基于mRNA生物信息学分析的肺腺癌免疫基因预后风险模型建立与评估

Construction and evaluation of prognostic risk model for immune genes in lung adenocarcinoma based on mRNA bioinformatics analysis

目的采用生物信息学方法构建并验证肺腺癌免疫基因预后风险模型,探究该模型对早期肺腺癌患者预后的预测潜力。方法将癌症基因组图谱(The Cancer Genome Atlas,TCGA)来源的肺腺癌及正常组织数据作为训练集,将基因表达综合数据库(Gene Expression Omnibus,GEO)来源的肺腺癌及正常组织数据作为测试集。根据免疫学数据库和分析平台(Immunology Database and Analysis Portal,ImmPort)中提供的免疫相关基因,利用生物信息学手段根据训练集数据建立预后风险模型并在测试集中进行外部验证。采用该模型对38例临床早期肺腺癌患者的转录组数据进行分析,评估高、低危组患者的临床病理参数差异。结果构建了包含12个差异表达免疫基因(CYBB、ARG2、UTS2、LIFR、SHC3、CTLA4、FGF2、SEMA7A、INHA、GPI、ANGPTL4、TNFRSF11A)的肺腺癌预后风险模型;在训练集中,该模型ROC曲线下面积(AUC^(ROC))为0.759;在测试集中,AUC^(ROC)为0.707。对于训练集及测试集中的早期肺腺癌患者,该模型也有良好的预后预测能力。在早期肺腺癌临床样本中,高风险患者与更大的肿瘤直径及更差的病理分型有关。结论该模型在训练集及测试集中都表现出良好的预后预测能力,并对临床早期肺腺癌患者预后有一定的提示作用。这些免疫基因能够为早期肺腺癌诊断、患者预后评估及新的治疗靶点研究提供方向。

Objectives To construct and validate a prognosis risk model for immune genes in lung adenocarcinoma(LUAD)patients by using bioinformatics methods,and explore its predictive potential for the prognosis of the patients with early lung adenocarcinoma.Methods The RNA-seq data and clinical data were downloaded from The Cancer Genome Atlas(TCGA)as the training set,and the RNA-seq data and clinical data were downloaded from Gene Expression Omnibus(Gene Expression Omnibus,GEO)as the test set.According to the immune-related genes from the Immunology Database and Analysis Portal(ImmPort),the prognostic risk model was established on these genes in the training set by the bioinformatics methods,and an external validation was performed by the data in the test set.The transcriptome RNA-seq data of 38 patients with early LUAD was analyzed by this prognostic risk model to evaluate the differences of clinicopathological characteristics between the high-risk and low-risk groups.Results We constructed a prognostic risk model for LUAD containing 12 differentially expressed immune genes(CYBB,ARG2,UTS2,LIFR,SHC3,CTLA4,FGF2,SEMA7A,INHA,GPI,ANGPTL4 and TNFRSF11A).The area under curve(AUC^(ROC))of the ROC curve was 0.759 in the training set,while the AUC of the ROC was 0.707 in the test set.The model also showed good predictive potential for prognosis in the patients with early in both training set and test set.The analysis for the clinical samples of LUAD found that the patients in high risk group were associated with large diameter of tumors and poor pathological classifications.Conclusion The model showed good prediction potentials in both training set and test set,and displayed certain cautionary implications for the prognosis of the patients with early LUAD in clinical practice.These immune genes may become new targets for diagnosis and treatment of LUAD and provide a direction to evaluate the immune state and prognosis of the patients.