Abstract
OBJECTIVES: To identify the key glycolysis-related genes (GRGs) in the occurrence and development of pancreatic ductal carcinoma (PDAC), and to construct a glycolysis-related gene model for predicting the prognosis of PDAC patients. METHODOLOGY: Pancreatic ductal carcinoma (PDAC) data and that of normal individuals were downloaded from the TCGA database and Genotype-Tissue Expression database, respectively. GSEA analysis of glycolysis-related pathways was then performed on PDAC data to identify significantly enriched GRGs. The genes were combined with other patient's clinical information and used to construct a glycolysis-related gene model using cox regression analysis. The model was further evaluated using data from the validation group. Mutations in the model genes were subsequently identified using the cBioPortal. In the same line, the expression levels of glycolysis related model genes in PDAC were analyzed and verified using immunohistochemical images. Model prediction for PDAC patients with different clinical characteristics was then done and the relationship between gene expression level, clinical stage and prognosis further discussed. Finally, a nomogram map of the predictive model was constructed to evaluate the prognosis of patients with PDAC. RESULTS: GSEA results of the training set revealed that genes in the training set were significantly related to glycolysis pathway and iconic glycolysis pathway. There were 108 differentially expressed GRGs. Among them, 29 GRGs were closely related to prognosis based on clinical survival time. Risk regression analysis further revealed that there were seven significantly expressed glycolysis related genes. The genes were subsequently used to construct a predictive model. The model had an AUC value of more than 0.85. It was also significantly correlated with survival time. Further expression analysis revealed that CDK1, DSC2, ERO1A, MET, PYGL, and SLC35A3 were highly expressed in PDAC and CHST12 was highly expressed in normal pancreatic tissues. These results were confirmed using immunohistochemistry images of normal and diseases cells. The model could effectively evaluate the prognosis of PDAC patients with different clinical characteristics. CONCLUSION: The constructed glycolysis-related gene model effectively predicts the occurrence and development of PDAC. As such, it can be used as a prognostic marker to diagnose patients with PDAC.