Predicting neoadjuvant chemotherapy benefit using deep learning from stromal histology in breast cancer

Abstract

Neoadjuvant chemotherapy (NAC) is a standard treatment option for locally advanced breast cancer. However, not all patients benefit from NAC; some even obtain worse outcomes after therapy. Hence, predictors of treatment benefit are crucial for guiding clinical decision-making. Here, we investigated the predictive potential of breast cancer stromal histology via a deep learning (DL)-based approach and proposed the tumor-associated stroma score (TS-score) for predicting pathological complete response (pCR) to NAC with a multicenter dataset. The TS-score was demonstrated to be an independent predictor of pCR, and it not only outperformed the baseline variables and stromal tumor-infiltrating lymphocytes (sTILs) but also significantly improved the prediction performance of the baseline variable-based model. Furthermore, we discovered that unlike lymphocytes, collagen and fibroblasts in the stroma were likely associated with a poor response to NAC. The TS-score has the potential to better stratify breast cancer patients in NAC settings.