Predicting Relative Efficacy of Anthracyclines vs. Taxanes in Breast Cancer NAC with a Longitudinal MRI Radiomics Model

Motivation: During the mid-phase of anthracycline and taxane-based neoadjuvant chemotherapy (NAC) for breast cancer, it is crucial to assess whether to prolong anthracycline and cyclophosphamide (AC) treatment. Goal(s): To clarify the feasibility of developing MRI radiomics models for predicting the relative efficacy between AC and taxane (T) treatments. Approach: 190 patients were enrolled. Relative efficacy was defined by comparing the tumor volume change rates between the AC and the T treatment stages. Radiomics models were developed to predict relative efficacy. Results: The multi-sequence, multi-region MRI radiomics model with longitudinal data achieved AUCs of 0.887 and 0.757 in the training and test sets, respectively. Impact: Our radiomics models underscored the value of multi-region, multi-sequence MRI data in longitudinal monitoring during neoadjuvant chemotherapy (NAC), revealing the potential to guide personalized NAC regimens for patients. It is necessary to confirm this potential through prospective comparative studies.