PD-L1 expression patterns in oral cancer as an integrated approach for further prognostic classification.

ABSTRACT

BACKGROUND: Despite the well-known role of programmed cell death ligand 1 (PD-L1) in promoting immune resistance in oral squamous cell carcinoma (OSCC), its potential utility as a prognostic biomarker is undetermined. We evaluated PD-L1 expression as predictor of survival in patients with OSCC and explored PD-L1 expression patterns. METHODS: We conducted a retrospective cohort study that assessed PD-L1 expression through immunohistochemistry in 123 surgical specimens of OSCC. A first approach evaluated tumor proportion scores (TPS) and combined proportion scores (CPS). Next, expression patterns were examined by evaluating PD-L1 localization in tumor nests, as well as the interfaces of tumor cells (TC) and immune cells (IC) in the tumor microenvironment. RESULTS: High-level PD-L1 expression determined by TPS and CPS using variable cutoffs was not associated with survival. Immunohistochemistry revealed that TC expressed PD-L1 in either patchy or diffuse patterns. The patchy pattern was an independent risk factor for overall survival. Furthermore, expression patterns in the tumor immune microenvironment showed that most cases expressed PD-L1 on both TC and IC, while PD-L1 non-expressors had the lowest overall survival. CONCLUSION: PD-L1 expression patterns in the context of localization in tumor nests and TC-IC interactions represent antitumor immune responses better than either TPS or CPS. Our suggested classification system may have important implications for the characterization of OSCC and for the use of PD-L1 as a prognostic biomarker.