基于TCGA数据库的胶质母细胞瘤LncRNA风险预测模型的建立

目的利用TCGA数据库建立胶质母细胞瘤患者预后的LncRNA风险评分模型。方法下载TCGA数据库中胶质母细胞瘤及正常神经组织的基因表达谱数据、临床相关数据,筛选差异表达LncRNA,采用单因素和多因素Cox风险回归模型筛选和建立LncRNA预后模型。结果从TCGA数据库中得到169份胶质母细胞瘤组织和5份正常神经组织的基因表达谱,使用R语言edgeR包进行差异基因分析(logFC≥2或≤-2,FDR<0.05)得到差异基因7 978个,其中差异LncRNA 1 643个。单因素Cox分析及多因素Cox回归分析得到基于4个LncRNA的多因素预后风险模型:风险评分=0.59×NDUFB2-AS1-0.41×ZEB1-AS1+0.31×AL139385.1+0.21×AGAP2-AS1。模型的ROC曲线下面积AUC=0.864。患者风险评分结果提示高评分患者预后较低评分患者差。结论 NDUFB2-AS1、ZEB1-AS1、AL139385.1和AGAP2-AS1的风险预测模型可有效预测胶质母细胞瘤患者的预后,有望用于指导临床治疗。

Establishment of LncRNA Risk Prediction Model for Glioblastoma Based on TCGA Database

Objective To establish a risk score model of LncRNA for the prognosis of glioblastoma patients using TCGA database. Methods The gene expression profiles and clinical data of glioblastoma and normal nerve tissues in TCGA database were downloaded to screen differentially-expressed LncRNA. The risk score model of LncRNA was screened and established by univariate and multivariate Cox regression models. Results The expression profiles of glioblastoma genes were obtained from TCGA database, including 169 glioblastoma tissues and 5 normal nerve tissues. The R software edgeR package was used for differentiallyexpressed gene analysis (logFC≥2 or ≤-2, FDR＜0.05, FDR<0.05). A total of 7978 differential expressed genes were obtained, of which 1643 were differential expressed lncRNAs. By univariate and multivariate Cox regression analyses, the prognostic risk model was obtained: Risk score=0.59×NDUFB2-AS1-0.41×ZEB1-AS1+0.31×AL139385.1+0.21×AGAP2-AS1. The area under ROC curve(AUC) of the model was 0.864. Risk scores results indicated that the prognosis of patients with high score was worse than that of patients with low score. Conclusion The risk prediction models of NDUFB2-AS1, ZEB1-AS1, AL139385.1 and AGAP2-AS1 mentioned above could effectively predict the prognosis of glioblastoma patients and are expected to be used for clinical treatment guidance.