摘要目的:利用癌症基因组图谱(TCGA)分析丙酮酸脱氢酶激酶1(PDK1)在脑胶质瘤中的表达和临床意义。方法:回顾性分析TCGA数据库中641例脑胶质瘤患者的临床资料。采用GEPIA(http：//gepia.cancer-pku.cn/)在线分析网站分析PDK1在脑胶质瘤患者中的表达情况。之后采用Pearson相关分析法创建与 PDK1相关的基因列表，再利用STRING在线分析工具对差异基因进行互作网络分析，根据可信度大小筛选出关键蛋白。进一步通过基因本体分析(GO)获得PDK1相关差异基因的功能及与京都基因和基因组百科全书(KEGG)通路的关系。根据PDK1蛋白表达量将所有患者分为PDK1蛋白低表达组(380例)和PDK1蛋白高表达组(261例)，采用Kaplan-Meier生存曲线比较两组患者的生存期差异。进一步采用单因素和多因素Cox回归分析法判断PDK1蛋白表达对脑胶质瘤患者预后的作用。 结果:PDK1蛋白在世界卫生组织(WHO)Ⅳ级胶质瘤中的表达明显高于Ⅱ、Ⅲ级(均 P<0.05)，且Ⅱ、Ⅲ级间的差异无统计学意义(Ⅱ、Ⅲ、Ⅳ级的蛋白相对表达量分别为7.24±0.60、7.34±0.80、8.86±0.90， P>0.05)。PDK1蛋白高表达组的脑胶质瘤患者累积生存率明显低于低表达组( P<0.05)。Pearson相关分析结果发现，96个基因与 PDK1表达密切相关，其中69个为正相关，27个为负相关。GO分析和KEGG通路分析结果显示，PDK1与缺氧微环境、有机环状化合物的反应、有机氮化合物的反应、cAMP反应及HIF-1信号通路密切相关。多因素Cox回归分析结果显示，年龄( HR＝1.579，95% CI：1.085～2.299， P＝0.017)、WHO分级( HR＝12.106，95% CI：6.521～22.474， P<0.001)及放疗( HR＝0.502，95% CI：0.325～0.775， P＝0.002)为胶质瘤患者生存期的独立影响因素(均 P<0.05)，可独立用于患者的预后判断。然而，PDK1蛋白表达对并不影响脑胶质瘤患者预后。 结论:不同WHO分级的脑胶质瘤患者PDK1的表达不同，且主要与缺氧和HIF-1信号通路密切相关。

abstractsObjective:To investigate the molecular and clinical features of PDK1 (pyruvate dehydrogenase kinase 1) in gliomas based on the Cancer Genome Atlas (TCGA) by bioinformatics analysis.Methods:The clinical data of 641 glioma patients in TCGA database were retrospectively analyzed. The online analysis website of GEPIA (http: //gepia.cancer-pku.cn/) was used to analyze PDK1 expression in glioma patients. Pearson correlation analysis was performed to create a gene list related to PDK1. The online analysis tool STRING was used to analyze the interaction network of the differential genes. According to the degree of confidence, the top key proteins were screened. The function of PDK1-related differential genes and its relationship with the kyoto Encyclopedia of Genes and Genomes (KEGG) pathway were explored by gene ontology (GO) analysis. According to PDK1 protein expression, all patients were divided into PDK1 protein low expression group (380 cases) and PDK1 protein high expression group (261 cases). Kaplan-Meier survival curves were used to compare the survival differences between the two groups of patients. Univariate and multivariate Cox regression analysis were used to determine the effect of PDK1 protein expression on the prognosis of patients with glioma. Results:The expression of PDK1 protein in 641 glioma patients in TCGA database showed that the expression level in the World Health Organization(WHO) Ⅳ glioma (8.86±0.90) was higher than that in grade Ⅱ (7.24±0.60) and grade Ⅲ (7.34±0.80)( P<0.05), and the difference between expression levels in grade Ⅱ and grade Ⅲ glioma patients was not statistically significant( P>0.05). The cumulative survival rate of glioma patients in the PDK1 high-expression group was significantly lower than that in the PDK1 low-expression group ( P<0.05). Pearson correlation test analysis revealed that 96 genes were closely related to PDK1 expression, of which 69 were positively correlated and 27 were negatively correlated. GO analysis and KEGG pathway analysis showed that PDK1 participated in many important biological processes, which were closely related to the hypoxia microenvironment, reaction to organic cyclic compounds, reaction to organic nitrogen compounds, response to cAMP and the HIF-1 signaling pathway. Multivariate analysis showed that age ( HR=1.579, 95% CI: 1.085-2.299, P=0.017), WHO grade ( HR=12.106, 95% CI: 6.521-22.474, P<0.001) and radiotherapy ( HR=0.502, 95% CI: 0.325-0.775, P=0.002) were independent influencing factors for the survival of glioma patients (all P <0.05) and could be used independently to explore the prognosis of patients.However, PDK1 protein expression does not affect the prognosis of glioma patients. Conclusion:PDK1 expression is different in glioma patients with different WHO pathological levels, and mainly related to hypoxia and HIF-1 signaling pathway.