AI-powered and manual assessment of tumor-infiltrating lymphocytes in early HER2-positive breast cancer in NSABP B-41

The assessment of tumor-infiltrating lymphocyte (TILs), together with gene expression signatures (GES), has the potential to guide personalized breast cancer therapy. We included 262 patients from the phase III NSABP B-41 trial, which evaluated neoadjuvant HER2-targeted therapies in combination with chemotherapy. We conducted a manual and artificial intelligence (AI)-based analyses of TILs, as well as GES from RNA sequencing. Higher manual TILs as a continuous variable were associated with pathologic complete response (pCR) in patients with estrogen receptor (ER)-negative disease. AI-based TILs were associated with pCR regardless of ER status. Immune GES (iGES) were associated with pCR. Manual TILs were not associated with event-free survival (EFS), while AI-TIL showed a marginal association. These results support the use of TIL assessment, complemented by GES, as a prognostic biomarker in HER2-positive breast cancer. Future studies are needed to evaluate their predictive utility to guide treatment decisions.