基于3个自噬相关微小核糖核酸表达信息学分析构建肺腺癌预后风险评分模型

Construction of Prognostic Risk Scoring Model for Lung Adenocarcinoma Based on Three Autophagy-related miRNAs Expression Bioinformatics

目的使用美国癌症基因组图谱(TCGA)数据库构建肺腺癌患者预后的自噬相关微小核糖核酸(miRNA)风险评分模型。方法结合TCGA数据库和人类自噬基因数据库(HADb)下载肺腺癌、正常肺组织的基因和miRNA表达谱的数据及临床相关资料,将总体人数随机均分为训练集和测试集,计算训练集中肺腺癌自噬基因的差异表达,筛选出和自噬基因相关的miRNA,采用单因素Cox风险回归和LASSO回归的方法筛选和建立自噬相关miRNA预后模型,并在测试集中验证,最后通过实验验证筛选出自噬相关miRNA在肺腺癌中的表达。结果从TCGA数据库中得到535例肺腺癌组织和59例正常肺组织的自噬基因表达谱,使用R软件中的edgeR包进行差异基因分析,得到差异表达的自噬基因30个,与这30个自噬基因相关的miRNA有12个。单因素Cox分析及LASSO回归分析得到基于3个miRNA的预后风险模型:风险评分=0.048×hsa-mir-31+0.201×hsa-mir-1293+0.174×hsa-mir-548f-1。训练集中模型预测3年总生存率的曲线下面积为0.796,5年总生存率的曲线下面积为0.837,提示模型准确率较高。多因素Cox回归分析得出,该预后模型可以作为独立的一个预后因子预测肺腺癌的风险(HR=2.100,95%CI=1.541~2.861,P<0.05),上述结论在测试集中得到验证。实时定量PCR提示,hsa-mir-31,hsa-mir-1293,hsa-mir-548f-1在肺腺癌中的表达较正常肺组织高(P<0.05),与生物信息分析结果一致。结论经过生物信息学分析处理,成功建立了基于3个自噬相关miRNA的肺腺癌风险预测模型,该模型有望对患者个体化治疗提供一定帮助,并提高肺腺癌患者个体化预测结果的准确度。

Objective To construct an autophagy related miRNA risk scoring model for the prognosis of lung adenocarcinoma patients by using The Cancer Genome Atlas(TCGA)database.Methods The gene and miRNA expression profile data and clinical data of lung adenocarcinoma and normal lung tissues were downloaded from TCGA database and Human Autophagy Database(HADb).The patients were randomly and equally divided into training set and test set.The differential expression of autophagy genes in lung adenocarcinoma in the training set was calculated,and the miRNAs related to autophagy genes were screened out.The univariate Cox risk regression and least absolute shrinkage and selection operator(LASSO)regression were used to screen and establish the prognosis model of autophagy related miRNA,which was verified in the test set.Finally,the expression of the screened autophagy related miRNAs in lung adenocarcinoma was verified through experiments.Results Autophagy gene expression profiles of 535 lung adenocarcinoma tissues and 59 normal lung tissues were obtained from TCGA database.The edgeR package in R software was applied for differential gene analysis,and 30 differentially expressed autophagy genes and 12 miRNAs correlative to the 30 autophagy genes were obtained.Univariate Cox analysis and LASSO regression analysis were used and the prognostic risk model based on three miRNAs was constructed:risk score=0.048×hsa-mir-31+0.201×hsa-mir-1293+0.174×hsa-mir-548f-1.The receiver operating characteristic curve(ROC)of the training set showed that the area under the curve(AUC)of 3-year survival rate predicted by the model was 0.796,and the AUC of 5-year survival rate was 0.837,suggesting the high accuracy of the model.Multivariate Cox regression analysis indicated that the prognostic model could be used as an independent prognostic factor to predict the risk of lung adenocarcinoma(HR=2.100,95%CI=1.541-2.861,P<0.05),which was verified in the test set.Real-time quantitative polymerase chain reaction(PCR)displayed that the expression of hsa-mir-31,hsa-mir-1293 and hsa-mir-548f-1 in lung adenocarcinoma was higher than that in normal lung tissue(P<0.05)and was concordant with the results of bioinformatics analysis.Conclusion A lung adenocarcinoma prognostic risk prediction model based on three autophagy related miRNAs is successfully established by bioinformatics analysis.This model is expected to provide some help for individualized treatment and improve the accuracy of individualized prediction of lung adenocarcinoma patients.