肺腺癌代谢相关基因的识别和预后模型构建

Construction of prognostic model based on metabolism-related gene in lung adenocarcinoma

目的运用生物信息学分析肺腺癌代谢发生发展的关键基因并构建预后模型。方法利用TCGA、KEGG、TCGA数据库筛选肺腺癌的基因表达、代谢相关基因、肺腺癌代谢相关的基因集和临床信息。采用Cox回归分析,lasso回归降维筛选预后相关基因。Cox回归构建肺腺癌风险模型,高风险与低风险组。结果成功筛选出4个基因,LDHA、GAPDH、GNPNAT1和HACD1。高风险组预后显著差于低风险组(P<0.01),根据生存时间受试者工作特征曲线,1年、2年、3年曲线下面积分别为0.64、0.637、0.645。实验和验证数据集均提示该预后模型有较好的预测能力。结论筛选4个关键基因与肺腺癌代谢相关,构建的风险模型对于肺腺癌治疗及预后判断提供依据,为肺腺癌的精准化治疗提供新思路。

Objective To analyze the key genes in the development of lung adenocarcinoma by bioinformatics and to establish a prognosis model of the disease.Methods TCGA,KEGG and TCGA database were used to screen the gene expression,metabolism-related genes,gene sets and clinical information related to lung adenocarcinoma metabolism.Cox regression analysis,lasso regression and dimensionality reduction were used to screen prognostic-related genes.Cox regression to construct lung adenocarcinoma risk models,high-risk and low-risk groups.Results Four genes,LDHA,GAPDH,GNPNAT1 and HACD1 were successfully screened.High-risk group had worse prognosisthan low-risk group(P<0.01).According to ROC of survival time,the areas under 1-year,2-year,and 3-year curves were respectively 0.64,0.637,and 0.645.Both experiment and validation data set indicated that the prognostic model possessed good prediction ability.Conclusions The screen confirmedfour key genes related to lung adenocarcinoma metabolism,which constructed risk model as a basis for the treatment and prognosis of lung adenocarcinoma,and laid new assumption for the precise treatment of lung adenocarcinoma.