多基因模型在肝细胞癌预后中的应用

Application of Multigene Model in Prognosis Prediction of Hepatocellular Carcinoma

利用较少分子信息预测肝细胞癌类型对患者的个性化治疗十分关键。探索已知的与肝细胞癌预后相关的信号通路,共发现41个关键基因。随后,运用机器学习的方法对其构建风险预测模型,并在4个肝细胞癌数据集上进行验证。结果显示,该模型能将肝细胞癌患者分成两个预后差异显著的类型:癌症基因图谱(The cancer genome atlas,TCGA)数据集交叉验证的平均log rank P值为0.03;其他测试数据集的log rank P 值分别为0.000 38、0.002 1和0.01。生物信息学分析显示肝细胞癌的预后与细胞周期等信号通路显著相关,并筛选出12个潜在的肝细胞癌分子标志物。研究结果表明,基于41个基因构建的肝细胞癌预后模型具有较好的稳健性和准确的风险预测能力。

Predicting the types of hepatocellular carcinoma(HCC)while using little molecular information may provide patients with more personalized therapy.Here,known HCC prognosis-related pathways were investigated,and 41 dominant genes were identified.Based on these genes,a risk prediction model was constructed using machine learning and validated with 4 HCC datasets.The results revealed that the model was proven to efficiently divide HCC patients into 2 subgroups with significantly different prognosis:average log rank P-values of cross-validation within TCGA dataset was 0.03,and the log rank P-values of validation on other external datasets were 0.000 38,0.002 1 and 0.01,respectively.After analyzing the HCC subgroups with bioinformatics pipeline,the prognosis of HCC significantly was related to signal pathways such as cell cycle,and 12 potential biomarkers of HCC were screened.In sum,the 41-gene based stratification model may robustly and accurately predict prognosis among HCC patients.