https://orcid.org/0000-0003-4346-473X
Figures & data
Figure 1. The expression of m6A-long noncoding RNAs (lncRNAs) and their role in the prognosis of hepatocellular carcinoma patients
(A): Network plot of correlation among m6A-related gene expression and lncRNAs. (B): Forest plot of univariate Cox regression analysis. Data on prognostic lncRNAs were extracted, and the confidence intervals and hazard ratios were calculated. Red represents high risk, whereas green represents low risk. (C): Boxplot of the differences in the expression of m6A-prognostic lncRNAs in the tumor and normal tissues. *p < 0.05; **p < 0.01; ***p < 0.001. (D): Heatmap of the differences in the expression of m6A prognostic lncRNAs between tumor and normal tissues. *p < 0.05; **p < 0.01; ***p < 0.001. Red represents high expression, whereas blue represents low expression. The abscissa represents the sample, whereas the ordinate represents prognostic lncRNAs.
Figure 2. The expression and relationship of m6A prognostic long noncoding RNAs (lncRNAs) and target gene
(A): Survival analysis according to subtypes of lncRNAs, the survival rate of cluster 2 was higher than that of cluster 1, p = 0.007. (B): Heatmap of differences in the expression of prognostic lncRNAs and the relationship with clinicopathological parameters in different clusters. *p < 0.05. Red represents high expression, whereas blue represents low expression. The abscissa represents the sample, whereas the ordinate represents prognostic lncRNAs. (C + D): Differences in the expression of the target gene in related subtypes and different types of tissues. The expression of CMTM3 in normal tissue was lower than that in tumoral tissues (p < 0.001), whereas the expression of the abovementioned gene was higher in cluster 1 (p < 0.001), which means that CMTM3 might be an oncogene for hepatocellular carcinoma. (E): Correlation analysis to analyze the correlation between target gene CMTM3 and prognostic m6A-lncRNAs in hepatocellular carcinoma. Red means a positive correlation, whereas blue means a negative correlation, *indicates a statistically significant difference.
Figure 3. Analysis of immune cell infiltration and the tumor microenvironment
(A): Vioplot of the analysis of differences in immune cell infiltration in different clusters. (B–G): Boxplot of the analysis of differences in immune cell infiltration in different clusters. Memory B cell, resting NK cells, and activated dendritic cells highly infiltrated cluster 1, whereas activated memory CD4 T cells, CD8 T cells, and follicular helper T cells highly infiltrated cluster 2, p < 0.05. (H-J): Analysis of differences in the tumor microenvironment in different subtypes. The ESTIMATE score and Stroma score were higher in cluster 2, which indicates a lower purity of tumor cells (p < 0.05).
Figure 4. Gene set enrichment analysis
To clarify the difference of related function or pathway in different samples, the top 6 enriched functions or pathways of each cluster were listed. The most enriched pathway was the Notch signaling pathway. Both FDR q-value and FWER p-value were <0.05.
Figure 5. Prognostic model and its influence on the prognosis of HCC patients
(A + B): A prognostic model was constructed via lasso regression. (C + D): Survival curve of different groups. In both groups, the survival rate of low-risk patients was higher than that of high-risk patients (C: test group, D: training group, p < 0.05). (E + F): Receiver operating characteristics curve to evaluate the accuracy of our model in predicting the survival of patients with the disease. (E: test group, F: training group, area under the curve >0.5). This model can accurately predict the prognosis of HCC patients.
Figure 6. m6A-lncRNA-related risk score and its influence on the prognosis of HCC patients
(+ D) Risk-related heatmap: m6A-lncRNAs, such as AC019080.5, AC145207.5, TMCC1-AS1, AL603839.2, LINC01138, AC099850.4, AL117336.2, AC026356.1, AC048344.4, and KDM4A-AS1, were highly expressed in the high-risk group, which means all of them might be detrimental to prognosis of HCC patients. (B + E) Risk-related curve and (C + F) risk-related spot plot. With an increase in the risk score, the number of deaths increases and the ratio of high risk increases. A + B + C: test group, D + E + F: training group.
Figure 7. Multivariate and univariate analyses of independent prognostic analysis
(A + B): in the test group, both stage and risk score were risk factors for the prognosis of HCC patients; (C + D): in the training group, grade, stage and risk score were risk factors, p < 0.05; (A + C): multivariate analysis, (B + D): univariate analysis.
Figure 8. Survival curve for model validation
Our model could be applied to different clinical groups: age (A + B), grade (G + L), lymphatic metastasis (I), M stage (H), stage (C + D), sex (J + K), and T stage (E + F), in different groups by age, sex, grade, lymph node metastasis, M stage, and T stage; the survival rate of the low-risk group was higher than that of the high-risk group, p < 0.05.
Figure 9. Correlation analysis of risk, immune cells, and clinical and genetic differences analysis of the target gene
A: Heatmap of risk and clinical correlation analysis, where the risk score was closely related to grade, cluster, and immunescore, p < 0.05(*p < 0.05, ***p < 0.001); B-H: Boxplot of risk and clinical correlation analysis. The risk score was closely related to grade, sex, T stage, stage, N stage, immunescore, and cluster; p < 0.05. I: Analysis of genetic differences in target gene. The expression of CMTM3 was higher in the high-risk group (p = 6.7e-11). J: Scatterplot of correlation analysis of risk score and immune cells. Regulatory T cells are negatively related to the risk score, R = −0.26 and p = 0.026.
Supplemental material
Supplemental Material
Download ()Data availability statement
The dataset supporting the conclusions of this article is available upon reasonable request from The Cancer Genome Atlas. https://portal.gdc.cancer.gov/
