基于TCGA数据库构建恶性胸膜间皮瘤基因预后模型及其生物信息学分析

Establishment and Bioinformatics Analysis of Gene Prognostic Model for Malignant Pleural Mesothelioma Based on TCGA Database

基于癌症基因组图谱(The Cancer Genome Atlas, TCGA)数据库构建恶性胸膜间皮瘤患者的基因预后风险模型及其验证分析.通过TCGA数据库获取恶性胸膜间皮瘤的基因表达数据以及临床信息,将数据集分为训练集和测试集,对训练集进行单因素Cox分析、单因素鲁棒性分析以及多因素Cox回归分析,建立风险预后模型,计算每个患者的风险评分,将数据集分成高风险组和低风险组,通过Kaplan-Meier生存曲线和受试者工作特征(Receiver Operating Characteristic, ROC)曲线评估模型的预测效能和准确性,利用TCGA数据库和GTEx数据库中的样本对预后模型中的基因进行表达验证,最后通过UALCAN数据库探索预后模型中基因的表达模式.本研究构建了一个由UHRF1、KIF4A和NEK2这三个基因组成的预后风险评估模型,预后风险模型风险评分=UHRF1表达量×1.452 5-KIF4A表达量×1.327+NEK2表达量×1.416 7.ROC曲线显示,该模型的ROC曲线下面积(Area Under the Curve, AUC)值为0.91,基于ROC曲线对截断值进行优化,最佳截止点为1.149,此时达到最大灵敏度和特异性.在最佳截止点处,患者被进一步分为高风险组和低风险组,Kaplan-Meier生存曲线显示,高风险组和低风险组之间的生存时间存在显著差异,与高风险组患者相比,低风险组患者总生存期显著延长.通过森林图对模型进行可视化,整个模型的Log-Rank p-value<0.000 1,表示其可以作为恶性胸膜间皮瘤的独立预后生物标志物.相比于正常肺组织,预后模型中的三个基因在恶性胸膜间皮瘤组织中均显著高表达,且在恶性胸膜间皮瘤患者的不同分期、肿瘤亚型、年龄和转移状态中表达有明显差异.本研究构建的恶性胸膜间皮瘤基因预后风险模型,能够有效地预测恶性胸膜间皮瘤患者预后,并且可以作为恶性胸膜间皮瘤的独立预后预测因子.预后模型中基因UHRF1、KIF4A和NEK2的表达与恶性胸膜间皮瘤患者的多个临床特征显著相关,提示其在恶性胸膜间皮瘤发生发展过程中的潜在作用.

The aim of our study is to establish and validate a prognostic risk model of genes in malignant pleural mesothelioma based on TCGA database.Gene expression data and clinical information of malignant pleural mesothelioma were obtained from TCGA database.The whole dataset was divided into training set and testing set.Univariate Cox regression analysis,univariate robustness analysis,and multivariate Cox regression analysis were performed in training set to establish the prognostic risk model of MPM.Risk score of each patient was calculated and datasets were divided into high-risk group and low-risk group.Effectiveness and accuracy of the model were evaluated with Kaplan-Meier survival curve and ROC curve.Tissue samples from TCGA and GTEx databases were used to validate the expression of genes in prognostic model.Exploring gene expression patterns in prognostic model through the UALCAN database.The prognostic risk model consisting of UHRF1,KIF4A and NEK2 was constructed and the model risk score=UHRF1 expression×1.4525-KIF4A expression×1.327+NEK2 expression×1.4167.The AUC value of the model is 0.91 and the cut-off value was optimized based on ROC curve.The optimal cut-off point was selected as 1.149,which was at the maximal sensitivity and specificity.With the optimal cut-off point,patients were further divided into high-risk and low-risk groups.The Kaplan-Meier survival curve indicated a significant difference in survival time between high-risk and low-risk groups.The survival time of patients in the low-risk group was significantly prolonged compared to that in the high-risk patients.The model was visualized using forest plot,Log-Rank p-value<0.0001 for the entire model indicates that it can be used as an independent prognostic biomarker for MPM.Compared to normal lung tissues,all three genes in the prognostic model were significantly highly expressed in MPM tissues.The expression of these three genes in tumor stages,histology,age and metastasis of MPM patients were significantly different.Our study constructed a prognostic risk model which can effectively predict the prognosis of MPM patients and can be used as an independent prognostic predictor of MPM.UHRF1,KIF4A and NEK2 in this prognostic model are significantly related to several clinical features of MPM patients,suggesting its potential role in the development of malignant pleural mesothelioma.