基于TCGA数据库的胃癌预后miRNAs筛选及预后评估模型构建

Construction of miRNA prognostic predicting model for gastric adenocarcinoma based on TCGA cohort

目的:利用TCGA数据库中胃癌miRNA测序数据筛选预后相关miRNA并构建预后评估模型。方法:利用R中RTCGA包下载胃癌miRNA表达谱数据和临床信息,基于R中Survival包依次进行单变量Cox回归分析筛选胃腺癌预后相关miRNA,进一步利用多变量Cox回归分析(比例风险模型)构建预后评估模型。利用在线数据库对预后相关miRNA进行靶基因预测和信号通路富集分析。结果:利用单变量Cox回归分析筛选到34个胃癌预后相关miRNA(P<0.05),进一步通过多变量Cox回归分析构建基于14-miRNA的胃腺癌预后评分模型,生存分析显示基于该评分模型的低风险组预后显著优于高风险组,差异有统计学意义(P<0.01),ROC曲线显示该模型具有良好的预测效果(AUC=0.793)。对14-miRNA的靶基因信号通路富集分析显示与癌症相关信号高度相关。结论:14-miRNA风险评分模型具有良好的预后预测效果。

Objective To screen prognostic related miRNAs using miRNA sequencing data in TCGA database and construct a prognostic predicting model.Method R script was used to download gastric cancer miRNA expression profiling data and clinical information.Based on the Survival package in R,univariate Cox regression analysis was used to screen the miRNAs related to the prognosis of gastric adenocarcinoma,and the multivariate Cox regression analysis was used to construct a prognostic evaluation model.Target gene prediction and pathway enrichment analysis of prognosis-related miRNAs were performed using online databases.Results Using univariate Cox regression analysis,34 gastric cancer prognostic miRNAs were screened(P<0.05).A multivariate Cox regression analysis was used to construct a 14-miRNA-based gastric adenocarcinoma prognostic predicting model.Survival analysis showed that the score based on this scoring model was low.The prognosis of the risk group was significantly better than that of the high-risk group(P<0.01).The ROC curve shows that the model had a good predictive effect(AUC=0.793).Enrichment analysis of 14-miRNA target gene signaling pathways showed a high correlation with cancer-related pathways.Conclusion The 14-miRNA risk model is useful for predicting prognosis.