Role of breast MR imaging for predicting malignancy of histologically borderline lesions diagnosed at core needle biopsy: prospective evaluation.

Purpose: To prospectively determine the diagnostic performance of breast magnetic resonance (MR) imaging in predicting the malignancy of breast lesions classified as borderline at core needle biopsy (CNB). Materials and Methods: This prospective study was approved by the local ethics committee, and all patients provided written informed consent. Between February 2007 and October 2009, 193 patients underwent ultrasonography (US)-guided CNB. Thirty-two lesions in 32 patients were classified as "borderline" ("lesions of uncertain malignant potential" according to the United Kingdom National Health Service Breast Screening program or those that are "probably benign" according to the Breast Imaging Reporting and Data System [BI-RADS] lexicon) at pathologic examination and were evaluated. All 32 patients underwent contrast agent-enhanced breast MR imaging 1-3 weeks after CNB; surgical biopsy was performed within 60 days of MR imaging. Results were compared with histologic findings. Sensitivity, specificity, accuracy, and positive and negative predictive values were calculated. Results: Findings from surgical biopsy confirmed the diagnosis of seven atypical ductal hyperplasia lesions, four atypical lobular hyperplasia (ALH) lesions, five papillary lesions, two fibroepithelial lesions, three radial scars, three lobular carcinomas in situ, three ductal carcinomas in situ (DCIS), four invasive ductal carcinomas, and one invasive lobular carcinoma. With regard to malignancy, breast MR imaging helped correctly classify 22 of the 32 lesions as negative and seven as positive. Two of the lesions confirmed to be ALH were incorrectly classified as suspicious (BI-RADS category IV; false-positive finding), whereas one small DCIS was incorrectly classified as nonsuspicious (BI-RADS category II; false-negative finding). The sensitivity, specificity, accuracy, and positive and negative predictive values for the correct demonstration of malignancy at MR imaging were 88% (seven of eight lesions), 92% (22 of 24 lesions), 91% (29 of 32 lesions), 78% (seven of nine lesions), and 96% (22 of 23 lesions), respectively. Conclusion: Breast MR imaging can improve the evaluation of lesions classified as borderline at CNB. In cases of nonsuspicious enhancement or no enhancement at breast MR imaging, follow-up rather than surgical biopsy might be performed. (C)RSNA, 2010