肝细胞癌相关的核编码线粒体基因及临床信息的综合预后模型

An integrated prognostic model of nuclear-encoded mitochondrial gene signature and clinical information for hepatocellular carcinoma

目的·建立一个基于线粒体基因和临床信息的肝细胞癌(hepatocellular carcinoma,HCC)总生存率(overall survival,OS)的预后模型。方法·从癌症基因组图谱(The Cancer Genome Atlas,TCGA)下载369例HCC患者和50例肝脏正常对照的基因表达谱和临床数据。核编码的线粒体基因(nuclear encoded mitochondrial gene,NEMG)从MitoCarta3.0数据库获得。使用“DEseq2”R包和单变量Cox分析选择与HCC患者OS相关并参与氧化磷酸化、三羧酸循环和细胞凋亡通路的NEMG[(泛素细胞色素C还原酶铰链蛋白(ubiquinol cytochrome C reductase hinge protein,UQCRH)、腺苷三磷酸柠檬酸裂解酶(ATP citrate lyase,ACLY)、磷酸烯醇式丙酮酸羧激酶2(phosphoenolpyruvate carboxykinase 2,PCK2)、Bcl-2同源拮抗剂1(Bcl-2 homologous antagonist/killer 1,BAK1)、Bcl-2相关X蛋白(Bcl-2-associated X protein,BAX)和Bcl-2/腺病毒E1B相互作用蛋白3样(Bcl-2/adenovirus E1B interacting protein 3-like,BNIP3L)]。应用多变量Cox回归来确定HCC OS的独立危险因素。建立一个基于独立危险因素(6个NEMG风险特征和TNM分期)的综合预后模型和预后列线图,计算中位预后评分。以中位预后评分作为分界点,将HCC患者分为高风险组和低风险组。进行Kaplan-Meier生存曲线分析,并进行对数秩检验来评估低风险组和高风险组患者OS的差异。使用“timeROC”软件包计算受试者操作特征(receiver operating characteristic,ROC)曲线下面积(area under the curve,AUC)。用基因表达数据库(Gene Expression Omnibus,GEO)下载HCC队列(GSE14520)验证综合预后模型对1、3、5年OS的预测性能。通过实时荧光定量聚合酶链反应(real-time quantitative polymerase chain reaction,qPCR)在来自上海交通大学医学院附属新华医院的34例HCC临床样本中验证6-NEMG的相对表达水平。结果·ROC分析结果显示,与仅6-NEMG风险特征(1、3、5年AUC分别为0.77、0.66、0.65)或仅TNM分期(1、3、5年AUC分别为0.66、0.67、0.63)相比,该综合预后模型对1年(AUC,0.78)、3年(AUC,0.73)和5年(AUC,0.69)HCC OS显示出更好的预测性能。Kaplan-Meier生存曲线分析结果显示高风险组患者的OS明显比低风险组差(P=0.001)。此外,在GEO外部队列中发现该预后模型的预测性能较好(1、3、5年AUC分别为0.67、0.66、0.74),高、低风险组患者的预后差异有统计学意义(P=0.001),与TCGA数据的结果一致。在临床HCC队列中,与癌旁肝脏组织相比,除BNIP3L外,其他5个NEMG在肿瘤组织的表达水平上调或者下调。相关性分析显示,在GSE14520与临床HCC队列中预后评分与HCC肿瘤的大小和数量呈正相关。结论·构建并验证了一个将6-NEMG风险特征与TNM分期相结合的HCC预后预测模型。该模型可能有助于HCC患者的预后预测。

Objective·To establish a prognostic model for the overall survival(OS)of hepatocellular carcinoma(HCC)based on mitochondrial genes and clinical information.Methods·The gene expression and the clinical data of 369 HCC patients and 50 controls with normal liver were downloaded from The Cancer Genome Atlas(TCGA)database.The nuclear-encoded mitochondrial genes(NEMGs)were obtained from the MitoCarta3.0 database.The"DESeq2"R package and univariate Cox analysis were used to select NEMGs[ubiquinol cytochrome C reductase hinge protein(UQCRH),ATP citrate lyase(ACLY),phosphoenolpyruvate carboxykinase 2(PCK2),Bcl-2 homologous antagonist/killer1(BAK1),Bcl-2-associated X protein(BAX)and Bcl-2/adenovirus E1B interacting protein 3-like(BNIP3L)]in HCC that were associated with OS of HCC and participated in dysregulation of oxidative phosphorylation,tricarboxylic acid cycle and cell apoptosis.Multivariate Cox analysis was applied to select independent risk factors for OS of HCC.A comprehensive prognostic model and a prognostic nomogram with 6-NEMG risk characteristics and TNM staging were established.By using the median of prognostic scores as a cut-off,HCC patients were classified into low-risk and high-risk group.Kaplan-Meier survival curve analysis was conducted and log-rank test was performed to evaluate the survival rates between the lowrisk and high-risk group.The area under the curve(AUC)values of receiver operating characteristic(ROC)curve were calculated via using the"timeROC"package.The prognostic model for HCC was validated by using the GEO HCC cohort(GSE14520)for 1,3 and 5 years.Finally,the relative expression level of 6-NEMG was validated in 34 clinical samples of HCC from Xinhua Hospital,Shanghai Jiao Tong University School of Medicine by using real-time quantitative polymerase chain reaction(qPCR)method.Results·Compared to 6-NEMG risk signature only(AUCs for 1,3 and 5 years were 0.77,0.66 and 0.65,respectively)or TNM stage only(AUCs for 1,3 and 5 years were 0.66,0.67 and 0.63,respectively),ROC curve analysis showed that this integrated prognostic model displayed better predictive performance for 1-year(AUC,0.78),3-year(AUC,0.73)and 5-year(AUC,0.69)OS of HCC.The Kaplan-Meier survival curve analysis showed that the OS of HCC patients in the high-risk group was significantly worse than that in the low-risk group(P=0.001).In addition,predictive performance of the prognostic model(AUC for 1,3 and 5 years is 0.67,0.66 and 0.74,respectively)and prognostic differences between the high-risk and low-risk group(P=0.001)were further validated in GEO(GSE14520)external cohort,and these results were consistent with the TCGA data.In addition to BNIP3L,dysregulation of five other NEMGs in the clinical HCC cohort was validated.The correlation analysis in GSE14520 and HCC clinical cohort showed a positive correlation between prognosis score and the size and number of tumors.Conclusion·A new prognostic model that combines 6-NEMG risk characteristics with TNM staging for predicting OS in HCC patients was constructed and validated.This model may help improve the prognosis prediction of HCC patients.