To seek survival-related genes in glioblastoma and establish a survival-gene signature for predicting prognoses of glioblastoma using public databases. Methods: Three independent glioma databases (GEO GSE53733, CGGA, TCGA) with whole genome expression data were included for analysis. Survival-related genes were obtained by comparing the long-term (>36 months) and short-term (<12 months) survivors in the database GSE53733. CGGA was used as the training set to develop the signature and TCGA was used as the validation set. Cox regression analysis and linear risk score assessment were conducted to look for prognostic signatures with survival-related genes. Principal components analysis, gene set enrichment analysis (GSEA), gene ontology (GO) and protein-protein interaction (PPI) analysis were performed to explore distinct expression profiles between risk grouped glioblastoma. Results: We totally found 211 survival-related genes and developed a signature with 17 survival-related genes for prognosis of glioblastoma. Based on this signature, the low-risk group had longer survival time while the high-risk group had shorter survival time. Additionally, the expression profiles between the high-risk and low-risk glioblastoma were different. Functional annotations revealed that the genes enriched in the high-risk glioblastoma were involved in immune systems and processes of extracellular matrix (ECM). Conclusion: The novel survival-gene signature can predict high-risk glioblastoma with shorter survival time, enhance immunosuppressive features, and increased invasion preferences.
目的：通过对多个公共数据库中胶质母细胞瘤转录组数据进行挖掘，筛选胶质母细胞瘤预后相关的基因，并构建预后分析模型。方法：利用生物信息学技术对GEO(GSE53733)中生存时间大于36个月和小于12个月的样本数据对比分析得到差异表达基因，即胶质母细胞瘤预后相关基因；采用Cox风险回归方法在CGGA和TCGA两个独立数据库中筛选与预后相关的标签基因并构建预后分析模型；采用基因探针富集分析(GSEA)，GO(gene ontology)功能富集分析和蛋白网络互作分析(PPI)等方法分析高、低风险胶质母细胞瘤的分子特征。结果：分析得到211个胶质母细胞瘤预后相关基因，并且从中筛选出17个标签基因。利用分子标签基因构建的模型能将胶质母细胞瘤划分为高风险组和低分险组，高风险组的患者预后较差，在分子特征上更具免疫抑制性和侵袭性。结论：通过挖掘公共数据库建立的分析模型对胶质母细胞瘤预后有良好的预测作用，判定为高风险组的胶质母细胞瘤预后更差，可作为潜在的胶质母细胞瘤预后指示标签。.