Integrated analysis of multi-omics data to identification of prognostic genes for pancreatic cancer

Abstract Background We aim to develop core modules related to pancreatic cancer (PC) to predict the prognosis of PC patients and explore their tumor microenvironment. Method: We merged 175 pancreatic cancer samples in the TCGA database with gene mutation expression, methylation level distribution, mRNA expression and pancreatic cancer-related genes into a public database, and then through weighted correlation network analysis (WGCNA), Two expression modules associated with pancreatic cancer are combined. Then, by integrating these selected genes identified from the first 10 genes of the two co-expression modules, a model risk score is established, and patients are divided into high-risk and low-risk subgroups. Kaplan-Meier survival analysis method is used to analyze differences, analyze the correlation of survival between subgroups, and analyze prognostic models. These selected core genes can divide early pancreatic cancer into two subgroups, compare the prognosis of these two groups, and screen for differentially expressed genes. Use GO and KEGG enrichment analysis to predict the function of differentially expressed genes. The differential expression level and immune cell infiltration level of these selected core genes were further analysis. Results Our findings shown nine core genes (MST1R, TMPRSS4, PTK6, KLF5, CGN, ABHD17C, MUC1, CAPN8, B3GNT3) were prognostic biomarkers of pancreatic cancer. These 9 genes could divide early pancreatic cancer into two subgroups, and the two subgroups had significant differences in prognosis, and were mainly different in functions such as digestion and extracellular cell adhesion. Further analysis revealed that the expression of these 9 genes were expressed at high levels in pancreatic cancer tissues. In addition, we validated pancreatic cancer cells and pancreatic epithelial cells through quantitative real-time PCR (qRT-PCR), suggesting that the MST1R, PTK6, ABHD17C and CGN expressed higher in PC cells. CIBERSORT analysis indicated that these genes expression were closely correlated with B-cell naive, CD8+ T cells, Macrophages M0 cells, suggesting that these genes may play a carcinogenic role in the preservation of immune-dominant status for tumor microenvironment. Conclusions Our research identified 9 key genes which may enhance our understanding of the molecular mechanisms associated with pancreatic cancer.