Analyzing and Validating the Role of Genes Related to Glucagon-like Peptide-1 Signaling in the Prognosis of Pancreatic Cancer.

Abstract

Aims: We aimed to develop a reliable prognostic tool related to glucagon-like peptide-1 (GLP-1) for guiding treatment of pancreatic cancer (PC).

Background: The treatment strategies for PC being greatly advanced the prognosis of cancer still remains unfavorable.

Objective: To develop a RiskScore model for evaluating PC prognosis.

Method: The bulk RNA-seq data of PC patients were obtained from the UCSCXena and GEO database, and the GSE156405 cohort was used for single-cell RNA-seq (scRNA- seq) analysis in the "Seurat" package. Firstly, the gene expression and mutation in the PC samples were analyzed to perform differentially expressed genes (DEGs) analysis using the "limma" package. The "survival" package was employed to conduct un/- multivariate Cox regression and Kaplan-Meier (KM) survival analysis. Secondly, a RiskScore model was developed and assessed using the "glmnet" and "timeROC" packages. Next, the CIBERSORT algorithm and the ssGSEA method were applied for immune infiltration analysis and calculation of the immune cell scores, respectively. Finally, pathway enrichment analysis was conducted using gene set enrichment analysis (GSEA).

Results: Most GLP-1 signaling genes were overexpressed in the PC samples with multiple mutation types. LASSO analysis selected 3 GLP-1 genes for the development of a RiskScore model with a high classification accuracy (AUC >0.6). Notably, high-risk patients showed a significantly shorter survival time in both training and validation sets. In addition, as an independent factor, the RiskScore was further used to establish a nomogram model for the survival prediction of PC in clinical practice. The tumor microenvironment (TME) analysis revealed that low-risk patients with more abundant immune and stroma components had higher levels of anti-tumor immune cell infiltration (such as activated B and T cells), while the proliferation pathways (E2F targets, G2M checkpoint) were significantly activated in the high-risk groups. The genes in the RiskScore model may affect the survival of PC patients through modulating the activities of NK cells and macrophages.

Conclusion: We demonstrated that the GLP-1 signaling affected PC development and developed a reliable RiskSocre model for the prognosis assessment in PC. Our findings are expected to improve PC diagnosis and treatment in clinical practice.