肺腺癌免疫相关基因预后模型的构建与应用

Construction and application of an immune-related gene prognostic model for lung adenocarcinoma

[目的]基于癌症基因图谱(TCGA)数据库和生物信息学方法构建免疫相关基因的肺腺癌患者预后模型。[方法]基于TCGA和IMMPORT数据库对肺腺癌中免疫相关基因进行LASSO-Cox分析构建模型,使用生存分析、ROC曲线分析和C指数(C-index)分析评估模型的预测性能,并在cBioPortal数据库下载的数据集中完成模型的外部验证。[结果]基于差异分析和LASSO-Cox筛选,一共得到26个预后免疫相关基因。训练集和验证集中的生存曲线都显示所构建的风险模型能够显著区分患者预后,高风险组患者预后更差。ROC曲线显示风险模型在训练集中预测1、3、5年生存率的曲线下面积(AUC)分别为0.752、0.772、0.736,在验证集中预测1、3、5年生存率的曲线下面积(AUC)分别为0.536、0.692、0.705,模型的C指数为0.715,显著高于第8版AJCC分期系统的C指数0.644。[结论]该模型的C指数显著高于第8版AJCC分期的C指数,能够较准确地预测肺腺癌患者的预后,可能是肺腺癌潜在的预后生物标志物。

[Objective]To construct a prognostic model of lung adenocarcinoma patients with immune-related genes based on the Cancer Tissue Gene Atlas(TCGA)database and bioinformatics methods.[Method]The model was constructed based on LASSO-Cox analysis of immune-related genes in lung adenocarcinoma based on TCGA and IMMPORT databases,and the predictive performance of the model was evaluated using survival analysis,ROC curve analysis and C-index analysis,and external validation of the model was completed in a dataset downloaded from the cBioPortal database.[Result]A total of 26 prognostic immune-related genes were obtained based on differential expressed analysis and LASSO-Cox screening.The survival curves in both the training and validation sets showed that the constructed risk model was able to significantly differentiate the prognosis of patients,with patients in the high-risk group having a worse prognosis.The ROC curves showed that the area under the curve(AUC)of the risk model predicting 1-,3-,and 5-year survival in the training set was 0.752,0.772,and 0.736,respectively,and the area under the curve predicting 1-,3-,and 5-year survival in the validation set(AUC)were 0.536,0.692,and 0.705,respectively,and the C-index of the model was 0.715,which was significantly higher than the C-index of 0.644 for the 8th edition of the AJCC staging system.[Conclusion]The C-index of this model was significantly higher than that of the 8th edition of AJCC staging,which can predict the prognosis of lung adenocarcinoma patients more accurately and may be a potential prognostic biomarker for lung adenocarcinoma.