Clinical significance of inter-assay discrepancy in PD-L1 evaluation for the efficacy of pembrolizumab in advanced NSCLC with high PD-L1 expression

Introduction Programmed cell death ligand-1 (PD-L1) expression is a predictive biomarker for the efficacy of anti-programmed cell death receptor-1/PD-L1 antibodies in advanced non-small cell lung cancer (NSCLC). Although several assays have been approved for evaluating PD-L1 expression status, inter-assay discordance has been observed between some assays. The clinical significance of these discrepancies is still unclear. Methods We retrospectively reviewed treatment-naïve NSCLC patients whose PD-L1 expression was evaluated using both 22C3 and SP142 assays. Among those, efficacy analysis was performed for patients with PD-L1 tumor proportion score (TPS) ≥ 50 % (22C3), who had received first-line pembrolizumab monotherapy. Additionally, transcriptome analysis was conducted in the available tumors with TPS ≥ 50 % to investigate the distinct immune profiles that accompany inter-assay discordance. Results In total, 611 patients were eligible. Among 198 patients with TPS ≥ 50 %, 91 (46 %) had tumor cell score ≤ 1 (SP142, i.e., inter-assay discrepancy). In the 52 patients who received first-line pembrolizumab monotherapy, treatment efficacy was significantly lower in patients with the discrepancy than that in those without (objective response rate: 18 % vs. 83 %, p < 0.001; median progression-free survival [months]: 3.2 vs. 8.3, p < 0.001). Transcriptome analysis revealed significantly more CD274 splice variants with aberrant 3′-terminal sequences in tumors with the inter-assay discrepancy than in those without. Conclusion The inter-assay discrepancy in the PD-L1 status of tumor cells between the 22C3 and SP142 assays, reflecting an imbalance in the CD274 splice variants, could be a biomarker for primary resistance against pembrolizumab monotherapy in high PD-L1-expressing NSCLCs.