Characterization of RNA modifications in gastric cancer to identify prognosis‐relevant gene signatures

BACKGROUND: Most human genes have diverse transcript isoforms, which mainly arise from alternative cleavage and polyadenylation (APA) at 3′ ends. N7‐methylguanosine (m(7)G) is also an essential epigenetic modification at the 5′ end. However, the contribution of these two RNA modifications to the development, prognosis, regulation mechanisms, and drug sensitivity of gastric cancer (GC) is unclear. METHODS: The expression data of 2412 patients were extracted from 12 cohorts and the RNA modification patterns of 20 marker genes were systematically identified into phenotypic clusters using the unsupervised clustering approach. Following that, we developed an RNA modification model (RMscore) to quantify each GC patient's RNA modification index. Finally, we examined the correlation between RMscore and clinical features such as survival outcomes, molecular subtypes identified by the Asian Cancer Research Group (ACRG), posttranscriptional regulation, and chemotherapeutic sensitivity in GC. RESULTS: The samples were categorized into two groups on the basis of their RMscore: high and low. The group with a low RMscore had a bad prognosis. Moreover, the low RMscore was associated with KRAS, Hedgehog, EMT, and TGF‐β signaling, whereas a high RMscore was related to abnormal cell cycle signaling pathway activation. The findings also revealed that the RMscore contributes to the regulation of the miRNA‐mRNA network. Drug sensitivity analysis revealed that RMscore is associated with the response to some anticancer drugs. CONCLUSIONS: The RMscore model has the potential to be a useful tool for prognosis prediction in patients with GC. A comprehensive investigation of APA‐RNA and m(7)G‐RNA modifications may reveal novel insights into the epigenetics of GC and aid in the development of more effective treatment strategies.