Scoring systems for PD-L1 expression and their prognostic impact in patients with resectable gastric cancer

The combined positive score (CPS) and tumor proportion score (TPS) have been developed to evaluate programmed death ligand-1 (PD-L1) expression, especially due to the potential benefit of the targeted therapy. However, the prognostic value of PD-L1 scoring systems in gastric cancer (GC) remains unclear. This study aimed to evaluate PD-L1 expression according to CPS and TPS in curative resected GC patients and its correlation with prognosis. We retrospectively evaluated 284 GC patients who underwent D2-gastrectomy by tissue microarray. PD-L1 expression was analyzed by immunohistochemistry. PD-L1 positivity by CPS and TPS was observed in 45 (15.8%) and 34 (12%) patients, respectively. Larger tumor size (p = 0.028), undetermined Lauren type (p < 0.001), and heavy inflammatory infiltrate (p = 0.009) were associated with CPS-positive GC. TPS-positive were more frequent in patients with larger tumor size (p = 0.004), undetermined type (p < 0.001), moderate/severe inflammatory infiltrate (p = 0.001), total gastrectomy (p = 0.036), and poorly differentiated histology (p = 0.025). No differences were observed in the pT, pN, and pTNM status according to the PD-L1 scores. Both scores were associated with Epstein-Barr virus positivity, microsatellite instability and p53-normal expression. The disease-free survival (DFS) was worse for CPS-negative compared to CPS-positive group (p = 0.052). No difference was observed between TPS-positive and negative groups (p = 0.436). Total gastrectomy, advanced pT status, and CPS-negative were independent factor for worse survival in GC. CPS was an independent prognostic factor for survival and could be used as a prognostic biomarker in patients with resectable GC.