Response Evaluation to Neoadjuvant Chemotherapy in Breast Cancer Patients: Sequential Dynamic Contrast-Enhanced MRI Using Computer-Aided Detection

Purpose: We evaluated whether there is an association between sequential changes in kinetic profiles by computer-aided detection (CAD) during neoadjuvant chemotherapy (NAC) and pathologic complete response (pCR) and residual cancer burden (RCB) in dynamic contrast-enhanced MRI (DCE-MRI) of patients with invasive breast cancer. Materials and Methods: This retrospective study involved 51 patients (median age, 48 years; range, 33–60 years) who underwent pre-, interim-, and post-NAC DCE-MRIs at 3 T. The tumor size and CAD-generated kinetic profiles (peak enhancement and delayed enhancement [persistent, plateau, and washout] components) were measured. Percent-age changes in pre-and interim-NAC (ΔMRI value1) and pre-and post-NAC (ΔMRI val-ue2) were compared between pCR and non-pCR cases, and according to RCB. Receiver operating characteristic curve analysis was performed to evaluate the association between pCR and MRI parameters (including CAD-generated kinetic profiles). Results: The pCR rate was 19.6% (10/51). There were statistically significant differences in Δtumor size2 (p < 0.01), Δpeak enhancement2 (p = 0.01), Δpersistent2 (p = 0.01), Δplateau2 (p = 0.02), and Δwashout2 (p = 0.03) between pCR and non-pCR. ΔTumor size2 provided very good diagnostic accuracy for pCR (cut-off,-90%; area under the curve, 0.88). There were differences in Δtumor size2, Δpeak enhancement2, Δplateau2, and Δwashout2 between RCB classes (p < 0.01). Conclusion: DCE-MRI using CAD has the potential for predicting pCR and RCB classes.