PD-L1 thresholds predict efficacy of immune checkpoint inhibition in first-line treatment of advanced gastroesophageal adenocarcinoma. A systematic review and meta-analysis of seven phase III randomized trials.

backgroundHigh expression of programmed death-ligand 1 (PD-L1) has been recognized as a marker of improved efficacy of immunotherapy in gastroesophageal adenocarcinoma (GEA); however, the optimal PD-L1 cut-off is still debated. The aim of the present review was to analyze available phase III trials and to identify the appropriate PD-L1 expression cut-off for GEA.

methodsPhase III trials investigating the efficacy of anti-programmed cell death protein 1 (PD-1) therapies in addition to standard chemotherapy versus standard chemotherapy in the first-line setting were selected. Progression-free survival (PFS), overall survival (OS) and objective response rate (ORR) were the analyzed outcome measures. Pooled treatment effects were assessed in the unselected population and in subpopulations with different levels of PD-L1 expression.

resultsPD-1 blockade efficacy was found to consistently increase in a linear manner with higher combined positive score (CPS) of PD-L1 expression: pooled hazard ratio (HR) for OS and PFS and pooled odds ratio (OR) for ORR of 0.80, 0.75 and 1.51, respectively, in the unselected population versus 0.67, 0.63 and 1.90, respectively, in the CPS ≥10 population (all P values < 0.0001). In the PD-L1-negative population (CPS <1) a significant benefit of anti-PD-1 agents could not be demonstrated in terms of OS and PFS (P = 0.28 and 0.12, respectively), but it was seen in terms of ORR (P = 0.03). PD-1 blockade was effective in the CPS <10 population (P value for pooled OS HR, PFS HR and response OR are all 0.01), while in the CPS <5 population the effect was of borderline significance for OS (P = 0.07) and significant for PFS and ORR (P = 0.02 and 0.03, respectively).

conclusionThe present meta-analysis confirmed that the benefit of PD-1 blockade in GEA patients is related to PD-L1 CPS, with increased benefit observed for higher CPS cut-offs and no OS benefit in the CPS <1 subset. Overall, data indicate that PD-L1 CPS ≥5 could represent an acceptable cut-off to optimize the risk/benefit ratio of such agents. Our data suggest a potential clinical benefit of immunotherapy in selected patients within the CPS 1-4 population which needs further investigation.