Developing Asymmetry Identified on Mammography: Correlation with Imaging Outcome and Pathologic Findings

Transcription

1 Asymmetry on Mammography Women s Imaging Original Research WOMEN S IMAGING Jessica W. T. Leung 1 Edward A. Sickles Leung JWT, Sickles EA Keywords: breast, breast cancer, mammography, screening, sonography DOI: /AJR Received March 21, 2006; accepted after revision July 31, Presented at the 2006 annual meeting of the American Roentgen Ray Society, Vancouver, BC, Canada. Supported by a grant from the Society of Breast Imaging and the American Roentgen Ray Society. 1 Both authors: Department of Radiology, University of California, San Francisco Medical Center, Box 1667, San Francisco, CA Address correspondence to J. W. T. Leung. AJR 2007; 188: X/07/ American Roentgen Ray Society Developing Asymmetry Identified on Mammography: Correlation with Imaging Outcome and Pathologic Findings OBJECTIVE. Developing asymmetry on mammography is a focal asymmetric deposit that has appeared or increased in size or conspicuity since a previous examination. We examined the frequency, imaging outcome, and pathologic significance of developing asymmetry. MATERIALS AND METHODS. This study was performed in a retrospective cohort manner. We searched for all cases of developing asymmetry consecutively entered in our mammography database from April 1985 to April We examined radiology records to determine whether sonography and MRI were used as adjunctive diagnostic tools and examined pathology records to determine tissue diagnosis. RESULTS. Developing asymmetry was present in 292 (0.16%) of 180,801 consecutive screening examinations and 32 (0.11%) of 27,330 consecutive diagnostic examinations. After exclusion for absent data, the study consisted of 281 screening and 30 diagnostic cases. In the 281 cases of screening-detected developing asymmetry, biopsy was recommended and was performed in 84 (29.9%) of the cases. Thirty-six cases of cancer were identified, resulting in a positive predictive value of 12.8%, obtained by division of the number of cases of cancers by the number of examinations with abnormal mammographic findings (PPV 1 ), and a PPV 2 of 42.9%, obtained by division of the number of cases of cancer by the number of mammographic examinations in which findings led to a recommendation for biopsy. Biopsy was recommended and performed in 26 (86.7%) of the 30 cases of diagnostic mammography. Eight cases of cancer were identified, resulting in a PPV 1 of 26.7% and a PPV 2 of 30.8%. Of the 44 cancers detected at screening and diagnostic mammography, 21 had available sonographic data. Five (23.8%) of these 21 cases of cancer had no correlate at sonographic examination. MRI was performed in only two cases, both with benign diagnoses. CONCLUSION. Developing asymmetry is an uncommon finding. When this sign is identified on screening and diagnostic mammography, the likelihood of malignancy is sufficiently high to justify recall and biopsy. Normal sonographic findings do not exclude malignancy in the case of developing asymmetry. eveloping changes on mammograms should be viewed with con- D cern because of the possibility of malignancy [1]. The developing density (neodensity) sign has been reported as being an infrequent mammographic indicator of malignancy [2]. Developing density is a focal asymmetric deposit of fibroglandular-density parenchyma not present on a previous mammographic examination or one that has increased in size or conspicuity. In a study of 300 consecutively detected cases of nonpalpable breast cancer, Sickles [2] found that 6% of the cases of cancer manifested with the developing density sign. Because of its infrequent occurrence, little is known about the clinical and pathologic importance of this sign. One recognized nonmalignant cause of developing density is the use of hormone replacement therapy by postmenopausal women [3, 4]. Pseudoangiomatous stromal hyperplasia (PASH) is another benign cause of developing density [5]. Part of the reason that the clinical importance of the developing density sign is not well known is that there is no widely accepted definition of the finding. None of the four editions of BI-RADS describes this sign [6 9]. The earlier editions differentiate mass, focal asymmetric density, and asymmetric breast tissue. In the fourth edition, the terms focal asymmetric density and asymmetric breast tissue are replaced with focal asymmetry and global asymmetry [9]. Therefore the mammographic finding previously known as developing density is more AJR:188, March

2 appropriately termed developing asymmetry, neoasymmetry, and enlarging asymmetry. We use the term developing asymmetry. Despite the lack of a widely accepted definition of developing asymmetry, a workable definition may be readily established among the several radiologists who work in a given mammography practice. Having a definition facilitates single-institution studies on the clinical importance of developing asymmetry. However, because the frequency of breast cancer among cases of developing asymmetry has been estimated as being low [2], a large number of such cases must be studied to produce statistically meaningful results. Furthermore, because developing asymmetry itself is an uncommon finding, it has been difficult for investigators to collect a sufficient number of cases from a single institution. Sonography often is useful for evaluation of mammographic findings [10]. It aids in the diagnosis of cysts, benign solid masses such as fibroadenoma, and cancer [11]. When developing asymmetry is identified on mammography, sonography targeted to the finding can be performed. Identification of a sonographic correlate then guides management. It is not known, however, whether lack of a sonographic correlate can be used to exclude malignancy and thus eliminate the need for biopsy of developing asymmetry [12]. Similarly, the role of MRI as an adjunctive imaging technique in the evaluation of developing asymmetry is undetermined. MRI is known to be sensitive for invasive carcinoma [13, 14], so perhaps the absence of an MRI correlate can be used to exclude malignancy and thus obviate biopsy of developing asymmetry. The goal of this study was to use our largescale clinical experience with mammography to determine the frequency, imaging outcome, and pathologic significance of developing asymmetry. We also investigated the roles of sonography and MRI in the evaluation of such imaging findings and whether demographic factors contribute to prediction of malignancy in cases of developing asymmetry. Materials and Methods This study was performed with approval of our institutional review board, which waived the requirement for written informed consent. The study was a retrospective observational analysis of consecutively registered cases of developing asymmetry prospectively collected over a 20-year period (April 1985 April 2005). Mammographic interpretations were prospectively entered into a computer database for clinical use and for internal quality assurance [15]. In particular, at interpretation the principal abnormal mammographic finding, if any, for each mammographic examination was recorded by the radiologist. From 1985 to 1998, data were collected for screening mammographic examinations performed in a mobile van operated by our institution. In addition, beginning in 1996, data also were collected for both screening and diagnostic mammographic examinations performed at all fixed locations within our institution [16]. The subjects in this study were women who underwent screening or diagnostic mammography at our institution whose mammograms were interpreted as having the finding of developing asymmetry. We identified this study cohort by searching our database for all cases prospectively coded with the finding developing asymmetry. Until November 2003, we coded using the term developing density. After publication of the fourth edition of BI-RADS, we used developing asymmetry, because the BI- RADS terminology for asymmetry changed in the fourth edition [9]. We also extracted from the database the following demographic information: patient age, menopause status, and family and personal history of breast cancer. We examined radiology records to obtain information about the use of sonography and MRI as adjunctive diagnostic tools and pathology records to determine tissue diagnosis. Samples for pathologic diagnosis were obtained through fine-needle aspiration biopsy, core biopsy, or surgical excision. Cancer was defined as invasive carcinoma or ductal carcinoma in situ (DCIS). Our database contained information on whether the biopsy result was benign or malignant along with the histologic findings for all malignant cases, whether or not biopsy was performed at our institution. For biopsies performed at our institution, we obtained information on the histologic features of benign diagnoses by retrospectively reviewing pathology records. Histologic data were not available for benign findings of biopsies performed at outside institutions. For cases in which biopsy was not performed, we ascertained that the finding was benign through long-term (at least 2 years) mammographic followup. We also performed computer linkage between our database and our regional Surveillance, Epidemiology and End Results (SEER) tumor registry. Such linkage and use of the computer-based outcome tracking system at our institution (also used in this study) aided in identification of more than 95% of breast cancer cases in our catchment area [17]. Thus linkage was important for ensuring the absence of malignancy in our study, because some of the patients who did not undergo biopsy underwent limited mammographic follow-up. All mammograms were screen-film. All mammography units, technologists, radiologists, and our several mammography facilities had been accredited by the American College of Radiology Mammography Accreditation Program since its inception in 1987 [18]. The characteristics of the interpreting radiologists have been previously reported [16]. Screening examinations involved craniocaudal and mediolateral oblique mammograms of each breast in women nominally free of symptoms. These examinations were performed with no radiologist on site and were batch-interpreted at a later time. All patients in whom developing asymmetry was identified on screening mammography were recalled for additional imaging assessment. Recall imaging included additional mammography (usually spot-compression magnification views) and occasionally sonography (if indicated after additional mammographic assessment was completed). Cases of developing asymmetry also were identified initially on diagnostic mammography. Some of these examinations were performed for evaluation of breast problems, such as palpable abnormality, nipple discharge, or focal breast pain. Other diagnostic examinations were performed for short-term followup of a probably benign mammographic finding, for short-term follow-up after benign and concordant imaging-guided percutaneous breast biopsy, and for follow-up of patients with breast cancer treated with breast conservation within 5 years of surgery. On diagnostic mammography, craniocaudal and mediolateral oblique views were interpreted by an on-site radiologist, who obtained additional mammographic views, sonograms, and MR images as needed for complete evaluation. All sonographic examinations were performed by a radiologist using a high-resolution linear array (7 12 MHz) transducer on one of two machines (Acuson 128 XP, Acuson, or Advanced Technology Laboratories 3500, Philips Medical Systems). The sonographic examinations were aimed at the area corresponding to the developing asymmetry, and transverse and longitudinal images were obtained. MRI was performed with a high-resolution 3D gradient-recalled echo sequence with fat saturation on a 1.5-T magnet (Signa, GE Healthcare). Images were obtained in the sagittal plane with a three-time-point protocol previously described [19]. Gadopentetate dimeglumine (Magnevist, Schering) was administered at a dose of 0.1 mmol/kg of patient body weight. All data were entered into and analyzed with an Excel spreadsheet (Microsoft). Positive predictive value (PPV) was defined as described in BI-RADS [9]. Specifically, PPV 1 was calculated by division of the number of cases of cancer by the number of mammographic examinations with findings interpreted as abnormal. PPV 2 was calculated by division of the number of cases of cancer by the number of mammographic examinations with findings that 668 AJR:188, March 2007

3 Asymmetry on Mammography led to recommendation for biopsy. Statistical analysis was performed with SAS version 9.2 (SAS Institute). Statistical significance was determined with Student s t test and 95% CI. Statistical significance was considered p 0.05, and correlation was reported with relative risk (RR). Results The data for this study were collected over a 20-year period from records of consecutive screening and diagnostic mammographic examinations. Among the 180,801 screening examinations, 292 cases of developing asymmetry were identified, representing a frequency of 0.16%. Among the 27,330 diagnostic examinations, 32 cases of developing asymmetry were identified, representing a frequency of 0.11%. Of the 292 cases of developing asymmetry identified on screening mammography, 11 were excluded from the study because the patient did not return for additional imaging (n = 4) and because the patient did not undergo the biopsy recommended after additional imaging (n = 7). Of the 32 cases of developing asymmetry identified on diagnostic mammography, two were excluded from the study because the patient did not undergo the biopsy recommended after additional imaging and because of lack of mammographic follow-up. Therefore, the final study cohort consisted of 281 screening mammography patients and 30 diagnostic mammography patients. The mean and median ages of the screening patients were 59 and 58 years, respectively (range, years). The mean and median ages of the diagnostic patients were 57 and 56 years (range, years). The outcome in the 281 screening-detected cases of developing asymmetry are presented in Table 1. Diagnostic mammographic views were obtained in all 281 cases and revealed summation artifacts (Fig. 1) rather than true lesions in 161 (57.3%) of the cases. In all but 11 (6.8%) of these 161 cases of summation artifacts, spot-compression magnification views alone (without sonography) were used to exclude the presence of a true lesion. Sonography showed cysts in another 24 (8.5%) of the 281 cases. Biopsy was recommended and performed in 84 (29.9%) of the 281 cases. The pathologic results are in Table 1. Thirty-six cases of cancer were identified, resulting in a PPV for abnormal screening interpretation (PPV 1 ) of 12.8% and a PPV for biopsy recommendation (PPV 2 ) of 42.9%. The outcome in the 30 cases of developing asymmetry identified on diagnostic TABLE 1: Outcome and Pathologic Findings in 281 Cases of Developing Asymmetry Identified on Screening Mammography Outcome n % No biopsy Summation artifact Cyst Miscellaneous benign finding Asymmetric breast tissue Skin lesion Intramammary lymph node Scar Dilated duct Generalized edema Biopsy Malignant Invasive carcinoma Ductal Lobular Tubular Mucinous Mixed ductal and lobular Ductal carcinoma in situ Benign Not available Fibrocystic change Fibrosis Fibroadipose tissue Fibroadenoma Fibroadenomatoid change Pseudoangiomatous stromal hyperplasia Radial scar Papilloma Note Percentages rounded up or down for summation. mammography is presented in Table 2. Biopsy was recommended and performed in 26 (86.7%) of the cases; the pathologic results are provided in Table 2. Eight cancers were identified, resulting in a PPV 1 (abnormal diagnostic interpretation) of 26.7% and a PPV 2 of 30.8%. Four (13.3%) of the 30 patients who underwent diagnostic mammography did not undergo biopsy. In one of these four cases, the recommendation for biopsy was based on mammographic findings alone (without sonography) and on the fact that the lesion was palpable. Biopsy was not performed, but there was no evidence of cancer after 4.5 years of mammographic follow-up. In the second case in which sonography was not performed, mastitis was suspected, and there was no evidence of cancer after 3 years of follow-up. Sonography was performed in two cases, revealing a benign simple cyst in one case and a vague hypoechoic focus in the second. The latter case was further evaluated with MRI. MRI was performed in two of the 30 diagnostic cases. In one of these cases, sonography showed a vague hypoechoic focus, no MRI correlate was found, and the patient underwent follow-up for 1.5 years without evidence of breast cancer. In the second case, sonography showed a vague hypoechoic focus, MRI showed regional enhancement with plateau kinetics, and biopsy showed PASH (Fig. 2). Sonographic data were not available in 23 of the 44 cases of cancer in this study because either sonography was not performed as part of the diagnostic evaluation (n = 13) or the diagnostic evaluation was performed at another institution and information was not available regarding whether sonography was performed (n = 10). Sonography data were available for the other 21 cases of cancer, and the findings were as follows: solid hypoechoic mass (n = 13) (Fig. 3), no sonographic correlate (n = 5), hyperechoic mass with necrosis (n = 1), hypoechoic focus (n = 1), and hypoechoic ducts (n = 1). In other words, five (23.8%) of the 21 cases of cancer in which sonographic data were available had no sonographic correlate. All five of these sonographically occult malignant tumors were nonpalpable and stage 1. There was no correlation between the year of examination and the sonographic findings. In the 201 cases in which biopsy was not performed for either screening or diagnosis, 98 (48.8%) of the patients underwent mammographic follow-up at our institution and 103 (51.2%) did not. The mean and median follow-up intervals were 5.6 and 5.0 years, respectively (range, years). Linkage with the SEER tumor registry and analysis with our institutional computer-based outcome tracking system revealed no cancer in any of the patients who did not undergo biopsy. In the 30 (22 benign, eight malignant) cases of diagnostic mammography, 13 patients presented with a palpable lesion and 17 underwent diagnostic mammography for an indication other than a palpable lesion. Twelve of the 13 patients with palpable findings had developing asymmetry corresponding to the palpable AJR:188, March

4 A B C D E F Fig year-old postmenopausal woman with nonpalpable developing asymmetry (visible in two projections, concave-outward contour, interspersed with fat) in left upper outer portion of breast detected on screening mammography. Diagnostic images show summation artifact. Follow-up mammograms (not shown) 24 months after screening and linkage with regional tumor registry showed no evidence of cancer. A, Mediolateral oblique projection mammogram of left breast obtained at previous screening shows no abnormal findings. B, Craniocaudal projection mammogram of left breast obtained at previous screening shows no abnormal findings. C, Mediolateral oblique projection mammogram of left breast 19 months after A and B suggests presence of asymmetry (arrow) not previously present in upper aspect. D, Craniocaudal projection mammogram of left breast in 19 months after A and B suggests presence of asymmetry (arrow) not previously present in outer aspect. E, Mediolateral oblique projection spot-compression magnification mammogram of developing asymmetry shows only superimposition of fibroglandular structures. F, Lateromedial projection spot-compression magnification mammogram of developing asymmetry shows only superimposition of fibroglandular structures. 670 AJR:188, March 2007

5 Asymmetry on Mammography finding. The other patient had developing asymmetry in the breast contralateral to the breast with the palpable finding. Therefore, 12 of the developing asymmetric lesions were palpable and 18 were nonpalpable. Among the 22 cases of diagnostic mammography of benign lesions, nine (40.9%) of the lesions were palpable, and 13 (59.1%) were nonpalpable. Three (37.5%) of the eight malignant lesions were palpable, and five (62.5%) were nonpalpable. Therefore, palpability did not correlate with malignancy (RR, 1.5; 95% CI, ). Age did appear to correlate with malignancy. Among the cohort of 44 patients with cancer, the mean and median ages were 69 (SD, 11.3) and 70 years (range, years). In contrast, the mean and median ages of patients with benign diagnoses were 58 (SD, 12.0) and 57 years (range, years). This difference was statistically significant (p < ). We also examined whether menopause status, family history of breast cancer, and personal history of breast cancer correlated with the presence of malignancy. The results are presented in Table 3. There was no correlation between family history and malignancy in the developing asymmetry (RR, 0.7; 95% CI, ). Being postmenopausal was associated with increased risk of cancer in the developing asymmetry (RR, 3.3; 95% CI, ), as was having a personal history of breast cancer (RR, 3.4; 95% CI, ). We found no correlation between year of detection and frequency of developing asymmetry on either screening or diagnostic mammography. Discussion Although a mass is defined as a lesion seen on at least two mammographic projections, densest in the center, and characterized by convex outward margins [7 10], focal asymmetry is defined as a two-projection lesion that has concave outward margins and may appear interspersed with fat [9]. Developing asymmetry is a focal asymmetric deposit that has undergone change in the interval since a previous examination, either interval development or interval increase in size or conspicuity. This interval change increases the likelihood of malignancy, just as similar interval change increases the likelihood of malignancy for masses and groups of calcifications [1]. Developing asymmetry is an infrequent but known mammographic sign of malignancy. Sickles [2] described the mammographic features of 300 consecutively detected nonpalpable malignant tumors. Most TABLE 2: Outcome and Pathologic Findings in 30 Cases of Developing Asymmetry Identified on Diagnostic Mammography Outcome n % No biopsy Biopsy Malignant Invasive carcinoma Ductal Lobular Mucinous Micropapillary Ductal carcinoma in situ Benign Fibrosis Pseudoangiomatous stromal hyperplasia Fibrocystic change Fibroadenoma Fibroadipose tissue Scar Note Percentages rounded up or down for summation. of the lesions were calcification clusters (42%) or dominant masses (39%), but 6% manifested as developing asymmetry. One of the reasons for the dearth of information about developing asymmetry is that the sign occurs very infrequently. Kopans et al. [20] reported that asymmetric breast tissue occurs in 221 (2.7%) of 8,048 mammograms, and Sickles [21] reported that focal asymmetry is found in 448 (1.3%) of 34,282 mammograms. Developing asymmetry differs from asymmetric breast tissue and focal asymmetry in that interval change is present in developing asymmetry, and to our knowledge there are no published data on the frequency of developing asymmetry. One [22] of the two published reports of developing asymmetry was limited to patients who had undergone biopsy. The second [5] did not give a denominator for calculation of frequency. In contrast, in our study of consecutive screening and diagnostic mammograms, the finding of developing asymmetry on screening and diagnostic mammography was rare (0.16% and 0.11%, respectively). For developing asymmetry, the PPV 1 on diagnostic mammography is higher than the PPV 1 at screening. This phenomenon is not unique for developing asymmetry; it applies to all abnormal mammographic findings [23]. Many definitely benign findings (e.g., summation artifacts) are considered abnormal on screening mammography, thereby diminishing the PPV 1 calculation by contributing only to the denominator, whereas on diagnostic mammography these findings are not included in the denominator because they are not considered abnormal after full diagnostic imaging evaluation. The finding of developing asymmetry on mammography should be evaluated with additional imaging to identify possible cancer. The initial evaluation usually involves diagnostic mammographic views, including 90 lateral, shallow oblique, and rolled views and spot-compression (with or without magnification) views [12]. These views are important in differentiating true findings from summation artifacts. At our institution, and for this study, spot-compression magnification mammographic views, usually in shallow oblique projection, were the diagnostic mammographic views of choice [24] (Figs. 1E, 2C, and 3C). Spot-compression magnification views alone were used to identify nearly all cases of summation artifact characterized as developing asymmetry on screening mammography. On subsequent mammographic follow-up, analysis with our institutional computer-based outcome tracking system, and computer linkage with the SEER tumor registry of these cases, no cancer was identified. Therefore, spot-compression magnification mammography was reliable for excluding cancer in the case of screen-detected developing asymmetry (representing summation artifact) without additional sonography or biopsy. This finding is clinically important because a sizable percentage of cases of developing asymmetry on screening mammography are found to represent summation artifacts on diagnostic imaging. Sonography is an invaluable adjunctive tool in breast imaging. In cases of developing asymmetry proved to represent true findings, sonography is helpful when a correlate (either definitely benign or suspicious for malignancy) is identified [12]. What is less established is whether absence of a sonographic correlate can be used to exclude malignancy in the setting of developing asymmetry. In a study of 16 cases of focal asymmetry and 20 of developing asymmetry, Shetty and Watson [22] found that two (28.6%) of seven malignant lesions had no sonographic correlate. One of these two tumors was nonpalpable developing asymmetry with interval enlarge- AJR:188, March

6 A B Fig year-old premenopausal woman with palpable lump in right upper breast. Core biopsy revealed pseudoangiomatous stromal hyperplasia. A, Mediolateral oblique projection mammogram of right breast at baseline shows no abnormal findings. B, Mediolateral oblique projection mammogram of right breast 1 year after A shows right breast with developing asymmetry (marked by metallic BB) in upper aspect that corresponds to presenting symptom of palpable lump. Developing asymmetry was located in central aspect in craniocaudal projection (not shown). Sonogram (not shown) showed vague hypoechoic focus. C, Mediolateral oblique projection spot-compression magnification mammogram of right upper breast shows persistence of asymmetry but no underlying mass, calcifications, or architectural distortion. Concave outward margins (arrows) and interspersed fat are evident. D, Contrast-enhanced maximum-intensity-projection sagittal MR image of right breast shows regional enhancement in upper aspect that corresponds to palpable lump and developing asymmetry on A C. C D ment from the previous study. The second tumor was palpable focal asymmetry identified on baseline mammography. In this study, five (23.8%) of 21 malignant tumors had no sonographic correlate. In other words, the absence of a sonographic correlate does not exclude malignancy in the setting of developing asymmetry. All five malignant tumors were nonpalpable and detected at screen- 672 AJR:188, March 2007

7 Asymmetry on Mammography ing. This 23.8% frequency is sufficiently high to justify biopsy despite the absence of a sonographic correlate for developing asymmetry. The frequency clearly is far too high to support alternative management with imaging surveillance because the likelihood of malignancy for probably benign (BI-RADS 3) lesions should be less than 2% [9, 21, 25]. On the basis of our findings and the recommendations of others [12, 22], we advocate biopsy in true cases of A B Fig year-old postmenopausal woman with nonpalpable developing asymmetry in left upper outer portion of breast detected on screening mammography (visible in two projections, concave-outward contour, interspersed with fat). Core biopsy revealed invasive ductal carcinoma. A, Mediolateral oblique projection mammogram of left breast at screening shows developing asymmetry (arrow) in upper aspect. This asymmetry was not present on previous screening mammography (not shown). B, Craniocaudal projection mammogram of left breast shows developing asymmetry (arrow) in outer aspect. This asymmetry was not present on previous screening mammography (not shown). C, Mediolateral oblique projection spot-compression magnification mammogram of left upper breast shows persistence of developing asymmetry (arrow) but no underlying mass, calcifications, or architectural distortion. D, Craniocaudal projection spot-compression magnification mammogram of left outer breast shows persistence of developing asymmetry (arrow) but no underlying mass, calcifications, or architectural distortion. E, Transverse sonogram shows irregular hypoechoic mass (arrow) with angular margin corresponding to developing asymmetry on A D. C D E developing asymmetry, even if there is no correlate on sonography. MRI is known to be highly sensitive in the detection of malignancy [13, 14], but there is no established role for MRI in the evaluation of breast asymmetry [12]. Lee et al. [13] examined the use of MRI in the evaluation of 86 problematic mammograms, and in 45% of the cases the finding was asymmetry seen in only one mammographic projection. No malignant tumor was seen in the 60 cases for which there was no MRI correlate. Developing asymmetry differs from one-view-only asymmetry in that developing asymmetry is seen in at least two projections and has undergone interval change. We attempted to determine the utility of MRI in assessment of developing asymmetry but drew no conclusion because MRI was performed in only two (0.64%) of 311 cases. In one of these two AJR:188, March

8 TABLE 3: Demographic Features of 311 Patients with the Mammographic Finding of Developing Asymmetry Benign (n = 267) Malignant a (n =44) Feature n % n % Family history of breast cancer No Yes Personal history of breast cancer No Yes Menopause status Premenopausal Postmenopausal a Invasive carcinoma or ductal carcinoma in situ. cases, PASH was seen as an area of regional enhancement (Fig. 2D). Cancer manifesting as developing asymmetry can be of any histologic type, including DCIS (Tables 1 and 2). Although the most common screening mammographic sign of DCIS is calcification, DCIS also can manifest as a noncalcified lesion. This finding typically would be a primarily circumscribed mass, but it also can be developing asymmetry. In a study of 190 cases of DCIS, Ikeda and Andersson [26] found that developing asymmetry was the mammographic finding in four (2.1%) of the cases. In our study of developing asymmetry, three (6.8%) of the 44 cases of cancer were DCIS (Tables 1 and 2). Although DCIS infrequently appears as developing asymmetry, it is nevertheless important to recognize that DCIS can manifest in this way, particularly because DCIS is regarded as an almost always curable disease. Invasive lobular carcinoma is known to manifest as subtle findings, including asymmetry, on mammography more frequently than does invasive ductal carcinoma [27, 28]. Most cases of breast cancer are of the ductal histologic type; lobular carcinoma accounts for approximately 10% of cases of cancer [29]. It is thought that invasive lobular carcinoma results in subtle mammographic findings because the cancer cells grow into adjacent tissues in a single-file pattern rather than by forming a cohesive mass of tumor cells [27]. In a study of oneview-only mammographic findings (e.g., asymmetry), Sickles [28] found that 33% of the cases of cancer found were invasive lobular carcinoma. This frequency is much higher than the reported 10% frequency of lobular carcinoma. In our series of cases of developing asymmetry, 15.9% of the cases of cancer were invasive lobular carcinoma (Tables 1 and 2), a frequency also somewhat higher than 10%. Therefore one should consider the possibility of invasive lobular carcinoma if developing asymmetry is found, just as one would for one-view-only asymmetry. Palpability has been known to increase the probability of malignancy in breast asymmetry [12, 20]. In our study of developing asymmetry, palpability did not correlate with malignancy. In other words, being nonpalpable did not negate or decrease the chance of malignancy. Therefore all cases of developing asymmetry, not only palpable lesions, necessitate biopsy. Similarly, the absence of a family history of breast cancer should not be used to obviate biopsy in the presence of developing asymmetry. Finally, even though statistically significant correlations were found when the determination of whether a case of developing asymmetry was malignant was compared with menopause status or personal history of breast cancer, a substantial number of cases of cancer would still be missed if biopsy was not always performed. Therefore we also recommend that biopsy be performed whenever developing asymmetry is found, regardless of menopause status or personal history of breast cancer. Patient age correlated with the presence of malignancy associated with developing asymmetry in a statistically significant manner. Developing asymmetry in older women was more likely to be malignant than developing asymmetry in younger women. These results are consistent with the widely reported general observation that the incidence of breast cancer increases with advancing age [30]. If developing asymmetry is identified on screening mammography, it is important to recall the patient for diagnostic imaging before recommending biopsy. In this series, 57.3% of screening-detected cases of asymmetry were found to be summation artifacts on diagnostic imaging, and 8.5% were found to be benign cysts (Table 1). These lesions do not necessitate biopsy or even short-interval follow-up imaging. PASH is a benign proliferative stromal lesion made up of myofibroblasts. The condition usually is found in premenopausal women or postmenopausal women taking hormone replacement therapy [5, 31]. Although the typical mammographic appearance of PASH is a circumscribed or partially circumscribed noncalcified mass [32, 33], the lesion can also appear as developing asymmetry [5]. In a study of 21 cases of nonpalpable developing asymmetry, Piccoli et al. [5] found no cases of cancer, and PASH was identified in all 13 cases retrospectively reviewed by a pathologist. PASH was prospectively identified as a specific diagnosis in only two (15.4%) of these 13 cases. PASH was not described in any of the 20 cases of developing asymmetry or 16 cases of focal asymmetry reported by Shetty and Watson [22]. This finding probably occurred because PASH is not always recognized as a specific entity at pathologic examination and because not all pathologists are making this diagnosis. We found PASH prospectively in seven (13.5%) of 52 biopsies with benign results (Tables 1 and 2) and agree with Piccoli et al. [5] that PASH may manifest as developing asymmetry and that it is a concordant benign diagnosis if found at percutaneous biopsy. Our findings about PASH differ from those of Piccoli et al. in that two (28.6%) of our seven patients with a PASH diagnosis were postmenopausal (one was taking hormone replacement therapy and the other was not), whereas all 21 patients in the study by Piccoli et al. were premenopausal. PASH in postmenopausal women not taking hormone replacement therapy also has been reported [34]. There were several limitations to our study. First, the data on sonography use and on the histologic features of the benign lesions were incomplete. A similar limitation was the lack of data on hormone replacement therapy because we did not have sufficient information on this subject to report or to draw conclusions. This study of consecutively registered subjects spanned more than 20 years, so we did not an- 674 AJR:188, March 2007

9 Asymmetry on Mammography alyze data that were not prospectively collected or recorded. It was necessary to collect data over a long time period to have a sufficient number of cases for statistical power, given the infrequent occurrence of developing asymmetry. Another limitation was that we did not have at least 2 years of follow-up data for all cases in which imaging findings were interpreted as negative (i.e., summation artifact) or benign and biopsy was not performed. To remedy this limitation, we linked our cohort to the SEER tumor registry and found no missed cases of cancer among the cases in which findings were interpreted as negative or benign. Last, MRI was performed in only two cases, so no meaningful conclusion can be drawn about the role of MRI in the evaluation of developing asymmetry. We set out to investigate the incidence and significance of developing asymmetry on mammography in a large-scale study because this sign is an infrequent but known indicator of malignancy. We conclude that all patients with developing asymmetry detected at screening need recall and that all findings that still represent developing asymmetry after full diagnostic imaging evaluation necessitate biopsy. Negative findings on sonographic examination do not preclude malignancy. Future studies should be performed to examine the role of MRI in excluding malignancy and obviating biopsy in the setting of developing asymmetry. Acknowledgments We thank Philip W. Chu for his immense contribution to statistical analysis and Yelena Borodina for her invaluable assistance in manuscript preparation. References 1. Hermann G. Keller RJ, Tartter P, et al. Interval changes in nonpalpable breast lesions as an indication of malignancy. Can Assoc Radiol J 1995; 46: Sickles EA. Mammographic features of 300 consecutive nonpalpable breast cancers. AJR 1986; 146: Cyrlak D, Wong CH. Mammographic changes in postmenopausal women undergoing hormonal replacement therapy. AJR 1993; 161: Stomper PC, Van Voorhis BJ, Ravnikar VA, Meyer JE. Mammographic changes associated with postmenopausal hormone replacement therapy: a longitudinal study. Radiology 1990; 174: Piccoli CW, Feig SA, Palazzo JP. Developing asymmetric breast tissue. Radiology 1999; 211: American College of Radiology. Breast imaging reporting and data system (BI-RADS). Reston, VA: American College of Radiology, American College of Radiology. Breast imaging reporting and data system (BI-RADS), 2nd ed. Reston, VA: American College of Radiology, American College of Radiology. Breast imaging reporting and data system (BI-RADS), 3rd ed. Reston, VA: American College of Radiology, American College of Radiology. ACR breast imaging reporting and data system (BI-RADS), 4th ed. Reston, VA: American College of Radiology, Mendelson EB. Problem-solving ultrasound. Radiol Clin North Am 2004; 42: Stavros AT, Thickman D, Rapp CL, Dennis MA, Parker SH, Sisney GA. Solid breast nodules: use of sonography to distinguish between benign and malignant lesions. Radiology 1995; 196: Samardar P, Shaw de Paredes E, Grimes MM, Wilson JD. Focal asymmetric densities seen at mammography: US and pathologic correlation. Radio- Graphics 2002; 22: Lee CH, Smith RC, Levine JA, Troiano RN, Tocino I. Clinical usefulness of MR imaging of the breast in the evaluation of the problematic mammogram. AJR 1999; 173: Kuhl CK, Schrading S, Leutner CC, et al. Mammography, breast ultrasound, and magnetic resonance imaging for surveillance of women at high familial risk for breast cancer. J Clin Oncol 2005; 23: Sickles EA. The usefulness of computers in managing the operation of a mammography screening practice. AJR 1990; 155: Sickles EA, Wolverton DE, Dee KE. Performance parameters for screening and diagnostic mammography: specialist and general radiologists. Radiology 2002; 224: Leung JWT, Sickles EA. Multiple bilateral masses detected on screening mammography: assessment of need for recall imaging. AJR 2000; 175: McLelland R, Hendrick RE, Zinninger MD, Wilcox PA. The American College of Radiology Mammography Accreditation Program. AJR 1991; 157: Hylton NM. Vascularity assessment of breast lesions with gadolinium-enhanced MR imaging. Magn Reson Imaging Clin N Am 2001; 9: Kopans DB, Swann CA, White G, et al. Asymmetric breast tissue. Radiology 1989; 171: Sickles EA. Periodic mammographic follow-up of probably benign lesions: results of 3,184 consecutive cases. Radiology 1991; 179: Shetty MK, Watson AB. Sonographic evaluation of focal asymmetric density of the breast. Ultrasound Q 2002; 18: Dee KE, Sickles EA. Medical audit of diagnostic mammography examinations: comparison with screening outcomes obtained concurrently. AJR 2001; 176: Sickles EA. Practical solutions to common mammographic problems: tailoring the examination. AJR 1988; 151: Sickles EA. Probably benign breast lesions: when should follow-up be recommended and what is the optimal follow-up protocol? Radiology 1999; 213: Ikeda DM, Andersson I. Ductal carcinoma in situ: atypical mammographic appearances. Radiology 1989; 172: Helvie MA, Paramagul C, Oberman HA, Adler DD. Invasive lobular carcinoma: imaging features and clinical detection. Invest Radiol 1993; 28: Sickles EA. Findings at mammographic screening on only one standard projection: outcomes analysis. Radiology 1998; 208: Arpino G, Bardou VJ, Clark GM, Elledge RM. Infiltrating lobular carcinoma of the breast: tumor characteristics and clinical outcome. Breast Cancer Res 2004; 6:R Search Cancer Statistics Review, Age distribution (%) of incidence cases by site, (Table I-10). Available at: cancer.gov/cgi-bin/csr/1975_2003/search.pl#results. Accessed March 15, Vuitch MF, Rosen PP, Erlandson RA. Pseudoangiomatous hyperplasia of mammary stroma. Hum Pathol 1986; 17: Polger MR, Denison CM, Lester S, Meyer JE. Pseudoangiomatous stromal hyperplasia: mammographic and sonographic appearances. AJR 1996; 166: Cohen MA, Morris EA, Rosen PP, Dershaw DD, Liberman L, Abramson AF. Pseudoangiomatous stromal hyperplasia: mammographic, sonographic, and clinical patterns. Radiology 1996; 198: Castro CY, Whitman GJ, Sahin AA. Pseudoangiomatous stromal hyperplasia of the breast. Am J Clin Oncol 2002; 25: AJR:188, March