Accuracy of digital mammography, ultrasound and MRI in predicting the pathological complete response and residual tumor size of breast cancer after completion of neoadjuvant chemotherapy

Background Neoadjuvant chemotherapy (NACT) is the standard of care for the treatment of locally advanced or non-metastatic breast cancer, which may increase the chances of breast conservative surgery (BCS) in place of radical mastectomy without compromising on the overall survival. The aim of this study was to evaluate the accuracy of mammography (MG), ultrasound (US), and magnetic resonance imaging (MRI) in predicting the complete response and to assess the extent of residual breast cancer in women treated with NACT. Materials and methods Fifty-six consecutive patients with stage II or III breast cancer, who underwent imaging evaluation of breast with digital mammogram, US, and MRI after NACT and before the breast surgery, were included in the study. For each patient, pathologic complete response (pCR) or residual tumor (non-pCR) was predicted and the maximum extent of the residual tumor was measured on each imaging modality. These measurements were subsequently compared with the final histopathology results. Results Of 56 patients, 22 showed pCR with MRI having better accuracy for predicting complete response than the MG and US (area under the receiver operating characteristic curve: 0.86, 0.68, and 0.65, respectively; p = 0.0001 for MRI; p = 0.06 for MG, and p = 0.02 for US). The sensitivity of MRI for detecting pCR was 72.7%; specificity and positive predictive value were 100%. For pathological residual tumor, the size measured on MRI showed significantly higher correlation with the pathologic size (correlation coefficient, r = 0.786), than the MG (r = 0.293) and US (r = 0.508) with P Conclusions Accuracy of MRI for predicting pathological complete response was significantly higher than the MG and US. Pathologic residual tumor size was also more precisely reflected by the longest tumor dimension on MRI with the strong positive correlation coefficient.