人脑胶质瘤预后相关基因模型的建立与验证  

Establishment and validation of human glioma prognosis-related genemodel

作　　者：何思远 覃玉娟 甘光磊 吴永明[4] 李光景 宋健 HE Si-yuan;QIN Yu-juan;GAN Guang-lei;WU Yong-ming;LI Guang-jing;SONG Jian(Basic Medical College;Graduate School,Youjiang Medical College for Nationalities,Baise 533000;Department of Neurology,Guangxi Zhuang Autonomous Region National Hospital,Nanning 533001;Human Sperm Bank,the First Affiliated Hospital of Guangxi Medical University,Nanning 530021,Guangxi,China)

机构地区：[1]右江民族医学院基础医学院,广西百色533000 [2]右江民族医学院研究生学院,广西百色533000 [3]广西壮族自治区民族医院神经内科,广西南宁530001 [4]广西医科大学第一附属医院人类精子库,广西南宁530021

出　　处：《川北医学院学报》2024年第4期433-438,共6页Journal of North Sichuan Medical College

基　　金：广西自然科学基金(2018GXNSFAA050115)。

摘　　要：目的:通过生物信息学分析方法筛选人脑胶质瘤预后不良相关基因,分析预后基因模型在人脑胶质瘤预后预测中的价值。方法:分别从GTEx数据库、TCGA数据库中下载正常人大脑皮层测序数据、人脑胶质瘤(胶质母细胞瘤和低级别胶质瘤)相关转录组测序数据和患者生存资料,验证集CGGA325和CGGA693测序数据来源于CGGA数据库,从人脑胶质瘤和正常人大脑皮层测序数据之间筛选出差异表达基因,对其进行单因素Cox回归、Lasso回归和CGGA325和CGGA693生存分析验证;筛选与人脑胶质瘤患者生存密切相关的基因,分析高、低风险评分预后基因模型对脑胶质瘤患者总生存期的影响;ROC曲线评估预后模型预测脑胶质瘤患者总生存期的价值。结果:在TCGA中,共筛选出12709个差异表达基因,其中表达上调的基因7120个,表达下调的基因5589个,经单因素Cox、Lasso回归分析结合验证集CGGA325和CGGA693生存分析验证,最终确定DDX47、DIABLO、GABARAP、SMARCE1、ZNF410均与脑胶质瘤患者预后相关(P<0.05),在TCGA中,高风险评分预后模型(DDX47+DIABLO+GABARAP+SMARCE1+ZNF410)组患者的预后较低风险评分组差(P<0.05),预后模型预测脑胶质瘤患者1年、3年和5年总生存期的曲线下面积(AUC)分别为0.779、0.750和0.742;在验证集CGGA325和CGGA693中,预后模型高风险评分组患者的预后均较低风险评分组差(P<0.05),预后模型在CGGA325中预测脑胶质瘤患者1、3和5年总生存期的AUC分别为0.673、0.778和0.830,在CGGA693中预测脑胶质瘤患者1、3和5年总生存期的AUC分别为0.680、0.700和0.709。结论:DDX47、DIABLO、GABARAP、SMARCE1和ZNF410可能为脑胶质瘤患者生存相关基因,具有作为人脑胶质瘤预后分子标志物的潜能,高风险评分预后模型(DDX47+DIABLO+GABARAP+SMARCE1+ZNF410)与患者预后不良相关,且对预测患者3、5年总生存期具有潜在价值。Objective:To screen the genes related to poor prognosis of human glioma by bioinformatics analysis,and to analyze the value of prognostic gene model in predicting the prognosis of human glioma.Methods:Normal human cerebral cortex sequencing data,human glioma(glioblastoma and low-grade glioma)related transcriptome sequencing data and patient survival data were downloaded from GTEx database and TCGA database,respectively.The validation set CGGA325 and CGGA693 sequencing data were obtained from CGGA database.The differentially expressed genes were screened from the sequencing data of human glioma and normal human cerebral cortex,and verified by univariate Cox regression,Lasso regression,CGGA325 and CGGA693 survival analysis.The genes closely related to the survival of human glioma patients were screened.The effect of prognostic gene models with high and low risk scores on the overall survival of glioma patients was analyzed,and the ROC curve was used to evaluate the value of prognostic models in predicting the overall survival of glioma patients.Results:In TCGA,a total of 12,709 differentially expressed genes were screened out,including 7,120 up-regulated genes and 5,589 down-regulated genes,which were verified by univariate Cox regression analysis,Lasso regression analysis combined with survival analysis of validation sets CGGA325 and CGGA693.DDX47,DIABLO,GABARAP,SMARCE1 and ZNF410 were finally determined to be related to the prognosis of glioma patients(P<0.05).In TCGA,the prognosis of patients in the high-risk score prognostic model(DDX47+DIABLO+GABARAP+SMARCE1+ZNF410)group was worse than that in the low-risk score prognostic model(P<0.05),and the AUC values of 1,3 and 5 year overall survival of glioma patients predicted by the prognostic model were 0.779,0.750 and 0.742,respectively.In the validation sets CGGA325 and CGGA693,the prognosis of patients in the high-risk score group of the prognostic model was worse than that in the low-risk score group(P<0.05).The AUC values of the prognostic model in predicting th

关 键 词：人脑胶质瘤 预后基因 模型验证 

分 类 号：R739.41[医药卫生—肿瘤]