Digital pathology to evaluate PD-L1 IHC scoring as a predictor of outcome with second-line avelumab treatment in patients with non-small cell lung cancer (NSCLC).

Abstract

e21539 Background: Assessment of programmed death ligand-1 (PD-L1) protein expression using immunohistochemistry (IHC)-based tests is currently the only approved biomarker guiding treatment of non-small cell lung cancer (NSCLC) with checkpoint inhibitors. Robust scoring guidelines and suitable cut-offs should be defined specifically for each PD-L1 IHC assay and are critical for appropriate treatment decisions. Methods: We retrospectively applied a novel digital pathology (DP) solution that mimics the conventional tumor proportion scoring (TPS) of PD-L1. The exploratory DP solution was developed and validated using samples from 340 patients enrolled in the first- and second-line (1L and 2L) NSCLC cohorts of the avelumab phase 1 JAVELIN Solid Tumor trial (NCT01772004) and 792 patients with NSCLC enrolled in the avelumab phase 3 JAVELIN Lung 200 trial (NCT02395172). Efficacy analyses were conducted for overall survival (OS) and progression-free survival (PFS) using the full analysis set with evaluable imaging data (n = 136 and n = 544, respectively). Results: Comparison of DP and conventional, semiquantitative pathologist scoring resulted in a high correlation overall (Spearman correlation coefficient, 0.86), with comparable performance for prediction of outcome to treatment with avelumab in 2L NSCLC. Consistent with conventional scoring, median OS and median PFS in avelumab-treated patients increased with higher PD-L1 expression cut-offs: in patients with ≥1%, ≥50%, and ≥80% PD-L1 expression on tumors cells, median OS was 10.0 months (95% CI: 8.6-14.3), 13.8 months (95% CI: 9.6-20.4) and 18.5 months (95% CI: 9.6-not estimable), respectively; median PFS, 3.1 months (95% CI: 2.7-4.9), 5.5 months (95% CI: 2.8-8.3) and 5.6 months (95% CI: 2.8-9.9). Conclusions: Our results demonstrate the technical feasibility, robustness, and utility of DP in scoring PD-L1 IHC in clinical trial samples, achieving comparable performance to conventional, semiquantitative pathologist scoring. Furthermore, our study supports the manual pathologist scoring algorithm (TPS scoring) in NSCLC and the selection of higher cut-offs for the PD-L1 IHC Ab clone 73-10.