Concordance Between the PD-L1 Tumor Area Positivity Score and Combined Positive Score for Gastric or Esophageal Cancers Treated With Tislelizumab.

Abstract

Tumor Area Positivity (TAP) score is an emerging score measuring programmed death-ligand 1 (PD-L1) expression in tumors. However, the availability of concordance data between TAP score and other immunohistochemistry scoring methods is limited. We investigated concordance between TAP score and combined positive score (CPS) and the relationship between PD-L1 status and clinical outcomes using gastric/gastroesophageal junction adenocarcinoma (GC/GEJC) and esophageal squamous cell carcinoma (ESCC) samples from patients subsequently treated with tislelizumab. Baseline tissue samples from RATIONALE-305 (GC/GEJC; NCT03777657), RATIONALE-302 (ESCC; NCT03430843), and RATIONALE-306 (ESCC; NCT03783442) were assessed for PD-L1 expression using an investigational-use only version of the VENTANA PD-L1 (SP263) CDx Assay. PD-L1 status was scored per protocol by TAP score and post-hoc by TAP score and CPS depending on the cutoff used. Concordance and correlation of both scores with clinical and safety outcomes were analyzed. Across all trials, agreement between TAP score and CPS was significant at PD-L1 cutoffs of ≥1%/≥1, ≥5%/≥5, and ≥10%/≥10 (Cohen's Kappa, 0.64-0.85). Similar outcomes for overall survival (OS), progression-free survival, objective response rate, and duration of response were observed between TAP score- and CPS-defined PD-L1-positive subgroups at analogous PD-L1 cutoffs. OS hazard ratios (HRs) in the PD-L1-high subgroups were similar between protocol-defined TAP score and the same numeric CPS value (OS HR [95% confidence interval]: RATIONALE-305, 0.72 [0.59-0.88] and 0.73 [0.60-0.89]; RATIONALE-302, 0.52 [0.35-0.76] and 0.54 [0.37-0.78]; RATIONALE-306, 0.63 [0.45-0.89] and 0.58 [0.42-0.81], respectively). Safety outcomes were generally comparable between all PD-L1 subgroups. In conclusion, TAP score and CPS are clinically comparable immunohistochemistry measures of PD-L1 expression in tislelizumab-treated patients with GC/GEJC or ESCC.