Clinical Interchangeability of 22C3 and 28 - 8 PD-L1 assays for patient selection in esophageal squamous cell carcinoma.

Esophageal squamous cell carcinoma (ESCC) comprises 84% of esophageal cancers, with a poor prognosis due to late diagnosis and limited treatment. Immunotherapy targeting programmed death-1 (PD-1) and programmed death ligand 1 (PD-L1) has improved survival outcomes; however, variability in PD-L1 testing platforms and scoring methods poses challenges. This study evaluated concordance between the PD-L1 immunohistochemistry assays, specifically 22C3 and 28 - 8, in ESCC using tumor proportion score (TPS) and combined positive score (CPS) methodologies to address inconsistencies in testing approaches. PD-L1 expression in 198 ESCC cases was assessed using biopsy or surgical resection specimens. Two standardized PD-L1 assays (Dako 22C3 pharmDx and Dako 28 - 8 pharmDx) were evaluated according to the manufacturer's protocols. Staining was independently scored by two pathologists using TPS and CPS with cutoffs of ≥ 1%, ≥ 10%, and ≥ 50%. Among 198 ESCC cases, PD-L1 expression assessed by the 22C3 and 28 - 8 assays showed comparable positivity rates across TPS and CPS cutoffs. Inter-assay concordance was high, with intraclass correlation coefficients (ICC) of 0.82 and 0.81 for TPS and CPS, respectively. Cohen's kappa and Gwet's AC1 showed substantial agreement across clinical cutoffs. Inter-observer reproducibility was excellent (ICC, 0.95‒0.97) for both assays. Higher T stage was associated with CPS positivity at the CPS ≥ 1 cutoff (p = 0.012 for 22C3; p = 0.045 for 28 - 8). Notable concordance was observed between the 22C3 and 28 - 8 assays, suggesting their potential interchangeability in selecting patients with ESCC for PD-1 inhibitor therapy.