An Investigation to Identify Tumor Microenvironment-Related Genes of Prognostic Value in Lung Squamous Cell Carcinoma Based on the Cancer Genome Atlas
Huan Liu,Boxuan Liu,Lei Zhang,Mingzhen Li,Cheng Chen,Shaohua He,Tingting Luo,Xiaohui He,Liming Tan
DOI: https://doi.org/10.21037/tcr-21-401
2021-01-01
Translational Cancer Research
Abstract:Background: Lung squamous cell carcinoma (LUSC) is a prevalent and lethal malignancy with a poor clinical prognosis. Major constituents of the tumor microenvironment (TME) include infiltrating immune cells and stromal cells, which play a pivotal role in the progression and growth of the disease. To improve the understanding of the prognostic influence of immune and stromal cell-related genes for patients with the disease, we performed a comprehensive bioinformatics analysis to identify TME-relevant biomarkers, and investigated the potential role of these candidate signatures in LUSC. Methods: Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data (ESTIMATE) assessed the samples of LUSC obtained from The Cancer Genome Atlas (TCGA). The samples were grouped according to their immune/stromal scores (high or low). Multivariate cox regression and receiver operating characteristic curves (ROC) were implemented to construct the risk assessment model for prognosis prediction. The co-upregulated differentially expressed genes (DEGs) in the immune and stromal groups were used for further analyses. Overall survival (OS) curves were used to determine the prognostic value of the DEGs, and the TME- related DEGs were verified with Gene Expression Omnibus (GEO) datasets. The functional assessments were performed include Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and protein-protein interaction (PPI) analyses. Results: The immune/ stromal scores calculated by ESTIMATE showed significant associations with OS (log-rank P<0.05). In addition, the prognostic risk score model based on immune and stromal scores also showed significant correlations with OS. A total of 94 TME-related genes were obviously related to poor OS. Among them, BHMT2, FES, HSPB7, NOVA2, LPAP2, and SEMA3B (BFHNLS) were confirmed using GSE4573 and GSE17710 datasets. The functional assessments exhibited those TME-related genes mostly participate in immune response, cytokine-cytokine receptor interaction, and metabolic pathways, which elucidated the probable correlation of TME with tumorigenesis in LUSC. Conclusions: In this study, 6 potential biomarkers named BFHNLS were identified as TME-related genes with prognostic value based on immune and stromal scores of LUSC patients of TCGA, and were verified using GEO datasets, which might serve as therapeutic targets.