Abstract 593: Cross-platform integrative analysis of NSCLC reveals association between PD-L1 expression patterns and tumor genetic profile

Abstract

Abstract Background: Immunotherapy with PD-1 inhibitors is associated with increased survival in NSCLC. While PD-L1 expression enriches for clinical benefit, additional predictive biomarkers are critically needed to further define patient subsets associated with specific response/resistance profiles and to develop optimal combination strategies. We hypothesized that integration of pathology features derived from H&E and PD-L1 immunohistochemistry with genetic/genomic data from NSCLC could identify phenotypic/genotypic associations that highlight specific immunosuppressive mechanisms. Towards that end, we present data from an initial pilot analysis. Methods: Twenty-nine NSCLC (23 adenocarcinoma; 6 squamous cell carcinoma) were included. H&E-stained sections were scored for tumor grade (1-3), relative proportion of tumor stroma (stroma score 1-3), percent necrosis and intensity of chronic inflammatory infiltrate (1-3+). PD-L1 immunohistochemistry was performed using the DAKO 28-8 antibody. PD-L1 expression was scored in tumor cells by modified H-score and by the predominant pattern observed - diffuse, heterogeneous, tumor-stroma interface or negative, and in immune cells using a semi-quantitative intensity scale (1-3+). DNA and RNA extracts from FFPE tissue were subjected to whole exome sequencing and RNAseq from which mutation load, oncogene/tumor suppressor genotypes and inflammation signatures were derived. Results: The highest PD-L1 H-scores were associated with diffuse expression patterns and were observed in high grade tumors (Grade 3). Deleterious mutations in TP53, STK11, KEAP1, KRAS, EGFR and MET occurred at expected frequencies. These mutations occurred across multiple PD-L1 expression patterns except for STK11 which was restricted to PD-L1 negative tumors (p=0.077, Fisher exact test). STK11-mutant tumors also displayed a lower PD-L1+ inflammation score (p=0.046, Fisher exact test). Trends toward an increase in mutation load with increasing levels of chronic inflammation on H&E stain as well as with PD-L1+ chronic immune infiltrates were noted. Inflammation signatures derived from RNAseq showed an association between diffuse PD-L1 expression by IHC and the highest levels of inflammation mRNA signatures comprising T-cells (CD8, Tregs), B-cells, macrophages and MDSCs. Conclusions: The potential association between tumor grade, PD-L1 expression, intensity of the immune infiltrate and mutation load raise the possibility that tumor morphology predicts mutation load and associated immune response. The finding that STK11 mutation is restricted to PD-L1 negative tumors suggests an immunosuppressive mechanism is invoked in this setting. This dataset is currently being expanded to establish the significance of these findings. Citation Format: Robin Edwards, Patrik Vitazka, Peter Szabo, Han Chang, John Cogswell, Hao Tang, Kaushal Desai, Darren Locke, Qiuyan Wu, Joseph Szustakowski, Steven Bernstein, Michele Cleary. Cross-platform integrative analysis of NSCLC reveals association between PD-L1 expression patterns and tumor genetic profile [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 593. doi:10.1158/1538-7445.AM2017-593