Early Prediction of Treatment Response in HER2-Positive Breast Cancer Using multiparametric MRI

Motivation: Imaging pre- and post- neoadjuvant chemotherapy (NAC) fails to adequately capture and quantify temporal heterogeneity and biological changes of tumors. Goal(s): To assess if longitudinal changes in multiparametric MRI can predict early response to neoadjuvant chemotherapy (NAC) in HER2-positive breast cancer and to establish quantitative models based on these features. Approach: Two predictive models were developed, one based on clinicopathologic features and another that combined clinicopathologic and MRI features. Results: The combined model performs optimally in all datasets. Changes observed in multiparametric MRI can predict early treatment responses in HER2-positive BC and assist in tailoring personalized treatment plans. Impact: The prediction model was simple and feasible, which was helpful for individualized treatment planning.