Finding Early Invasive Breast Cancers: APracticalApproach

Finding Early Invasive BreastCancers: A Practical Approach1

Jennifer A. Harvey, MDBrandi T. Nicholson, MDMichael A. Cohen, MD

Detection of early invasive breast cancer is important, aspatient survival is high when the cancer is 2 cm or smaller.Invasive breast cancers typically manifest mammographi-cally as focal asymmetries or masses. Strategies for detect-ing focal asymmetries and masses on screening mammo-grams include side-by-side comparison, looking for parenchy-mal contour deformity, close inspection of the retromammaryfat, identifying the presence of associated findings, andcomparison with prior mammograms. Focal asymmetriesare often normal but are concerning when there is distor-tion of the normal breast architecture. Masses and focalasymmetries are best evaluated in the diagnostic setting byusing spot compression and true lateral views and, fre-quently, ultrasonography. Management of a lesion de-pends on the worst imaging feature. Indications for anassessment of probably benign findings are very specificbut are often misapplied. This review for residents pro-vides a practical approach to the detection and manage-ment of breast masses and focal asymmetries.

� RSNA, 2008

1 From the Department of Radiology, University of VirginiaHealth Sciences Center, PO Box 800170, Charlottesville,VA 22908. Received February 21, 2006; revision re-quested April 24; revision received January 5, 2007; ac-cepted February 9; final version accepted July 24; finalreview by J.A.H. February 27, 2008. Address correspon-dence to J.A.H. (e-mail: [email protected]).

� RSNA, 2008

REVIEWS

ANDCOM

MENTARY

�REVIEW

FORRESIDENTS

Radiology: Volume 248: Number 1—July 2008 61

Note: This copy is for your personal non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.

Invasive breast cancer typically man-ifests as a mass at mammography.Before developing into a mass, a can-

cer may manifest as a focal asymmetry.Early detection of these masses andasymmetries that represent invasivecarcinoma is important in reducingbreast cancer mortality. Women withinvasive cancers of 1 cm or smaller havea 95% chance of survival at 10 years,while those with invasive cancers 1–2cm and 2–5 cm in size have, respec-tively, 85% and 60% survival at 10years (1). For large invasive cancers,survival is dependent on tumor grade(1). However, survival is high for womenwith small invasive cancers, whetherthe cancer is low or high grade (2).Thus, detection of small invasive can-cers substantially affects breast cancermortality.

The Breast Imaging Reporting andData System (BI-RADS) (3) defines amass as “a space occupying lesion seenin two different projections.” The BI-RADS lexicon states that a focal asym-metry “is visible as a confined asymme-try with a similar shape on two views,but completely lacking borders and theconspicuity of a true mass” (3). A poten-

tial mass is termed a focal asymmetry ifit does not meet the three-dimensionalcriteria of a mass. Thus, if a lesion hasdefinable borders but is only visualizedin one projection, it is considered a focalasymmetry.

The detection of breast masses andfocal asymmetries on screening mam-mograms is largely a challenge of per-ception. Once a lesion is detected, man-agement is relatively straightforward.This is in contrast to breast calcifica-tions, where perception of the finding isnot as difficult, but disposition is morechallenging. Diagnostic evaluation of breastmasses and focal asymmetries includesadditional mammographic views and of-ten breast ultrasonography (US).

This Review for Residents is intendedto provide a practical approach to the de-tection and management of breast massesand focal asymmetries. We will describean approach to screening mammographywith the goal of improving identificationof breast masses and focal asymmetries,develop strategies for evaluating breastmasses and focal asymmetries in the di-agnostic setting, and describe properapplication of the BI-RADS assessmentand recommendation categories. Thetechniques and findings that we de-scribe pertain primarily to invasive duc-tal carcinoma, which is the most com-mon type of breast cancer. We havepreviously reviewed the findings of un-usual breast cancers, which may pro-vide clues to a diagnosis other than in-vasive ductal carcinoma (4).

An Approach to Detecting Masses andFocal Asymmetries at ScreeningMammography

Screening mammograms may be inter-preted “online” while the patient waitsfor results or may be batch interpreted,often on the next business day followingthe examination. Online interpretationof screening mammograms reduces pa-tient stress for those women with ab-normalities (5). Inefficiencies due to on-line interpretation of screening mam-mograms result in an increase in cost ofnearly $30 per study that most womenare not willing to pay (6). Batch inter-pretation is more efficient and results in

improved specificity with equal cancerdetection rates (7,8). The percentage ofU.S. facilities using batch interpretationincreased from 20% in 1992 (9) to 93%in 2002 (10).

A screening mammogram is twoviews of each breast: the craniocaudal(CC) view and the mediolateral oblique(MLO) view. Our protocol is to reviewthe current CC views side-by-side (withthe chest wall of the left breast next tothat of the right breast) and the MLOviews (also side-by-side, as for the CCviews) on the lower panel of the roller-scope, with a comparison study ob-tained 2 or more years earlier hung inan identical fashion on the panel above.Your protocol may be different, but itshould be consistent. Consistency aidsin lesion detection as one becomes ac-customed to reviewing studies with aparticular visual pattern.

A consistent approach to readingwill ensure that all aspects of the mam-mogram are reviewed. Ensure that youare reading in an optimal environmentusing high-luminance viewboxes (3500nits) for screen-film mammography orhigh-resolution monitors for digitalmammography, mask any extraneouslight around the images, and reduceroom light as much as possible. First,confirm that the name and other patientidentifiers are correct (ie, you are read-ing the correct study) and that the im-ages are of adequate quality. In mam-mography, the images must have goodcontrast, compression, positioning, andlack of blur or artifacts. In reviewing themammogram, an overview is helpful tocheck for obvious abnormalities andchanges from the previous study. Next,compare the MLO views side-to-sideand then the CC views (or vice versa) tolook for asymmetry in the breast paren-

Published online10.1148/radiol.2481060339

Radiology 2008; 248:61–76

Abbreviations:BI-RADS � Breast Imaging Reporting and Data SystemCC � craniocaudalMLO � mediolateral oblique

Authors stated no financial relationship to disclose.

Essentials

� Look for asymmetry, contour de-formity, lesions in the retromam-mary fat, and associated findingsto identify early breast cancer.Use studies 2 or more years oldfor comparison when available,but beware of stability. Suspiciousmorphology outweighs stability.

� Perform diagnostic evaluationwhen a concerning mass or focalasymmetry is identified at screen-ing mammography and include atrue lateral view, spot compres-sion views in both craniocaudaland mediolateral oblique projec-tions, and, often, a US study.

� Management depends on theworst imaging feature. The as-sessment of “probably benign” hasvery specific indications and isfrequently misapplied. Biopsyshould be considered when suspi-cious features are present.

REVIEW FOR RESIDENTS: Finding Early Invasive Breast Cancers Harvey et al

62 Radiology: Volume 248: Number 1—July 2008

chymal pattern, paying particular atten-tion to the retromammary fatty regions.Evaluate the contour of the breast pa-renchyma where it interfaces with fat.Finally, with the magnifying glass,closely examine the breast parenchymafor architectural distortion and calcifi-cations. If a finding is identified, com-pare with the prior mammograms to as-sess whether the finding is new, growing,or stable by using studies more than 2years older if need be.

Asymmetric Breast Tissue versus FocalAsymmetryAsymmetry in the breast may be globalor focal. Asymmetric breast tissue, orglobal asymmetry, in which the paren-chyma is overall greater in volume inone breast compared with the other, iscommon and usually normal. However,global asymmetry is of concern whenthere is associated architectural distor-tion, an apparent decrease in breastsize at mammography, skin thickening,axillary adenopathy, or clinical findings.Breast cancers manifesting as globalasymmetry are likely to be large.

Global asymmetry can be due to un-

derlying invasive lobular carcinoma,particularly when associated with archi-tectural distortion without a centralmass or an apparent decrease in breastsize (the “shrinking breast”) (11). Inva-sive lobular carcinoma accounts for6%–9% of all breast cancers (12,13).Most breast cancers enlarge en mass,like a ball getting larger, whereas inva-sive lobular carcinoma spreads in sin-gle-layer sheets of tumor cells similar toa spider web. The breast density mayincrease and the breast tissue becomesless compliant. This may result in archi-tectural distortion, manifested asstraight lines, typically in a radial pat-tern in the breast. Use of the magnifyingglass helps to identify subtle architec-tural distortion.

Detecting Focal AsymmetriesFocal asymmetry is also frequently a be-nign finding in the breast, representingsummation of normal tissue. However,early breast cancer may manifest as afocal asymmetry on screening images. Amass may become apparent on diagnos-tic images. The goal is to find breastcancers at this early stage.

Comparison.—If you wanted tocompare two different clothing items,you would put them next to each otherto look at the differences. Look for earlycancers by comparing the breasts fromside to side in the manner mentionedpreviously (Fig 1). This can be done asthe initial step in reading the mammo-gram after ensuring proper identifica-tion and quality.

Contour deformity: look for thehook, slow down for the speed bump.—Women with dense breast tissue typi-cally have an outward convex margin totheir breast parenchyma. Cancers maymanifest with distortion to this outwardmargin, similar to taking a crochet hookand pulling in the tissue toward the can-cer (Fig 2) (14). The subcutaneous fat isdrawn into this space, creating straightlines (architectural distortion) with anangular fat density invagination intobreast parenchyma. This may be theonly sign of a cancer hidden withinopaque breast tissue. When identified,additional imaging including spot viewsand US will usually unmask the hiddenlesion.

In contrast, women with scattered

Figure 1

Figure 1: Side-by-side comparison. Bilateral screening mammogram in a 69-year-old woman: CC (left) and MLO (right) views are hung side-by-side to evaluate forasymmetry. When comparing side to side (rectangles), asymmetry is identified in the left breast at 9 o’clock position (arrows), with associated architectural distortion. Atsubsequent diagnostic imaging, a suspicious mass was identified. Biopsy specimen showed invasive lobular carcinoma.



fibroglandular densities or heteroge-neously dense breast tissue typicallyhave a scalloped or concave margin tothe breast parenchyma. A breast massmay be detected by looking for an out-ward convex margin along the paren-chymal margin (Fig 3). Think of this as aspeed bump when you are following theedge of the parenchyma. Slow downand take a closer look at the area.

Living in the wrong neighborhood(triangle).—There are particular trian-gle-shaped zones of the breast that jus-tify a higher level of suspicion when afocal asymmetry is identified (Fig 4). Onthe CC view, these zones are the retro-mammary fat between the breast tissueand the chest wall and the medial orinner triangle (Fig 4). On the MLOview, these are the central space be-tween the breast tissue and the chestwall and the lower triangle (Fig 4).These are all typically fatty areas, so thepresence of a focal asymmetry in theseareas should raise suspicion. The retro-mammary fat area on the MLO view isoften referred to as “the Milky Way,”because we are searching for white

Figure 2

Figure 2: Contour deformity. (a) The contour ofdense breast parenchyma is typically outwardly con-vex. Invasive breast cancer may cause contour distor-tion, with the parenchymal margin appearing to havebeen pulled in as if with a crochet hook. (b) Left CCview in a 28-year-old woman with a palpable abnor-mality demonstrates transition from outward convexdense parenchyma (straight arrows) to a breast masscausing distortion in the contour (curved arrow). Corebiopsy specimen showed invasive ductal carcinoma.

Figure 3

Figure 3: Contour deformity. (a) The contour of the parenchymal margin in women with scattered fi-broglandular densities or heterogeneously dense breast tissue is typically scalloped. (b) This is shown on thelateral portion of a CC view, where the contour of the parenchyma is scalloped (line) except where a mass isvisualized with a convex margin (arrows). US demonstrated a simple cyst.



stars on an otherwise featureless nor-mal inky background (Fig 4).

The retroareolar area of the breastis very difficult to evaluate because ofthe complex anatomy of the region.Mammography is less sensitive forbreast cancer detection in this area(15). To improve detection of a subareo-lar cancer, having the nipple in profile onone of the two screening views (CC orMLO) may be helpful (Fig 5) (16). Oncewell-positioned images have been ob-tained, assess the retroareolar regionsfor asymmetry compared to the oppo-site side. Also be aware that not allround masses and densities in the ret-roareolar area represent nipples out ofprofile. True masses may look like nip-ples. Spot compression views of thesubareolar region with a radiopaquemarker on the nipple or with the nipplein profile will typically resolve whether amass is present.

Associated findings: looking formore clues.—When a potential lesion isidentified, the level of suspicion is in-creased when associated findings arealso identified. Architectural distortionthat is present with a focal asymmetry issuspicious (Fig 6). Calcifications associ-ated with a focal asymmetry may repre-sent ductal carcinoma in situ (calcifica-tions) associated with invasive carci-noma (focal asymmetry) (Fig 7). Thepresence of ipsilateral adenopathy like-wise increases suspicion, as the focalasymmetry may represent an invasivecancer with metastasis to axillary lymphnodes (Fig 8). Occasionally, axillary orintramammary adenopathy may be mis-taken for a primary breast carcinoma(Fig 9).

Use of prior studies: beware of sta-bility!—Breast cancers may growslowly. Therefore absence of substan-tial growth or change on a mammo-gram when compared to the previousyear’s mammogram may mislead oneto a false assumption that the lesion isbenign (Fig 10). Our standard practiceis to compare the current mammogramto studies that are 2 or more years old.If there is a finding for which change isdifficult to assess by using that previousmammogram, older studies are re-viewed. It is typical that comparison

Figure 4

Figure 4: Living in the wrong neighborhood (triangle). Masses or focal asymmetries in the retromammaryfat should be viewed with suspicion. This includes triangles at the bottom of the images. (a, b) A 10-mm inva-sive ductal carcinoma (arrow) is present in the (a) medial triangle on the CC view and (b) inferior triangle onthe MLO view of the left breast in a 66-year-old woman. (c) A 12-mm invasive tubular carcinoma (arrow) in theretromammary fat triangle on the CC view of the left breast in a 48-year-old woman. (d) An 8-mm invasiveductal carcinoma (arrow) in the retromammary fat triangle on the right MLO view in a 46-year-old woman.



Figure 5

Figure 5: Subareolar region. Bilateral mammogram obtained in a 67-year-old woman with new right nipple retraction. Left: A small mass (arrow) with architecturaldistortion is seen in the right subareolar region of the CC view. Right: The lesion is difficult to perceive on the MLO view because the nipple (arrow) is not in profile.

Figure 6

Figure 6: Architectural distortion. Left breast mammogram in a 68-year-old woman after remote right breast mastectomy with no current breast complaints. (a) Leftmammogram shows a focal asymmetry (arrows) in the central left breast. (b) Spot compression view in MLO projection shows numerous straight lines (arrows) associ-ated with the focal asymmetry consistent with architectural distortion. The lesion was assigned BI-RADS category 5 (highly suggestive of malignancy). Core-needle bi-opsy specimen showed an invasive ductal carcinoma, which measured 7 mm at excision.



Figure 7

Figure 7: Mass with calcifications. Close up ofa left CC view from a screening mammogram in a53-year-old woman shows a 6-mm oval, spicu-lated, equal-density mass (straight arrows), withassociated amorphous microcalcifications(curved arrow). Stereotactic-guided core-needlebiopsy specimen showed invasive ductal carci-noma with associated ductal carcinoma in situ.

Figure 8

Figure 8: Abnormal lymph node. (a) Right MLO view from a screening mammogram in a 67-year-oldwoman shows a 20-mm round ill-defined mass (straight arrow). An abnormally dense axillary lymph node isalso noted (curved arrow). (b) US image of the right axilla confirms focal areas of thickening (outward contour)of the lymph node cortex (arrows). Findings at US-guided core-needle biopsy of the right breast mass showedinvasive ductal carcinoma. Findings at US-guided core-needle biopsy with an 18-gauge Temno needle (Alle-giance HealthCare, McGaw Park, Ill) showed metastatic adenocarcinoma.

Figure 9

Figure 9: Bilateral mammogram in a 29-year-old woman with palpable lump in the right breast at 10 o’clock position. The obvious axillary mass represents metastaticadenopathy (straight arrows). Side-to-side comparison (rectangles) reveals focal asymmetry in the right upper outer quadrant (curved arrows), corresponding to thepalpable lump. Biopsy specimen of the focal asymmetry showed invasive ductal carcinoma, grade III.



with studies that are 4 or 5 years oldermakes clear that a finding has beenpresent for some time and is un-changed, and thus likely benign, or thatthe finding is indeed developing and re-call is indicated. The exception is that

when a round or oval mass is identifiedand a cyst is suspected, comparisonwith studies from the previous year ishelpful if the mass was shown at US tobe a simple cyst at that time and has notchanged in size since then. Note, how-

ever, that a finding that looks suspiciousshould not be discounted just becausesomeone worked it up last year and saidit was benign. In all cases, lesion mor-phology takes precedence over stability.If the morphology is suspicious, work up

Figure 10

Figure 10: Comparison mammograms. MLO viewsof the left breast demonstrate a slow-growing mass(arrows). The current mammogram (left) was not sub-stantially different from that of the previous year (mid-dle) but was considerably different from that of 3 yearsearlier (right). Biopsy specimen showed invasive lobu-lar carcinoma, grade I.

Figure 11

Figure 11: Normal focal asymmetry. (a) Bilateral digital mammogram obtained in a 42-year-old woman for baseline screening. A focal asymmetry is present in the leftbreast at 12 o’clock position, posterior third (arrows). The asymmetry respects normal architecture of the breast, with no straight lines or outward convex margins. (b)Looking at the black fat lobules (circles) rather than the white breast parenchyma can be helpful in assessing whether a focal asymmetry is respecting normal breast struc-tures. This was assigned BI-RADS assessment category 2 (benign finding) and has remained stable for 2 years of follow-up.



the lesion regardless of apparent stabil-ity.

When to ignore a focal asymme-try: respect for normal breast archi-tecture.—When a focal asymmetryrepresents summation of normal tis-sue, there will be no distortion of Coo-per ligaments or the regular, symmet-ric, undulating interface between glan-dular tissue and subcutaneous fat (Fig11). In other words, normal tissue willrespect normal boundaries. One ap-proach is to look at the black on theimage instead of the white. You shouldsee oval fat lobules, with soft curvedCooper ligaments separated bystrands of white breast tissue. Invasivebreast cancer grows through normalstructures and may result in abnormal

outward convex margins and straight-ening of Cooper ligaments. A focalasymmetry should be recalled for eval-uation, even if it appears to be re-specting normal breast structures, if itis new in comparison to prior mammo-grams, since early breast cancers maycause little distortion. Spot compres-sion views will help define if a focalasymmetry is behaving respectfully.

A focal asymmetry that has outwardconvex margins or is associated withstraight lines is not respecting the nor-mal architecture of the breast andshould increase suspicion (Fig 12).Breast cancer does not respect normalbreast structures but grows throughthem instead creating abnormal lines.These straight lines represent early

spiculation around a mass or focalasymmetry (Fig 12).

Leave-alone masses.—Fat-contain-ing masses in the breast are benign withrare exception if circumscribed andround or oval in shape (17,18). Thesemasses include: lymph node, lipoma, oilcyst, hamartoma, and galactocoele. If amass identified at screening clearly rep-resents one of these five, no additionalwork-up is needed. However, fat-con-taining masses are not necessarily be-nign if the margins are irregular or spic-ulated, since cancers can engulf fat asthey grow.

Volume counts.—As with any task,practice increases skill. Read as manymammograms as possible during yourresidency. At our institution, residents

Figure 12

Figure 12: Abnormal focal asymmetry. (a) Bilateral screening mam-mogram in a 62-year-old woman. A focal asymmetry (arrows) is presentin the left breast at 3 o’clock position, middle third. (b) At close evaluation,the focal asymmetry demonstrates lack of respect of normal breast archi-tecture, with outward convex margins (straight arrows) and straighteningof Cooper ligament lines (curved arrows). At subsequent diagnosticimaging, lesion appeared masslike on spot compression views, and acorresponding hypoechoic solid mass was seen at US (not shown). Core-needle biopsy specimen showed invasive ductal carcinoma.



spend their first month of breast imag-ing seeing diagnostic imaging patientsso they will see many breast cancers.The second month focuses on screeningmammography and breast procedures.The third month focuses on managingdifficult diagnostic cases, as well as fur-ther experience in screening and proce-dures. We expect our residents to readat least 500 screening mammogramsduring each of their second and thirdmonths. Even those who are planning toperform a fellowship in another area ofradiology are likely to read mammo-grams if they are in private practice.You will be much more confident if youhave first read at least 1000 mammo-grams with a mentor.

Double reading and computer-aideddetection.—Both double reading of mam-mograms and computer-aided detection,or CAD, improve sensitivity (19,20). ACAD system places marks on potentiallysuspicious calcifications and possiblemasses. Freer and Ulissey (20) found a19.5% increase in cancer detection withthe use of CAD in a community breastcenter, with an increase in screeningrecall rate from 6.5% to 7.7%. How-ever, the benefit can differ (21). If youare a resident at an institution that usesCAD, gain experience using the tool sothat you will be accustomed to its poten-tial benefits and pitfalls before enteringpractice.

Evaluating a Mass or Focal AsymmetryIdentified at Screening

The next step after identifying a mass orfocal asymmetry is diagnostic evalua-tion. Biopsy should not be recom-mended based on the findings of screen-ing mammography. Additional imagingin the diagnostic setting allows charac-terization of the mass or focal asymme-try to evaluate if the finding is real andto permit the imager to develop a levelof suspicion. If a lesion is highly suspi-cious for breast cancer on the screeningimages, recall for diagnostic evaluationis still useful to evaluate for the extent ofdisease, multifocality, and staging of theaxilla with US. In addition, diagnosticevaluation allows the radiologist to dis-cuss the findings with the patient, an-

Figure 13

Figure 13: Spot compression in two views. (a) Leftbreast mammogram in a 63-year-old woman with nobreast complaints. A focal asymmetry (solid circle) isseen superiorly on the MLO view, posterior third. Noobvious corresponding lesion is seen in the CC pro-jection. However, the most likely location can be local-ized by closely evaluating the region equal in distancefrom the nipple as the finding in the MLO view (ar-rows). The most likely location is in the lateral breast(dotted circle). (b) Spot compression view in CC pro-jection shows corresponding 8-mm oval equal-den-sity mass (arrow) with ill-defined margins at 2 o’clockposition in left breast. A corresponding oval hypo-echoic solid mass with ill-defined margins was identi-fied at US (not shown). Core-needle biopsy specimenshowed invasive ductal carcinoma.



swer questions, and arrange for biopsy.An important advantage to this ap-proach is that scheduling of proceduresis very efficient, since biopsies willrarely be cancelled if the lesion has beencharacterized at imaging prior to theprocedure. If a lesion is highly suspi-cious on the screening mammogram,recall of the patient should be priori-tized, however, so that minimal delayoccurs between screening and diagnos-tic evaluation.

Diagnostic Evaluation of Masses andFocal AsymmetriesSpot compression views.—All screeningrecall examinations with a mass or focalasymmetry should include spot com-pression views as well as a true lateralview. Spot compression views should beobtained in both CC and MLO projec-tions without coning. Obtaining spotcompression views in both projectionsis important because some cancers maynot appear masslike on a spot compres-sion view in one projection but may ap-pear spiculated in the other projection.If a focal asymmetry is seen on only oneview, a “best guess” at location may beapproximated by using the distance ofthe lesion from the nipple (Fig 13).Many breast imagers prefer to obtain

spot compression views with magnifica-tion, although this is not routine at ourinstitution.

Go to the lateral view for more in-formation.—The true lateral view maybe either a mediolateral or lateromedialview. Our standard protocol is to obtaina mediolateral view in all patients re-called for an abnormal screening exam-ination. However, if the lesion is locatedin the medial breast, then a laterome-dial view may provide better visualiza-tion of the lesion, since it will be closerto the film or digital receptor. Othermammographic views such as rolled,exaggerated CC lateral, and cleavageviews may also be useful in localizing alesion and to evaluate if a finding repre-sents summation of normal tissue or asuspicious lesion. US is frequently usefulfor confirming that a focal asymmetryrepresents normal tissue or characteriz-ing a mass as benign or malignant.

Lesions seen on only one view.—If alesion is seen only on the CC view, itmay be obscured on the MLO view ormay be medial and superior, since thisis where the MLO view is most likely tomiss tissue. Obtaining a true lateralview may reveal the finding. Rolledviews, in which the breast is rolled ei-ther medial or lateral in the CC projec-tion, may be helpful in localizing thelesion. The direction in which the lesionmoves on each rolled CC view providesa clue to the true lesion location.

If a lesion is seen only on the MLOview, it may be obscured on the CCview or may be far lateral, since thisarea may not be visualized on the CCview. An exaggerated CC lateral view isfrequently helpful because most breastcancers are located in the lateral breast.A true lateral view (mediolateral or lat-eromedial) is very useful for localizingthe lesion if not seen on the CC or exag-

Figure 14

Figure 14: Shift in position for medial andlateral lesion from MLO to mediolateral view. Me-dial lesions (star) will appear lower on the MLOview (line) with relationship to the nipple (F) thanthe true location. Lateral lesions (irregular massshape) will appear higher on the MLO view thanthe true location. Lesions centrally located in thebreast will shift little in position between MLO andmediolateral views, whereas far lateral or far me-dial lesions will shift considerably.

Figure 15

Figure 15: Use of true lateral view to localize lesions. Left: Right MLO view in a 66-year-old woman withhistory of prior benign breast biopsy findings (scar is marked by wire). A focal asymmetry with architecturaldistortion is present superiorly (circle). Right: Mediolateral view shows that wire marking the scar moveslower (solid arrow), indicating it is lateral in location, whereas the focal asymmetry rises, indicating it is in themedial breast (dashed arrow). Spot compression views (not shown) of the focal asymmetry revealed a spicu-lated mass; subsequent core-needle biopsy specimen showed invasive ductal carcinoma.



gerated CC lateral views (Fig 14). Le-sions in the lateral breast project higheron the MLO view than they are actuallylocated in the breast, because the MLOview is obtained at an oblique angle.Likewise, lesions in the medial breastproject lower on the MLO view thanthey are actually located in the breast.Thus, lesions that shift lower in positionon the mediolateral view are located inthe lateral breast while lesions that shifthigher in position on the mediolateralview are located in the medial breast

(Fig 15). This leads to the well-used ad-age, “Lead (lateral) sinks, muffins (me-dial) rise.” Note, however, that lesionsthat are more central in the breast(slightly lateral or slightly medial to thenipple) will shift little or not at all be-tween the MLO and mediolateral views.The best approach is to think about thegeometry of the views rather than byrelying on a saying. Lesions seen in onlyone projection can also be localized byusing step-oblique mammographic views,where mammographic views are ob-

tained at 15° intervals between the truelateral and CC views. Pearson et al (22)successfully localized 50 of 50 truemasses in three dimensions by usingthis technique.

Occasionally a lesion may be seenwell only in one projection, despite ob-taining additional mammographic views. Ifa lesion is suspicious at mammography,biopsy can often still be performed inmost cases by using stereotactic guid-ance. If the lesion is too far posterior,US or magnetic resonance (MR) imag-

Figure 16

Figure 16: Use of US with negative spot com-pression views. (a) Bilateral mammogram in a48-year-old woman with no breast complaintsshows focal asymmetry in the left breast at 12o’clock, posterior third (circles). (b) Spot com-pression views show no persistent abnormality.(c) US image shows 10-mm irregular solid masswith posterior acoustic shadowing (arrow). Core-needle biopsy specimen showed invasive ductalcarcinoma.



ing may be useful for localization. Place-ment of a marking clip at the time ofbiopsy and a subsequent mammogramdocumenting clip location can confirmthat the lesion identified at US or MRimaging represented the mammo-graphic finding.

Using US in the diagnostic setting.—Early use of US in the diagnostic breastevaluation setting was to differentiatebenign simple cysts from solid masses.The work of Stavros et al (23) definedbenign and malignant characteristics ofbreast masses, allowing more appropri-ate disposition of breast findings. At ourinstitution, we frequently use US as a

second check of a focal asymmetry iden-tified at screening mammography whenthe spot compression views do not showan underlying mass, especially withdense tissue (Fig 16). If a mass is notidentified at US, focal areas of shadow-ing or hypoechogenicity may be identi-fied that would increase suspicion or,alternatively, a corresponding island ofnormal breast tissue may be identifiedthat may reduce the level of concern(Fig 17). However, if a lesion is con-cerning at mammography, perhaps be-cause it is new, a negative US findingshould not influence one to not recom-mend biopsy.

MR imaging for problem solving.—Current indications for breast MR im-aging may include evaluation of extentof a known breast cancer, detection ofan unknown primary breast carci-noma in patients manifesting malig-nant adenopathy with a normal mam-mogram, or screening in high-riskwomen (24,25). Breast MR imaging isnot ordinarily indicated for routineevaluation of masses or focal asymme-tries on mammograms, and it shouldnever replace a thorough diagnosticevaluation including additional mammo-graphic views and US images. However,breast MR imaging can be useful insome circumstances, particularly whena lesion is quite suspicious on one viewbut cannot be localized despite appro-priate and complete additional imaging(Fig 18).

Management of Breast Masses andFocal Asymmetries

The Worst Feature WinsLesions in the breast should be managedbased on their worst features. If the mam-mogram finding is suspicious but the USexamination is normal, or vice versa, bi-opsy should still be considered.

Malignant FeaturesBiopsy should be performed if malig-nant features are present at mammog-raphy or US. On mammograms, thesefeatures include irregular shape, spic-ulated or irregular margins, or highdensity. On US images these includespiculation, angular margins, markedhypoechogenicity, posterior acoustic shad-owing, calcifications, duct extension,branch pattern, or microlobulation(23). For lesions with these character-istics, assessment of stability has norole—for example, “stable” and “spicu-lated” do not belong in the same sen-tence in a report. Lesions should beconsidered suspicious when the mar-gins are ill defined or microlobulatedwith any shape or density of lesion; bi-opsy should be considered.

For highly suspicious lesions (BI-RADS category 5), US may be usefulfor evaluating for additional masses

Figure 17

Figure 17: Use of US to confirm presence of normal tissue. (a) Left digital MLO view in a 42-year-oldwoman referred for biopsy of a left 12-o’clock focal asymmetry (arrow) on baseline mammogram. (b) Focalasymmetry appears to respect normal breast architecture on spot compression view, as it does not appearmasslike or have straight lines. (c) On US image, an island of normal hyperechoic breast tissue (arrows) isidentified that corresponds to the focal asymmetry in location. Biopsy was cancelled.



that may represent multifocal invasionand for abnormal lymph nodes in theaxilla. Remember that lymph entersthe node on the capsule, not the hi-lum, so an abnormal outward contourof the lymph node should be consid-ered with suspicion (Fig 8). The mostpredictive signs of axillary metastasisin lymph nodes are maximum cortex

thickness greater than 2 mm and ap-pearance of the cortex (26).

Benign FeaturesMasses have benign features if round,oval, or lobular in shape with circum-scribed margins. A focal asymmetryhas benign features when there are no out-ward convex margins or associated

straightening of Cooper ligaments.Having prior mammograms for com-parison is key in the disposition of le-sions with benign features. Routinefollow-up can be performed if the le-sion is stable for 2 or more years (BI-RADS 2: benign finding). However,when a lesion with benign features isnew, biopsy should be considered (BI-

Figure 18

Figure 18: Use of MR imaging to evaluate a “problem mammogram.” (a) Bilateral digital screening mammogram in 52-year-old woman shows small focal asymmetrysuperior on right MLO view, posterior third (circle); lesion could not be identified on the CC view (left). The lesion shifted up on the mediolateral view (not shown), indi-cating the lesion was in the medial breast. Despite considerable effort by experienced breast imagers, the lesion could not be localized at US. Stereotactic biopsy wasthought difficult due to far posterior location. MR imaging was performed to aid in localizing the lesion. (b) T1-weighted gradient-echo sagittal MR image with fat sup-pression (repetition time msec/echo time msec, 15.0/4.5) obtained 2 minutes after intravenous administration of gadolinium-based contrast agent shows irregular lesion(arrow) very high on the chest wall at 2 o’clock position in right breast. (c) Second-look US image shows a corresponding 5-mm hypoechoic solid mass (arrow). US-guided core needle biopsy specimen showed invasive ductal carcinoma.



RADS 4: suspicious) unless shown tobe a simple cyst at US examination(BI-RADS 2). If this was the patient’sfirst mammogram (baseline) or it isnot possible to obtain earlier studies,short-term follow-up imaging in 6months (BI-RADS 3: probably benign)is reasonable since the risk of malig-nancy is less than 2% (27).

Probably BenignThe BI-RADS 3 (probably benign) cat-egory is a frequently misapplied cate-gory (28). Probably benign does notequate to “I am not sure what to dowith this lesion.” If you are not sure,get more information by obtainingmore mammographic views or US im-ages or get input from someone else.The two primary indications for aprobably benign assessment are around or oval mass or grouped roundcalcifications on a baseline mammo-gram. It is also okay to consider alesion probably benign if it is far pos-terior and that portion of the breastwas not visualized previously. A roundor oval mass that is new is not “prob-ably benign” unless it represents aspecific benign diagnosis such as asimple cyst or lymph node.

The key to successful applicationof the BI-RADS 3 category is rigorousevaluation of the lesion characteristicsand strict adherence to the criteriadifferentiating benign from possiblymalignant. Many lesions that are as-signed a BI-RADS 3 classification andare ultimately determined to be malig-nant at follow-up examination were, inretrospect, misclassified as BI-RADS 3initially (29). Findings recalled fromscreening for diagnostic evaluationshould be assigned a category of BI-RADS 0 (needs additional evaluation),not BI-RADS 3, even if the level ofsuspicion is low that the finding repre-sents a cancer.

Summary

Asymmetric breast tissue is usually nor-mal but is worrisome when new orwhen there are associated findings, in-cluding architectural distortion, a palpa-ble lump, or thickening at clinical exam-

ination, or when the breast appears tobe smaller at mammography (“the shrink-ing breast”).

Look for asymmetry (side-by-sidecomparison), contour deformities (lookfor the hook, slow down for the speedbump), lesions in the retromammary fat(living in the wrong triangle), and asso-ciated findings to identify early breastcancer.

Use studies that are 2 or more yearsold for comparison when available, butbeware of stability. Suspicious morphol-ogy outweighs stability (“stable” and“spiculated” do not belong in the samesentence).

Diagnostic evaluation should be per-formed when a concerning mass or focalasymmetry is identified at screeningmammography and should include atrue lateral view, spot compressionviews in both CC and MLO projections,and often US.

If in doubt about the finding afterobtaining additional mammographic views,consider performing US. This will oftenlower or raise the level of suspicion. Ifstill in doubt, biopsy can often be per-formed with stereotactic guidance, evenif the lesion is seen only on one view.

References1. Tabar L, Duffy SW, Vitak B, Chen H,

Prevost TC. The natural history of breastcancer. Cancer 1999;86:449–462.

2. Tabar L, Chen H, Duffy SW, et al. A novelmethod for prediction of long-term outcomeof women with T1a, T1b, and 10–14 mminvasive breast cancers: a prospective study.Lancet 2000;355:429–433.

3. American College of Radiology. Breast imag-ing reporting and data system (BI-RADS) Re-ston, Va: American College of Radiology,2003.

4. Harvey JA. Unusual breast cancers: usefulclues to expanding the differential diagnosis.Radiology 2007;242:683–694.

5. Lindfors KK, O’Connor J, Parker RA. False-positive screening mammograms: effect ofimmediate versus later work-up on patientstress. Radiology 2001;218:247–253.

6. Raza S, Rosen MP, Chorny K, Mehta TS,Hulka CA, Baum JK. Patient expectationsand costs of immediate reporting of screen-ing mammography: talk isn’t cheap. AJRAm J Roentgenol 2001;177:579–583.

7. Burnside ES, Park JM, Fine JP, Sisney GA.The use of batch reading to improve the per-formance of screening mammography. AJRAm J Roentgenol 2005;185:790–796.

8. Ghate SV, Soo MS, Baker JA, Walsh R,Gimenez EI, Rosen EL. Comparison of recalland cancer detection rates for immediateversus batch interpretation of screeningmammograms. Radiology 2005;235:31–35.

9. Houn F, Brown ML. Current practice ofscreening mammography in the UnitedStates: data from the National Survey ofMammography Facilities. Radiology 1994;190:209–215.

10. Hendrick RE, Cutter GR, Berns EA, et al.Community-based mammography practice:services, charges, and interpretation meth-ods. AJR Am J Roentgenol 2005;184:433–438.

11. Harvey JA, Fechner RE, Moore MM. Appar-ent ipsilateral decrease in breast size onmammography: a sign of infiltrating lobularcarcinoma. Radiology 2000;214:883–889.

12. Berg JW, Hutter RV. Breast cancer. Cancer1995;75(1 suppl):257–269.

13. Hilleren DJ, Andersson IT, Lindholm K,Linnell FS. Invasive lobular carcinoma:mammographic findings in a 10-year expe-rience. Radiology 1991;178:149–154.

14. Tabar L, Dean PB. Teaching atlas of mam-mography. 3rd ed. New York, NY: Thieme,2001.

15. Giess CS, Keating DM, Osborne MP, Ng YY,Rosenblatt R. Retroareolar breast carcinoma:clinical, imaging, and histopathologic features.Radiology 1998;207:669–673.

16. Bassett LW, Hirbawi IA, DeBruhl N, HayesMK. Mammographic positioning: evaluationfrom the view box. Radiology 1993;188:803–806.

17. Lee EH, Wylie EJ, Bourke AG, Bastiaan DeBoer W. Invasive ductal carcinoma arising ina breast hamartoma: two case reports and areview of the literature. Clin Radiol 2003;58:80–83.

18. Breucq C,Verfaillie G, Perdaens C, VermeirenB, Stadnik T. Lobular carcinoma located in abreast hamartoma. Breast J 2005;11:508–509.

19. Thurfjell EL, Lernevall KA, Taube AA. Bene-fit of independent double reading in a popu-lation-based mammography screening pro-gram. Radiology 1994;191:241–244.

20. Freer TW, Ulissey MJ. Screening mammog-raphy with computer-aided detection: pro-spective study of 12,860 patients in a com-munity breast center. Radiology 2001;220:781–786.

21. Birdwell RL, Bandodkar P, Ikeda DM. Com-



puter-aided detection with screening mam-mography in a university hospital setting.Radiology 2005;236:451–457.

22. Pearson KL, Sickles EA, Frankel SD, LeungJW. Efficacy of step-oblique mammographyfor confirmation and localization of densitiesseen on only one standard mammographicview. AJR Am J Roentgenol 2000;174:745–752.

23. Stavros AT, Thickman D, Rapp CL, DennisMA, Parker SH, Sisney GA. Solid breastnodules: use of sonography to distinguish be-tween benign and malignant lesions. Radiol-ogy 1995;196:123–134.

24. Orel SG, Schnall MD. MR imaging of thebreast for the detection, diagnosis, and stag-ing of breast cancer. Radiology 2001;220:13–30.

25. Kriege M, Brekelmans CT, Boetes C, et al.Efficacy of MRI and mammography forbreast-cancer screening in women with a fa-milial or genetic predisposition. N EnglJ Med 2004;351:427–437.

26. Deurloo EE, Tanis PJ, Gilhuijs KGA, et al.Reduction in the number of sentinel lymphnode procedures by preoperative ultra-sonography of the axilla in breast cancer.Eur J Cancer 2003;39:1068–1073.

27. Sickles EA. Periodic mammographic fol-low-up of probably benign lesions: results in3184 consecutive cases. Radiology 1991;179:463–468.

28. Taplin SH, Ichikawa LE, Kerlikowske K,et al. Concordance of breast imaging report-ing and data system assessments and man-agement recommendations in screeningmammography. Radiology 2002;222:529–535.

29. Rosen EL, Baker JA, Soo MS. Malignant lesionsinitially subjected to short-term mammographicfollow-up. Radiology 2002;223:221–228.



Finding Early Invasive Breast Cancers: APracticalApproach

Documents

Transcript of Finding Early Invasive Breast Cancers: APracticalApproach