Exploring the potential of combination immune checkpoint strategies in non-small cell lung cancer (NSCLC).

Abstract

9037 Background: The clinical outcomes for patients with NSCLC remain suboptimal despite recent advancements in anti-PD-1/PD-L1 immunotherapies. Thus, identification of rational combinations of novel immune checkpoint (IC) strategies is imperative. This study aimed to comprehensively characterize the expression of IC targets to inform drug development and design effective combinatorial strategies, to advance the search for optimized therapies for NSCLC. Methods: We analyzed the differential gene expression of IC targets in 537 primary lung adenocarcinoma (LUAD) samples and 59 normal lung tissue samples pre-immunotherapy receipt using RNA-sequencing data from The Cancer Genome Atlas. Gene expression of IC-targets including PDCD1 (PD-1), CD274 (PD-L1), CTLA4, TIGIT, LAG3, HAVCR2 (TIM-3), VSIR (VISTA), ICOS, CD276 (B7-H3), BTLA, IDO1, KLRD1 (CD94/NKG2A), NT5E (CD73), TNFRSF4 (OX-40), TNFRSF18 (GITR), and TNFRSF9 (4-1BB) were obtained. Differential gene expression was determined intra-tumorally by comparison of transcripts per million and inter-tumorally using the DESeq2 algorithm. Log2 fold change >0.5 and false discovery rate adjusted P<0.05 were used as criteria for significant differential expression. Hierarchical clustering was used to identify subsets of tumors with similar profiles of IC-targetable gene expression. Results: Tumors showed significantly increased expression of TNFRSF18 (GITR), NT5E (CD73), TNFRSF9 (4-1BB), LAG3, TIGIT, CTLA4, PD-1, BTLA, B7-H3, and OX-40 (Table 1). On the other hand, tumors exhibited significant downregulation of CD274 (PD-L1), HAVCR2 (TIM-3), ICOS, KLRD1 (CD94/NKG2A), and VSIR (VISTA). Hierarchical clustering revealed recurrent patterns of IC gene overexpression which were characterized by: 1) IDO1-hi, 2) CD276-mid + TNFRSF18-mid, 3) VSIR-hi, 4) CD276-hi, 5) NT5E-hi, and 6) CD276-mid + NT5E-mid. Interestingly, patterns of IC gene expression distinguish tumors from controls. Of note, high VSIR expression was specific to normal tissue rather than tumors. Conclusions: Our study identified many IC targetable genes that were significantly overexpressed in tumors. However, subsets of tumors showed distinct and segregable combinations of overexpressed IC genes. These results suggest heterogeneous mechanisms of immune evasion in NSCLC, which may explain suboptimal outcomes with anti-PD-1/PD-L1 based strategies in this patient population. Our findings emphasize the importance of biomarker-guided approaches for the optimal selection of ICs and the development of rational combinatorial immunotherapies in NSCLC.