生物信息学分析甲状腺癌免疫基因构建预后评估模型及对免疫细胞浸润的影响  

Construction of a prognostic evaluation model based on bioinformatics analysis of immune genes in thyroid cancer and its influence on immune cell infiltration

作　　者：余幼林 余军林 沈雄山[1] 胡超华[1] 卢婷 张文杰[1] 杨青青 刘程浩 杨峰 Yu Youlin;Yu Junlin;Shen Xiongshan;Hu Chaohua;Lu Ting;Zhang Wenjie;Yang Qingqing;Liu Chenhao;Yang Feng(Department of Thyroid and Breast Surgery,Xiaogan Hospital Affiliated Wuhan University of Science and Technology,Xiaogan 432000,China;Department of Oncology,Xiaogan Hospital Affiliated Wuhan University of Science and Technology,Xiaogan 432000,China;Department of Clinical Medicine,Graduate School,Jinzhou Medical University,Jinzhou 121000,China)

机构地区：[1]武汉科技大学附属孝感医院甲状腺乳腺外科,湖北孝感432000 [2]武汉科技大学附属孝感医院肿瘤科,湖北孝感432000 [3]锦州医科大学研究生院临床系,辽宁锦州121000

出　　处：《实用肿瘤杂志》2022年第1期50-59,共10页Journal of Practical Oncology

基　　金：湖北省卫生计生委联合基金项目(WJ2018H0098);武汉科技大学附属孝感医院(孝感市中心医院)院级科研项目(201818)。

摘　　要：目的基于生物信息学方法构建甲状腺癌免疫基因预后评估模型及危险分层系统,分析模型评分对免疫细胞浸润的影响和免疫调控网络。方法从癌症基因组图谱(The Cancer Genome Atlas,TCGA)数据库及ImmoPort Resources数据库下载甲状腺癌患者临床、转录组及免疫基因数据,筛选出差异表达的免疫相关基因,从Cistrome Project数据库中下载转录因子数据,构建差异表达的免疫相关基因及差异表达的转录因子之间的调控网络。采用单因素和多因素Cox分析筛选出独立的预后免疫相关基因并构建预后评估模型,分析预后评估模型风险评分与临床病例特征及预后的相关性。从TIMER2.0数据库下载肿瘤相关性免疫细胞浸润数据,分析预后评估模型风险评分与肿瘤相关性免疫细胞(B细胞、CD4;T细胞、CD8;T细胞、中性粒细胞、巨噬细胞及树突状细胞)浸润丰度的相关性。结果基于R语言共筛选出272个在甲状腺癌中差异表达免疫相关基因及36个差异表达的转录因子[错误发现率(false discovery rate,FDR)<0.05]。单因素Cox分析筛选出13个免疫相关基因与预后相关(均P<0.01)。与转录因子相关性分析结果显示,共13个转录因子与11个预后相关的免疫基因相关(均∣r∣>0.3,均P<0.05),并构建调控网络。多因素Cox分析结果显示,6个免疫基因(CXCL5、COLEC10、S100A9、MMP12、APOD和FGF7)为独立预后因子构建预后评估模型(均P<0.05)。基于模型风险评分分为高风险组和低风险组,高风险组和低风险组患者10年总生存(overall survival,OS)率为95.6%和85.4%。该预后评估模型具有较高准确度[曲线下面积(area under curve,AUC)=0.992]。单因素及多因素Cox回归分析结果显示,预后评估模型风险评分是OS的独立预测因子(P<0.05),高风险评分是患者OS不良的危险因素,且与肿瘤微环境中的中性粒细胞增多及CD8;T细胞减少相关。结论基于生物信息学方法构建了甲状Objective To construct an immune gene prognostic evaluation model and risk stratification system for thyroid cancer based on bioinformatics methods, and to analyze the influence of model score on immune cell infiltration and immune regulatory network. Methods The clinical, transcritome and immune gene data of patients with thyroid cancer from The Cancer Genome Atlas(TCGA) database and ImmoPort Resources database were used to screen the differentially expressed immune genes, and transcription factor data from the Cistrome Project database were used to construct a regulatory network between the differentially expressed immune related genes and differentially expressed transcription factors. Univariate and multivariate Cox analysis were used to screen out the independent prognosis immune genes and construct a prognostic evaluation model. The correlation between the risk score of the prognostic evaluation model and the clinical characteristics and prognosis of patients was analyzed. The tumor-related immune cell infiltration data were downloaded from TIMER2.0 database to analyze the correlation between the risk score of the prognostic evaluation model and the abundance of tumor-related immune cells(B cells, CD4;T cells, CD8;T cells, neutrophils, macrophages and dendritic cells). Results A total of 272 differentially expressed immune-related genes and 36 transcription factors in thyroid cancer were screened out by R language [false discovery rate(FDR)<0.05]. Univariate Cox analysis showed that 13 immune-related genes were related to prognosis(all P<0.01). The results of correlation analysis with transcription factors showed that a total of 13 transcription factors were related to 11 prognostic immune genes(all |r|>0.3,all P<0.05), and their regulatory network was constructed. Multivariate Cox analysis showed that 6 immune genes(CXCL5, COLEC10,S100 A9, MMP12, APOD, and FGF7) were independent prognostic factors to construct a prognostic evaluation model(all P<0.05). The patients were divided into the high-and low-risk g

关 键 词：甲状腺癌 预后评估模型 调控网络 生物信息学 

分 类 号：R736.1[医药卫生—肿瘤] Q811.4[医药卫生—临床医学]