Computational measurement of tumor immune microenvironment in gastric adenocarcinomas

The use of four groups of tumor immune microenvironments (TME) based on PD-L1 and tumor-infiltrating T lymphocytes (TIL) is a reliable biomarker for anti-PD-1/PD-L1 inhibitor therapy. We classified the TME in 241 gastric cancers which were subdivided according to 40 EBV+, 76 microsatellite instability-high (MSI-H), and 125 EBV-/microsatellite-stable (MSS) subtypes by quantitative image analysis (QIA) and correlated the results with mRNA expression levels. The mean PD-L1 ratio and CD8 ratio in EBV+, MSI-H, and EBV−/MSS GCs were significantly different (P < 0.006). The PD-L1 ratio and CD8 ratio obtained by QIA correlated well with the RNA levels of PD-L1 (r = 0.63) and CD8 (r = 0.67), respectively. The TME were type I (PD-L1^H/CD8^H) in 45, type II (PD-L1^L/CD8^L) in 106, type III (PD-L1^H/CD8^L) in 8, and type IV (PD-L1^L/CD8^H) in 82 cases. The type I TME was significantly associated with high TIL (P = 3.0E-11) and EBV+ status (P = 8.55E-08). In conclusion, QIA results correlated well with the mRNA expression levels and classified TME of gastric cancers.