Concordance among three programmed death-ligand 1 (PD-L1) scoring methods and their association with clinical outcomes of tislelizumab (TIS) monotherapy in esophageal squamous cell carcinoma (ESCC).

Abstract

390 Background: Multiple scoring methods and cutoffs have been developed to evaluate tumor PD-L1 expression status in patients with ESCC, and PD-L1 expression level has been associated with the degree of response to anti-programmed cell death protein 1 (PD-1)/PD-L1 therapy. Here, we retrospectively investigated concordance between three PD-L1 scoring methods and their association with clinical outcomes in RATIONALE-302, a phase 3 study of the anti-PD-1 antibody TIS vs investigator-chosen chemotherapy (ICC) as second-line treatment for advanced unresectable/metastatic ESCC (NCT03430843). Methods: Patients enrolled in RATIONALE-302 with evaluable PD-L1 expression by the tumor area positivity (TAP) score (based on visual estimation of positive tumor cells [TCs] and tumor-associated immune cells [ICs]) using the VENTANA PD-L1 (SP263) assay were categorized at a 10% cutoff. Stained slides from those patients were rescored post hoc using both combined positive score (CPS; based on counting positive TCs and ICs) at cutoff 10 and TC (based on counting positive TCs only) score at a 1% cutoff, thresholds currently used in ESCC for anti-PD-(L)1 therapy. Concordance at these thresholds was investigated. Clinical efficacy (overall survival [OS]) for PD-L1 subgroups was assessed. Results: Of 512 pts enrolled, 364 had evaluable TAP scores (TIS, n=180; ICC, n=184), of whom 355 had evaluable post-hoc CPS and TC scores (TIS, n=175; ICC, n=180). TAP score and CPS showed high concordance in terms of overall percentage agreement (OPA; 90% [95% confidence interval (CI): 86, 93]) and Cohen’s Kappa (0.79 [95% CI: 0.72, 0.85]), while TAP and TC scores had lower concordance (OPA: 78% [95% CI: 73, 82]; Cohen’s Kappa: 0.56 [95% CI: 0.47, 0.64]). OS benefit with TIS vs ICC in PD-L1 subgroups defined by TAP, CPS, and TC score cutoffs were generally similar (Table). Conclusions: OS subgroup analysis showed comparable treatment effect by TAP score at 10% cutoff, CPS at cutoff 10, and TC score at 1% cutoff based on SP263 staining. TAP score and CPS at these cutoffs exhibited substantial concordance. The results indicate that the less time-consuming, visually estimated TAP score and CPS may be interchangeable for clinical measurement of PD-L1 expression in patients with ESCC. Clinical trial information: NCT03430843 . [Table: see text]