基于IDH突变相关代谢基因的胶质瘤分型探讨

Discussion on the classification of glioma based on IDH mutation related metabolic genes

目的探讨异柠檬酸脱氢酶(IDH)突变相关代谢基因表达水平、胶质瘤分型。方法采用加权相关网络分析方法对来自中国脑胶质瘤基因组图谱计划(CGGA)294例脑胶质瘤转录组数据进行分析,筛选与IDH状态(IDH突变型/野生型,IDH-mut/wt)相关差异表达基因(DEG)。参照KEGG在DEG中筛选出代谢通路相关基因,对IDH-wt胶质瘤进行聚类分型。应用生存分析验证新分型的临床意义,应用随机森林分析筛选新分型的特征基因。结果根据聚类结果将IDH-wt胶质瘤分为C1、C2和C3型。1年及3年生存率分析显示,IDH-mut和C1型胶质瘤患者生存率明显较C2及C3型患者高。应用随机森林分析筛选出C1型前5个最显著特征基因,分别为TK1、GPX8、RRM2、NNMT和GPX7,其MDG值分别为13.98、13.94、10.48、10.37和9.54。应用五基因模型诊断C1型,训练组(随机选择CGGA数据库中的2/3数据)的分类准确度为95.12%,灵敏度为93.75%,特异性为95.60%。结论胶质瘤代谢亚型区分有助于准确判断预后,指导脑胶质瘤患者的个体化治疗。

Objective To investigate the expression level of metabolic genes associated with isocitrate dehydrogenase(IDH)mutation and the classification of glioma.Methods The transcriptome data of 294 gliomas from the Chinese glioma genome atlas(CGGA)were analyzed by weighted correlation network analysis to screen the differentially expressed genes(DEG)associated with IDH status(IDH mutant type/IDH wild type,IDH-mut/wt).Genes related to metabolic pathways were screened from DEG with reference to KEGG,and cluster typing of IDH-wt gliomas was performed.Survival analysis was used to verify the clinical significance of the new types,and random forest analysis was used to screen the characteristic genes of the new types.Results According to the clustering results,IDH-wt gliomas were divided into C1,C2 and C3 types.The one-year and three-year survival rate analysis showed that IDH-mut and C1 glioma patients had significantly higher survival rate than C2 and C3 patients.Random forest analysis was used to screen out the top five most significant characteristic genes of C1 type,namely TK1,GPX8,RRM2,NNMT and GPX7,and their MDG values were 13.98,13.94,10.48,10.37 and 9.54,respectively.The C1 type was diagnosed by five-gene model.The classification accuracy,sensitivity and specificity of the training group(2/3 of the data from CGGA database were randomly selected)were 95.12%,93.75% and 95.60%.Conclusion The differentiation of metabolic subtypes of glioma is helpful for predicting prognosis and guiding the individualized treatment of glioma patients.