Abstract
Background: Pancreatic cancer is a malignancy with a high mortality rate and worse prognosis. Recently, public databases and bioinformatics tools make it easy to develop the prognostic risk model of pancreatic cancer, but the aging-related risk signature has not been reported. The present study aimed to identify an aging-related risk signature with potential prognostic value for pancreatic cancer patients. Method: Gene expression profiling and human clinical information of pancreatic cancer were derived from The Cancer Genome Atlas database (TCGA). Aging-related gene sets were downloaded from The Molecular Signatures Database and aging-related genes were obtained from the Human Ageing Genomic Resources database. Firstly, Gene set enrichment analysis was carried out to investigate the role of aging process in pancreatic cancer. Secondly, differentially expressed genes and aging-related prognostic genes were screened on the basis of the overall survival information. Then, univariate COX and LASSO analysis were performed to establish an aging-related risk signature of pancreatic cancer patients. To facilitate clinical application, a nomogram was established to predict the survival rates of PCa patients. The correlations of risk score with clinical features and immune status were evaluated. Finally, potential therapeutic drugs were screened based on the connectivity map (Cmap) database and verified by molecular docking. For further validation, the protein levels of aging-related genes in normal and tumor tissues were detected in the Human Protein Atlas (HPA) database. Result: The genes of pancreatic cancer were markedly enriched in several aging-associated signaling pathways. We identified 14 key aging-related genes related to prognosis from 9,020 differentially expressed genes and establish an aging-related risk signature. This risk model indicated a strong prognostic capability both in the training set of TCGA cohort and the validation set of PACA-CA cohort and GSE62452 cohort. A nomogram combining risk score and clinical variables was built, and calibration curve and Decision curve analysis (DCA) have proved that it has a good predictive value. Additionally, the risk score was tightly linked with tumor immune microenvironment, immune checkpoints and proinflammatory factors. Moreover, a candidate drug, BRD-A47144777, was screened and verified by molecular docking, indicating this drug has the potential to treat PCa. The protein expression levels of GSK3B, SERPINE1, TOP2A, FEN1 and HIC1 were consistent with our predicted results. Conclusion: In conclusion, we identified an aging-related signature and nomogram with high prediction performance of survival and immune cell infiltration for pancreatic cancer. This signature might potentially help in providing personalized immunotherapy for patients with pancreatic cancer.