胶质母细胞瘤铜死亡相关lncRNA预后模型的构建

目的 采用生物信息学方法分析胶质母细胞瘤差异表达的铜死亡相关长链非编码RNA(lncRNA)并构建预后模型。方法 计算机检索TCGA数据库下载胶质母细胞瘤的RNA-seq数据及其临床信息，应用单因素Cox分析和LASSO回归筛选铜死亡相关lncRNA，并计算风险评分=基因系数×基因表达量，以风险评分中位数分为低风险组和高风险组；将胶质母细胞瘤样本随机分为测试组和训练组，再进行整合生物信息学分析。结果 共下载胶质母细胞瘤样本169例，对照样本5例，共表达分析发现791个铜死亡相关lncRNA，其中正相关lncRNA有753个，负相关lncRNA有38个。单因素Cox分析发现22个显著差异表达的lncRNA(P<0.05),LASSO回归分析筛选9个差异表达的lncRNA；多因素Cox回归分析显示风险评分增高是胶质母细胞瘤预后不良的独立危险因素（P<0.05）。生存曲线分析显示，无论是训练组，还是测试组，高风险组总体生存期和无进展生存期显著缩短（P<0.05）。ROC曲线分析显示模型对1、2、3年生存期具有良好的预测价值（P<0.05），而且风险评分预测效能显著优于...

Construction of a prognostic model using cuproptosis-associated lncRNAs for patients with glioblastomas based on bioinformatics analysis

Objective To explore the differential expression of cuproptosis-associated long non-coding RNAs (lncRNAs) in glioblastomas by bioinformatics methods and to construct a prognosis model using cuproptosis-associated lncRNAs for patients with glioblastomas. Methods RNA-seq data and clinical information of glioblastoma patients were retrieved from TCGA database. Univariate Cox analysis and LASSO regression were used to screen cuproptosis-associated lncRNAs. Risk score (gene coefficient×gene expression) was calculated, and the glioblastoma patients were divided into low risk and high risk groups according to the median risk score. Glioblastoma patients were randomly divided into test and training groups, and then integrated bioinformatics analyses were performed. Results A total of 169 glioblastoma patients and 5 control samples were downloaded. Co-expression analysis revealed 791 cuproptosis-associated lncRNAs, including 753 lncRNAs with positive correlation and 38 with negative correlation. Univariate Cox analysis revealed 22 differentially expressed lncRNAs (P<0.05), and LASSO regression analysis screened 9 differentially expressed lncRNAs. Multivariate Cox regression analysis showed that increased risk score was an independent risk factor for poor prognosis of glioblastoma patients (P<0.05). Survival curve analysis showed that the overall survival and progression-free survival were significantly shortened in both the training and test groups (P<0.05). ROC curve analysis showed that the model had good predictive value for 1, 2 and 3 year survival (P<0.05), and the predictive efficacy of risk score was significantly better than other clinical features (P<0.05). Conclusions We successfully construct a prognostic model using cuproptosis-associated lncRNAs for patients with glioblastomas by bioinformatics methods, which has good prognostic value for glioblastoma patients.