Development and validation of a metastasis-related gene signature for predicting the overall survival in patients with pancreatic ductal adenocarcinoma

Abstract

Background: Pancreatic ductal adenocarcinoma (PDAC) is a highly fatal, aggressive cancer characterized by invasiveness and metastasis. In this study, we aimed to propose a gene prediction model based on metastasis-related genes (MTGs) to more accurately predict PDAC prognosis. Methods: Differentially expressed MTGs (DE-MTGs) were identified via integrated analysis of gene expression omnibus (GEO) datasets and Human Cancer Metastasis Database (HCMDB). Overall survival (OS) related DE-MTGs were then identified and a prognostic gene signature was established using Lasso-Cox regression with TCGA-PAAD datasets. Tumor immunity was analyzed using ESTIMATE and CIBERSORT algorithms. Finally, a nomogram predicting 1-year, 2-year, and 3-year OS of PDAC patients was established based on the prognostic gene signature and relevant clinical parameters using a stepwise Cox regression model. Results: A total of 36 DE-MTGs related to OS were identified in PDAC. Consequently, an MTG-based gene signature comprising of RACGAP1, RARRES3, TPX2, MMP28, GPR87, KIF14, and TSPAN7 was established to predict the OS of PDAC. The MTG-based gene signature was able to distinguish high-risk patients with significantly poorer prognosis and accurately predict OS of PDAC in both the training and external validation datasets. Cox regression analysis indicated that the MTG-based gene signature was an independent prognostic factor in PDAC. The gene set enrichment analysis (GSEA) showed that molecular alterations in the high-risk group were associated with multiple oncological pathways. Moreover, analysis of tumor immunity revealed significantly higher levels of follicular helper T cells and M0 macrophage infiltration, and lower levels of infiltrating naïve B cells, CD8 T cells, monocytes, and resting dendritic cells in the high-risk group. Immune cell infiltration levels were significantly associated with the expression of the seven DE-MTGs. Finally, a nomogram was established by incorporating the prognostic gene signature and clinical parameters, which was superior to the AJCC staging system in predicting the OS of PDAC patients. Conclusions: The DE-MTGs we identified were closely associated with the progress and prognosis of PDAC and are potential therapeutic targets. The MTG-based gene signature and nomogram may serve to improve the individualized prediction of survival, assisting in clinical decision-making.