Differential analysis of RNA methylation regulators in gastric cancer based on TCGA data set and construction of a prognostic model.

Background Methylation is one of the common forms of RNA modification, which mainly include N6-methyladenosine (m6A), C5-methylcytidine (m5C), and N1-methyladenosine (m1A). Numerous studies have shown that RNA methylation is associated with tumor development. We aim to construct prognostic models of gastric cancer based on RNA methylation regulators. Methods The transcriptome and clinical data of gastric cancer and normal samples were obtained from the National Cancer Institute Genome Data Commons (NCI-GDC). Use Least Absolute Shrinkage and Selection Operator (LASSO) Cox regression analysis to construct risk models for different types of RNA methylation. Receiver operating characteristic (ROC) curves were generated to evaluate the predictive efficiency of risk characteristics. Cluster heat maps are used to assess the correlation with clinical information. Univariate and multivariate Cox analyses were used to analyze prognostic effects of risk scores. Gene Set Enrichment Analysis (GSEA) analyzes the functional enrichment of RNA methylation genes. And make a separate analysis of the data of Asians. Results The expression of most of the 30 RNA methylation regulators were significantly different in cancer and paracancerous tissues (P<0.05). Three methylated genes (FTO, ALKBH5, and RBM15) were screened from m6A by LASSO Cox regression analysis. Five methylated genes (FTO, ALKBH5, TRMT61B, RBM15, and YXB1) were selected from the population, and were used to construct two risk ratio models. Survival analysis showed that the survival rate of patients in the low-risk group was significantly higher than that in the high-risk group (P<0.05). All ROC curves indicated that the predictive efficiency of risk characteristics was good [area under the ROC curve (AUC): 0.6-1].Cluster analysis reveals differences in clinical data between the two groups. Univariate and multivariate Cox regression results show that the risk score has independent prognostic value. GSEA showed that pathways such as cell cycle were significantly enriched in the low-risk group, while pathways such as calcium signaling pathway were significantly enriched in the high-risk group. In addition, three methylation models that can predict the prognosis of Asian gastric cancer patients were obtained. Conclusions The methylation prognosis model constructed in this study can effectively predict the prognosis of gastric cancer patients.