Integration of tumour infiltrating lymphocytes, programmed cell-death ligand-1, CD8 and FOXP3 in prognostic models for triple-negative breast cancer: Analysis of 244 stage I-III patients treated with standard therapy.

BACKGROUND: Tumour infiltrating lymphocytes (TILs) are an established prognostic biomarker for triple-negative breast cancer (TNBC). We evaluated the role of programmed cell-death ligand-1 (PD-L1), CD8 and FOXP3 expression in refining a prognostic model for non-metastatic TNBC beyond classic factors and TILs. METHODS: Primary tumour samples from 244 early patients with TNBC, all treated with surgery and chemotherapy, were collected. Stromal TILs were evaluated on haematoxylin-eosin slides according to guidelines. PD-L1, CD8 and FOXP3 were assessed by immunohistochemistry and evaluated by digital pathology. RESULTS: TILs, PD-L1, CD8 and FOXP3 were positively correlated with each other (P < 0.001). TILs were confirmed as an independent prognostic factor. When PD-L1, CD8 and FOXP3 were added to multivariable models including classic factors (age, stage, histologic grade) and TILs, PD-L1 provided the largest amount of additional prognostic information: likelihood ratio χ2 4.60, P = 0.032 (in a model including classic factors and TILs 10% increments) and likelihood ratio χ2 6.50, P = 0.011 (in a model including classic factors and TILs >30% versus <30%). In the subset of patients treated with neoadjuvant chemotherapy, FOXP3 provided further prognostic information beyond classic factors, TILs and pathological complete response (pCR) (likelihood ratio χ2 5.01, P = 0.025). For patients who did not achieve a pCR, the expression of CD8 and PD-L1 was significantly increased from baseline to residual disease. CONCLUSIONS: Beyond clinicopathological factors and TILs, other immune biomarkers may add prognostic information for early TNBC. The increased PD-L1 expression on residual disease after neoadjuvant chemotherapy strengthens the rationale of testing immune checkpoint inhibitors in the post-neoadjuvant setting.