基于癌症基因组图谱数据库肺腺癌预后风险模型的建立

Establishment and analysis of prognostic risk model for lung adenocarcinoma based on the cancer genome atlas database

目的:构建基于微小RNA(miRNA,miR)表达量的肺腺癌预后列线图风险模型,研究其对肺腺癌患者生存的预测价值。方法:从癌症基因组图谱(TCGA)数据库下载获取肺腺癌miRNA表达数据,包括miRNAseq和临床数据。应用R 3.6.2软件中的edgeR包筛选肺腺癌组织和正常肺组织的差异基因,以│logFC│>2,P<0.05为筛选条件。应用单因素回归分析、LASSO回归分析、多因素Cox回归分析筛选与预后明显相关的miRNAs,建立风险评分(risk score)方程,构建列线图模型。应用内部验证法和图像校准法、受试者工作特征曲线(ROC)评价模型的特异性和敏感性。通过风险评分及生存曲线对患者进行生存分析。分析miRNAs在各种族表达差异。结果:识别127个差异miRNA,下调16个,上调111个。单因素回归分析得到14个与预后相关的miRNAs,LASSO回归分析得到13个与预后相关的miRNAs(P<0.05)。多因素回归分析结果显示hsa-miR-1293、hsa-miR-450a-1、hsa-miR-5571、hsa-miR-3189、hsa-miR-490、hsa-miR-142、hsa-miR-31、hsa-miR-548v是显著的预测因素(P<0.05)。应用这8个miRNA构建列线图模型,Concordance值(0.701)及图像校准显示该列线图预测能力较好。1、3、5年曲线下面积(AUC)为0.721、0.748、0.733。生存分析结果显示高风险组较低风险组生存率低(P<0.05),随着风险评分数值的升高,死亡人数增多。KM生存曲线显示hsa-miR-1293、hsa-miR-5571、hsa-miR-490、hsa-miR-142、hsa-miR-31、hsa-miR-548v与肺腺癌生存预后相关(P<0.05)。另外这8个miRNAs表达种族差异无统计学意义(P>0.05)。结论:成功构建基于miRNAs表达预测肺腺癌预后的列线图模型,其预测准确性较高。

Objective To construct a predictive nomogram risk model of Lung adenocarcinoma based on the expression of microRNA(miRNA,miR)and to study its predictive value.Methods Download the miRNA expression data from The Cancer Genome Atlas(TCGA),including miRNAseq and clinical data.Use the edgeR package to screen the differential genes between normal lung tissues and lung adenocarcinoma tissue,and use│logFC│>2,P<0.05 as the screening conditions.Use single factor regression analysis,LASSO regression analysis,and multi-factor COX regression analysis to screen miRNAs significantly related to prognosis,establish a risk score equation,and construct a nomogram model.Internal verification method(C-statistics)and image calibration method,receiver operating characteristic curve(ROC)were used to evaluate the specificity and sensitivity of the model.Analyze the patient’s survival through risk scores and survival curves.Ethnic expression analysis.P<0.05 indicates that there is statistical difference.Results 127 differential miRNAs were identified,16 down-regulated,and 111 up-regulated.Firstly,a single factor regression analysis was performed,14 miRNAs related to prognosis were obtained.Then LASSO regression analysis was applied,and 13 miRNAs related to prognosis were obtained(P<0.05).Finally,a multivariate regression analysis of these 13 miRNAs showed that hsa-miR-1293,hsa-miR-450a-1,hsa-miR-5571,hsa-miR-3189,hsa-miR-490,hsa-miR-142,hsa-miR-31,hsa-miR-548v are significant predictors(P<0.05),and these eight miRNAs are used to construct a nomogram.The concordance value(0.701)and the image calibration method show that the nomogram has a good predictive ability.The area under curve(AUC)for one year,three years,and five years is 0.721,0.748,and 0.733,indicating that the model has a high accuracy rate.The results of the survival analysis revealed that the high-risk group had a low survival rate.KM survival curve were showed that hsa-miR-1293,hsa-miR-5571,hsa-miR-490,hsa-miR-142,hsa-miR-31,hsa-miR-548v are correlated with survival prognosis(P<0.05).There is no apparent racial difference in the expression of these eight miRNAs(P>0.05).Conclusion Successfully constructed a risk assessment model based on miRNA expression,which can effectively predict the prognosis of lung adenocarcinoma.