Ovarian cancer subtypes based on the regulatory genes of RNA modifications: Novel prediction model of prognosis.

Background: Ovarian cancer (OC) is a female reproductive system tumor. RNA modifications play key roles in gene expression regulation. The growing evidence demonstrates that RNA methylation is critical for various biological functions, and that its dysregulation is related to the progression of cancer in human. Method: OC samples were classified into different subtypes (Clusters 1 and 2) based on various RNA-modification regulatory genes (RRGs) in the process of RNA modifications (m1A, m6A, m6Am, m5C, m7G, ac4C, m3C, and Ψ) by nonnegative matrix factorization method (NMF). Based on differently expressed RRGs (DERRGs) between clusters, a pathologically specific RNA-modification regulatory gene signature was constructed with Lasso regression. Kaplan-Meier analysis and receiver operating characteristic (ROC) curves were used to evaluate the prognostic ability of the identified model. The correlations of clinicopathological features, immune subtypes, immune scores, immune cells, and tumor mutation burden (TMB) were also estimated between different NMF clusters and riskscore groups. Results: In this study, 59 RRGs in the process of RNA modifications (m1A, m6A, m6Am, m5C, m7G, ac4C, m3C, and Ψ) were obtained from TCGA database. These RRGs were interactional, and sample clusters based on these regulators were significantly correlated with survival rate, clinical characteristics (involving survival status and pathologic stage), drug sensibility, and immune microenvironment. Furthermore, Lasso regression based on these 21 DERRGs between clusters 1 and 2 constructed a four-DERRG signature (ALYREF, ZC3H13, WTAP, and METTL1). Based on this signature, 307 OC patients were classified into high- and low-risk groups based on median value of riskscores from lasso regression. This identified signature was significantly associated with overall survival, radiation therapy, age, clinical stage, cancer status, and immune cells (involving CD4+ memory resting T cells, plasma cells, and Macrophages M1) of ovarian cancer patients. Further, GSEA revealed that multiple biological behaviors were significantly enriched in different groups. Conclusions: OC patients were classified into two subtypes per these RRGs. This study identified four-DERRG signature (ALYREF, ZC3H13, WTAP, and METTL1) in OC, which was an independent prognostic model for patient stratification, prognostic evaluation, and prediction of response to immunotherapy in ovarian cancer by classifying OC patients into high- and low-risk groups.