Bioinformatic Analysis Revealing Independent Predictors of Poor Prognosis and Immunologic/Metastatic Behaviors in Squamous Cell Lung Cancer

Abstract

Abstract Background Squamous cell carcinoma of the lung (LUSC) is a malignant tumor with a high mortality rate and few treatment options. Methods This study explored novel biomarkers and the related mechanisms in LUSC development via integrated bioinformatics analysis based on multiple databases. TCGA-LUSC, GSE30219, and GSE37745 datasets were selected, and univariate Cox analysis was applied to each. The candidate prognostic genes were those survival hazardous genes that overlapped in the three datasets and were further applied in Kaplan-Meier log-rank test. Receiver-operating characteristics (ROC) curves were made to compare the candidate prognostic genes’ predictive accuracy and traditional clinicopathological parameters (CPPs). Moreover, fast preranked gene set enrichment analysis, single-sample gene set enrichment analysis, gene set variation analysis, and correlation analysis was sequentially performed to explore the candidate prognostic genes’ potential functions. Results Based on the data for stage I–III LUSC, aldehyde dehydrogenase 7 family member A1 (ALDH7A1), progestogen-associated endometrial protein (PAEP), and vav guanine nucleotide exchange factor 2 (VAV2) were identified as prognostic candidate biomarkers that could improve the accuracy of traditional CPPs for predicting overall survival (OS). Moreover, ALDH7A1, PAEP, and VAV2 were, respectively, correlated with the immune microenvironment and epithelial–mesenchymal transition (EMT), but not with tumor mutation burden, microsatellite instability, or CPPs. The EMT biomarker vimentin (but not E-cadherin) was consistent with ALDH7A1, PAEP, and VAV2. Conclusion This study elucidated the prognostic roles and potential functions of ALDH7A1, PAEP, and VAV2 in LUSC.