基于生物信息学预测甲状腺癌预后相关高风险糖酵解基因及其与患者预后的关系

Prediction of prognosis-related genes with high risk of glycolysis in thyroid cancer and their relation with patients′prognosis:a bioinformatics-based study

目的基于生物信息学预测甲状腺癌预后相关的高风险糖酵解基因,并分析其与患者预后的关系。方法(1)从TCGA数据库中下载甲状腺癌相关的转录组数据及临床数据,包括癌旁样本和肿瘤样本。从基因集富集分析(GSEA)数据库中搜索所有与糖酵解相关的基因集并进行GSEA,筛选出P<0.05的基因集。(2)应用Perl程序语言提取两种样本中上述基因集的表达量,并采用Wilcoxon检验验证表达量差异,筛选出P<0.05的差异性表达基因;对差异性基因的表达量与生存数据进行相关性分析,得到与甲状腺癌预后相关的糖酵解基因。(3)将预后相关糖酵解基因纳入Cox回归模型构建糖酵解预后模型并计算各临床样本患者的风险值及中位值,根据风险值及中位值将患者划分为高、低风险组,绘制出两组的生存曲线图;根据Cox回归分析结果获得各基因的风险比,筛选出甲状腺癌预后相关的高风险糖酵解基因。(4)采用R软件绘制出糖酵解预后模型预测甲状腺癌患者预后的受试者工作特征(ROC)曲线、糖酵解预后模型预测得到的基因的热图,以及所有样本的时间-生存散点图,并进行预后分析。结果共获得501例临床样本(56753个基因)。通过GSEA数据库共搜索到5个与糖酵解相关的基因集,仅有1个基因集的P<0.05。获得与甲状腺癌相关的糖酵解差异性表达基因147个,其中与甲状腺癌预后相关的糖酵解基因48个;构建糖酵解预后模型后得到与甲状腺癌预后相关的糖酵解基因15个,其中高风险基因8个(TGFBI、SLC16A3、PFKP、PPP2CB、PKM、DDIT4、CHPF2、QSOX1)。生存曲线显示,在疾病的早期,随着时间的延长患者的生存率降低,但在疾病后期,随着时间的延长患者的生存率逐渐趋于稳定,高风险组的总体生存率低于低风险组(P<0.05)。ROC曲线结果显示,糖酵解预后模型预测患者预后的曲线下面积为0.783;热图显示,表达量最高的4个基因分别是PKM、PPP2CB、PFKP、DDIT4;时间-生存散点图显示,随着风险值的增加,死亡的患者数增多。结论甲状腺癌糖酵解途径与TGFBI、PKM、DDIT4、SLC16等高风险基因密切相关,抑制高风险基因的表达可能有助于改善甲状腺癌患者的预后。

Objective To predict the prognosis-related genes with high risk of glycolysis in thyroid cancer,and to analyze their relation with patients′prognosis based on bioinformatics.Methods(1)The thyroid cancer-related transcriptome data and clinical data were downloaded from the TCGA database,including paracancerous and neoplastic samples.In the database of the Gene Set Enrichment Analysis(GSEA),all of the gene sets related to glycolysis were searched,and GSEA was conducted to screen out the gene sets which satisfied P<0.05.(2)The expression of the aforesaid gene sets from the two samples were extracted using Perl programming language,and the differences in the expression were validated by employing Wilcoxon test to screen out the differentially expressed genes which satisfied P<0.05;the correlation analysis was conducted on the expression and survival data of differential genes to obtain the glycolysis genes related to the prognosis of thyroid cancer.(3)The prognosis-related glycolysis genes enrolled in the Cox regression model were used to establish a model of glycolysis prognosis,and patients′value-at-risk and median of various clinical samples were calculate;moreover,the patients were assigned to high-or low-risk group according to the value-at-risk and median,drawing the survival curve of the two groups.The risk ratio of various genes was obtained based on the Cox regression analysis results,screening out thyroid cancer prognosis-related genes with high risk of glycolysis.(4)The receiver operating characteristic(ROC)curve of thyroid cancer patients′prognosis predicted by the glycolysis prognosis model,the heat map of the genes predicted by the glycolysis prognosis model,as well as a time-survival scatter diagram of all samples were drawn by R software,and prognostic analysis was conducted.Results A total of 501 clinical samples(56753 genes)were obtained.A total of five gene sets related to glycolysis were retrieved by the GSEA database,with only one gene set which satisfied P<0.05.A total of 147 differentially expressed genes of glycolysis related to thyroid cancer were acquired,thereinto 48 glycolysis genes were related to thyroid cancer prognosis.Fifteen thyroid cancer prognosis-related glycolysis genes were acquired by establishing the glycolysis prognosis model,among which eight of the genes(TGFBI,SLC16A3,PFKP,PPP2CB,PKM,DDIT4,CHPF2,QSOX1)expressed in high risk.The survival curve revealed that the survival rate of patients decreased over time in the early stage of diseases;however,the survival rate of patients tended to be stable over time in the later stage of diseases,and the total survival rate of the high-risk group was lower than that of the low-risk group(P<0.05).The ROC curve results depicted that the area under the curve of patients′prognosis predicted by the glycolysis prognosis model was 0.783.The heat map interpreted that the four genes with the highest expression were PKM,PPP2CB,PFKP,and DDIT4.The time-survival scatter diagram implied that as the value-at-risk increased,the number of patients died increased.Conclusion The glycolytic pathway of thyroid cancer is closely related to the high-risk genes,including TGFBI,PKM,DDIT4,and SLC16.Inhibiting the expression of high-risk genes may be beneficial to ameliorating the prognosis of patients with thyroid cancer.