Construction of a prognostic risk model for Stomach adenocarcinoma based on endoplasmic reticulum stress genes.

OBJECTIVE

Stomach adenocarcinoma (STAD) is caused by malignant transformation of gastric glandular cells and is characterized by a high incidence rate and a poor prognosis. This study was designed to establish a prognostic risk model for STAD according to endoplasmic reticulum (ER) stress feature genes as cancer cells are susceptible to ER stress.

METHODS

The TCGA-STAD dataset was downloaded to screen differentially expressed genes (DEGs). By intersecting DEGs with ER stress genes retrieved from GeneCards, ER stress-related DEGs in STAD were obtained. Kmeans cluster analysis of STAD subtypes and Single sample gene set enrichment analysis (ssGSEA) analysis of immune infiltration were performed. Cox regression analysis was utilized to construct a risk prognostic model. Samples were split into high-risk and low-risk groups according to the median risk score. Survival analysis and Receiver Operating Characteristic (ROC) curves were conducted to assess the validity of the model. Gene set enrichment analysis (GSEA) was performed to investigate differential pathways in the two risk groups. Cox analysis was performed to verify the independence of the risk model, and a nomogram was generated.

RESULTS

A total of 162 ER stress-related DEGs in STAD were identified by bioinformatics analysis. Kmeans cluster analysis showed that STAD was divided into 3 subgroups. The ssGSEA showed that the levels of immune infiltration in subgroups 2 and 3 were significantly higher than subgroup 1. With 12 prognostic genes (MATN3, ATP2A1, NOX4, AQP11, HP, CAV1, STARD3, FKBP10, EGF, F2, SERPINE1, CNGA3) selected from ER stress-related DEGs using Cox regression analysis, we then constructed a prognostic model. Kaplan-Meier (K‑M) survival curves and ROC curves showed good prediction performance of the model. Significant enrichment of genes in the high-risk group was found in extracellular matrix (ECM) receptor interaction. Cox regression analysis combined with clinical factors showed that the risk model could be used as an independent prognostic factor. The prediction correction curve showed that the good prediction ability of the nomogram.

CONCLUSION

The STAD could be divided into three subgroups, and the 12-gene model constructed by ER stress signatures had a good prognostic performance for STAD patients.