膀胱癌miRNAs风险预后模型建立与应用  

Construction and application of a miRNA risk prognosis model for bladder cancer

作　　者：华金骏[1] 李兴[2] 高杨 赵建华[1] HUA Jinjun;LI Xing;GAO Yang;ZHAO Jianhua(Department of Urology,Shuguang Hospital Affiliated to Shanghai University of Traditional Chinese Medicine,Shanghai 201210,China;Department of Anesthesiology,Shuguang Hospital Affiliated to Shanghai University of Traditional Chinese Medicine,Shanghai 201210,China;Department of Pathology,Shuguang Hospital Affiliated to Shanghai University of Traditional Chinese Medicine,Shanghai 201210,China)

机构地区：[1]上海中医药大学附属曙光医院泌尿外科,上海201210 [2]上海中医药大学附属曙光医院麻醉科,上海201210 [3]上海中医药大学附属曙光医院病理科,上海201210

出　　处：《现代肿瘤医学》2021年第8期1344-1349,共6页Journal of Modern Oncology

基　　金：国家自然科学基金面上项目(编号:81973652)。

摘　　要：目的:通过筛选与膀胱癌生存预后相关的非编码小分子核糖核酸(microRNAs,miRNAs),构建风险预后模型评价患者预后。方法:从人类癌症基因图谱(The Cancer Genome Atlas,TCGA)数据库下载膀胱癌miRNAs与mRNAs数据及对应的临床相关信息,筛选差异的miRNAs和mRNAs。对差异的miRNAs进行Kaplan-Meier(K-M)生存分析和单因素Cox回归分析评价筛选得到与预后相关的miRNAs。采用多因素Cox回归进一步分析获得风险模型目标miRNAs,建立最佳风险预后模型,评价模型效能。通过在线数据库和差异mRNAs数据对结合的miRNA靶基因进行筛选,然后进行KEGG和GO富集分析,预测miRNAs靶基因功能。结果:筛选得到膀胱癌差异miRNAs有473个,差异mRNAs有7106个。运用K-M生存分析及单因素Cox回归筛选得到24个与膀胱癌患者生存预后相关的miRNAs,进一步通过多因素Cox回归得到11个与风险预后模型紧密相关的miRNAs,成功构建miRNAs风险预后模型。模型风险评分=(-0.4×Exp_(hsa-miR-145-5p))+(-0.928×Exp_(hsa-miR-99a-5p))+(-0.295×Exp_(hsa-miR-7705))+(-0.566×Exp_(hsa-miR-590-3p))+(0.201×Exp_(hsa-miR-337-3p))+(0.576×Exp_(hsa-miR-125b-2-3p))+(0.345×Exp_(hsa-miR-652-3p))+(-0.718×Exp_(hsa-miR-143-5p))+(0.286×Exp_(hsa-miR-3065-3p))+(0.282×Exp_(hsa-let-7c-3p))+(0.145×Exp_(hsa-miR-216a-5p))。模型的受试者工作特征曲线(receiver operating characteristic curve,ROC)曲线下面积(the area under the ROC curve,AUC)为0.781,提示风险预后模型具有较好的预测效能。K-M生存分析提示模型评分越高的患者5年生存率越低。GO和KEGG富集分析显示,模型miRNAs能够调控多种泌尿系统活性信号和相关通路,介导膀胱癌疾病进展。结论:成功构建膀胱癌miRNAs风险预后评估模型,为临床膀胱癌患者生存预测提供了一定的价值。Objective:To screen biomarkers of non-coding microRNAs(miRNAs)related to survival and prognosis of bladder cancer,and construct a risk prognosis model to evaluate the prognosis of patients.Methods:The number of miRNAs and mRNAs of bladder cancer and the corresponding clinical related information were downloaded from The Cancer Genome Atlas,and the differential miRNAs and mRNAs were screened.The miRNAs related to prognosis were screened by Kaplan-Meier(K-M)survival analysis and univariate Cox regression analysis.Multi-factor Cox regression was used to further analyze the risk model target miRNAs,then establish the best risk prognosis model,and evaluate the model effectiveness at last.Online database and differential mRNAs data was used to predict the binding miRNA target gene,then KEGG and GO enrichment analysis was carried out to predict the function of miRNA target gene.Results:There were 473 differential miRNAs and 7106 differential mRNAs in bladder cancer.24 miRNAs related to the survival and prognosis of bladder cancer patients were obtained by K-M survival analysis and univariate Cox regression screening.Eleven target miRNAs closely related to the risk prognosis model were acquired by multivariate Cox regression,and the miRNAs risk prognosis model of bladder cancer was successfully constructed.The model risk score=(-0.4×Exp_(hsa-miR-145-5p))+(-0.928×Exp_(hsa-miR-99a-5p))+(-0.295×Exp_(hsa-miR-7705))+(-0.566×Exp_(hsa-miR-590-3p))+(0.201×Exp_(hsa-miR-337-3p))+(0.576×Exp_(hsa-miR-125b-2-3p))+(0.345×Exp_(hsa-miR-652-3p))+(-0.718×Exp_(hsa-miR-143-5p))+(0.286×Exp_(hsa-miR-3065-3p))+(0.282×Exp_(hsa-let-7c-3p))+(0.145×Exp_(hsa-miR-216a-5p)).The area under the receiver operating characteristic curve of the model was 0.781,which indicated that the risk prognosis model had good prediction efficiency.K-M survival analysis suggested that the higher the model score,the lower the 5-year survival rate.GO and KEGG enrichment analysis showed that model miRNAs could regulate a variety of urinary system activity signa

关 键 词：膀胱癌 MIRNAS 风险预后模型 TCGA 

分 类 号：R737.14[医药卫生—肿瘤]