Comparison of PD-L1 immunohistochemical assays in advanced gastric adenocarcinomas using endoscopic biopsy and paired resected specimens

Abstract

Immunohistochemical (IHC) assays for programmed death ligand 1 (PD-L1) expression are crucial for guiding immune checkpoint inhibitor therapies in advanced gastric adenocarcinoma (AGC). The results from clinical trials of various PD-L1 antibody clones are variable and the exchangeability of these assays is a highly sought goal. The aim of this study was to determine whether three different PD-L1 assays (SP263 and 22C3 on the Dako and Ventana platforms) are interchangeable through analysis of their concordance rate within samples between biopsy and paired resected specimens. One hundred pairs of biopsied and resected AGC specimens were collected and stained for PD-L1. The combined positive score (CPS) was used for the IHC analysis and a four tiered system was applied, i.e., <1, 1 to < 5, 5 to 50, and >50. The agreement for the different IHC assays was low across all cut-offs with the biopsied or resected specimens (biopsy, κ=0.17-0.453; resection, κ=0.02-0.311). The overall positive agreement (OPA) for the PD-L1 results from the biopsy and resection tissues was 100% (SP263, κ=1), 86% (22C3 on the Dako platform, κ=0.693) and 93% (22C3 on the Ventana platform, κ=0.82) at the CPS1 cut-off. The low concordances among the three PD-L1 IHC assays indicated that they cannot be used interchangeably in clinical practice. The results of the SP263 assay using CPS1 showed the highest agreement between the biopsy and resection specimens, suggesting SP263 may provide the most representative approach for the evaluation of PD-L1 status in gastric cancer.