基于TCGA数据库预测结肠癌预后基因及其临床应用价值  

Predicting colon cancer prognosis genes and clinical application value based on TCGA database

作　　者：甄秋来 吕欣然 叶辉[1] 丁绪超 柴小雪 胡辛 周明 曹莉莉[1,3] ZHEN Qiulai;LYU Xinran;YE Hui;DING Xuchao;CHAI Xiaoxue;HU Xin;ZHOU Ming;CAO Lili(Department of Oncology,Shandong Qianfoshan Hospital,Cheeloo College of Medicine,Shandong University,Shandong Provincial Key Laboratory for Rheumatic Disease and Translational Medicine,Jinan 250014,Shandong,China;Department of Blood Transfusion,Central Hospital of Zibo MiningGroup Co.,Ltd.,Zibo 255120,Shandong,China;Department of Oncology,The First Affiliated Hospital of Shandong First Medical University,Jinan 250014,Shandong,Chinp)

机构地区：[1]山东大学附属山东省千佛山医院肿瘤科,山东省风湿免疫病转化医学重点实验室,山东济南250014 [2]淄博矿业集团有限责任公司中心医院输血科,山东淄博255120 [3]山东第一医科大学第一附属医院肿瘤科,山东济南250014

出　　处：《山东大学学报（医学版）》2021年第1期64-71,共8页Journal of Shandong University:Health Sciences

基　　金：山东省重点研发计划(2019GSF108180);济南市科技发展计划(201907119);山东省中医药科技发展计划(2019-0378);山东第一医科大学第一附属医院培育基金(QYPY2019NSFC1015)。

摘　　要：目的利用生物分析学方法对癌症基因组图谱(TCGA)数据库的结肠癌数据进行挖掘分析,筛选预后基因,识别结肠癌患者死亡的高低风险,并预测其预后。方法访问TCGA并下载结肠癌患者RNA表达数据和临床信息。通过单因素Cox和多因素Cox回归分析,构建比例风险回归模型并形成风险评分公式。根据风险评分中位值将患者分为高风险组和低风险组,识别结肠癌患者死亡风险。采用接收者操作特征曲线(ROC)及曲线下面积(AUC)验证该模型的评估性能。利用R语言对预后相关基因进行生存分析,并对差异基因进行GO功能和KEGG通路富集分析。结果结肠癌5 544个差异表达基因中,有27个基因与患者整体生存率相关。从中筛选出GABRD、FAM132B、LRRN4、RP11-400N13.2、RP11-108K3.2、RNU6-403P、RP11-429J17.8、LINC01296、RP11-190J1.3、AC002076.10和CTC-573N18.1共11个基因,构建结肠癌患者的Cox预后模型。ROC分析显示,高风险组5年期生存率为39.5%(95%CI:29.5~53.0),低风险组为89.6%(95%CI:82.2~97.7),AUC=0.827,该模型可以较好地区分高低风险的结肠癌患者。结论通过Cox比例风险模型基因获得风险得分并结合临床信息,用作结肠癌患者的预后及生存时间的评估。Objective To screen the prognostic genes, identify risks and predict prognosis by excavating colon cancer data from TCGA database. Methods The RNA expression data and clinical information of colon cancer patients were downloaded from TCGA database. A proportional hazard regression model was constructed and a risk scoring formula was formed after univariate Cox and multivariate Cox regression analyses. The patients were divided into high-risk and low-risk groups based on the median risk score to determine the mortality risk. The receiver operating characteristic(ROC) curve and area under the curve(AUC) were used to verify the evaluation performance of the model. Survival analysis of prognosis-related genes was performed using R language. The differentially expressed genes were analyzed using GO function and KEGG pathway enrichment. Results Of the 5 544 differentially expressed genes, 27 were associated with overall survival, and 11 were screened to construct the prognostic model, including GABRD, FAM132 B, LRRN4, RP11-400 N13.2, RP11-108 K3.2, RNU6-403 P, RP11-429 J17.8, LINC01296, RP11-190 J1.3, AC002076.10 and CTC-573 N18.1. ROC analysis showed that the 5-year survival rate was 39.5%(95%CI: 29.5-53.0) in the high-risk group and 89.6%(95%CI: 82.2-97.7) in the low-risk group, with AUC being0.827, indicating that the model could effectively distinguish patients with high and low risks. Conclusion The risk score obtained from the Cox proportional hazard model genes combined with clinical information can be used to evaluate the prognosis and survival of patients with colon cancer.

关 键 词：TCGA数据库 结肠癌 RNA COX比例风险模型 生存时间 

分 类 号：R574.62[医药卫生—消化系统]