Comprehensive analyses of m6A regulators and interactive coding and non-coding RNAs across 32 cancer types

Abstract

N6-Methyladenosine (m6A) is an RNA modification that interacts with numerous coding and non-coding RNAs and plays important roles in the development of cancers. Nonetheless, the clinical impacts of m6A interactive genes on these cancers largely remain unclear since most studies focus only on a single cancer type. We comprehensively evaluated m6A modification patterns, including 23 m6A regulators and 83 interactive coding and non-coding RNAs among 9,804 pan-cancer samples. We used clustering analysis to identify m6A subtypes and constructed the m6A signature based on an unsupervised approach. We used the signatures to identify potential m6A modification targets across the genome. The prognostic value of one target was further validated in 3,444 samples from six external datasets. We developed three distinct m6A modification subtypes with different tumor microenvironment cell infiltration degrees: immunological, intermediate, and tumor proliferative. They were significantly associated with overall survival in 24 of 27 cancer types. Our constructed individual-level m6A signature was associated with survival, tumor mutation burden, and classical pathways. With the signature, we identified 114 novel genes as potential m6A targets. The gene shared most commonly between cancer types, BCL9L, is an oncogene and interacts with m6A patterns in the Wnt signaling pathway. In conclusion, m6A regulators and their interactive genes impact the outcome of various cancers. Evaluating the m6A subtype and the signature of individual tumors may inform the design of adjuvant treatments.

Keywords: Multi-omics; N6-Methyladenosine; Pan-cancer; Survival outcome.

Fig. 1

a Study workflow. b Circos plot of the selected genes on the chromosome. c Gene co-expression network of the m6A-related genes. The gene pairs with Pearson r > 0.3 are considered to have co-expression correlation. d Discrimination analyses of the tumor and adjacent normal tissues in pan-cancer (cancer types with ≥5 tumor-normal pairs included) based on the m6A gene panel. In the principal components plot of the m6A gene panel in lung cancer tumor/normal tissues, the area under the curve (AUC) of distinguishing between tumor and normal tissues is 0.96 (95% CI: 0.93–0.98). e Heatmap of the somatic mutation frequency of the genes across pan-cancer

Fig. 2

a-c Associations of the m6A subtypes and pan-cancer overall survival (OS), Progression-free interval (PFI) and disease-specific survival (DSS). The hazard ratios (HR) are evaluated by the trend association (clusters 1–3) in the Cox proportional hazard models. d Kaplan-Meier plots of the m6A subtypes and overall survival. e Associations between m6A subtypes and the ssGSEA scores of tumor microenvironment cell infiltration. Cluster 1 is used as a reference group. f Summary of the characteristics of different m6A subtypes. g Correlation of the m6A signature and m6A subtypes. h Distributions of the m6A signature across different clinical stage (I-IV). i Kaplan-Meier plot of the m6A signature and overall survival. We categorized the signature into five subgroups with equal sample sizes and adjusted the curves with age, gender, stage, cancer types and probable estimations of expression residual (PEER) factors. j Correlation plot of the m6A signature and median survival time (MST) of each cancer type. k The top potential m6A targets associated with the m6A signature in ≥12 cancer types. We used the OncoScore system to evaluate their relationships with cancer based on previous literature reports