PD-L1 is prognostic and a negative predictor of response to immunotherapy in gastric cancer.

Cancer cells evade immune detection via programmed cell death 1/programmed cell death-ligand 1 (PD-1/PD-L1) interactions that inactivate T cells. PD-1/PD-L1 blockade has become an important therapy in the anti-cancer armamentarium. However, some patients do not benefit from PD-1/PD-L1 blockade despite expressing PD-L1. Here, we screened 101 gastric cancer (GC) patients at diagnosis and 141 healthy control subjects and reported one such subpopulation of GC patients with rs17718883 polymorphism in PD-L1, resulting in a nonsense P146R mutation. We detected rs17718883 in 44% of healthy control subjects, and rs17718883 was associated with a low susceptibility to GC and better prognosis in GC patients. Structural analysis suggests that the mutation weakens the PD-1:PD-L1 interaction. This was supported by co-culture experiments of T cells, with GC cells showing that the P146R substitution results in interferon (IFN)-γ secretion by T cells and enables T cells to suppress GC cell growth. Similar results with animal gastric tumor models were obtained in vivo. PD-1 monoclonal antibody treatment did not enhance the inhibitory effect of T cells on GC cells expressing PD-L1in vitro or in vivo. This study suggests that rs17718883 is common and may be used as a biomarker for exclusion from PD-1/PD-L1 blockade therapy.