基于糖酵解风险基因老年结直肠癌患者预后风险模型构建  

Construction of A Prognostic Risk Model for Elderly Colorectal Cancer Patients Based on Glycolysis Risk Genes

作　　者：何耀明[1] 李尚仁[1] 李晓平[2] 王成兴 谭群英 梁伟俊 周朝荣 赵景林[1] HE Yaoming;LI Shangren;LI Xiaoping;WANG Chengxing;TAN Qunying;LIANG Weijun;ZHOU Chaorong;ZHAO Jinglin(Gastrointestinal Surgery,Jiangmen Guangdong 529000,China;Breast Surgery, Jiangmen Central Hospital, Jiangmen Guangdong 529000, China)

机构地区：[1]江门市中心医院胃肠外科,广东江门529000 [2]江门市中心医院乳腺外科,广东江门529000

出　　处：《河南医学高等专科学校学报》2022年第3期259-266,共8页Journal of Henan Medical College

基　　金：广东省医学科学技术研究基金项目(B2021057);江门市基础与应用基础研究类重点项目(2019030102420012926);江门市医疗卫生领域科技计划项目(2019020200050000723)。

摘　　要：目的根据糖酵解相关风险基因建立老年人结直肠癌预后模型,为高危老年结直肠癌患者进行提早干预提供依据。方法从TCGA数据库、MsigDB数据库获得老年结直肠癌患者转录组数据、临床资料和糖酵解相关基因集。通过GSEA分析,获得差异富集糖酵解相关基因集,使用edgeR包进一步对基因集内糖酵解基因进行结直肠癌肿瘤组织和正常组织的基因差异分析,获得糖酵解相关差异表达基因。使用单因素COX回归生存分析,LASSO回归模型分析,筛选出风险基因,构建风险分数模型,通过受试者工作特征(ROC)曲线分析,计算曲线下面积(AUC)评价风险模型效能,通过GEO数据(GSE17536)外验证检验风险分数模型效能。最后结合患者风险得分和淋巴管浸润、脉管浸润、T分期、N分期、M分期等临床指标构建列阵图,ROC曲线进行预测效能验证效能。结果从TCGA老年结直肠癌患者数据库中获得340例肿瘤样品和32例正常组织样品,MsigDB数据库获得糖酵解相关基因集9个,通过GSEA富集分析结果发现,HALLMARK_GLYCOLYSIS和REACTOME_GLYCOLYSIS两个糖酵解相关基因集差异富集显著,从中提取糖酵解相关基因254个,进一步使用edgeR包分析获得糖酵解差异基因46个,其中上调基因37个,下调基因9个。单因素COX回归生存分析,LASSO回归模型分析确定10个与预后相关的糖酵解相关差异表达基因。根据该10个风险基因构建风险分数模型,以风险分数中位值作为临界值将老年结直肠癌患者分为高风险组和低风险组,低风险组的总体生存时间长于高风险组(P<0.001),1、3、5 a生存的AUC值分别为0.705、0.730、0.674,GSE17536模型外验证提示,低风险组患者总体生存期长于高风险组(P=0.023)。结合风险分数和临床指标的预测模型列阵图C-index=0.730,ROC曲线评价提示模型预测1、3、5 a生存率的AUC值分别为0.766、0.742、0.783。结论基于糖酵解相关风险基因建立�Objective To establish a prognostic risk model of colorectal cancer based on glycolysis-related risk genes to provide evidence for early intervention in high-risk elderly patients with colorectal cancer.Methods Transcriptomic data,clinical information and glycolysis-related gene sets of elderly colorectal cancer patients were obtained from TCGA database and MsigDB database.We obtained the differentially enriched glycolysis-related gene sets by GSEA analysis,and further analyzed the glycolysis-related genes within these gene sets by using the edgeR package to get the differentially expressed glycolysis-related genes(DEGRGs)between tumor tissues and normal tissues.Using univariate COX regression and LASSO regression model,we screened out the prognosis related DEGRGs.A risk-scoring model based on DEGRGs related to prognosis was constructed.Receiver operating characteristic(ROC)analysis and calculation of the area under the curve(AUC)were used to evaluate the performance of the model.The model was verified using an external dataset(GSE17536).Finally,combined with the patient’s risk score and clinical indicators such as lymphatic invasion,vascular invasion,T stage,N stage,and M stage,an array chart was constructed,and the ROC curve was used to verify the predictive efficacy.Results 340 tumor samples and 32 normal tissue samples were obtained from the TCGA elderly colorectal cancer patient database,and 9 glycolysis-related gene sets were obtained from the MsigDB database.The results of GSEA enrichment analysis showed that the HALLMARK_GLYCOLYSIS and REACTOME_GLYCOLYSIS gene sets were significantly enriched differently.254 glycolysis-related genes were extracted from it,and 46 differentially glycolysis genes were obtained by further analysis using the edgeR package,including 37 up-regulated genes and 9 down-regulated genes.Univariate COX regression and LASSO regression indicated that 10 of these genes were related to prognosis.A risk score model was constructed based on the 10 risk genes,and the median risk score was

关 键 词：结直肠癌 老年人 糖酵解 列阵图 

分 类 号：R735.3[医药卫生—肿瘤]