Objective: Cluster classification based on m6A methylation regulators and construct prognostic evaluation model. Methods: Utilizing consensus cluster to classify the liver cancer samples form TCGA based on the expression of 13 m6A methylation regulators, and verify the function and prognostic significance of the clustered subtypes. Marker genes were further screened to construct a risk prediction model for evaluating the prognosis of liver cancer patients. Results: The two clustered subtypes based on m6A methylation regulators showed significant differences in the prognosis value of liver cancer patients (P=0.048), and 38 prognostic markers related to m6A methylation in liver cancer were screened from the subgroup with poor prognosis. Two m6A regulatory genes, YTHDF1 and YTHDF2, are proved with adverse prognosis by univariate cox analysis (P<0.05, Hazard ratio>1). We used Lasso regression method to build risk assessment model and effectively predicted the prognosis status of liver cancer patients within 4 years (4-year AUC=0.685, 3-year AUC=0.669). Moreover, the assessment model was validated in another dataset of Asia liver cancer patients. Conclusion: The study provided ideas for studying m6A methylation in liver cancer, and the risk prediction model can be used to evaluate the short-term prognosis of liver cancer patients.
目的： 基于m6A甲基化调节基因对肝癌进行聚类分型并构建模型评估预后风险。 方法： 基于13个m6A甲基化调节基因在肝癌中的表达情况，使用一致性聚类分析对来自TCGA数据库的肝癌患者进行分型，并对所得亚型的功能和预后意义进行研究。进一步筛选标记基因构建风险预测模型，用于评估肝癌患者的预后。 结果： 基于m6A甲基化调节基因进行分型所得到的2个亚组在肝癌患者预后方面差异具有统计学意义（P=0.048），并在预后较差的亚组中筛选得到了38个在肝癌中与m6A甲基化相关的预后标志物；利用单因素cox分析鉴定出2个与肝癌不良预后有关的m6A调节基因：YTHDF1和YTHDF2（风险比>1，P<0.05）；通过Lasso回归构建了风险预测模型能够有效预测肝癌患者4年内的预后状态（4年和3年的曲线下面积分别为0.685、0.669），并使用来自亚洲肝癌患者的数据集对该模型进行了验证。 结论： 此研究为基于m6A甲基化的肝癌诊断和治疗提供新思路，且风险预测模型可以用于评估肝癌患者的短期预后情况。.