The pattern of expression and prognostic value of key regulators for m7G RNA methylation in hepatocellular carcinoma

Abstract

N7-methylguanosine (m⁷G) modification on internal RNA positions plays a vital role in several biological processes. Recent research shows m⁷G modification is associated with multiple cancers. However, in hepatocellular carcinoma (HCC), its implications remain to be determined. In this place, we need to interrogate the mRNA patterns for 29 key regulators of m⁷G RNA modification and assess their prognostic value in HCC. Initial, the details from The Cancer Genome Atlas (TCGA) database concerning transcribed gene data and clinical information of HCC patients were inspected systematically. Second, according to the mRNA profiles of 29 m⁷G RNA methylation regulators, two clusters (named 1 and 2, respectively) were identified by consensus clustering. Furthermore, robust risk signature for seven m⁷G RNA modification regulators was constructed. Last, we used the Gene Expression Omnibus (GEO) dataset to validate the prognostic associations of the seven-gene risk signature. We figured out that 24/29 key regulators of m⁷G RNA modification varied remarkably in their grades of expression between the HCC and the adjacent tumor control tissues. Cluster one compared with cluster two had a substandard prognosis and was also positively correlated with T classification (T), pathological stage, and vital status (fustat) significantly. Consensus clustering results suggested the expression pattern of m⁷G RNA modification regulators was correlated with the malignancy of HCC strongly. In addition, cluster one was extensively enriched in metabolic-related pathways. Seven optimal genes (METTL1, WDR4, NSUN2, EIF4E, EIF4E2, NCBP1, and NCBP2) were selected to establish the risk model for HCC. Indicating by further analyses and validation, the prognostic model has fine anticipating command and this probability signature might be a self supporting presage factor for HCC. Finally, a new prognostic nomogram based on age, gender, pathological stage, histological grade, and prospects were established to forecast the prognosis of HCC patients accurately. In essence, we detected association of HCC severity and expression levels of m⁷G RNA modification regulators, and developed a risk score model for predicting prognosis of HCC patients' progression.

Keywords: bioinformatics; hepatocellular carcinoma; m7G; prognosis; risk signature.

FIGURE 1

The levels of m⁷G modification regulators in hepatocellular carcinoma (HCC). (A)The heatmap shows the mRNA level of m⁷G RNA modification regulators in each sample. “N” represents normal sample, and “T” represents cancer sample. Green stands for low expressions, and red stands for high expressions. (B) The vioplot shows the differential regulators of m⁷G RNA modification in HCC. Blue stands for normal samples, and red stands for HCC samples. White spots indicate the median value of the expression. (C) Spearman correlation analyses of 29 m⁷G RNA modification regulators in HCC. ***p < 0.001.

FIGURE 2

Consistent clustering analyses of the HCC. (A) The correlations between subgroups when the number of clusters is k = 2. (B) Cumulative distribution function (CDF) for k = 2–9 is displayed. (C) The relative variation of the area under the CDF curve of k = 2–9. (D) Principal component analysis of the RNA-seq data. Red dots stand for cluster 1, and cyan dots stand for cluster 2.

FIGURE 3

Differences between cluster one and cluster two in clinicopathological features and overall survival. (A) Heatmap and clinicopathological features of the two clusters. Green stands for low expressions, and red stands for high expressions. (B) Compare the overall survival (OS) distribution of cluster one and cluster 2. *p < 0.05, ***p < 0.001.

FIGURE 4

Analyses of the differentially expressed genes between the two clusters by the Kyoto Encyclopedia of Genes and Genomes (KEEG) and Gene Ontology (GO). The highly expressed genes in cluster one were functionally annotated applying the GO terms (A,B) and KEGG pathway (C,D).

FIGURE 5

The prognostic risk model established on basis of m⁷G RNA modification regulator genes. (A) Univariate Cox regression analyses of m⁷G RNA methylation regulators. (B–D) The process of constructing signatures applying least absolute shrinkage and selection operator (LASSO) Cox regression. (E) Distribution of risk scores of the patients in the prognostic model. (F) Distributions of survival status of the patients in the prognostic model.

FIGURE 6

Kaplan–Meier survival analysis of the prognostic model. Invalids in both datasets were divided into high-risk (red) and low-risk (blue) groups, applying the median risk score as the threshold. (A,B) In the TCGA cohort, the survival probability of the low-risk group was higher than that of the high-risk group (p < 0.001). The AUCs at 1-, 3-, and 5-years were 0.788, 0.628, and 0.634, respectively. (C,D) This prognostic model was verified in the GEO cohort. The survival probability of low-risk group was higher than that of high-risk group (p = 0.0002). The AUCs at 1 year, 2 years, and 3 years were 0.735, 0.699, and 0.73, respectively.

FIGURE 7

Effects of the risk score and the clinicopathological characteristics on the prognosis of invalids with HCC. (A) Heatmap manifests the distribution of clinicopathological characteristics and the expressions of seven m7G RNA modification regulators in high- and low-risk groups. (B) Univariate Cox regression analyses of the clinicopathological parameters and OS. (C) Multivariate Cox regression analyses of the clinicopathological parameters and OS. **p < 0.01, ***p < 0.001.

FIGURE 8

Prognostic nomogram established via the combination with the clinicopathologic features and risk score.