AbstractPATIENTS AND METHODS
Eligible patients with locally advanced breast cancer received doxorubicin 60 mg/m2 and docetaxel 60 mg/m2 (with granulocyte colony stimulating factor support) every 14 days for a maximum of six cycles. Breast magnetic resonance imaging was performed at baseline and repeated every two cycles. Surgery (either local excision or mastectomy) was performed after six cycles in responding or stable patients. Residual tumor size on pathology and preoperative magnetic resonance imaging was compared; concordance was defined as a < 0.5-cm difference.RESULTS
To date, three of 17 enrolled subjects (17.6%) attained pathologic complete response, and three additional patients attained near pathologic complete response, with residual foci of ≤ 1 mm. Of these six patients, only one was disease free by magnetic resonance imaging. Discordance between magnetic resonance imaging findings and pathologic evaluation was found in four of six patients (66.6%) who obtained pathologic complete response or near pathologic complete response. In the three patients in whom four axillary lesions were followed with magnetic resonance imaging, discordance was found in all four lesions, with magnetic resonance imaging overestimating pathologic disease in all cases.DISCUSSION
Our findings caution that magnetic resonance imaging may frequently overestimate residual invasive carcinoma after neoadjuvant chemotherapy. These results contradict previous studies suggesting that postchemotherapy magnetic resonance imaging may underestimate residual cancer. The use of magnetic resonance imaging in evaluating response to therapy in locally advanced breast cancer should be further studied.