MA03.07 Interferon-Gamma Mediated Immune Evasion: A Potential Mechanism of Resistance to Immunotherapy in Non-Small Cell Lung Cancer

Abstract

Immunotherapy has unprecedentedly improved survival rates in patients with non-small cell lung cancer. Some patients with high tumor mutation burder, PD-1 and PD-L1 shows limited benefit from immunotherapy. Immune evasion described by immune contexture in tumor microenvironment could be potential mechanism of resistance to immunotherapy in non-small cell lung cancer. Data in our study were accessed from the Gene Expression Omnibus (GSE30219, GSE50081, GPL570) , The Cancer Genome Atlas (TCGA) and Immlnc database. Unsupervised hierarchical clustering was conducted to establish novel classification for NSCLC based on immune-related lncRNAs. Kaplan-Meier survival plots were utilized for survival analysis and overall survival (OS) estimation. Single sample gene set enrichment analysis and gene set variation analysis were performed to explore the landscape of tumor immune microenvironment. Gene instability and potential therapeutic effects were analyzed among three cohorts. The Tumor Immune Dysfunction and Exclusion (TIDE) algorithm mapping was applied to predict the therapeutic response to ICIs. Comparison of single nucleotide variation (SNV) and tumor mutational burden (TMB) among three identified clusters were performed through Kruskal-Wallis test. We found that cluster A had the best prognosis by Kaplan-Meier analysis, followed by cluster B and cluster C significantly (p=0.0048). With a favorable prognosis, cluster A was marked by distinct infiltration of activated B cell, activated dendritic cell, activated CD8 T cell, central memory CD4 T cell, presented as immune-inflammed phenotype. However, in cluster B, the infiltration density of plasma cells, M1 and M2 macrophages, CD8 T cells, gamma delta T cells, myeloid-derived suppressor cells and memory CD4 T cells infiltration was significantly high as well, corresponding to immune-evaded phenotype. Cluster C had the lowest existence of immune infiltrating lymphocytes, correlated with immune-desert phenotype. Tumor mutation burden in cluster B was highest (p < 0.001), while lowest in the cluster A. We subsequently investigated the expression of crucial immune checkpoints among three clusters, such as PD-1, CTLA-4, CD28, and TNFRSF14, that higher expression of immune checkpoint molecules was observed highest in cluster A, followed with cluster B (p < 0.001).With respect to predicted response to immunotherapy, cluster B had the higher tumor immune dysfunction and exclusion score and lower predicted responders than cluster A (p<0.001). Cluster B had high mutation of KEAP1, LRRC7, PAPPA2, ABCA13, APOB, MUC17, NRXN1, DNAH9, SORCS1 and ZNF804A. The hub genes analyzed by Cytoscape software and String database in cluster B were STAT1, PARP9, TRIM21, OAS3. To explore potential mechanisms, activation of interferon−gamma and MHC class I signaling pathways could be detected in cluster B by Gene Ontology functional annotation analysis. In conclusion, we established a novel prognostic classification for NSCLC corresponding with classic immune inflammation, immune evasion and immune desert phenotypes, which describe tumor features in terms of immune contexture, genomic characteristics, and immune checkpoints, targeting the tumor immune microenvironment comprehensively rather than immune checkpoint and TMB alone. Our result suggested interferon-gamma mediated immune evasion could be a potential mechanism of therapeutic resistance to immunotherapy in non-small cell lung cancer.