Consistency analysis of PD-L1 immunohistochemistry antibodies in esophageal squamous cell carcinoma

Background and purpose: Immunohistochemistry (IHC) expression of programmed death ligand-1 (PD-L1) is the most widely used biomarker for predicting the efficacy of esophageal cancer immunotherapy, and accurate and reliable PD-L1 testing is crucial for screening potential beneficiaries of immunotherapy. This study aimed to investigate the expression of PD-L1 in patients with esophageal squamous cell carcinoma (ESCC) and the consistency and correlation of IHC detection of seven different clones of PD-L1 antibody, and to provide reference for the gradual standardization of PD-L1 IHC detection and follow-up clinical research of esophageal cancer. Methods: This study collected surgically resected and pathologically confirmed specimens from 146 ESCC patients in Fudan University Shanghai Cancer Center from January 2020 to December 2021, and specimens were stained with seven antibodies: PD-L1 22C3, SP263, 28-8, E1L3N XP, CST E1L3N, BP6099 and MXR006, respectively. The consistency of the antibodies was statistically analyzed in different cut-off values using three scoring criteria: comprehensive positive score (CPS), tumor cell proportional score (TPS) and immune cell score (IPS). Results: The results of PD-L1 22C3, SP263 and 28-8 antibodies demonstrated that the consistency between the three antibodies was substantial (Kappa: 0.66-0.80) when CPS cut-off was 10, and the consistency of PD-L1 28-8 with 22C3 and SP263 antibodies was substantial (Kappa=0.80, 0.65) whereas the consistency between PD-L1 22C3 and SP263 was moderate (Kappa=0.49) when CPS cut-off was 1. The three antibodies showed a high consistency when TPS cut-off was 10% and 1% (Kappa: 0.79-0.89, 0.71-0.91). The companion diagnostic PD-L1 22C3 clone antibody and the four domestic clone antibodies E1L3N XP, CST E1L3N, BP6099 and MXR006 consistency study showed that the consistency between the five antibodies was substantial (Kappa: 0.67-0.88, 0.65-0.79) when CPS cut-off was 10 and TPS cut-off was 10%, and the consistency between the five antibodies was substantial or moderate when CPS cut-off was 1 (Kappa: 0.58-0.88). When TPS cut-off was 1%, there was an excellent consistency between the five antibodies (Kappa: 0.85-0.92). Using IPS cut-off of 1% or 10%, the five antibodies showed moderate or poor consistency (Kappa: 0.30-0.62). Conclusion: PD-L1 22C3, SP263 and 28-8 antibodies showed high consistency. PD-L1 22C3 and four domestic antibodies PD-L1 E1L3N XP, CST E1L3N, BP6099 and MXR006 antibodies showed moderate consistency in CPS and TPS, while the consistency was poor in IPS. These antibodies are interchangeable with each other, providing more selectivity for clinical testing needs.