胶质瘤患者的生存风险预测模型

Survival risk prediction model for patients with glioma

摘要:

目的 探索与胶质瘤患者预后相关的RNA，并以这些RNA建模以预测患者的生存状况。方法 对TCGA数据库中653个胶质瘤的RNA测序数据作单因素生存分析，筛选与患者预后相关的基因；对所得到的基因，利用Lasso回归建模，获得可预测患者生存状况的模型并加以验证；根据模型所得的风险分数，结合临床特征做多因素Cox回归分析，验证模型是否有效且独立于临床特征。结果 筛选得到31641个与预后相关的基因，Lasso回归模型中共包含40个基因表达量，多因素Cox回归分析证明模型有效（P<0.05）且独立于临床特征（P<0.05）。结论 利用RNA测序数据和Lasso回归建模所得模型可预测患者的生存状况。

Abstract:

Objective To explore the RNAs related to the prognosis of patients with glioma, and to establish a model based on the RNAs to predict the survival status of the patients.Methods A univariate survival analysis was performed on the RNA-Seq data of 653 gliomas from The Cancer Genome Atlas (TCGA) to screen out the genes related to the survival of the patients. Using the obtained genes, a model that can predict the survival status of the patients was established through Lasso regression, and then the model was validated. Based on the risk scores derived from the model and with reference to the clinical features, a multivariate Cox regression analysis was conducted to validate whether the model was effective and independent of the clinical features.Results Totally, 31641 prognosis-related genes were screened out, and 40 expressed genes were included in the Lasso regression model; the multivariate Cox regression analysis demonstrated that the model was effective (P<0.05) and independent of the clinical features (P<0.05).Conclusions The model established with RNA-Seq data through Lasso regression can predict the survival status of patients.