The impact of high PD-L1 expression on the surrogate endpoints and clinical outcomes of anti-PD-1/PD-L1 antibodies in non-small cell lung cancer

•The hazard ratio of OS was strongly correlated with hazard ratio of PFS in all patients in ICI trials.•Especially in patients with high PD-L1 expression, ORR and PFS was correlated with OS.•These surrogate endpoints may be considered in accordance with the design of the trials to evaluate the efficacy of ICIs. Recent reports have indicated that the objective response rate (ORR) and progression-free survival (PFS) cannot serve as surrogates for predicting overall survival (OS) in immune checkpoint inhibitor (ICI) trials. We performed a trial-based correlative analysis to evaluate conventional endpoints as surrogates for predicting OS in ICI-treated non-small cell lung cancer (NSCLC) patients. A systematic electronic literature search for randomized clinical trials using ICI monotherapies for NSCLC revealed 7 trials. The correlative analysis to clarify the correlations among clinical outcomes used a weighted Spearman rank correlation coefficient (wS), weighted Pearson correlation coefficient (wP), and weighted linear regression model (wL) in all patients and patients with high PD-L1 expression. The correlative analysis of the total population revealed that the odds ratio of the ORR (OR-ORR) and the hazard ratio of OS (HR-OS) were strongly correlated with the hazard ratio of PFS (HR-PFS) (R for wP and wS, R2 for wL; −0.869, −0.968, 0.756 between OR-ORR and HR-PFS; 0.923, 0.959, 0.851 between HR-PFS and HR-OS). The strongest correlation was observed between one-year overall survival (1y-OS) and the HR-OS (R for wP and wS, R2 for wL; 0.985, 1.000, R2: 0.968). In those with high PD-L1 expression, the ORR and PFS were strongly associated with OS (R2: 0.842 between ORR and OS; 0.771 between PFS and OS). The OR-ORR and HR-PFS could serve as surrogate endpoints for predicting the HR-OS in randomized trials using ICIs for NSCLC, while the ORR and PFS could be useful endpoints for predicting OS in trials with patient selection based on high PD-L1 expression.