肺腺癌干性相关分子预后预测模型的构建及验证  

Construction and validation of a stemness-related molecular model for predicting the prognosis of lung adenocarcinoma

作　　者：蔡杰 蔡剑桥 朱余明[1] 赵晓刚[1] CAI Jie;CAI Jianqiao;ZHU Yuming;ZHAO Xiaogang(Department of Thoracic Surgery,Shanghai Pulmonary Hospital,School of Medicine,Tongji University,Shanghai 200433,China)

机构地区：[1]同济大学附属上海市肺科医院胸外科,上海200433

出　　处：《同济大学学报（医学版）》2022年第4期495-502,共8页Journal of Tongji University(Medical Science)

基　　金：上海市“科技创新行动计划”启明星项目扬帆专项(22YF1437600)。

摘　　要：目的构建肺腺癌干性相关分子预后预测LASSO模型并验证其有效性。方法应用R语言包“DESeq2”筛选TCGA肺腺癌癌旁间差异表达基因,R语言包“WGCNA”分析差异表达基因和干性指数mRNAsi相关性以筛选中枢干性相关差异表达基因,R语言包“glmnet”构建肺腺癌干性相关分子预后预测LASSO模型,R语言包“rms”构建基于该LASSO风险模型和各临床病理特征的列线图。结果共筛选得8010个差异表达基因和1437个中枢干性相关差异表达基因。当λ=0.03879,获得20个带有非零风险系数的基因组成最优干性LASSO风险模型。Kaplan-Meier曲线和ROC曲线结果表明该模型在试验组、内部验证组和外部验证组(GSE72094)中均能有效预测肺腺癌预后。相比单纯临床病理指标列线图(C指数=0.668,95%CI:0.6080.727),基于该风险模型的列线图(C指数=0.803,95%CI:0.7650.841)预测效率显著提高(P<0.001)。临床决策曲线进一步表明基于该风险模型的列线图可能给肺腺癌患者带来更高的临床获益度。结论本研究成功构建的肺腺癌干性相关分子预后预测LASSO模型,可能是对目前肺腺癌TNM临床分期的有效补充。Objective To construct and verify a stemness-related LASSO model for predicting the prognosis of lung adenocarcinoma.Methods The R language package“DESeq2”was used to screen the differentially expressed genes between lung adenocarcinoma and adjacent tissues based on TCGA database,and the R package“WGCNA”was used to analyze the correlation between differential expression genes and stemness index mRNAsi for screening the differentially expressed stemness-related genes.The R package“glmnet”was used to construct a stemness-related LASSO(Least Absolute Shrinkage and Selection Operator)model for predicting the prognosis of lung adenocarcinoma,and the R package“rms”was used to construct a nomogram based on the LASSO risk model and various clinicopathological characteristics.Results A total of 8010 differentially expressed genes and 1437 hub stem-related differentially expressed genes were obtained from the screening.Usingλ=0.03879,we obtained the optimal stemness-related LASSO risk model composed of 20 genes with non-zero risk coefficients.The Kaplan-Meier curve and ROC curve results showed that the model effectively predicted the prognosis of lung adenocarcinoma in the development cohort,internal validation cohort and external validation cohort(GSE72094).Compared with the simple clinicopathological nomogram(C index=0.668,95%CI:0.6080.727),the prediction efficiency of the nomogram based on this risk model(C index=0.803,95%CI:0.7650.841)was significantly improved(P<0.001).The clinical decision curve further indicated that the nomogram based on this risk model might bring higher clinical benefit to patients with lung adenocarcinoma.Conclusion The stemness-related LASSO model for predicting prognosis of lung adenocarcinoma has been successfully constructed in this study,which may be an effective supplement to the current TNM clinical staging of lung adenocarcinoma.

关 键 词：肺腺癌 干性相关差异表达基因 LASSO模型 列线图 

分 类 号：R734.2[医药卫生—肿瘤]