基于TCGA数据库胆管癌自噬-临床预后模型的构建和评估

Construction and validation of autophagy-clinical prognostic model in patients with cholangio-carcinoma based on TCGA database

目的探讨自噬相关基因(autophagy-related genes,ATGs)在临床预后方面的价值,并构建自噬基因的预后风险模型,评估胆管癌(cholangiocarcinoma,CCA)患者的生存及预后状况。方法在癌症基因组图谱数据库(TCGA)中下载CCA的转录组数据和临床数据,从人类自噬数据库(HADb)中提取自噬相关基因,筛选出CCA样本中差异表达的自噬相关基因(DEATGs);GO富集分析与KEGG通路分析进一步阐明DEATGs在CCA中的作用;通过Cox回归分析构建自噬预后风险评分模型,并运用生存分析、风险曲线、ROC曲线及独立预后分析验证预后模型的准确性。结果CCA组织样本和非肿瘤组织样本一共筛选出49个DEATGs(上调的基因48个,下调的基因1个);通过单因素Cox回归分析及多因素Cox回归分析筛选出5个最有预后价值的基因(RHEB、PPP1R15A、ATG101、BNIP3、NRG1),并基于这5个基因构建自噬预后风险评分模型,模型公式为:风险值=NRG1表达量×1.1207+BNIP3表达量×1.4002+ATG101表达量×1.3457-PPP1R15A表达量×1.1613-RHEB表达量×1.3279。生存分析表明低风险组的生存率要显著高于高风险组,风险曲线表明患者风险值与生存时间及生存状态显著相关,ROC曲线提示自噬预后模型具有良好的准确性,是一个独立的预后因素(HR 1.427,95%CI 1.161~1.756,P<0.001)。结论5个自噬基因(RHEB、PPP1R15A、ATG101、BNIP3、NRG1)构建的自噬相关预后风险模型,可以有效预测CCA患者的预后情况。

Objective To explore the value of autophagy-related genes(ATGs)in clinical prognosis,and to build a prognostic risk model of autophagy genes to evaluate the survival and prognosis of CCA patients.Methods Transcriptome data and clinical data of CCA were downloaded from cancer genome map database(TCGA),autophagy related genes were extracted from human autophagy database(HADb),and autophagy-related genes(DEATGs)differentially expressed in CCA samples were screened out.GO enrichment analysis and KEGG pathway analysis further clarified the role of DEATGs in CCA;Cox regression analysis was used to construct autophagy prognostic risk scoring model,and survival analysis,risk curve,ROC curve and independent prognostic analysis were used to verify the accuracy of the prognostic model.Results A total of 49 DEATGs were selected from CCA tissue samples and non-tumor tissue samples(48 up-regulated genes and 1 down-regulated gene).By univariate Cox regression analysis and multivariate Cox regression analysis,5 most important prognostic genes were selected(RHEB,PPP1R15A,ATG101,BNIP3,NRG1).The autophagy prognostic risk score model was based on these 5 genes,the model formula was:risk value=NRG1 expression×1.1207+NRG1BNIP3 expression×1.4002+NRG1ATG101 expression×1.3457-PPP1R15A expression×1.1613-RHEB expression×1.3279.Survival analysis showed that the survival rate of low risk group was significantly higher than that of high risk group.The risk curve showed that the CCA risk value is significantly related to the survival time and survival state.ROC curve suggested good accuracy of autophagy prognostic model,which is an independent prognostic factor(HR 1.427,95%CI 1.161~1.756,P<0.001).Conclusion In this study,autophagy-related prognostic risk model based on 5 autophagy genes(RHEB,PPP1R15A,ATG101,BNIP3,NRG1)can effectively predict the prognosis of patients with CCA.