A bioinformatics study of autophagy biomarkers associated with the prognosis of endometrial carcinoma
Abstract Background: Autophagy plays a critical role in endometrial carcinoma (EC), but prognosis studies of differentially expressed autophagy-related genes (DEARGs) in EC are lacking. This study aimed to access the prognostic value of autophagy-related genes (ARGs) in EC and to identify the potential characteristics of ARGs in predicting patient survival and guiding treatment.Methods: The RNA-Seq data and clinical data were obtained from TCGA databases. The DEARGs were identified by “limma” package in R software. Clusterprofler was used for functional analysis. Using STRING databases to construct a protein-protein interaction (PPI) network and plug-in MCODE to screen hub modules in Cytoscape. The degrees method of cytoHubba was used to select important hub genes. Co-expression analysis was performed using “limma” package and Metascape for functional analysis. Univariate and multivariate Cox proportional hazards regression analyses were performed to construct a prognostic signature. Kaplan–Meier curve analysis, ROC curve analysis and Gene set enrichment analysis (GSEA) were also performed. Validation was executed by UALCAN, CCLE and HPA databases.Results: In total, 45 DEARGs were identified. Functional analysis showed that DEARGs were strikingly enriched in autophagy pathway. PPI network showed that CASP3 was the most central protein. CASP3 co-expression analysis revealed that co-expressed genes were mostly enriched in cell cycle. We identified a novel prognostic signature consisting of 3 genes (CDKN2A, PTK6 and GRID2). The risk score based on the prognostic signature was able to classify patients into high-risk and low-risk groups with significantly different overall survival. Furthermore, the prognostic signature is an independent prognostic predictor of survival and demonstrates superior prognostic performance compared with the clinicopathologic features for predicting 5-year survival. CDKN2A and PTK6 were further validated in UALCAN. GSEA suggested that the two genes played crucial roles in drug metabolism.Conclusion: Using bioinformatics analyses, we identified DEARGs and determined CDKN2A, PTK6 and DRID2 are tumor-associated genes and can be utilized as possible biomarkers in EC treatment.