superior to the implant

as Sl in extent. Typically surgery. In this case, and

as evidence Precontrast

a mastectomy

was

Sl lesions the tumor

images (arrow),

are amenable to breastextended to the implant

performed.

demonstrate on the fat-

specific for silicone and can be used to characterize suspected The coronal reformation (B) of the silicone-suppressed images around

categorized conservation

show the linguine sign (arrow) but no evidence of free silicone.

im-

leak by (right)

smaller

leaks

the trials. Long-term results with follow-up times of 8 to 10 years demonstrate overall survival and relapsefree survival between the arms were not statistically different in any of the trials. Recurrence in the breast after breast-conservation treatment ranged from 3% to 19% compared with local recurrence after primary mastectomy of 2% to 9%. The clinical determination of which patients are candidates for breast-conservation treatment

versus mastectomy is highly dependent on the imaging findings. Women with malignancies in separate quadrants of the breast would not generally be candidates for breast conservation. Although tumor size is not an absolute contraindication to breast conservation, there is little evidence to support conserving surgery in tumors > 4 to 5 cm. A large tumor in a small breast is a relative contraindication because of the significant cosmetic alteration.Accurate definition of tumor size, number, and margins is highly critical in determining the staging for successful breast-conservation treatment. Tumor Size and Multiple Tumors. Accurate determination of the extent of disease is important in the planning for breast-conservation surgery. The size of Curr

Probl

Diagn

Radiol,

November/December

1996

FIG. 7. Recurrent tumor versus scar. In a patient with a previous lumpectomy surgical clips on the mediolateral oblique mammogram (A). Th e sagittal nonenhancing mass (curved solitary mass

mass anterior to the surgical clips (straight arrow) as evidence of scar. However, posterior to the clips is a spiculated enhancing arrow) that was not noted on the mammogram. This mass was found to represent recurrent infiltrating ductal carcinoma. The would be categorized as Sl in extent. A salvage mastectomy was performed.

the primary tumor in breast-conservationsurgerydoes not seemto haveany effect on recurrencerate.Kurtz et a1.61 reported,in a seriesof 783patients,equalrecurrence ratesfor Tl (12 cm diameter) andT2 (2 to 5 cm diameter).Veronesiet al.56observedno difference in localregional recurrencesin tumors 14.5 cm in diameter. The incidenceof multicentric breastcanceris reported to rangefrom 9% to 75%.62-6g In patientswith clinically occult,nonpalpablebreastcancersdetectedon mammography, Schwartzet a1.64 reporteda multicentricity incidencerate of 44%. Of the tumors found to be microinvasive on pathologic examination, 57% were multifocal. In another series from Memorial Sloan Kettering CancerCenter,60% of patientsin whom mastectomieswere performed for in situ cancerhad multicentric disease.65 Lagios et al.@examinedmastectomy specimensfor cancerswith tumor foci outside a 5 cm radius of the referencetumor, the hypothetical border of abreastquadrant.Multifocality was found in 20% of cases.Their findings closely approximated those of Rosenet a1.,67 who reportedresidualtumor in 26% and 38% of patientswith referencetumors smaller than and largerthan 2 cm in diameter,respectively.The presence Curr

Probl

and radiation therapy, a spiculated mass was noted anterior to the precontrast (left) and postcontrast [right) RODEO images (B) show a

Diagn

Radiol,

November/December

1996

of multiple primary cancershasbeenassociatedwith an increasedrisk of local recurrenceafterbreastconservation. Leopold et a1.68 reporteda recurrencerate of 40% with multiple lesions comparedwith 11% with one lesion. Similar findings were observedby Kurtz et a1.69 when 25% of patents with multiple tumors had local recurrencecomparedwith 11%recurring in thosewith single lesions. These findings indicate a fundamental differencein the capability of breast-conservationsurgery when multiple primary tumors arepresent. The powerof MRI in the determinationof multifocal diseaseis shown in Fig. 8. The additional information provided by breastMRI may allow better definition of treatment categoriesfor more efficient and effective delivery of therapy.The determinationof diseaseextent by MRI staging may be used to define the treatment protocols, as shownin Fig. 9. Extensive Intraductal Component.The presenceof extensiveintraductalcomponent(EIC) hasbeenassociatedwith an increasedrisk of recurrence.Recht et a1.7o reportedthatin patientswith early-stage breastcancer25% of the EIC-positive tumors recurredcomparedwith 5% in the EIC-negativegroup. Kurtz et a1.69 found that the 203

FIG. 8. Multifocal disease. This postmenopausal oblique mammogram (A) showed no focal mass, postcontrast lesions

around

(right)

RODEO

the larger

images mass

(8) show

OS evidence

woman had an axillary mass found to represent and no palpable masses were noted on physical

a 3 cm spiculated of multifocal

disease.

mass.

local recurrence rate of EIC-positive patients was 18%, and the EIC-negative patients had recurrences of only 7.9%. Lindley et a171had similar results with rates of 22% for EIC positive and 10% for EIC negative. Holland et a1.72 compared the histologic findings in mastectomy specimens of EIC-positive and EIC-negative patients. For EIC-positive patients, 59% had residual carcinoma present >2 cm from the index lesion compared with 29% for EICnegative specimens. In addition, EIC-positive patients had residual carcinoma at the 4 cm distance from the index tumor in 32% of cases compared with only 12% for EICnegative cases.At 6 cm farther from the index lesion, 21% of the EIC-positive group had residual carcinoma compared with 8% for the EIC-negative group. Vicini et a1.73 compared local recurrence rates forT1 andT2 lesions for EIC status depending on the size of the resection. For Tl tumors with small resections, the EIC-positive patients with small resections had a local recurrence rate of 29% compared with 10% for EIC-negative patients. ForT2 lesions, EIC-positive patients with small resections had a local recurrence rate of 3%, but if a large resection was done, the local recurrence rate was 9%. The EIC status is an indicator of the potential for increased subclinical disease and a higher potential for positive margins. For these reasons, many surgeons prefer to perform a larger resection or mastectomy with EIC-positive histologic results. This problem might more appropriately be addressed by a more accurate imaging assessment of disease extent before surgery so that the size of the resection can be matched to the extent of disease. MRI may be useful in determining the extent of intraductal disease and infiltrating carcinoma to provide 204

The

The disease

sagittal extent

projection

metastatic adenocarcinoma. examination. The sagittal image

(C) demonstrates

in this case would

be classified

The mediolateral precontrast(/eft) and several

small

satellite

as S2 in MRI extent.

more appropriate therapy. Spiculation and ring enhancement are features that are more commonly associated with infiltrating carcinoma. Coalescent clumped enhancement is usually associated with intraductal carcinoma. Other features seen with intraductal carcinoma include the alignment of enhancement along a ductal ray and microcalcifications. An example of an infiltrating cancer and DCIS in the same breast is shown in Fig. 10. Ductal Carcinoma In Situ. The increased use of screening mammography during the past 2 decades has led to a marked increase in the number of patients diagnosed with DCIS and today, in many centers, one DCIS lesion is diagnosed for every two to three mammographically detected invasive breast cancers. Multiple DCIS lesions are often present. However, these lesions are typically in a segmental pattern and are thus multifocal (within the same quadrant) rather than multicentric (more than one quadrant). The distinction between DCIS and DCIS with microinvasion or even frankly invasive cancer is problematic. The histologic distinction between DCIS and in lobular carcinoma in situ (LCIS) is sometimes difficult. Microinvasion can be mimicked by artifact, duct sclerosis, epithelial entrapment, microvessel proliferation, periductal fibroblasts, and mechanical implantation in needle tracts. Microinvasion is one of the most commonly revised diagnoses on pathologic review. This difficulty in histologic categorization has led to controversy in the interpretation of several treatment trials for DCIS.74 The use of MRI for the detection of DCIS is controversial.‘2,20 Dynamic imaging often cannot demonstrate Curr

Probl

Diagn

Radiol,

November/December

1996

Suspicious mammographic or clinical lesion I

I

FIG. 9. Algorithm

for MRI selection

of treatment

DCIS, whereasstudies that use high-resolution imaging reportsuccessfulimaging. This differenceis best attributed to the larger voxel and lower contrastusedin most dynamic techniquesthat result in substantialvolume averagingwhen imaging the microscopic distention of ducts by DCIS. We routinely usedRODEO for the demonstrationof DCIS. An example of the ability of RODEO to demonstratethe extentof DCIS is shown inFig. 11. Surgical Margins. To improve breastconservation, many surgeonsnow preferto matchthelumpectomysize to the estimatedsize of the mass,including a margin of normal parenchyma. The inability to accurately determinethe extentof diseaseby clinical and conventional breastimaging oftenresultsin aninadequatetumor resection.At leastfour large,retrospectivereviewshave demonstratedthat involved marginsleadto anincreased risk of local-regional recurrence,73,75-77 whereasother studiesindicateno effectfrom positivesurgicalmargins. Thesediscrepanciesprobably reflect differencesin the size of the initial tumor resection.Ghosseinet a1.76 reported recurrencerates of 41% for tumorectomy, 15% Curr

Probl

Diagn

Radiol,

November/December

1996

protocol

Needle biopsy

by disease

extent.

for wide local excision, and 14% for quadrantectomy. Thus the larger resectionsprovide greatercertainty for negative margins that is reflected in a decreasein the numberof local recurrences.Many of thestudiesdid not perform rigorous pathologic analysis to guaranteeadequateexcision. The fact that most breastrecurrences occur close to the original tumor site seemsto confirm this impression.The incidenceof positivemarginsin the lumpectomy specimenrangesfrom 30% to >95%. The wide variation is attributedto the differencesin the size of the specimen(lumpectomy vs quadrantectomy)and the extentof the pathologic examination.A reportfrom Japanindicatedthat when thoroughpathologic examinationwasperformed,thepositive-marginratewas95%. The occurrenceof positive margins could not be predictedby tumor size or distancefrom thenipple but did correlatewith the presenceof intraductaldiseaseandtumor necrosis.78The study of Schmidt-Ullrich et a1.79 evaluated108women with AK stageI andII invasive carcinomafor adequacyof thehistopathologicmargins. Inadequatemarginswerefoundin theinitial lumpectomy specimenin 32% of the Tl carcinomasand49% of the 205

FIG. 10. Infiltrating RODEO image(A) an infiltrating enhancement

and intraductal carcinoma.The sagittal postcontrast shows a spiculated, ring-enhancing mass typical of

carcinoma. A linear arrangement extends from the ring-enhancing

of coalescent clumped mass in a ray-like pat-

tern toward the nipple. This pattern is typical of an infiltrating carcinoma with extensive intraductal component. The extent of the abnormal enhancement is well demonstrated on the sagittal postcontrast RODEO

projection

infiltrating ductal ductal carcinoma Because quadrant,

image

(B). Pathologic

examination

demonstrated

carcinoma in the region of ring enhancement and in situ in the region of the clumped enhancement.

the disease the extent

was multiple and extended was categorized as M3.4’

over

more

than

one

T2 carcinomas. For patients with tumors >2 cm in diameter in which reexcision was required, 7 1% had residual carcinoma. Incomplete tumor excisions and residual microscopic carcinoma may be associated with higher recurrence rates, as suggested by the tendency of larger tumors to recur more frequently.79 The problem of positive or close margins is being addressed at many centers with the use of a radiation therapy boost to the primary site, but no randomized prospective studies validate this concept. Although false-positive scanning with MRI has been reported in the postoperative breast of 4 year,52 we have found RODEO to be a helpful adjunct in patients with positive lumpectomy mar206

FIG.

11. Ductal

carcinoma

in situ. The craniocaudal

mammogra

shows a small cluster of microcalcifications (arrow) typical other areas of abnormality were seen on the mammogram. tal precontrast (left) and large area of coalescent of the images

breast. The demonstrates

The extent enhancing

axial

of disease foci typical

Curr

postcontrast (right) RODEO (B) images clumped enhancement in the superior reformation multicentric

was classified of infiltrating

Probl

Diagn

m (4

of DCI S. No The Ijagits1low a a spect

(C) of the postcontrast RC lDE0 coalescent clumped enhancer nent. as M3, but no spiculated or ringcarcinoma were identified.

Radiol,

November/December

1996

gins.38This hasbeenparticularly useful in patientsreferredfor mastectomybecauseof positivemargins.MRI provides information on the size, location, and extent of the residual diseaseand may be used to guide the reexcision andavoid mastectomy (Fig. 12).The ability of MRI to define andlocalize tumors may be best employed for the more accurateremovalof primary tumors to preventpositive lumpectomy margins.This information may be used to developan algorithm for the more effectivemanagementof treatmentresources(Fig. 13). MI&Directed Biopsy and Localization. Coreneedle biopsies approacha failure rate of only 1%. The accuracy of coreneedlebiopsies is comparablewith that of surgery.A variety of prototypestereotacticdeviceshave been built for MRI-directed breastbiopsy and needle localization.80-85 Thesedevicesgenerally consist of the following components:breast immobilization, lesion localizationfrom MRI coordinatesto spatialcoordinates, and needleguidance.Localization methodsare highly variable.All methodsuse some form of fiducial markersthat referencethebiopsy systemto the MRI coordinate system. Corrections for gradient nonlinearity are neededfor accurateneedlelocalization. MRI-compatible localization wires andbiopsy needlesarecurrently underinvestigationandarenot yet FDA approved;however,approvalis expectedsoon. MRI localization is expectedto become an integral part of breastcancermanagement.A potentialalgorithm for the use of MRI information with stereotacticMRI localization is shownin Fig. 14. Radiation Therapy. The theoreticreasonfor radiation therapyin breast-conservation treatmentis to supplement minimal surgerywith amoderateradiationdoseto eradicateresidualmicroscopic foci of diseasewhile preserving the cosmetic appearanceof the breast.It is preferable to limit the extent of breast surgery as much as possible,

Consistent

with

obtaining

lOCal

tumor

Control

with radiation theraov.Killing of cells bv radiation is anexponentialfunction.Thereforethe doserequiredfor a certain level of tumor control is directly proportional to thelogarithm of the numberof clonogeniccells.Thus microscopic extensionsof diseasecanbe controlled by a lower dosethan macroscopicdisease.In addition,microscopicdiseaseis much lesslikely to containhypoxic foci that arelesssusceptibleto irradiation.It is estimated that a radiation doseof 5000 cGy given in 5 weeks to the entirebreastis sufficient to control 75% of subclinical disease.86 Given that the radiation dosetoleratedby atissueis inverselyrelatedto the volume irradiated,the radiation therapist must choosebetween the risks of I,

Curr

Probl

Diagn

Radiol,

November/December

,

1996

FIG. 12. Recent lumpectomy formed 2 weeks before this tive

margins

on the

with positive examination

lumpectomy

specimen.

margins. revealed Because

A lumpectomy histologically the surgery

perposiwas

performed at an outside institution, the extent and location of the positive margins were unclear. The sagittal precontrast (left) and postcontrast (right) RODEO images (A) and axid reformatted postcontrast RODEO image (B) clearly enhancing mass

show the seroma (straight arrows) (curved arrow) representing the

and the spiculated residual infiltrating

ductal carcinoma. The residual tumor was categorized as Sl in extent. Successful reexcision of the mass was performed instead of a mastectomy.

marginalrecurrencesaroundsmall volumesof high dose, centralrecurrencesin largevolumes of low dose,or excessivenormal tissue damagein largevolumes of high dose.It is a common practice to supplementthe radiation treatmentto the whole breastwith a boost doseto the primary site such that the tumor area (region of lumpectomy) receives6000 cGy. This strategyderives 207

FIG.

13. Algorithm

for

the care

lumpectomy is 10%. With Sl -extent tumors, radiation

of clinical

and

mammographically

MRI, it is hoped that this rate of recurrence therapy may be avoided.

determined

from the assumption that the “subclinical” disease consists of microscopic foci that are effectively treated with radiation therapy.87 However, as stated previously, more extensive subclinical disease may be present. Several treatment trials confirmed the benefits of radiation therapy. The NSABP-BOG protocol was a prospective, randomized study designed to examine the outcome of invasive breast carcinoma treated with mastectomy, lumpectomy alone, or lumpectomy and irradiation. This trial showed no difference in survival and recurrence rate for women treated with mastectomy compared with lumpectomy with radiation therapy. The recurrence rate in the lumpectomy with radiation therapy group was 12%, whereas the lumpectomy-alone group had a 43% local recurrence rate. The incidence of distant recurrences for both lumpectomy groups was the same. These findings confirm that radiation therapy can be used effectively to treat the residual subclinical disease.8s Although the randomized trials demonstrated a local control benefit of radiation therapy for some, the cancer of the majority of patients treated with excision alone does not recur. The NSABP-06, which demonstrated the most improvement for radiation therapy only, benefited -30% of patients (the disease in -10% recurred even with radiation therapy, and that in 60% did not recur with excision alone). These analyses of the study results and the costs, inconvenience, morbidity, and potential late side effects of radiation therapy have prompted some groups to evaluate the selection of patients for which excision alone may be sufficient treatment. Previous trials indicated that even when the treatment fails, the recurrence will be 208

lumpectomy

can be maintained

candidates.

or reduced

even

Without when

MRI, the recurrence

less therapy

is administered.

rate

for For

local. In the event of a recurrence, the radiation therapy could be reserved for the recurrent disease. Randomized clinical trials of excision alone are under way for the evaluation of invasive tumors c2 cm.8s-94MRI may have substantial benefit for the exclusion of significant subclinical disease and in patient follow-up. As discussed previously, substantial disease can be present yet remain clinically occult. The probability of treatment failure due to more extensive residual tumor can be predicted by the presence of grossly positive or diffusely positive microscopic margins. Other factors that are associated with an increased recurrence rate because of substantial residual disease include histologic patterns such as EIC, invasive lobular carcinoma, and comedo carcinoma. An imaging method is needed to estimate the extent of residual disease and to predict the potential for treatment of subclinical disease with radiation therapy. An example of the ability of those who use breast MRI to define the extent of disease and better determine the appropriate therapy is shown in Fig. 15. Radiation Therapy for DCIS. The NSABP-B 17 trial, which randomized lumpectomy with and lumpectomy without radiation for DCIS, found the 5-year recurrence rate to be lower in the women treated with irradiation (7.5% vs 10.4%). Factors that could be used to predict recurrence included comedo-type and involved/uncertain margins.89-95 Subsequent pathologic review of the NSABP-B 17 trial found the inclusion of atypical ductal hyperplasia and DCIS with microinvasion. Another criticism of the NSABP-B 17 trial is the lack of systemCurr

Probl

Diagn

Radiol,

November/December

1996

LUMPECTCW LOcALlzAnON

I I

I

I

Ml, M3 EXTENT

Sl, SZ,S3 EXTENT

LOCALIZATION 1

I

L I

FIG. would Better

14. Algorithm allow

better

for the use of stereotactic

MRI localization

categorization

extent

MRI localization

will probably

of disease result

in fewer

and positive

compared

potentially margins

avoid and

atic recording of size and extent of DCIS. The shortterm benefitsof radiation therapyin reducingthe number of local recurrencesafterexcisionalbiopsy for DCIS havebeenwell documented,but this benefit appearsto decreasewith longerfollow-up. Lagioset a1.97,98 reported that DCIS of ~25 mm are not associatedwith occult invasion or axillary metastasisat mastectomy,suggesting that excision alone may be sufficient treatmentfor selectedpatients.Severalinvestigatorshave examined the results of breast-conservingtreatmentof DCIS using surgical resection alone.74,89-95.99-103 These studies indicated that when using surgerywith careful evaluation of marginsandthoroughspecimenexaminationfor evidence of microinvasion, treatment results can be achievedthat arecomparablewith thoseof excisionwith radiation therapy.Lagios concluded,“Given the recurrencerates availablefrom the published literature at 8 and 10 yearsit may be more appropriateto reserveradiation therapy for invasive recurrencesshould they occur.“74 Induction Chemotherapy. “Locally advancedcancer of the breast” refers to breastcarcinomaswith significantprimary or nodaldiseasein which distantmetastases cannot be documented (International Union against Cancer [UICC] and American Joint Commission on Cancer Staging and End Results Reporting [AJC] stage-T3b to T4, N2 or N3, MO). These cancershave been shown to be poorly controlled by surgery alone. Patients with locally advanceddiseasemay undergo combination chemotherapybeforemastectomy(inducCurr

Probl

Diagn

Radiol,

November/December

1996

with

current

localization reduce

localization for

the need

methods.

lumpectomies for

repeated

The that

will

use of MRI before likely

have

positive

localization margins.

surgery.

tion chemotherapy).lo4-lo7 The NSABP recently performed atrial (B-21)comparinginductionchemotherapy beforebreast-conservationsurgerywith mastectomy.108 One of the conceptsof induction chemotherapyin this applicationis to reducethetumor burdensothat breastconservationsurgeryis feasible.Unfortunately,chemotherapy often results in breast edema and softensthe tumor, which confoundsthe estimationof responseand extentof residualdiseaseby clinical andmammographic methods.108-110 An imaging examination that can accurately determine tumor bulk may have a role in determining the responseto chemotherapyandthe extentof residualdiseasebeforebreast-conservation surgery.This examinationcould be usedto adjustchemotherapeutic regimensif a responseis unsatisfactory.109-111 In some patients, the diseaseis a solitary mass that canbe palpatedand shownon mammograms.On MRI these would be Sl or Ml in extent. These massesoften reducein size concentrically when treatedwith induction chemotherapy(Fig. 16). The patient with this appearancewould typically follow a protocol similar to the NSABP B-21 in which the therapy is designed to reducetumor size to facilitate breastconservation. Most patientswith locally advanceddiseasedo not have solitary massesbut instead have diffuse diseasethat cannot be readily palpated or assessed by mammograms.Whenthis diseaseis treatedwith induction chemotherapy,it does not concentrically reduce in size. Instead,the respondingbreasttransforms into fibrosis and necrosis with islands of residual tumor 209

FIG. 15. Improved detection of subclinical disease with breast MRI. The mediolateral oblique mammogram ing shows a small spiculated mass posteriorly in the breast. The oblique reformatted postcontrast mammographically additional spiculated intraductal

detected mass (arrow) and another larger enhancing lesion in a different quadrant

carcinoma.

The

larger

yet mammographically

and

spiculated mass that was slightly clinically

graphic information, this patient would qualify for breast for no radiation therapy in some protocols. This decision

conservation. would have

treated.

the patient

However,

with the MRI information

of M2 extent,

elected

(Fig. 17). This posttreatment breast can be very difficult to evaluate by clinical and conventional imaging methods. MRI has considerable promise as a method for accurately evaluating chemotherapeutic response and residual disease. RODEO MRI was evaluated as a tool for determining tumor response and extent of residual disease after neoadjuvant chemotherapy. Forty breasts in 39 patients with stage II, III, or IV breast carcinoma were prospectively evaluated before and after neoadjuvant chemotherapy by MRI, physical examination, and mammography. Assessment of response determined by the three methods was compared. In addition, detailed pathologic correlation of residual disease was determined by serial sectioning of 3 1 mas210

silent The small left behind

anteriorly [curved smaller than that

lesions

were

(A) obtained RODEO image

arrow). shown.

infiltrating

during routine screen(B) demonstrates the

Another section demonstrated Biopsy of the smaller lesion

ductal

carcinomas.

From

an found

the mammo-

mammographically detected lesion would allow the potential an infiltrating carcinoma that may not have been adequately

to have

mastectomy

and

reconstruction.

tectomy specimens from 30 patients. Nine patients had breast conservation and were included in the response evaluation only. Estimates of tumor response were made by both surgical and medical oncologists. Independent interpretations of MRI studies without knowledge of clinical response were made by three radiologists. The surgical oncologists assessed complete response (CR), partial response (PR), and no response (NR) in 11,22, and 7 cases, respectively. The medical oncologists assessed CR, PR, and NR in 12,21, and 7 cases, respectively. The surgical and medical oncologists’ clinical assessment of response agreed with the results of MRI in 52% and 55% of cases, respectively, and with each other in 30 (75%) of 40 cases. Mammography correCurr Probl

Diagn

Radiol,

November/December

1996

before breast conservation. The precontrast (left) and postcontrast (right) sagittal FIG. 16. lndl Jction chemotherapy ring-enhancing mass with Sl extent. The ratio of lumpectomy to breast size was considered strut ‘e a spit ulated, chemotherapy before lumpectomy was selected as the appropriate treatment in a effort cosr netic res ult. Induction improve ! the cosmetic action in the abnormal response

result. The precontrast (left) and postcontrast (right) sagittal RODEO (B) images enhancement. Reduction >50% in the volume of the abnormal enhancement

Probl

(A) ir nages

demonfor an acceptable the tumor burden after chemotherapy st iow concentric would be categorizt :d as a partial

to chemotherapy.

lated to MRI response in only 52% of cases. However, MRI accurately predicted the pathologic determination of residual disease in 30 (97%) of 31 cases. There was no disagreement in the assessments of residual disease or response among the three radiologists. RODEO breast MRI accurately estimates residual disease after induction chemotherapy. It better assesses response to neoadjuvant chemotherapy than traditional methods of physical examination or mammography. The information obtained from this MRI technique may be used as an objective tool during clinical trials and better to select patients for breast conservation after neoadjuvant chemotherapy for locally advanced disease. Curr

RODEO

too large to reduce

Diagn

Radiol,

November/December

1996

Summary MRI can significantly improve the detection of breast cancer. The information provided by MRI can supplement a conventional breast diagnostic and assist in the management of the disease. The enhanced understanding of breast cancer gained with MRI could be used to improve therapeutic methods for breast cancer. A wide range of success with breast MRI has been reported in the radiologic literature. As with many MRI methods, the clinical capability varies with the quality of the images. Unfortunately, commercial packages continue to stress low-resolution, rapid-scanning methods that are easily implemented. Techniques that use fat211

FIG.

17. lndu ction chemotherapy demonstrate (A) it nages

RODEO disease

exten t is classified

before diffuse

as M3 in extent.

The precontrc 1st (/effj and postcontrast abnormal enl hancement and would however, imaging enhancemeni

has not changed. met1 lads because

mastectomy abnormal

in addition

(right) sagittal be categorized

This replacement the size of the

by MRI corresponds

Notice

for locally enhancement

advanced breast cancer. throughout the breast

that the enhancement

RODEO (9) images as a partial response

extends

to the skin as evidence

after chemotherapy to chemotherapy.

show a >50% The distribution

of viable tumor by fibrosis and necrosis is often difficult palpable mass or density of the mass does not change.

with the extent

of residual

tumor

on serial-section

suppressiol I, magnetization transfer, rapid scanning, high resolution, and high SNR are difficult to obtain on most 212

The precontrast (/eff,l and postcontrast (right) sagittal as evidence of locally advanced breast cancer. This

pathologic

of inflammatory

overall reduction of the abnormal

carcinoma. in the volume enhancement,

to assess by clinical and The reduction in volume

of

conventional of abnormal

analysis.

systems available today. Perhaps the greatest challenge for the future is the integration of low-cost, high-qualCurr

Probl

Diagn

Radiol,

November/December

1996

ity MRI breast-imaging to existing hardware.

protocols that can be adapted

139-55.

Acknowledgement This work was supported in part by a grant from National Cancer Institute (Grant No. ROl-CA9735-01).

the

References 1. American Cancer Society. Cancer facts and figures 1994. Atlanta: American Cancer Society, 1994: 13. 2. Miller BA, Reis LAG, Hankey BF, Kosary CL, Edwards BK, editors. Cancer statistics review: 1973-1989. Bethesda (MD): National Cancer Institute, NIH Pub. No. 92-2789, 198. 3. Miller BA, Feuer EJ, Hankey BE Recent incidence trends for breast cancer in women: an update. CA Cancer J Clin 1993;43:27-41. 4. Bovee WM, Creyghton JH, Getreuer KW, Korbee D, Lobregt S, Smidt J, et al. NMR relaxation and images of human breast tumours in vitro. Trans R Sot Biol 1980;289:535-6. 5. Mansfield P, Morris PG, Ordidge R. Carcinoma of the breast imaged by NMR. Br J Radio1 1979;52:242-3. 6. El Yousef SJ, Alfidi RJ, Duchesneau RH, Hubay CA, Haaga JR, Bryan PJ, et al. Initial experience with nuclear magnetic resonance (NMR) imaging of the human breast. J Comput Assist Tomogr 1983;7:215-18. 7. El Yousef SJ, Duchesneau RH, Alfidi RJ, Haaga JR, Bryan PJ, LiPuma JP. Magnetic resonance imaging of the breast: work-in-progress. Radiology 1984;150:761-6. 8. El Yousef SJ, Duchesneau RH. Magnetic resonance imaging of the human breast: a phase I trial. Radio1 Clin North Am 1984;22:859-68. 9. Stelling CB, Wang PC, Lieber A, Mattingly SS, Griffen WO, Powell DE. Prototype coil for magnetic resonance imaging of the female breast: work-in-progress. Radiology 1985;154:457-62. 10. Turner DA, Alcorn FS, Adler YT. Nuclear magnetic resonance in the diagnosis of breast cancer. Radio1 Clin North Am 1988;26:673-87. 11. Kopans DB. Breast imaging, Philadelphia: JB Lippincott, 1989:356-9. 12. Kaiser WA, Zeitler E. MR imaging of the breast: fast imaging sequences with and without Gd-DTPA. Radiology 1989;170:681-6.

13. Heywang SH, Wolf A, Pruss E, Hilbertz T, Eiermann W, Permanetter W. MR imaging of the breast with Gd-DTPA: use and limitations. Radiology 1989;171:95-103. 14. Stack JP, Redmond OM, Codd MB, Dervan PA, Ennis JT. Breast disease: tissue characterization with Gd-DTPA enhancement profiles. Radiology 1990;174:491-4. 15. Parker SH, Lovin JD, Jobe WE, Burke BJ, Hopper KD, Yakes WE Nonpalpable breast lesions: stereotactic automated largecore biopsies. Radiology 1991;187:507-511. 16. Parker SH, Jobe WE, Dennis MA, Stavros AT, Johnson KK, Yakes WE US-guided automated large-core breast biopsy. Radiology 1993;187:507-11. 17. Jackman RJ, Nowels KW, Shepard MJ, Finkelstein SI, Marzoni FA. Stereotaxic large-core needle biopsy of 450 nonpalpable breast lesions with surgical correlation: I. lesions with cancer or atypical hyperplasia. Radiology 1994;193:91-5. 18. Pierce WB, Harms SE, Flamig DP, Griffey RH, Evans WP, Hagans JE. Gd-DTPA enhanced MR imaging of the breast: a new fat suppressed three-dimensional imaging sequence. Radiology 1991;181:757-63. Curr Probl

Diagn

Radiol,

19. Harms SE, Flamig DP, Hesley KL, Evans WP Magnetic resonance imaging of the breast. Magn Reson Q 1992;8(3):

November/December

1996

20. Harms SE, FIamig DP, Hesley K.L, Evans WP III, Cheek JH, Peters GN, et al. Fat suppressed three-dimensional MR imaging of the breast. Radiographics 1993;13:247-67. 21. Harms SE, Flamig DP, Hesley KL, et al. Breast MRI: rotating delivery of excitation off-resonance: clinical experience with pathologic correlations. Radiology 1993;187:493-501. 22. Harms SE, Flamig Dl? Staging of breast cancer with magnetic resonance. MRI Clin North Am 1994;2:4. 23. Bydder GM, Steiner RE, Blumgart LH, Khenia S, Young IR, et al. Imaging of the liver using short Tl inversion recovery sequences. J Comput Assist Tomogr 1985;9(6):1084-88. 24. Monticciolo DL, Nelson RC, Dixon WT, Bostwick J III, Mikundan S, Hester TR. MR Detection of Leakage from silicone breast implants: value of a silicone-selective pulse sequence. AJR 1994;163:51-9. 25. Gorczyca DP, Litwer CA, DeBruhl ND, McCombs M, Ahn CY, Bassett LW. Silicone breast implant ruptures: comparison between fast multiplanar IR and fast SE MR imaging [abstract]. Radiology 1994;193(P):318. 26. Keller PJ, Hunter WW, Schmalbrock P. Multisection fat-water imaging with chemical shift selective presaturation. Radiology 1987;164:539-41. 27. Dixon WT. Simple proton spectroscopic imaging. Radiology 1984;153:189-94. 28. Schneider E, Chan TW. Selective MR imaging of silicone with the three-point Dixon technique. Radiology 1993;187:89-94. 29. Derby KA, Frankel SD, Kaufman L, Carlson JW, Mineyev MI, Occhipinti KA, et al. Differentiation of silicone gel from water and fat in MR phase imaging of protons at 0.064 T. Radiology 1993;189(2):617-20. 30. Gorczyca DP, Schneider E, DeBruhl ND, Foo TKF, Ahn CY, Sayre JW, et al. Silicone breast implant rupture: comparison between three-point Dixon and fast spin-echo MR imaging. AJR 1994;162:305-10. 31. Maudsley AA, Hilal SK, Perman WH, Simon HE. Spatially revolved high resolution spectroscopy by “four dimensional” NMR. J Magn Reson 1983;51:147-51. 32. Frahm J, Haase A, Hanickle W, Matthaei D, Bonsdorf H, Helzel T. Chemical shift selective MR imaging using a wholebody magnet. Radiology 1985;156:441-4. 33. Hasse A, Frahm J. Multiple chemical shift selective NMR imaging using stimulated echoes. J Magn Reson 1985;64: 94-102. 34. Garrido L, Kwong KK, Pfleiderer B, Crawley AP, Hulka CA, Whitman GL, et al. Echo-planar chemical shift imaging of silicone gel prostheses. Magn Reson Imaging 1993;11(5): 625-34. 35.

36.

37. 38. 39.

Meyer CH, Pauly JM, Macovski A, Nishimura DG. Simultaneous spatial and spectral selective excitation. Magn Reson Med 1990;15:287-304. Abraham DC, Jones RC, Jones SE, Cheek JH, Peters GN, Knox SM, et al. Evaluation of neoadjuvant chemotherapeutic response in locally advanced breast cancer by magnetic resonance imaging. Cancer. In press. Soderstrom CE, Harms SE, Copit DS, Evans WP, Krakos PA, Farrell RS, et al. 3D RODEO breast MR imaging of lesions containing ductal carcinoma in situ. Radiology. In press. Soderstrom CE, Harms SE, Farrell RS, Prnneda JM, Flamig DP MRI detection of residual tumor in the recent postoperative breast. AJR. In press. Rodenko GN, Harms SE, Pnmeda JM, Farrell RS, Evans WP, Copit DS, et al. MRI in the pre-operative management of 213

40. 41. 42. 43. 44. 45. 46. 47. 48. 49.

50.

51.

52.

53.

54. 55.

56.

57.

58.

214

lobular carcinoma of the breast: comparison with mammography. AJR. In press. Baker LH. Breast cancer detection demonstration project: five year summary report. Cancer 1982;32(4):194-225. Tabar L, Dean PB. Mammographic parenchymal patterns: risk indicator for breast cancer. JAMA 1982;247:185-9. Jackson VP, Hendrick RE, Feig SA, Kopans DB. Imaging of the radiographically dense breast. Radiology 1993;188: 297-301. American College of Radiology. American College of Radiology breast imaging reporting and data system. Reston (VA): American College of Radiology. In press. Grace G, Roberts C, Cohen I. The role of mammography in detecting breast cancer in augmented breasts. Ann Plast Surg 1990;26:119-29. Hays H, Vandergrift J, Diner W. Mammography and breast implant: discussion. Plast Reconstr Surg 1988;82:7-8. Handel N, Silverstein M, Gamagami P, Jensen J, Collins A. Factors affecting mammographic visualization of the breast after augmentation mammoplasty. JAMA 1992;268:14-9. Eklund GW, Busby R, Miller S, Job J. Improved imaging of the augmented breast. AJR 1988; 151:469-73. Dershaw D, Chaglassian T. Mammography after prosthesis placement for augmentation or reconstructive mammoplasty. Radiology 1989;170(1):69-74. Harms SE, Jensen RA, Meiches MD, Flamig DP, Evans WP. Silicone-suppressed 3D MRI of the breast using rotating delivery of excitation off-resonance. J ComputAssistTomogr 1995;19:394-9. Dao TH, Rahmouni A, Campana F, Laurent M, Asselain B, Fourquet A. Tumor recurrence versus fibrosis in the irradiated breast: differentiation with dynamic gadoliniumenhanced MR imaging. Radiology 1993;187(3):7.51-5. Hansmann J, Brado M, Richter GM, von Fournier D, Kauffmann GW. MR mammography: Gd-DTPA uptake by breast tissue after breast conserving therapy [abstract]. Radiology 1994;193(P):121. Heywang-Kobrunner SH, Schlegel A, Beck R, Wendt T, Kellner W, Lommatzsch B, et al. Contrast-enhanced MRI of the breast after limited surgery and radiation therapy. J Comput Assist Tomogr 1993;17(6):891-900. Lewis-Jones GH, Whitehouse GH, Leinster SJ. The role of magnetic resonance imaging in the assessment of local recurrent breast carcinoma. Clin Radio1 1991;43(3): 197-204. Moore NR, Dixon AK, WheelerTK, Freer CE, Hall LD, Sims C. Axillary fibrosis or recurrent tumour: an MRI study in breast cancer. Clin Radio1 1990;42( 1):42-6. Fisher B, Redmond C, Poisson R, Margolese R, Wolmark N, Wickerham L, et al. Eight-year results of a randomized clinical trial comparing total mastectomy and lumpectomy with or without irradiation in the treatment of breast cancer. N Engl J Med 1989;320:822-8. Veronesi U, Banfi A, Del Vecchio M: Saccozzi R, Clemente C, Greco M, et al. Comparison of Halsted mastectomy with quadrantectomy, axillary dissection, and radiotherapy in early breast cancer: long-term results. Eur J Cancer Clin Oncol 1986322: 1085-g. Sarrazin D, Le MG, Arriagada R, Contesso G, Fontaine F, Spielmann M, et al. Ten-year results of a randomized trial comparing a conservative treatment to mastectomy in early breast cancer. Radiother Oncol 1989;14:177-84. Blichert-Toft M. A Danish randomized trial comparing breast conservation with mastectomy in mammary carcinoma [abstract]. Br J Cancer 1990;62 Suppl 12: 15.

59. Blichert-Toft M, Brincker H, Andersen JA, Andersen KW, Axelsson CK, Mouridsen HT, et al. A Danish randomized trial comparing breast-preserving therapy with mastectomy in mammary carcinoma: preliminary results. Acta Oncol 1988;27:671-7. 60. Bader J, Lippman ME, Swain SM. Preliminary report of the NC1 early breast cancer (BC) study: a prospective randomized comparison of lumpectomy (L) and radiation (XRT) to mastectomy (M) for stage I and II BC [abstract]. Int J Radiat Oncol Biol Phys 1987;13 Suppl 1:160. 61. Kurtz JM, Amalric R, Brandone H, Ayme Y, Spitalier JM. Local recurrence after breast-conserving surgery and radiotherapy. Helv Chir Acta 1989;55:837-42. 62. Lagios MD, Westdahy PR, Rose MR. The concept and implications of multicentricity in breast carcinoma. In: Sommers SG, Rosen PP, editors. Pathology annual. New York: Appleton-Century-Crofts, 1981:83-102. 63. Holland R, Veling SHJ, Mravunac M, Hendricks JHCL. Histologic multifocality of Tis, Tl-2 breast carcinomas: implication for clinical trials of breast-conserving surgery. Cancer 1985;56:979-90. 64. Schwartz GF, Patchesfsky AS, Feig SA, Shaber GS, Schwartz AB. Multicentricity of nonpalpable breast cancer. Cancer 1980;45:2913-16, 65. Rosen PP, Braun DW Jr, Kinne DE. The clinical significance of preinvasive breast carcinoma. Cancer 1980;46:919-25. 66. Lagios MD, Richards VE, Rose MR, Yee E. Segmental mastectomy without radiotherapy: short-term follow-up. Cancer 1983;52:2153-79. 67. Rosen PP, Fracchia AA, Urban JA, Schottenfeld D, Robbins GE “Residual” mammary carcinoma following simulated partial mastectomy. Cancer 1975;35:739-47. 68. Leopold KA, Recht A, Schnmitt SJ, Connolly JL, Rose MA, Silver B, et al. Results of conservative surgery and radiation therapy for multiple synchronous cancers of one breast. Int J Radiat Oncol Biol Phys 1989;16:11-16. 69. Kurtz JM, Jacquemier J, AmaIric R, Brandone H, Ayme Y, Hans D, et al. Breast-conserving therapy for macroscopitally multiple cancers. Ann Surg 1990;212:38-44. 70. Recht A, Pierce SM, Abner A, Vicini F, Osteen RT, Love SM, et al. Regional nodal failure after conservative surgery and radiotherapy for early-stage breast carcinoma. J Clin Oncol 1991;9:998-996. 71. Lindley R, Connolly J, Geman R. Histologic features predictive of an increased risk of early local recurrence after treatment of breast cancer by local tumor excision and radical radiotherapy. Surgery 1989; 105: 13-20. 72. Holland R, Connolly JL, Gelman R, Mravunac M, Hendricks JH, Verbeek AL, et al. The presence of an extensive intraductal component following a limited excision correlates with prominent residual disease in the remainder of the breast, J Clin Oncol 1990;8(1):113-18. 73. Vicini FA, Eberlein TJ, Connolly JL, Recht A, Abner A, Schnitt SJ, et al. The optimal extent of resection for patients with stages I or II breast cancer treated with conservative surgery and radiotherapy. Ann Surg 1991;214:200-5. 74. Lagios MD, Margolin FR, Westdahl PR, Rose MR. Mammographically detected duct carcinoma in situ: frequency of local recurrence following tylectomy and prognostic effect of nuclear grade on local recurrence. Cancer 1989;63:618-24. 75. Veronesi U, Volterrani F, Luini A, Saccozzi R, Del Vecchio M, Zucali R, et al. Quadrantectomy versus lumpectomy for small size breast cancer. Eur J Cancer 1990;26:671-3. 76. Ghossein NA. Alpert S, Barba J, Pressman P, Stacey P, Lorenz Curr

Probl

Diagn

Radiol,

November/December

1996

77.

78.

79.

80.

81.

82.

83. 84. 85.

86. 87.

88.

89.

90. 91. 92.

93.

E, et al. Importance of adequate surgical excision prior to radiotherapy in the local control of breast cancer in patients treated conservatively. Arch Surg 1992;127:41 l-15. Kurtz JM, Amalric R, Delouche G, Pierquin B, Roth .I, spitalier JM. The second ten years: long term risk of breast conservation in early breast cancer. Int J Radiat Oncol Biol Phys 1987;13:1327-32. Haga S, Makita M, Shimizu T, Watanabe 0, Imamura H, Kajiwara T, et al. Histopathological study of local residual carcinoma after simulated lumpectomy. Surg Today 1995;25:329-33. Schmidt-Ullrich R, Wazer DE, Tercilla 0, Safaii H, Mar-chant DJ, Smith TJ, et al. Tumor margin assessment as a guide to optimal conservation surgery and irradiation in early stage breast carcinoma. Int J Radiat Oncol Biol Phys 1989;17: 733-8. Heywang-Koebrunner SH, Halle MD, Requardt H, Oellinger HJ, Fischer U, Viehweg P, et al. Optimal procedure and coil design for MR imaging-guided transcutaneous needle localization and biopsy [abstract]. Radiology 1994;193(P):267. Fischer U, Vosshenrich R, Bruhn H, Funke M, Oestmann JW, Grabbe EH. Breast biopsy guided with MR imaging: experience with two stereotaxic systems [abstract]. Radiology 1994;193(P):267. Fischer U, Vosshenrich R, Keating D, Bruhn H, Doler W, Oestmann JW, et al. MR-guided biopsy of suspect breast lesions with a simple stereotaxic add-on device for surface coils. Radiology 1994; 192( 1):272-3. Hussman K, Renslo R, Phillips JJ, Fischer HJ, Khalkhali I, Braslau DL, et al. MR mammographic localization. Radiology 1993;189(3):915-7. Schnall MD, Ore1 SG, Connick TJ. MR guided biopsy of the breast. MRI Clin North Am 1994;4:585-90. Ore1 SG, Schnall MD, Newman RW, Powell CM, Torosian MH, Rosato EF. MR imaging-guided localization and biopsy of breast lesions: initial experience. Radiology 1994;193:97-102. Fletcher GH. Clinical dose-response curves of human malignant epithelial tumours. Br J Radio1 1973;46:1-12. Solin LJ, Fowble BL, Schultz DJ, Goodman RL. The significance of the pathology margins of the tumor excision on the outcome of patients treated with definitive irradiation for early stage breast cancer. Int J Radiat Oncol Biol Phys 1991;21:279-87. Fisher B, Redmond C, Poisson R, Margolese R, Wolmark N, Wickerham L, et al. Eight-year results of a randomized clinical trial comparing total mastectomy and lumpectomy with or without irradiation in the treatment of breast cancer. N Engl J Med 1989;320:822-8. Veronesi U, Saccozzi R, Del Vecchio M. Comparing radical mastectomy with quadrantectomy, axillary dissection, and radiotherapy in patients with small cancers of the breast. N Engl J Med 1981;305:1097-101. Silverstein MJ, Furmanski M, Gierson ED. Cosmetic quadrantectomy: a new surgical approach for mid-sized breast cancer. Proc Surg Sot Oncol 1993;46:159. Gelber RD, Goldhirsch A. Radiotherapy to the conserved breast: is it avoidable if the cancer is small? J Nat1 Cancer Inst 1994;86:652-4. Liljegren G, Holmberg L, Adami H-O, Westman G, Graffman S, Bergh J. Sector resection with or without postoperative radiotherapy for stage I breast cancer: five year results of a randomized trial. J Nat1 Cancer Inst 1994;86:717-22. Cady B, Stone MD, Wayne J. New therapeutic possibilities

Curr Probl

Diagn

Radiol,

November/December

1996

in primary invasive breast cancer. Ann Surg 1993;218: 338-49. 94. Moffat FL, Ketcham AS, Robinson DS, Legaspi A, GhandurMnaymneh L, Hilsenbeck S. Segmental mastectomy without radiotherapy for Tl and small T2 breast carcinomas. Arch Surg 1990;125:364-9. 9.5. Reed MW, Morrison JM. Wide local excision as the sole primary treatment in elderly patients with carcinoma of the breast. Br J Surg 1989;76:898-900. 96. Fisher B, Costantino J, Redmond C, Fisher E, Margolese R, Dimitrov N, et al. Lumpectomy compared with lumpectomy and radiation therapy for the treatment of intraductal breast cancer. N Engl J Med 1993;328:1581-6. 97. Lagios MD, Westdahl PR, Margolin FR, Rose MR. Duct carcinoma in situ: relationship of extent of noninvasive disease to the frequency of occult invasion, multicentricity, lymph node metastases, and short-term treatment failures. Cancer 1982;63:619-24. 98. Lagios MD. Duct carcinoma in situ; pathology and treatment. Surg Clin North Am 1990;70:853-71. 99. Amesson LG, Smeds S, Fagerberg G, Grontolt 0. Follow-up to two treatment modalities for ductal cancer in situ of the breast. Br J Surg 1989:76:672-5. 100. Carpenter R, Boulter PS, Cooke T, Gibbs NM. Management of screen detected ductal carcinoma in situ of the female breast. Br J Surg 1989;76:564-7. 101. Silverstein MJ. Cohlan BJ, Gierson ED, Furmanski M, Gamagami P, Colburn WJ, et al. Duct carcinoma in situ: 227 cases without microinvasion. Eur J Cancer 1992;28:630-4. 102. Schwartz GF, Finkel GC, Garcia JG, Patchefsky AS. Subclinical ductal carcinoma in situ of the breast: treatment by local excision and surveillance alone. Cancer 1992;70: 2468-74. 103. Amesson LG, Smeds S, Fagerberg G. Follow-up of two treatment modalities for ductal cancer in situ of the breast. Br J Surg 1989;76:672-5. 104. Bonadonna G, Conceptual and practical advances in the management of breast cancer. J Clin Oncol 1989;7(10):1380-97. 105. Schwartz GF, Cantor RI, Biermann WA. Neoadjuvant chemotherapy before definitive treatment for stage III carcinoma of the breast. Arch Surg 1987;122:1430-4. 106. Hortobagyi GN, Ames FC, Buzdar AU, Kau SW, McNeese MD, Paulus D, et al. Management of stage III primary breast cancer with primary chemotherapy, surgery, and radiation therapy. Cancer 1983;51:2507-16. 107. Lopez MJ, Andriole DP, Kraybill WG, Khojasteh A. Multimodal therapy in locally advanced breast carcinoma. Am J Surg 1990;160:669-74 108. Fisher B, Anderson S. Conservative surgery for the management of invasive and noninvasive carcinoma of the breast: NSABP trials. World J Surg 1994;18:63-9. 109. Singletary SE, McNeese MD, Hortobagyi GN. Feasibility of breast conservation surgery after induction chemotherapy for locally advanced breast carcinoma. Cancer 1992;69: 2849-52. 110. Mauriac L, Durand M, Avril A, Dilhuydy J-M. Effects of primary chemotherapy in conservative treatment of breast cancer patients with operable tumors larger than 3 cm. Ann Oncol 1991;2:347-54. 111. Gilles R, Guinebretiere JM, Toussaint C, Spielman M, Rietjens M, Petit JY, et al. Locally advanced breast cancer: contrast-enhanced subtraction MR imaging of response to preoperative chemotherapy, Radiology 1994;191:633-8.

215