Ultrasound-Guided Fine-NeedleAspiration Biopsy Remains aValid Approach in the Evaluationof Nonpalpable Breast LesionsJim Liao, M.D., Ph.D.,1 Diane D. Davey, M.D.,1 Graham Warren, Ph.D.,1Joseph Davis,1 Angela R. Moore, M.D.,2 and Luis M. Samayoa, M.D.

1,3*

The use of ultrasound-guided fine-needle aspiration (FNA) biopsyfor nonpalpable breast lesions varies considerably. This retro-spective study stresses the role of breast FNA in evaluating sono-graphically suspicious nonpalpable breast masses using a proba-bilistic reporting system. One hundred and eight consecutiveultrasound-guided FNA biopsies diagnosed as positive (32), sus-picious (8), atypical (11), benign (55), and unsatisfactory (2) wereanalyzed and correlated with 61 subsequent surgical specimens.All positive cytologies showed carcinoma on histology; suspiciouscases were followed by 5 carcinomas, 2 fibroadenomas, and 1papillary lesion. Follow-up of atypical cases included 4 carcino-mas, 3 fibroadenomas, and 2 papillary lesions, while all 10 biop-sies following benign cytology showed fibrocystic changes. Twocases with suspicious sonographic findings but unsatisfactory cy-tology had lobular carcinoma. The remainder of the benign andatypical cases were followed clinically and radiographically for atleast 10 months and had no evidence of carcinoma. Positivepredictive values were positive, 100%; suspicious, 63%; atypical,36%; benign, 0%. Most (40/43; 93%) carcinomas were invasive.In conclusion, ultrasound-guided FNA for nonpalpable breastlesions is highly accurate, and probabilistic reporting helps directpatient management. Diagn. Cytopathol. 2004;30:325–331.© 2004 Wiley-Liss, Inc.

Key Words: breast; ultrasound; probabilistic reporting; nonpal-pable lesion; carcinoma; cytology

Screening mammography remains the standard of care forthe detection of nonpalpable breast abnormalities. However,its sensitivity and specificity vary according to the type of

lesion under evaluation. The introduction of breast ultra-sound to the workup of nonpalpable lesions has improvedthe sensitivity and specificity of this screening process, butthere are a number of cases requiring definitive tissue con-firmation. Nonpalpable breast abnormalities usually presentas new clusters of microcalcifications, new irregular densi-ties/masses, or a combination of both. Presently, these le-sions are evaluated by radiology-assisted (stereotactic andultrasound-guided) core needle biopsy (CNB) and fine-needle aspiration (FNA) biopsy as alternatives to excisionalbiopsies.1–3 Since microcalcifications are often related tointraductal processes,4,5 initial cytologic evaluation of theselesions becomes extremely difficult, and stereotactic CNB isgenerally superior. On the other hand, cytologic evaluationof lesions presenting as irregular densities or masses isgenerally comparable to CNB.3–5

A consensus statement on breast FNA for palpable andnonpalpable breast abnormalities was developed and ap-proved at a 1996 U.S. National Cancer Institute (NCI)-sponsored consensus conference.6,7 The recommendationstated that one may use FNA in women with mammogramsthat are “highly suggestive of malignancy, suspicious formalignancy, and some lesions at low risk for malignancybut for which the recommended follow-up with imaging isnot feasible or accepted by the patient.”6 Use of the tripletest (FNA, clinical, and imaging impression) is stronglyrecommended when the FNA technique is used.6 Despitethis statement, the practice of percutaneous CNB versusFNA as primary diagnostic tools for evaluating breast le-sions remains controversial.1–3

Breast ultrasound (US) relies on the transmission of spe-cialized sound waves using a high-frequency transducer topenetrate through and image the underlying breast tissue.The formation of the US image results from the varyingrates of reflection and refraction between the sound waves,tissue structures, and interfaces encountered by the wave as

1Department of Pathology and Laboratory Medicine, University of Ken-tucky Medical Center, Lexington, Kentucky

2Department of Diagnostic Radiology, University of Kentucky MedicalCenter, Lexington, Kentucky

3Department of Pathology, VA Medical Center, Lexington, Kentucky*Correspondence to: Dr. Luis M. Samayoa, Department of Pathology

and Laboratory Medicine, 800 Rose Street MS 157, Lexington, KY 40536.E-mail: lmsama1@uky.edu

Received 20 October 2003; Accepted 2 December 2003DOI 10.1002/dc.20068Published online in Wiley InterScience (www.interscience.wiley.com).

© 2004 WILEY-LISS, INC. Diagnostic Cytopathology, Vol 30, No 5 325

it penetrates the breast tissue. Therefore, the characteristicsof nonpalpable breast abnormalities presenting mammo-graphically as irregular densities, solid masses, or cysticprocesses are enhanced under this method.4 Microcalcifica-tions are rarely detected by US, but, when present, areusually related to malignant processes.5

Results from early reports indicate that when experiencedcytopathologists are available, US-FNA is an effective firstdiagnostic modality for nonpalpable breast lesions.7–9 Fur-thermore, use of FNA as a diagnostic modality substantiallydecreases complications associated with tissue biopsiessuch as hematoma formation, scarring, and the need toexcise skin and underlying breast tissue containing the coretract to prevent the seeding phenomenon. In addition, pro-cedure costs and diagnostic turnaround time are reduced.10

However, the effectiveness of US-FNA may vary with thetype of lesion targeted and the level of suspicion. Manyinstitutions have already applied the probabilistic classifi-cation of breast FNA endorsed by the NCI for breastFNA.7,11,12 In this system, FNA diagnoses are classified intoone of four diagnostic categories (benign, atypical/indeter-minate, suspicious/probably malignant, and malignant)based on the probability of underlying malignancy.6 Thissystem is believed to provide more meaningful informationto physicians in terms of patient care. This study investi-gates the value of US-FNA of nonpalpable breast lesionsusing well-defined ultrasonographic criteria for biopsy andthe probabilistic cytologic classification in a universityteaching hospital.

Materials and MethodsFrom January 2000 to August 2002, 108 consecutive casesof radiographically identified nonpalpable breast lesionsunderwent US-FNA with 22-gauge needles at the Univer-sity of Kentucky Medical Center Breast Clinic by special-ized radiologists. Lesions selected for US-FNA were non-palpable masses with specific sonographic featuresrequiring further evaluation. These features included lesionsshowing architectural distortion represented by irregular,angular, or microlobulated borders; lesions with taller thanwide dimensions; rapidly growing circumscribed masses;new lesions less than 1 cm in size suspected to representfibroadenomas; new lesions with a low index of suspicion inpatients with a strong family history of cancer; and lesionsin which the patient requested additional evaluation. Breastlesions in which the main finding was suspicious microcal-cifications, even those with architectural distortion, wereevaluated by core biopsy rather than by US-FNA.

Usually two to four passes were performed to ensureproper sampling; however, the appropriate number ofpasses was determined on site by the cytopathologist andradiologist. The cytopathologist prepared the air-dried andalcohol-fixed smears, and the needle was rinsed in a bal-anced saline solution (Plasmalyte). The air-dried smears

were stained by the Diff-Quik method and assessed imme-diately. The cytopathologist determined if sufficient sam-pling had been performed, thereby minimizing the numberof passes required. The aspirates were diagnosed accordingto the probabilistic approach as positive for malignancy,suspicious for malignancy, atypical, benign, and nonrepre-sentative/unsatisfactory.

Women with malignant cytology findings proceeded tosurgical excision or other treatment while those with suspi-cious findings were advised to have tissue examination,usually in the form of CNB. Most patients with atypicalreports also had tissue examination, although a few werefollowed clinically depending on imaging and clinical find-ings. Core biopsy occurred in the majority of cases wherethe immediate cytologic evaluation was discordant with theclinical, sonographic, and mammographic findings; whenthe procedure failed to provide enough cytologic materialfor diagnosis; and when the radiographic findings werehighly suspicious and a definitive diagnosis of malignancycould not be rendered based on the cytology material ob-tained. Patients with benign cytologic findings and incon-spicuous sonographic findings were advised to return forfollow-up mammogram in 6 months to assess for intervalchange. If interval change was identified, the patient wasreferred for appropriate follow-up, usually by CNB. Be-cause few tissue specimens were obtained in the benigngroup, clinical/mammographic follow-up was obtained inmost cases. The statistical calculations were based on his-tology (61 cases) and clinical and imaging outcome (47cases). Comparison of the FNA categories with tissue ormammographic follow-up was used to determine the pre-dictive value for malignancy as well as the sensitivity,specificity, positive and negative predictive values.

ResultsTable I shows the results and final outcome for the 108FNAs. The cytologic interpretations included 2 unsatisfac-tory (2%), 55 benign (51%), 11 atypical (10%), 8 suspicious(7%), and 32 malignant (30%). Two cases with suspiciousmammograms but scant cellularity were called unsatisfac-tory/nonrepresentative on FNA and were subsequently di-agnosed on CNB as invasive lobular carcinoma. Histologiccorrelation occurred for 61 patients (Table I) and is dis-cussed in detail below. The remaining benign and atypicalcases were followed clinically/mammographically; 21 caseshad more than 1-year follow-up.

The probability of malignancy was related to the diag-nostic category of the US-FNA and increased from 0% inthe benign group to 36% in the atypical group, 63% in thesuspicious group, and 100% in malignant specimens (Fig.1). If the suspicious and malignant FNA categories arecombined, atypical cases grouped with negative, and unsat-isfactory cases not included, the sensitivity is 90% (37/41)and specificity 95% (62/65). However, when a threshold of

LIAO ET AL.

326 Diagnostic Cytopathology, Vol 30, No 5

atypical was used to define a case as potentially malignant,100% sensitivity was achieved.

Histologic and Imaging CorrelationsIn retrospect, all histologically proven malignancies hadsuspicious/malignant ultrasonographic features (Fig. 2), in-cluding the two cases cytologically classified as nonrepre-sentative and the four invasive carcinomas in the atypicalcytology category. The invasive carcinomas ranged from 6to 30 mm in size, and all had less than 25% associatedductal carcinoma in situ (DCIS) with no microcalcifications(Fig. 3). Of the three cases with a histologic diagnosis ofDCIS (suspicious and malignant cytology categories), allpresented as suspicious masses on US. One was DCISassociated with a mucocele-like lesion without microcalci-fications, and the other two cases were DCIS presenting asmasses greater than 1 cm (intermediate- to high-grade DCIScolonizing areas of tumoral adenosis, and a radial scar).None of the DCIS cases in our series had mammographi-cally detected microcalcifications.

Four cytologically atypical cases were carcinoma on ex-cision (Fig. 4); all four malignancies were small (less than1 cm) low-grade mammary carcinomas with little (� 10%)or no DCIS component. All presented as ultrasonographi-cally suspicious masses lacking microcalcifications in theircorresponding mammograms.

There were three cases with histologic diagnoses of pap-illary lesions (one in the suspicious category and two in theatypical category); in two of these cases, the probability ofa papillary lesion was suggested in the cytology report. Two

cases had associated florid hyperplasia of the usual type andthe third had atypical lobular hyperplasia cells involving thepapilloma. One was a partially cystic papilloma, and theother two were partially sclerosed clustered micropapillo-mas. All of the remaining histologically benign cases in-cluded in the suspicious and atypical cytologic categorieswere fibroadenomas. The atypical and suspicious cases withbenign follow-up showed loss of cellular cohesion includingsingle cells or loose groups. The papillary lesions includedsingle-columnar-type cells.

The cases with benign cytologic diagnoses (Fig. 5) thathad surgical follow-up all showed a histologic spectrum offibrocystic changes with significant areas of aggregate andsclerosing adenosis. The suspicious radiologic index in themajority of the cytologic benign cases was low. There wereonly two cases showing incidental atypical lobular hyper-plasia on histology and no cases of atypical ductal hyper-plasia.

Table I. FNA Cytology Results vs. Final Outcome

FNAdiagnoses

Number(%)

Surgical diagnosesClinical

follow-upMalignant Benign

Malignant 32 (30%) 32 (2 DCIS) 0Suspicious 8 (7%) 5 (1 DCIS) 3 (2 fibroadenomas, 1 papillary lesion)Atypical 11 (10%) 4 5 (3 fibroadenomas, 2 papillary lesions) 2Benign 55 (51%) 0 10 (10 fibrocystic changes) 45Unsatisfactory 2 (2%) 2 (lobular) 0 0Total 108 (100%) 43 18 47

Fig. 1. Bar graph showing positive predictive values (PPVs) for differentcytologic diagnoses categories in probabilistic reporting scheme.

Fig. 2. Ultrasound illustration of a case of invasive carcinoma presentingwith suspicious sonographic features (vertical dimension greater than hor-izontal and shadowing).

US-FNA OF NONPALPABLE BREAST LESIONS

Diagnostic Cytopathology, Vol 30, No 5 327

Regarding radiologic versus histologic size, only a 1–3mm size difference was noted in all cases of ductal adeno-carcinoma. More significant size differences (10 mm ormore) were noted in cases of lobular carcinoma or carci-noma with lobular features.

DiscussionIn the last 10 years, the popularity of the FNA technique hasbeen challenged by the introduction of CNB. A question-naire mailed to Papanicolaou Society members in late 1998documented an increasing use of CNB in the precedingyears.12 In brief, the performance of these two methods ofsampling breast tissue appears comparable.6,13–16 When USguidance is used, both have sensitivities generally in therange of 90% or higher.6,13 FNA has certain advantages overCNB because the former is less invasive, less costly forpatients, more quickly performed, and usually provides amore timely diagnosis. FNA cytology frequently obviatesthe need for other costly biopsy procedures in many patientswith carcinomas and benign lesions. Complications of FNAare rare and seldom serious, and there are minimal effectson subsequent histology.

One criticism of image-guided FNA of nonpalpablemasses has been the high rate of insufficient samples.17 In amulticenter trial, the rate of insufficient specimens appeared

to be greater with calcified versus mass lesions, benignversus malignant lesions, and with use of sterotactic versusUS guidance. In this trial, the insufficient rate was 8.6% formass lesions aspirated under US guidance.17 The rate ofinsufficient specimens was very low in our series, and thislikely related to the types of lesions selected for US-FNA bythe radiologist and the presence of the cytopathologist at theprocedure for immediate assessment.

The NCI consensus conference recommended a probabi-listic categorization of FNA diagnoses so that breast cytol-ogy specimens could be stratified according to the risk ofunderlying malignancy.6 This classification system has thepotential to recognize FNAs with atypical or suspiciousfeatures in a manner not allowed by a dichotomous positiveand negative classification scheme.11 According to the NCIrecommendations, atypical specimens generally requireclinical and imaging correlation, and suspicious findingsshould be followed by tissue biopsy prior to definitivetherapy. Any cases with mixed or inconclusive findings bytriple test also usually proceed to biopsy.6

Our study demonstrates that this classification systemaccurately stratifies US-FNAs of nonpalpable breast lesionsinto groups that correlate with the likelihood of underlyingmalignancy. US-FNAs classified as atypical had a 36%

Fig. 3. A: Cytology specimen interpreted as suspicious for mammarycarcinoma with lobular features. Specimen showed loose clusters and a fewsingle small atypical cells (Diff-Quik). B: Corresponding surgical excisionshowing classic lobular carcinoma.

Fig. 4. A: Cytology specimen interpreted as atypical showing mostlygroups with some crowding and disorder (Diff-Quik). B: Correspondingsurgical excision showing low-grade invasive carcinoma.

LIAO ET AL.

328 Diagnostic Cytopathology, Vol 30, No 5

frequency of underlying malignancy. This increased to 63%in suspicious cases and 100% in cases with a definitivecytologic diagnosis of malignancy. The varying probabili-ties for detecting malignancy may help guide additionalpatient evaluation. For example, a patient with an atypicalcytology report but benign clinical and imaging findingsmay be followed for a limited time period or may undergoa small histologic biopsy procedure. Women with suspi-cious diagnoses have excisional or other definitive biopsyprocedures prior to definitive treatment or axillary sam-

pling, and women with positive diagnoses can proceed todefinitive treatment. Suspicious cytologic diagnoses canalso be immediately followed by CNB, although some pap-illary lesions and complex sclerosing lesions may requireexcisional biopsy for diagnosis.

The positive predictive value for the suspicious categoryis somewhat lower than observed in previous studies, whichcite positive rates of 80% or more.11,18,19 This may be due tothe relatively small number of cases in this category and thedifficulty in differentiating small fibroadenomatoid and pap-illary lesions from low-grade malignancy, especially whenimaging studies are indeterminate. Some series have in-cluded mostly larger or palpable lesions that may havedifferent imaging characteristics. Furthermore, cytologiccriteria and thresholds for separating suspicious and atypicalcases are not well established. The fact that the majority ofsuspicious cases were malignant on excision still supportsthe recommendation for confirmatory tissue studies prior todefinitive treatment. Concerning the atypical and suspiciouscategories, Page et al.11 states that “each individual labora-tory may be expected to have a variable likelihood ofmalignancy in this category.” Laboratories should in turncommunicate the results of their probabilistic reporting cat-egories to clinicians and radiologists managing the patients.

In the past, the atypical category has generally beencombined with the negative category for statistical pur-poses. However, the goal of the breast triple test is toseparate patients who require further management fromthose who can be followed clinically. In the field of cervicalcytology, it is well recognized that the atypical categorycannot be combined with negative specimens in terms ofpatient management. Achieving sufficient sensitivity in de-tection of biopsy-proven high-grade cervical squamous le-sions and cancers is possible only through use of atypicalsquamous cells as the abnormal threshold.20 While cervicalcytology screening has obvious differences from US-FNA,attempts to eliminate the use of the atypical category inbreast cytology, or to minimize the need for correlation andfollow-up, will likely endanger patient safety.

Only a minority of patients with benign cytology reportshad histologic biopsies, and all were benign, indicating thehigh sensitivity of this procedure. According to the triple-test model, benign cases should be correlated with theclinical and imaging findings, with any noncorrelating casesproceeding to biopsy. The two unsatisfactory cases werescantly cellular and cytology did not correlate with clinicaland imaging findings; both proved to be lobular carcinomaon histology. These two cases illustrate the importance ofdistinguishing benign/negative from nondiagnostic/unsatis-factory cases. This distinction is easier to make when thecytopathologist is on site and can review clinical and im-aging features with the radiologist and other clinicians.

The choice of biopsy method for a mammographic ab-normality may affect the ability to render a diagnosis suf-

Fig. 5. A: Cytology specimen diagnosed as benign showing ductal groupswith apocrine change (Diff-Quik). B: Corresponding surgical excisionshowing florid ductal hyperplasia of the usual type.

US-FNA OF NONPALPABLE BREAST LESIONS

Diagnostic Cytopathology, Vol 30, No 5 329

ficient for patient management using the least number ofprocedures. The value of breast FNA has been questionedregarding the lack of specificity for benign diagnoses, theinability to separate in situ versus invasive carcinomas, andthe inability to render diagnoses of atypical ductal andlobular hyperplasia. While some fibrocystic processes, in-cluding radial scars, papillary lesions, atypical hyperplasias,and adenosis, are not specifically diagnosed on cytologyspecimens, few of these lesions present as sonographicallysuspicious masses, as most will present with associatedindeterminate microcalcifications on mammogram.4,5 Fur-thermore, many such lesions are also difficult to diagnose incore biopsies, often requiring excision for complete charac-terization. In our series, only 10 of the 55 cases with benigncytologic diagnoses required further histologic confirma-tion.

The management of breast cancer depends in large parton whether the carcinoma is invasive or in situ.21,22 Veryfew of the carcinomas in this series were in situ variants(3/41), and most of the invasive carcinomas had no signif-icant in situ component. These findings suggest that masslesions detected by ultrasound are much more likely to beinvasive than in situ carcinomas, similar to observationsreported in clinically palpable breast masses.23 As withpalpable breast abnormalities, those DCIS cases presentingas masses were usually greater than 1 cm in size and hadhigh nuclear grades with frequent comedo component.High-grade DCIS cases are more likely to exhibit microin-vasion and have a low but significant risk of axillary me-tastases.24 Managing such patients with lumpectomy andsentinel node biopsy is thus not precluded by US-FNA,assuming that a mass lesion is present on US. CNB does notovercome this problem completely since some women withinvasive carcinoma have only atypical or in situ lesions onCNB. Clinical and radiologic correlation is useful in deter-mining further management in most cases, but some lesionswill require excision for definitive diagnosis before pro-ceeding to axillary dissection.

In conclusion, both FNA and CNB have a role in theassessment of clinically nonpalpable lesions, and the biopsymodality can be decided on an individual basis. Radiolo-gists in our institution judge the location of the lesion andthe imaging characteristics. When the lesion is suitable forUS-FNA, this modality is selected because of accuracy ofthe cytology diagnosis, the availability of immediate eval-uation, and minimal trauma to the patient. CNB is mainlyreserved for abnormalities with microcalcifications or inother situations in which CNB is judged to have a greaterlikelihood of providing a definitive diagnosis. The probabi-listic approach to breast FNA interpretation, when com-bined with other components of the triple test, accuratelystratifies women so that the appropriate follow-up may bedetermined. Many institutions have seen a shift to increas-ing use of US guidance instead of stereotactic guidance

technique for nonpalpable masses,25 and US-FNA appearsto lead to fewer insufficient results than stereotactic FNA.Laboratories with breast cytology expertise may wish toeducate clinicians on the potential advantages of US-FNAversus CNB techniques for patients with suitable mammo-graphic abnormalities.

References1. Jackman RJ, Nowels KW, Rodriguez-Soto J, Marzoni FA Jr, Finkel-

stein SI, Shepard MJ. Stereotactic, automated, large-core needle bi-opsy of nonpalpable breast lesions: false negative and histologic un-derestimation rates after long-term follow-up. Radiology 1999;210:799–805.

2. Parker S. Stereotactic core biopsy: a return of the nattering nabobs ofnegativism. Acad Radiol 2000;7:225–227.

3. Westenend PJ, Sever AR, Beekman-De Volder HJ, Liem SJ. A com-parison of aspiration cytology and core needle biopsy in the evaluationof breast lesions. Cancer 2001;93:146–150.

4. Kopans DB. Breast imaging, 2nd ed. Philadelphia: Lippincott-Raven;1998.

5. Tohno E, Cosgrove DO, Sloane JP. Ultrasound diagnosis of breastdiseases. New York: Churchill Livingstone; 1994.

6. Abati A, Abele J, Bacus S, et al. The uniform approach to breastfine-needle aspiration biopsy. Diagn Cytopathol 1997;16:295–311.

7. Boerner S, Fornage BD, Singletary E, Sneige N. Ultrasound-guidedfine-needle aspiration of nonpalpable breast lesions: a review of 1885FNA cases using the National Cancer Institute-supported recommen-dations on the uniform approach to breast FNA. Cancer 1999;87:19–24.

8. Saarela AO, Kiviniemi HO, Rissanen TJ, Paloneva TK. Nonpalpablebreast lesions: pathologic correlation of ultrasonographically guidedfine-needle aspiration biopsy. J Ultrasound Med 1996;15:549–553.

9. Apesteguia L, Pina L, Inchusta M, et al. Nonpalpable, well-defined,probably benign breast lesion: management by fine-needle aspirationbiopsy and long-interval follow-up mammography. Eur Radiol 1997;7:1235–1239.

10. Ciatto S, Rosselli del Turco M, Ambrogetti D, et al. Solid nonpalpablebreast lesions: success and failure of guided fine-needle aspirationcytology in a consecutive series of 2444 cases. Acta Radiol 1997;38:815–820.

11. Page DL, Johnson JE, Dupont WD. Probabilistic approach to thereporting of fine-needle aspiration cytology of the breast. Cancer1997;81:6–9.

12. Tabbara SO, Frost AR, Stoler MH, Sneige N, Sidawy MK. Changingtrends in breast fine-needle aspiration: results of the PapanicolaouSociety of Cytopathology Survey. Diagn Cytopathol 2000;22:126–130.

13. Ballo MS, Sneige N. Can core needle biopsy replace fine-needleaspiration cytology in the diagnosis of palpable breast carcinoma: acomparative study of 124 women. Cancer 1996;78:773–777.

14. Liberman L, Cohen MA, Dershaw DD, Abramson AF, Hann LE,Rosen PP. Atypical ductal hyperplasia diagnosed at stereotaxic corebiopsy of breast lesions: an indication for surgical biopsy. Am JRoentgenol 1995;164:1111–1113.

15. Liberman L, Dershaw DD, Rosen PP, et al. Stereotaxic core biopsy ofbreast carcinoma: accuracy at predicting invasion. Radiology 1995;194:379–381.

16. Symmans WF, Weg N, Gross J, et al. A prospective comparison ofstereotaxic fine-needle aspiration versus stereotaxic core needle biopsyfor the diagnosis of mammographic abnormalities. Cancer 1999;85:1119–1132.

17. Pisano ED, Fajardo LL, Tsimikas J, et al. Rate of insufficient samplesfor fine-needle aspiration for nonpalpable breast lesions in a multi-center clinical trial: the Radiologic Diagnostic Oncology Group 5Study. Cancer 1998;82:679–688.

LIAO ET AL.

330 Diagnostic Cytopathology, Vol 30, No 5

18. Sneige N, Fornage BD, Saleh G. Ultrasound-guided fine-needle aspi-ration of nonpalpable breast lesions: cytologic and histologic findings.Am J Clin Pathol 1994;102:98–101.

19. Wang HH, Ducatman BS. Fine needle aspiration of the breast: aprobabilistic approach to diagnosis of carcinoma. Acta Cytol 1998;42:285–289.

20. Kinney WK, Manos MM, Hurley LB, Ransley JE. Where’s the high-grade cervical neoplasia? the importance of minimally abnormal Pa-panicolaou diagnoses. Obstet Gynecol 1998;91:973–976.

21. Kurtz JM, Jacquemier J, Torhorst J, et al. Conservation therapy forbreast cancers other than infiltrating ductal carcinoma. Cancer 1989;63:1630–1635.

22. Hortobagyi GN, Ames FC, Buzdar AU, et al. Management of stage IIIprimary breast cancer with primary chemotherapy, surgery and radia-tion therapy. Cancer 1998;62:2507–2516.

23. Chhieng DC, Fernandez G, Cangiarella JF, et al. Invasive carcinoma inclinically suspicious breast masses diagnosed as adenocarcinoma byfine-needle aspiration. Cancer 2000;90:96–101.

24. Lagios MD, Margolin FR, Westdahl PR, Rose MR. Mammographi-cally detected duct carcinoma in situ: frequency of local recurrencefollowing tylectomy and prognostic effect of nuclear grade on localrecurrence. Cancer 1989;63:618–624.

25. Fornage BD. Sonographically guided needle biopsy of nonpalpablebreast lesions. J Clin Ultrasound 1999;27:385–398.

US-FNA OF NONPALPABLE BREAST LESIONS

Diagnostic Cytopathology, Vol 30, No 5 331