Programmed death-ligand 1 (PD-L1) expression is a critical predictive biomarker for immune checkpoint inhibitor therapies in non-small cell lung cancer (NSCLC). Although various antibody clones are used for immunohistochemical (IHC) detection, differences in platforms and scoring algorithms raise concerns regarding analytical agreement at clinically relevant thresholds. This study evaluated the analytical concordance between VENTANA SP142, VENTANA 22C3, and DAKO 22C3 PD-L1 IHC assays in NSCLC, focusing on therapeutic decision-making thresholds.
Non-small cell lung cancer samples from 205 patients were stained using VENTANA SP142, VENTANA 22C3, and DAKO 22C3 assays. Programmed death-ligand 1 expression was assessed in tumor cells using predefined thresholds (< 1%, 1–49%, ≥ 50%). Inter-assay agreement was evaluated using intraclass correlation coefficients (ICC).
Strong analytical agreement (ICC > 0.90) was observed between SP142 and both DAKO and VENTANA 22C3 assays in the < 1% and 1–49% expression groups. Agreement decreased between VENTANA SP142 and DAKO 22C3 at the  ≥ 50% threshold (ICC = 0.704). Conversely, concordance between VENTANA and DAKO 22C3 was high across all thresholds, particularly in the < 1% and ≥ 50% subgroups (ICC = 1.0 and 0.867, respectively).
VENTANA and DAKO 22C3 assays demonstrate high analytical agreement. While SP142 is comparable at low expression levels, its reliability declines at higher thresholds. These findings underscore the necessity of standardized assay selection for accurate immunotherapy stratification in NSCLC.