基于凋亡相关长非编码RNA的肺腺癌预后风险模型的构建与验证

Establishment and validation of a prognostic signature for lung adenocarcinoma based on apoptosis-related long non-coding RNAs

目的检测凋亡相关长非编码RNA(lncRNA)对肺腺癌(LUAD)预后的预测能力, 并探讨其分子机制。方法从癌症基因组图谱(TCGA)数据库和高通量基因表达数据库(GEO)分别下载535例和125例LUAD患者转录组和临床信息, 筛选出与161个凋亡基因高度相关的lncRNA。通过Cox回归和lasso分析获取预后相关lncRNA并构建预测模型, 并验证模型的稳定性。通过肿瘤突变负荷(TMB)、药物敏感性、免疫细胞相关性和基因集变异分析探讨该模型可能的生物学途径及潜在的分子机制。结果研究构建一个由19个凋亡相关的lncRNA组成的LUAD预后预测模型。通过模型风险评分可将LUAD患者分为高低风险组, 高评分与患者总生存率呈负相关(P<0.05)。训练集和测试集中1、3、5年的受试者工作特征曲线下面积(AUC)均>0.7。该评分与患者的临床指标如总分期[Ⅰ/Ⅱ/Ⅲ+Ⅳ分别为-0.069(-0.256, 0.169)比0.063(-0.156, 0.309)比0.099(-0.099, 0.463), H=16.9, P<0.01]、T分期[T1/T2/T3+T4分别为-0.083(-0.389, 0.130)比0.037(-0.165, 0.282)比0.180(-0.066, 0.392), H=20.8, P<0.01]和N分期[N0/N1/N2+N3分别为-0.055(-0.252, 0.201)比0.043(-0.143, 0.357)比0.119(-0.064, 0.457), H=13.1, P<0.01]成正相关, 可作为LUAD的独立预后因素(风险比=3.402, 95%可信区间:2.338~4.952, P<0.01)。高评分组患者对Gefitinib的药物敏感性高于低评分组[2.168(2.101, 2.239)比2.204(2.153, 2.251), Z=11.60, P<0.01], 同时对Dasatinib[0.989(0.353, 2.056)比1.463(0.809, 2.402), Z=11.60, P<0.01]以及Imatinib[4.925(4.873, 4.971)比4.937(4.892, 4.972), Z=10.60, P<0.05]的药物敏感性也高于低评分组。评分影响肿瘤免疫细胞含量。高风险组TMB高于低风险组[6.105(2.395, 12.421)比4.342(1.579, 10.237), Z=8.41, P<0.05], 且主要富集glycolysis, myc targets v1和mtorc1 signaling等信号通路。结论建立基于凋亡相关lncRNA的LUAD预后模型, 并验证了该模型的性能。

Objective To examine the predictive ability of apoptosis-related long non-coding RNAs(lncRNA)in the prognosis of lung adenocarcinoma(LUAD)and to explore its molecular mechanism.Methods The transcriptome profiling and clinical information of LUAD patients were retrieved from The Cancer Genome Atlas database(535 cases)and Gene Expression Omnibus(125 cases)respectively.The apoptosis-related lncRNAs were identified.Then,a prognostic signature was developed by Cox regression and lasso regression analysis based on the apoptosis-related lncRNAs.Survival analysis was used to validate the robustness of the model internally and externally.In addition,possible biological functions and potential molecular mechanisms were explored by tumor mutational burden,drug sensitivity,immune cell infiltration and gene set variation analysis.Results A prognostic signature of LUAD consisting of 19 apoptosis-related lncRNA was established.Patients with LUAD could be stratified into high-risk group and low-risk group according to risk score of the signature.And high-risk score was significantly correlated with poor prognosis of patients(P<0.05).Area under receiver operating characteristic(ROC)curve(AUC)of 1,3,and 5 years was more than 0.7 for training and testing set.The risk score was positively correlated with clinical indicators such as total stage[Ⅰ/Ⅱ/Ⅲ+Ⅳwere-0.069(-0.256,0.169)/0.063(-0.156,0.309)/0.099(-0.099,0.463),H=16.9,P<0.01],T stage[T1/T2/T3+T4 were-0.083(-0.389,0.130)/0.037(-0.165,0.282)/0.180(-0.066,0.392),H=20.8,P<0.01]and N stage[N0/N1/N2+N3 were-0.055(-0.252,0.201)/0.043(-0.143,0.357)/0.119(-0.064,0.457),H=13.1,P<0.01],and could be identified as an independent prognostic factor for LUAD(hazard ratio=3.402,95%confidence interval,2.338-4.952,P<0.01).Risk score significantly affected the abundance of infiltrating immune cells.The drug sensitivity of patients in high risk group was significantly higher than that in low risk group,such as Gefitinib[2.168(2.101,2.239)versus 2.204(2.153,2.251),Z=11.60,P<0.01],Dasatinib[0.989(0.353,2.056)versus 1.463(0.809,2.402),Z=11.60,P<0.01]and Imatinib[4.925(4.873,4.971)versus 4.937(4.892,4.972),Z=10.60,P<0.05].The tumor mutational burden was significantly higher in the high-risk group than in the low-risk group[6.105(2.395,12.421)versus 4.342(1.579,10.237),Z=8.41,P<0.05].Gene set variation analysis suggested that glycolysis,myc targets v1,and mtorc1 signaling pathways were screened as the mostly enriched pathways in the high-risk group.Conclusion A reliable prognostic model based on apoptosis-related lncRNA for LUAD was identified and validated for its performance.