Identification of Immune-Desert Phenotypes in Head and Neck Squamous Cell Carcinoma With a Nine-Gene Signature Prediction Model Using Bioinformatics Methods

Abstract

Background: Research into anti-programmed cell death-1 and anti-programmed cell death ligand-1(anti-PD-1/PD-L1) immunotherapy has increased exponentially in the past years. However, immune check inhibitor (ICI) therapies are not effective in all patients. In this study, we explored biomarkers within the TME to identify specific immunophenotypes, and to predict the efficacy of immunotherapy in patients with head and neck squamous cell carcinoma (HNSCC) through bioinformatics analysis. Method: Data on HNSCC patients were derived from TCGA and GEO databases and re-annotated through BLAST analysis. The composition of immune cells in HNSCC was estimated by CIBERSORT and ssGSEA. Weighted correlation network analysis (WGCNA) was used to measure co-expression relationships. The Lasso-logistic model was used to screen biomarkers associated with specific cancer immunophenotypes. A bootstrapping algorithm was used to select optimal cut-off values. Gene function analysis was performed using the Reactome pathway databases. The ESTIMATE algorithm was used to infer the tumor purity. Target drugs which might be used in the immune-desert patient population were screened by a connectivity mapping analysis. Findings: A high degree of similarity in the genes and pathways enriched from the immune-related modules of TCGA HNSCC and GSE65858 datasets was observed. Sixty hub genes were identified through the intersection of two modules to distinguish between immune-desert and non-immune-desert phenotypes. Furthermore, a nine-gene signature prediction model (ALOX5, ARHGDIB, GIMAP1, ITGAL, MAP4K1, RASSF2, SH2D1A, SLAMF6, UCP2) was selected which had a prediction accuracy of 83%, and was related to the efficacy of ICI therapies. Finally, it was found that at least five of these nine genes with negative expression could predict the tumor samples as immune-desert phenotype, and protein kinase C (PKC) activators might be useful drugs for immune-desert patient population in HNSCC. Interpretation: Our nine-gene model may predict the immunophenotype in HNSCC. At least five of these nine genes with negative expression could distinguish immune-desert phenotypes. These findings provide a new strategy to identify the immune state of the TME in HNSCC, and the nine genes may serve as potential biomarkers for clinical applications. The PKC activators might be a potential treatment for immune-desert phenotype tumor. Funding Statement: This work was supported by The National Key Research and Development Program of China (2017YFC1308900), Technological Special Project of Liaoning Province of China (2019020176-JH1/103), Science and Technology Project of Liaoning Province (NO.2013225585), The General Projects of Liaoning Province Colleges and Universities (LFWK201706), National Natural Science Foundation of China (NO.81302023) and China Medical University Youth Backbone Support Plan Project (QGZ2018047). Declaration of Interests: The authors declare no conflict of interests.