To Develop and Validate the Combination of RNA Methylation Regulators for the Prognosis of Patients with Gastric Cancer

Abstract

Background: Gastric cancer (GC) accounts for high mortality. RNA methylation has recently gained interest as markers in specific tumors. This study aimed to uncover the function of the roles of 25 RNA methylation regulators in GC.

Methods: RNA sequence and clinical data were downloaded from The Cancer Genome Atlas (TCGA) database. "STRING" and R were performed to analyze the correlation among the methylase. COX and LASSO were performed to screen for prognostic associated RNA methylation regulators. A prognostic model was established based on the expression of methylase. RT-PCR and immunohistochemistry detected the expression of methylase in GC cells and tissue. Kaplan-Meier curve and Cox analysis were applied to evaluate the effectiveness of the model.

Results: The prediction model was established based on the expression of m6A RNA methylation regulators FTO (fat mass and obesity-associated) and RBM15 (RNA binding motif protein 15). Based on the model, GC patients were divided into "high risk" and "low risk" groups to compare the differences in survival. The model was re-evaluated with the clinical data of our center.

Conclusion: The two-methylase combination model was an independent prognostic factor of GC.

Keywords: RNA methylation; gastric cancer; least absolute shrinkage and selection operator; prognosis; survival analysis.

Figure 1

The workflow of the study.

Figure 2

The expression of RNA methylation regulators in GC. (A). Heatmap showed the expression of the RNA methylation regulators in 375 GC and 32 para-carcinoma tissues; *P<0.05, **P<0.01, ***P<0.001. (B). Vioplot visualized the expression of 25 RNA methylation regulators in different tissue samples in GC and para-carcinoma tissues.

Figure 3

Identification of co-expressed gene clusters of RNA methylation regulators. (A). Protein-protein interaction network constructed by STING database. (B). Spearman analyzed the correlation of RNA methylation modified regulator in GC; Spearman correlation analysis of the 13 m6A modification regulators in gastric cancer. (C). RNA methylated genes could be clustered into two consistency matrices. (D). The principal component analysis was performed on the expression profile of total mRNA in the TCGA dataset. (E). The Kaplan–Meier curve was used to analyze the overall survival of the two subgroups. (F). The heatmap showed the correlation between the two subgroups and clinicopathological data.

Figure 4

Risk signature with 25 RNA methylation regulators. (A). Univariate Cox regression calculated the hazard ratios (HR) and 95% confidence intervals (CI) of RNA methylation regulators. (B). LASSO coefficient values of the 25 RNA methylation regulators in the TCGA cohort. (C). L1-penalty of LASSO-COX regression. The dotted vertical lines at optimal log(Lambda) value: 2. (D). Patients were divided into high-risk and low-risk groups based on risk scores, and survival curves were plotted.

Figure 5

Relationship between risk prediction model and clinicopathological features and prognostic value. (A). The ROC curve evaluates the predictive efficiency of the risk prediction model. (B). Heatmap showed the expression of two m6A RNA methylation regulators in GC. The distribution of clinicopathological features was compared between high-risk and low-risk groups. (C). Univariate Cox regression analysis of clinicopathological factors and risk score associated with OS. (D). Univariate Cox regression analysis of clinicopathological factors and risk score associated with OS.

Figure 6

The expression of FTO and RBM15 in GC cells and tissue. (A). FTO was overexpressed in GC cells. (B). RBM15 was weak-expressed in GC cells. (C). FTO was overexpressed in GC tissue. (D). RBM15 was weak-expressed in GC tissue. (E). Typical IHC images showed FTO was overexpressed in GC tissues and weakly expressed in para-cancer groups (a vs b). Typical IHC images showed RBM15 was weak-expressed in GC tissues and overexpressed in para-cancer groups (c vs d). Magnifications are 200×. Data are shown as the mean±SD, n=3. The Student’s t-test assessed the statistical significance of the data. *P<0.05.

Figure 7

Kaplan-Meier DFS and OS curve for GC patients in the high-risk and low-risk group. (A): Disease-free survival curves stratified by risk signature. (B): Overall survival curves stratified by risk signature.