基于GEO数据库构建肝细胞癌预后模型

Construction of a prognostic model of hepatocellular carcinoma based on GEO datebase

目的基于基因表达数据库(GEO)筛选肝细胞癌(HCC)差异表达基因,探索HCC发生、发展的通路及分子机制,构建肝细胞癌预后模型。方法筛选GEO数据库中的HCC基因表达芯片“GSE76427”,利用R语言的“limma”“clusterprofiler”包筛选HCC差异基因并行京都基因和基因组百科全书(KEGG)分析,用“survival”包行单因素COX分析并用“glmnet”包构建最小绝对收缩和选择算法(LASSO)惩罚的回归模型建立预后模型;用“survival”“survminer”“timeROC”包构建风险评分的受试者工作特征曲线(ROC);用“RMS”包绘制预后相关列线图;用癌症基因组图谱(TCGA)数据库相关数据验证。结果得到425个HCC差异基因及15个HCC预后相关基因(P<0.05)。KEGG分析发现,差异基因主要富集在化学致癌、色氨酸代谢、脂肪酸降解等通路。LASSO回归得到8个关键基因,并建立预后模型。低风险组总体生存(OS)显著高于高风险组(P=0.00097);TCGA验证队列结果显示,低风险组OS显著高于高风险组(P=0.021)。结论本研究建立8个关键基因的预后模型可有效预测HCC患者预后风险,为HCC患者的临床决策和个性化治疗提供理论基础。

Objective To screen differentially expressed genes(DEGs)in hepatocellular carcinoma(HCC)based on GEO database and explore pathways and mechanisms of the occurrence and development of HCC and contruct a prognostic model for HCC patients.Methods The HCC gene expression chip"GSE76427"was screened in GEO database,DEGs were screened by the"limma"package of the R language between HCC and liver tissue,KEGG analysis of DEGs was performed by using the"cluster profiler"package.Univariate COX analysis of DEGs was performed by using"survival"package and the regression model of LASSO punish-ment was constructed with"glmnet"package to establish the prognostic model.ROC curves of risk score were constructed by using"survival","survminer"and"timeROC"packages.The nomogram was drawn by using RMS package.And we used TCGA database related data to validate the prognostic model.Results The total of 425 DEGs of HCC were identified from HCC and 15 prognostic-related genes of HCC were identified(P<0.05).KEGG analysis results of differentially genes showed that DEGs were mainly enriched in the pathways of chemical carcinogenesis,tryptophan metabolism,fatty acid degradation etc.The LASSO regression analy-sis showed that eight key genes,were identified to construct the prognostic model.The overall survival(OS)of the low-risk group was significantly higher than that of the high-risk group(P=0.00097).The OS of the low-risk group was significantly higher than that of the high-risk group(P=0.021)by using the TCGA data-base.Conclusion We found that the prognostic model based on eight key genes can accurately predict the survival risk for HCC patients.It can give basis theory for clinical decision and individualized treatment for HCC patients.