False-Positive Findings at Contrast- Enhanced Breast MRI: A BI-RADS Descriptor Study

Transcription

1 Women s Imaging Original Research Baltzer et al. BI-RADS MRI Women s Imaging Original Research WOMEN S IMAGING Pascal A. T. Baltzer 1 Matthias Benndorf 1 Matthias Dietzel 1 Mieczyslaw Gajda 2 Ingo B. Runnebaum 3 Werner A. Kaiser 1 Baltzer PAT, Benndorf M, Dietzel M, Gajda M, Runnebaum IB, Kaiser WA Keywords: breast cancer, breast MRI, false-positive, MR mammography, sensitivity, specificity DOI: /AJR Received August 14, 2009; accepted after revision November 26, P. A. T. Baltzer and M. Benndorf contributed equally to this work. 1 Institute of Diagnostic and Interventional Radiology, Friedrich-Schiller-University Jena, Erlanger Allee 101, D Jena, Germany. Address correspondence to P. A. T. Baltzer (pascal.baltzer@med.uni-jena.de). 2 Institute of Pathology, Friedrich-Schiller-University Jena, Jena, Germany. 3 Clinic of Gynaecology, Friedrich-Schiller-University Jena, Jena, Germany. AJR 2010; 194: X/10/ American Roentgen Ray Society False-Positive Findings at Contrast- Enhanced Breast MRI: A BI-RADS Descriptor Study OBJECTIVE. Breast MRI has high sensitivity in breast cancer detection, and the BI- RADS MRI lexicon was a step toward standardized description of lesions. However, falsepositive findings occur and lead to unnecessary biopsy. The purpose of this investigation was to identify criteria for false-positive findings in clinical practice. MATERIALS AND METHODS. Eligible for investigation were all breast MRI examinations from a consecutive 16-month time period that had histopathologic verification and findings classified as BI-RADS category 4 6 in the initial MRI report. Accordingly, 132 patients with 120 malignant and 31 benign lesions were enrolled. Two blinded observers categorized lesions into mass or nonmass and used BI-RADS to identify descriptor distribution differences between the benign and malignant subgroups. RESULTS. The ratio of mass to nonmass lesions differed significantly (p < 0.001) between benign (1.2:1) and malignant (7:1) findings. Seventeen mass and 14 nonmass lesions were falsepositive, and 105 mass and 15 nonmass lesions were true-positive. Among mass lesions, it was possible to differentiate malignant and benign lesions on the basis of margin (smooth, irregular, or spiculated) and dynamic enhancement features (p < 0.05). Among nonmass lesions, only stippled enhancement had a significant difference between the subgroups (p < 0.05). Tumor diameter had no influence on the correct diagnosis of nonmass lesions (p = 0.301). Conversely, among mass lesions, false-positive lesions were smaller than true-positive lesions (p = 0.01). CONCLUSION. Nonmass lesions were the major cause of false-positive breast MRI findings. BI-RADS descriptors are not sufficient for differentiating benign and malignant nonmass lesions. S ensitivity and specificity contribute to the accuracy of a diagnostic tool. In the case of contrastenhanced breast MRI, there is strong evidence that the sensitivity is greater than the sensitivity of other techniques of imaging the breast [1]. The controversy about the specificity of breast MRI may be resolved, because studies [1 3] have shown specificity values of breast MRI similar to that of the most common technique of imaging the breast: mammography. However, for the further implementation of diagnostic breast MRI in clinical practice, a reduced overall number of false-positive findings remains a major aim. For sufficient reliable and standardized differential diagnosis of malignant and benign lesions, the characterization of specific features of the lesions is vital. A number of investigations have been conducted to introduce and describe the morphologic and kinetic features of breast lesions [4 18]. To standardize breast MRI technique and lesion classification, the BI-RADS MRI lexicon was introduced [19]. In BI-RADS, attention is given to biologic differences in lesion growth patterns through differentiation of solid mass and nonsolid nonmass lesions [19]. Recent publications mame specificities of 81 97% [2, 20]. To our knowledge, however, no research has been conducted to systematically examine the characteristics of false-positive breast MRI findings compared with true-positive findings. Consequently, this retrospective classification analysis of biopsy-proven lesions suspicious on breast MR images was performed to identify BI-RADS criteria and subgroups that have false-positive findings. Materials and Methods Study Subjects From January 2005 through April 2006, 828 consecutive breast MRI investigations were performed 1658 AJR:194, June 2010

2 BI-RADS MRI at our institution. Of those, the inclusion criteria for this study were the presence of biopsy-proven lesions, referral of the patient to our institution from the department of gynecology and obstetrics of our university hospital, and all lesions described as suspicious in the initial breast MRI report. Initial MRI was performed by trained radiologists not specialized in breast MRI (> 500 breast MRI examinations). Nonenhancing lesions were omitted. Local ethical review board approval was obtained for this evaluation of previously acquired images. The inclusion criteria were met by 132 women (mean age, 54.4 ± 12.6 years [SD] at breast MRI; range, years) in whom 151 lesions were biopsy proven. Reasons for referral to MRI were unclear (n = 16) or suspicious (n = 58) mammographic findings, screening because of previous breast cancer (n = 37), suspicious ultrasound findings (n = 24), follow-up of probably benign lesions (n = 9), contralateral findings (n = 5), and suspicious findings at galactography (n = 1). In one patient, the reason for referral was not confirmed retrospectively, but the single lesion was included in the study. The 151 lesions suspicious for malignancy at MRI consisted of 120 malignant (truepositive) and 31 benign (false-positive) lesions. Histopathologic examination was defined as the reference standard. Figures 1 4 show examples of both false-positive and false-negative mass and nonmass lesions. MRI All examinations were performed with a clinical whole-body 1.5-T MRI unit (Symphony or Sonata, Siemens Healthcare). A phased-array receive-only breast coil (Breast Array, Siemens Healthcare) was used with the patient in the prone position. The standardized imaging protocol started with an unenhanced axial T1-weighted gradient-echo sequence (FLASH 2D, generalized autocalibrating partially parallel acquisition [GRAPPA] factor 2; TR/TE, 113/5; flip angle, 80 ; spatial resolution, mm; acquisition time, 1 minute per measurement). An IV bolus injection of 0.1 mmol/kg body weight gadopentetate dimeglumine (Magnevist, Bayer Schering Pharma) was administered at 3 ml/s with an automatic injector (Spectris, Medrad) and followed immediately by 20 ml of saline solution. Thirty seconds after contrast administration, dynamic MRI was performed with the same sequence parameters and identical tuning conditions, for a total of seven measurements. After dynamic acquisition, a bilateral axial T2-weighted turbo spin-echo sequence (GRAPPA factor 2; 8,900/207; flip angle, 90 ; spatial resolution, mm; acquisition time, 2 minutes 15 seconds) and a bilateral axial turbo spin-echo inversion recovery sequence Fig year-old woman with true-positive breast MRI finding. Contrast-enhanced subtraction MR image obtained in first minute after IV contrast injection shows lobulated mass lesion with irregular margins and heterogeneous enhancement pattern and pathologic vascularization with feeding vessels. Signal intensity time curve type showed fast washout pattern. Histopathologic examination revealed G3 invasive ductal carcinoma 23 mm in diameter. Fig year-old woman with true-positive breast MRI finding. Contrast-enhanced subtraction MR image obtained in first minute after IV contrast injection shows segmental nonmass lesion with heterogeneous enhancement pattern. Curve type was medium and persistent. Histopathologic examination showed 14-mm G2 ductal carcinoma in situ. with magnitude reconstruction (GRAPPA factor 2; 8,420/70; inversion time, 150 milliseconds; flip angle, 180 ; spatial resolution, mm; acquisition time, 2 minutes 33 seconds) were performed in identical slice positions. Image Interpretation Two examiners, one experienced (more than 2,000 breast MRI investigations) and one trained (more than 500 breast MRI investigations) classified the selected breast MR images anew. Neither was involved in rating the initial MR images, and Fig year-old woman with false-positive breast MRI finding. Contrast-enhanced subtraction MR image obtained in seventh minute after IV contrast injection shows lobulated mass lesion with smooth margins and somewhat heterogeneous enhancement pattern. Signal intensity time curve was slow and persistent. Irregular-appearing margin parts are caused by partial volume effects at slice thickness of 3 mm. Histopathologic examination revealed fibroadenoma. Fig year-old woman with false-positive breast MRI finding. Contrast-enhanced subtraction MR image obtained in first minute after IV contrast injection shows focal nonmass lesion with stippled internal enhancement pattern. Curve type was medium and persistent. Histopathologic examination revealed fibrocystic disease. both were blinded to the initial MRI report and to the results of the histopathologic examinations. Rating was performed in consensus. In cases of discordance, the rating was discussed until agreement was reached. First, identified lesions were categorized as mass or nonmass lesions. The entity focus was skipped, and lesions smaller than 5 mm in diameter were classified in the remaining categories. Tumor diameter was assessed on axial images with electronic calipers. Table 1 shows the compilation of morphologic lesion features assessed in this study according to the BI-RADS MRI lexicon [19]. AJR:194, June

3 Baltzer et al. Dynamic contrast enhancement was assessed by placement of a region-of-interest of 3 6 voxels in the most suspect part of the tumor and consecutive signal intensity analysis of unenhanced and contrast-enhanced images. The increase in signal intensity in the first minute after IV contrast injection (SI 1min ) was calculated in percentage relative to the unenhanced signal intensity (SI native ) as follows: (SI 1min SI native )/SI native. According to the BI-RADS lexicon, initial enhancement was ordinally classified as slow (< 50%), intermediate (50 100%), or fast (> 100%). Type of delayed enhancement curve type was classified according to the BI-RADS lexicon [21]. Reference Standard We chose the results of histopathologic examination as the reference standard for lesion evaluation. Findings could either be benign or malignant, and the tumor type was recorded. All histopathologic examinations were performed by board-certified breast pathologists at the institute of pathology of our university hospital. Statistical Analysis All statistical analyses were performed with software (SPSS version 15.0, SPSS; and MedCalc version , MedCalc). Differences in tumor diameter were tested with the Mann-Whitney U test. For testing the hypothesis that categoric lesion features differ between true-positive and false-positive findings, a two-tailed chi-square test was used. In expected frequencies less than 5, the two-tailed Fisher s exact test was performed. For cross-tab analysis, each lesion feature was dichotomized as present or not present and tested against reference standard categories malignant and benign. To address the single descriptor value for lesion differentiation, positive and negative likelihood ratios and their 95% CIs were calculated. Unlike predictive values, these parameters are not influenced by pretest probability [22]. Results The ratio of mass to nonmass lesions differed significantly (p < 0.001) between truepositive and false-positive findings (Table 1). For true-positive findings the ratio was 105:15 (7:1), whereas for false-positive findings the ratio was 17:14 (1.2:1). The positive predictive value for mass lesions (86.1%; 95% CI, %) was significantly higher (p < 0.001) than that for nonmass lesions (51.7%; 95% CI, %). The 15 malignant nonmass lesions were nine invasive ductal carcinomas, one invasive lobular carcinoma, two ductal carcinomas in situ (DCIS), one sarcoma, one inflammatory TABLE 1: True-Positive () and False-Positive () Findings on Mass and Nonmass According to BI-RADS MRI Lexicon Finding Mass Nonmass Total False-positive True-positive Positive predictive value (95% CI) a 86.1 ( ) 51.7 ( ) 79.5 ( ) Total Note Values in parentheses are percentages. a Difference between mass and nonmass lesions, p < carcinoma (with clinical inflammatory symptoms), and one carcinoma not classified. The 14 benign nonmass lesions were eight cases of fibrocystic disease, two fibroadenomas, three papillomas, and one case of inflammation. Atypical ductal hyperplasia, atypical lobular hyperplasia, and lobular carcinoma in situ were not found. The 105 malignant mass lesions were 76 invasive ductal carcinomas, 14 invasive lobular carcinomas, three DCIS, two lymphomas, one tubular carcinoma, one mucinous carcinoma, one medullary carcinoma, one mixed-type carcinoma, and six carcinomas not classified. The 17 benign mass lesions were 13 cases of fibrocystic disease, two fibroadenomas, and two papillomas. There was no significant difference in the prevalence of pure DCIS among malignant TABLE 2: Morphologic Characteristics of Mass Shape Descriptor (n = 105) mass lesions and malignant nonmass lesions (p = 0.117, Fisher s exact test). However, malignant nonmass lesions were accompanied by DCIS more often (seven DCIS components among the 15 [two malignant nonmass lesions and two pure DCIS subtracted]) than were malignant mass lesions (22 DCIS components among the 105 [three malignant mass lesions subtracted]) (p < 0.05, chisquare test). Tumor diameter had no influence on the correct diagnosis of nonmass lesions (p = 0.301). The mean diameter of false-positive nonmass lesions was 30.4 ± 26.9 mm, whereas the mean diameter of true-positive nonmass lesions was 39.2 ± 29.0 mm. Conversely, false-positive mass lesions were significantly smaller than true-positive mass lesions (mean diameter of false-positive mass lesions, 16.9 ± 16.9 mm; (n = 17) Negative Likelihood Round ( ) 1.27 ( ) Oval ( ) 1.03 ( ) Lobulated ( ) 0.86 ( ) Irregular ( ) 0.88 ( ) Margin Smooth a ( ) 2.34 ( ) Irregular b ( ) 0.5 ( ) Spiculated c ( ) 0.76 ( ) Internal enhancement pattern Homogeneous 1 0 NA 0.99 ( ) Heterogeneous ( ) 1.66 ( ) Rim ( ) 0.8 ( ) Dark septations 1 0 NA 0.99 ( ) Enhancing septations 3 0 NA 0.97 ( ) Central, centrifugal ( ) Note Values in parentheses are 95% CI. Difference between benign (false-positive) and malignant (true-positive) lesions was significant at a p < 0.001, b p < 0.05, and c p < NA = not applicable AJR:194, June 2010

4 BI-RADS MRI TABLE 3: Morphologic Characteristics of Nonmass TABLE 5: Initial Enhancement of Mass and Nonmass Pattern Mass lesions Descriptor Spatial distribution Negative Likelihood Slow ( ) 1.11 ( ) Intermediate ( ) 1.21 ( ) Strong ( ) 0.73 ( ) Total Nonmass lesions (n = 15) Slow ( ) 1.4 ( ) Intermediate ( ) 0.93 ( ) Strong 3 0 NA 0.8 ( ) Total (n = 14) Negative Likelihood Focal ( ) 1.4 ( ) Ductal ( ) 1.09 ( ) Segmental ( ) 0.59 ( ) Regional ( ) 1.01 ( ) Multiple regional ( ) 1.01 ( ) Internal enhancement pattern Heterogeneous ( ) 0.70 ( ) Stippled a ( ) 1.63 ( ) Clumped ( ) 1.03 ( ) Reticular ( ) 0.86 ( ) Note Values in parentheses are 95% CI. Difference between benign (false-positive) and malignant (true-positive) lesions was significant at a p < TABLE 4: Delayed Enhancement Curve Type of Mass and Nonmass Pattern Mass lesions Negative Likelihood Persistent a ( ) 1.38 ( ) Plateau ( ) 0.97 ( ) Washout ( ) 0.72 ( ) Total Nonmass lesions LR+ (95% CI) LR- (95% CI) Persistent ( ) 1.60 ( ) Plateau ( ) 0.53 ( ) Washout 1 0 NA 0.93 ( ) Total Note Values in parentheses are 95% CI. Difference between benign (false-positive) and malignant (true-positive) lesions was significant at a p < Note Values in parentheses are 95% CI. No significant differences between benign (false-positive) and malignant (true-positive). mean diameter of true-positive mass lesions, 23.7 ± 15.5 mm) (p = 0.01). Tables 2 and 3 show the distribution of morphologic features with respect to truepositive and false-positive findings. In evaluation of mass lesions, margin (smooth, irregular, or spiculated) and contrast agent kinetics were statistically significant variables in differentiating malignant and benign lesions (smooth and irregular margins, p < 0.05; spiculated margins, p = 0.068). In the evaluation of nonmass lesions, the only feature that significantly differentiated malignant and benign lesions was the stippled enhancement pattern (p < 0.05). Initial enhancement was somewhat stronger in mass than in nonmasslike lesions (benign subgroup, p = 0.05; malignant subgroup, p = 0.08). versus malignant mass lesions and benign versus malignant nonmass lesions did not differ significantly in this respect. Significant differences in delayed curve type were found for mass lesions only (persistent, p < 0.05). The initial and delayed enhancement features are shown in Tables 4 and 5. Discussion Nonmass lesions were identified as a challenging subgroup causing a high proportion of false-positive diagnoses at diagnostic breast MRI. Although representing a minority, nonmass lesions made up nearly one half (14 of 31) of all false-positive findings in routine diagnostics. The ratio of nonmass to mass lesions has been described previously [17, 23]. Our study sample included 29 suspicious nonmass lesions and 122 suspicious mass lesions. This proportion is in concert with the earlier reports. Higher ratios of nonmass to mass lesions have been reported, reaching 40% [11]. Tozaki and Fukuda [17] also found a high fraction of false-positive findings among 30 nonmass lesions (40%, 12 of 30). This value is comparable with our results. We identified 48% false-positive findings among all nonmass lesions. However, our study showed that nonmass lesions make up a high fraction of all false-positive findings despite their low prevalence among all lesions detected with breast MRI. DCIS lesions have been found to exhibit nonmass enhancement at a disproportionately high rate, ranging between 69% and 90% [11, 17, 24]. This finding was not supported in our study because only two of five (40%) pure DCIS lesions had nonmasslike enhancement. However, DCIS components have been found significantly more frequently in association AJR:194, June

5 Baltzer et al. with nonmass lesions than with mass lesions. This finding is in concordance with the assumption that DCIS often grows in nonmasslike patterns [15, 16, 24]. Because a high percentage of DCIS lesions exhibit suspicious microcalcifications on mammograms [25], biopsy is frequently performed before MRI. The presence of pure DCIS so far has not been a common indication for breast MRI [1]. Accordingly, the prevalence of DCIS in our study was low, and our findings in terms of DCIS should be regarded as preliminary. Liberman et al. [26] found that the positive predictive value (PPV) of breast MRI in the detection of malignancy increases with increasing lesion size. This finding is in accordance with our finding that smaller lesions are more likely to be false-positive. This rule, however, is applicable only to mass lesions. The size of nonmass lesions does not affect the PPV of breast MRI in the detection of malignancy. Three lesion features margin and contrast agent kinetics of mass lesions and stippled enhancement of nonmass lesions enabled differentiation of false-positive and true-positive lesions. The significance of stippled enhancement had been reported [11, 16, 17]. Tozaki and Fukuda [17] found segmental enhancement a lesion feature exclusive to nonmass malignant lesions. We found the PPV of segmental enhancement was 73% (8/11), comparable with findings in a previous publication on DCIS [15]. Another study [2] put forward a PPV of 34% for segmental enhancement. Sakamoto et al. [16] found no value of segmental enhancement pattern in the differential diagnosis of pathologic changes in the breast. Dynamic contrast enhancement patterns can be used to differentiate malignant and benign lesions [10]. Our findings suggest that nonmass lesions do not follow the reported kinetic patterns because a typical washout pattern occurred only once among all malignant lesions. Initial enhancement was less in nonmass lesions but did not differ significantly in the benign and malignant subgroups. However, a tendency toward less enhancement of benign lesions was observed. The causes may be biologic differences, such as lower levels of neoangiogenesis, and technical limitations of the spatial resolution of the imaging technique used. According to these results, kinetic enhancement features cannot be recommended as a diagnostic criterion for nonmass lesions. One limitation of our study pertains to the applicability of our results to routine conditions, that is, the methodologic exclusion of lesions for which no biopsy was recommended (findings that could be either true-negative or false-negative). Thus the evidence of certain lesion features indicating benign or malignant breast disease might have been overestimated or underestimated in our study. Predictive values depend on the prevalence of the examined feature and do not consider the relative frequency of the feature among benign and malignant lesions [22]. Accordingly, we did not perform PPV calculations for single lesion features. For example, in our study rim enhancement had a PPV of 92.3% for malignancy of mass lesions (39 lesions showed rim enhancement, 36 of them were malignant). The corresponding p value was 1, implying that the feature is not helpful for further lesion differentiation. This result occurred because the PPV among the study patients collectively was high (mass lesions, 86.1%; nonmass lesions, 51.7%; total, 79.5%) because only BI-RADS category 4 and 5 lesions (suspicious abnormality and highly suggestive of malignancy) were investigated. The power of descriptors such as rim enhancement is weakened in this situation. In addition, the three papillary lesions in our patient collective were classified as benign lesions. Most of these lesions are excised because of their uncertain malignant potential [27]. However, all papillary lesions in this study were managed with open surgery and proved benign. It was the aim of this investigation to identify whether BI-RADS criteria accumulate in false-positive lesions. We found limited accuracy of morphologic and dynamic BI-RADS descriptors in further discrimination of truepositive and false-positive MRI findings and thus in decreasing the number of false-positive results. Because the inclusion criteria in this study represent current clinical practice (i.e., routine indications, trained observers, up-to-date technology), these findings have to be expected at least in part. Therefore, to further reduce the rate of false-positive findings, future investigations should concentrate on more accurate lesion features in the differential diagnosis of pathologic changes in the breast, especially nonmasslike lesions. Regarding initial research on alternative MRI techniques, proton MR spectroscopy has had high diagnostic values for nonmass lesions, but diffusion-weighted imaging has not [23, 28]. According to our results, false-positive findings of breast MRI show disproportionately frequent nonmasslike enhancement. The size of true-positive nonmass lesions did not differ from that of false-positive nonmass lesions, but false-positive mass lesions were significantly smaller than true-positive mass lesions. Among the existing BI-RADS descriptors, only the stippled enhancement pattern was helpful in differentiating benign from malignant nonmass lesions in our investigation. For mass lesions, margin characteristics should be given strong attention because more than one half of all false-positive findings in mass lesions had a smooth margin. We conclude that more specific lesion descriptors are necessary to eliminate false-positive ratings, especially of nonmass lesions. References 1. Kuhl CK. Current status of breast MR imaging. Part 2. Clinical applications. Radiology 2007; 244: 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: Warner E, Plewes DB, Hill KA, et al. Surveillance of BRCA1 and BRCA2 mutation carriers with magnetic resonance imaging, ultrasound, mammography, and clinical breast examination. JAMA 2004; 292: Facius M, Renz DM, Neubauer H, et al. Characteristics of ductal carcinoma in situ in magnetic resonance imaging. Clin Imaging 2007; 31: Fischer DR, Baltzer P, Malich A, et al. Is the blooming sign a promising additional tool to determine malignancy in MR mammography? Eur Radiol 2004; 14: Fischer DR, Malich A, Wurdinger S, Boettcher J, Dietzel M, Kaiser WA. The adjacent vessel on dynamic contrast-enhanced breast MRI. AJR 2006; 187:338; [web]w147 W Fischer U, von Heyden D, Vosshenrich R, Vieweg I, Grabbe E. Signal characteristics of malignant and benign lesions in dynamic 2D-MRT of the breast [in German]. Rofo 1993; 158: Kaiser WA. False-positive results in dynamic MR mammography: causes, frequency, and methods to avoid. Magn Reson Imaging Clin N Am 1994; 2: Kuhl CK, Klaschik S, Mielcarek P, Gieseke J, Wardelmann E, Do Schild HH. T2-weighted pulse sequences help with the differential diagnosis of enhancing lesions in dynamic breast MRI? J Magn Reson Imaging 1999; 9: Kuhl CK, Mielcareck P, Klaschik S, et al. Dynamic breast MR imaging: are signal intensity time course data useful for differential diagnosis of enhancing 1662 AJR:194, June 2010

6 BI-RADS MRI lesions? Radiology 1999; 211: MRI. Breast Cancer 2008; 15: fusion-weighted imaging (DWI) in MR mam- 11. Liberman L, Morris EA, Lee MJ, et al. Breast le- 17. Tozaki M, Fukuda K. High-spatial-resolution mography (MRM): clinical comparison of echo sions detected on MR imaging: features and posi- MRI of non-masslike breast lesions: interpreta- planar imaging (EPI) and half-fourier single-shot tive predictive value. AJR 2002; 179: tion model based on BI-RADS MRI descriptors. turbo spin echo (HASTE) diffusion techniques. 12. Malich A, Fischer DR, Wurdinger S, et al. Poten- AJR 2006; 187: Eur Radiol 2009; 19: tial MRI interpretation model: differentiation of benign from malignant breast masses. AJR 2005; 185: Matsubayashi R, Matsuo Y, Edakuni G, Satoh T, Tokunaga O, Kudo S. Breast masses with peripheral rim enhancement on dynamic contrast-enhanced MR images: correlation of MR findings with histologic features and expression of growth factors. Radiology 2000; 217: Morakkabati-Spitz N, Leutner C, Schild H, Traeber F, Kuhl C. Diagnostic usefulness of segmental and linear enhancement in dynamic breast MRI. Eur Radiol 2005; 15: Neubauer H, Li M, Kuehne-Heid R, Schneider A, Kaiser WA. High grade and non-high grade ductal carcinoma in situ on dynamic MR mammography: characteristic findings for signal increase and morphological pattern of enhancement. Br J Radiol 2003; 76: Sakamoto N, Tozaki M, Higa K, et al. Categorization of non-mass-like breast lesions detected by 18. Kaiser WA, Zeitler E. MR imaging of the breast: fast imaging sequences with and without Gd-DT- PA preliminary observations. Radiology 1989; 170: Ikeda DM, Hylton NM, Kinkel K, et al. Development, standardization, and testing of a lexicon for reporting contrast-enhanced breast magnetic resonance imaging studies. J Magn Reson Imaging 2001; 13: Leach MO, Boggis CR, Dixon AK, et al. Screening with magnetic resonance imaging and mammography of a UK population at high familial risk of breast cancer: a prospective multicentre cohort study (MARIBS). Lancet 2005; 365: American College of Radiology. BI-RADS: MRI, 1st ed. In: Breast imaging reporting and data system: BI-RADS atlas. Reston, VA: American College of Radiology, Altman DG, Bland JM. Diagnostic tests 2: predictive values. BMJ 1994; 309: Baltzer PA, Renz DM, Herrmann KH, et al. Dif- 24. Orel SG, Mendonca MH, Reynolds C, Schnall MD, Solin LJ, Sullivan DC. MR imaging of ductal carcinoma in situ. Radiology 1997; 202: Mendez A, Cabanillas F, Echenique M, Malekshamran K, Perez I, Ramos E. Mammographic features and correlation with biopsy findings using 11-gauge stereotactic vacuum-assisted breast biopsy (SVABB). Ann Oncol 2004; 15: Liberman L, Mason G, Morris EA, Dershaw DD. Does size matter? Positive predictive value of MRI-detected breast lesions as a function of lesion size. AJR 2006; 186: Putti TC, Pinder SE, Elston CW, Lee AH, Ellis IO. Breast pathology practice: most common problems in a consultation service. Histopathology 2005; 47: Bartella L, Thakur SB, Morris EA, et al. Enhancing nonmass lesions in the breast: evaluation with proton ( 1 H) MR spectroscopy. Radiology 2007; 245:80 87 AJR:194, June