Abstract CT143: Pooled analysis of PD-L1 expression across 6 tumor types in the nivolumab clinical program

Abstract

Abstract Introduction: Programmed death ligand 1 (PD-L1) is a key biomarker for PD-1 checkpoint inhibitors; PD-L1 evaluation is incorporated across the nivolumab clinical trial program. Understanding relative expression of PD-L1 for a specific tumor type is important in evaluating its predictive and prognostic value. In this analysis, we present prevalence data from 13 050 patients across 6 tumor types from 23 nivolumab clinical trials. Methods: PD-L1 expression on tumor cells was evaluated with the Dako PD-L1 IHC 28-8 pharmDx assay in select clinical studies: melanoma (MEL), non-small cell lung cancer (NSCLC), renal cell carcinoma (RCC), small-cell lung cancer (SCLC), urothelial cancer (UC), and squamous cell carcinoma of the head and neck (SCCHN). All tumor types included previously treated patients; MEL, NSCLC, and RCC trials also included treatment-naive. PD-L1 IHC staining and interpretation was conducted by reference laboratories. Interpretation was based on a scoring guideline developed for each tumor type, consistent with Dako development standards. Cut-off for data collection was November 23, 2016. Results: Across all tumor types, 93% of samples were evaluable for PD-L1 (Table). PD-L1 expression was evaluated in individual tumor types; proportion of samples with PD-L1 expression ≥1% varied by type (9.0% [SCLC] to 62.8% [MEL]). Proportion of samples at other expression levels also varied, with the highest proportion of ≥50% PD-L1 expression in NSCLC (28.4%). Evaluating the 3 largest nivolumab programs (MEL, NSCLC, and RCC), the profile of PD-L1 expression was significantly different between tumor types (P<0.0001). Conclusion: Prevalence of PD-L1 expression ≥1% and expression levels for screened patients vary significantly by tumor type with minor differences between treatment-naive and overall population. PD-L1 prevalence in nivolumab NSCLC trials is consistent with previously published results, including those with other PD-L1 assays (Aggarwal et al, Ann Oncol 2016;27[s6]). Table 1.PD-L1 expression by tumor type in 23 nivolumab clinical trials evaluated for PD-L1 expressionTumor typeMELNSCLCSCCHNUCRCCSCLCTotal, nNumber of trials, n67133323Patients with PD-L1 tests, n40525241309685204571813 050Treatment-naive3424372600109808248Previously treated62815153096859477184802Total in all trials, %Evaluable PD-L1 tests, % (n)96.6 (3915)91.1 (4777)88.7 (273)92.3 (632)96.0 (1964)79.0 (568)93.0Quantifiable PD-L1 in total population analyzed, %Mean, %≥1% expression62.861.956.942.724.49.043.0<1% expression37.238.143.157.375.691.057.1Total in all trials, %Non-evaluable PD-L1 tests, % (n)3.4 (137)8.9 (464)11.3 (35)7.7 (53)4.0 (81)20.9 (150)7.0Indeterminate PD-L1 tests, % (n)6.3 (247)0.1 (6)0000.4 (2)2.1 Citation Format: Gabriel Krigsfeld, James Novotny, Emin Oroudjev, Josette Carnahan, Songlan Zuo, Steven D. Averbuch, Virginia Burns. Pooled analysis of PD-L1 expression across 6 tumor types in the nivolumab clinical program [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr CT143. doi:10.1158/1538-7445.AM2017-CT143