Comprehensive immunogenomic landscape analysis of prognosis-related genes in head and neck cancer

Abstract

AbstractHead and neck cancer is the sixth most common malignancy around the world, and 90% of cases are squamous cell carcinomas. In this study, we performed a systematic investigation of the immunogenomic landscape to identify prognostic biomarkers for head and neck squamous cell carcinoma (HNSCC). We analyzed the expression profiles of immune‐related genes (IRGs) and clinical characteristics by interrogating RNA-seq data from 527 HNSCC patients in the cancer genome atlas (TCGA) dataset, including 41 HPV+ and 486 HPV− samples. We found that differentially expressed immune genes were closely associated with patient prognosis in HNSCC by comparing the differences in gene expression between cancer and normal samples and performing survival analysis. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were performed to annotate the biological functions of the differentially expressed immunogenomic prognosis-related genes. Two additional cohorts from the Oncomine database were used for validation. 65, 56 differentially expressed IRGs was associated with clinical prognosis in total and HPV- samples, respectively. Furthermore, we extracted 10, 11 prognosis-related IRGs from 65, 56 differentially expressed IRGs, respectively. They were significantly correlated with clinical prognosis and used to construct the prognosis prediction models. The multivariable ROC curves (specifically, the AUC) were used to measure the accuracy of the prognostic models. These genes were mainly enriched in several gene ontology (GO) terms related to immunocyte migration and receptor and ligand activity. KEGG pathway analysis revealed enrichment of pathways related to cytokine−cytokine receptor interactions, which are primarily involved in biological processes. In addition, we identified 63 differentially expressed transcription factors (TFs) from 4784 differentially expressed genes, and 16 edges involving 18 nodes were formed in the regulatory network between differentially expressed TFs and the high-risk survival-associated IRGs. B cell and CD4 T cell infiltration levels were significantly negatively correlated with the expression of prognosis-related immune genes regardless of HPV status. In conclusion, this comprehensive analysis identified the prognostic IRGs as potential biomarkers, and the model generated in this study may enable an accurate prediction of survival.