整合多芯片数据的肺癌诊断关键基因的发现与初步验证  

Development and preliminary verification of lung cancer diagnostic marker genes based on the joint analysis of multiple chips

作　　者：王立鹏 李义慧[2] 李唯源 牛超 李玉凤 王建功[2] WANG Lipeng;LI Yihui;LI Weiyuan;NIU Chao;LI Yufeng;WANG Jiangong(School of Clinical Medicine,North China University of Science and Technology,Hebei 063000;The Eighth Department of Radiation and Chemotherapy,Tangshan People's Hospital;The First Department of Oncology,Tangshan Central Hospital;Department of Infection,Zunhua People's Hospital;Hebei Provincial Key Laboratory of Molecular Oncology;Tangshan Key Laboratory of Cancer Prevention and Treatment;Cancer Research Institute of Tangshan People's Hospital)

机构地区：[1]华北理工大学临床医学院,河北063000 [2]唐山市人民医院放化八科 [3]唐山中心医院肿瘤一科 [4]遵化市人民医院感染科 [5]河北省分子肿瘤学重点实验室 [6]唐山市肿瘤防治重点实验室 [7]唐山市人民医院肿瘤研究所

出　　处：《南通大学学报（医学版）》2024年第4期307-312,共6页Journal of Nantong University(Medical sciences)

基　　金：河北省自然科学基金资助项目(H2024105019);唐山市科技计划项目(24150220C)

摘　　要：目的:通过GEO多芯片联合分析筛选出一组与肺癌发生密切相关的基因,作为预测肺癌的关键标志基因并进行初步验证。方法:从GEO数据库下载GSE89047、GSE108055与GSE116959肺癌表达数据集并进行合并,采用R语言中sva程序包ComBat矫正批次效应,limma程序包进行基因差异表达分析从中筛选出肺癌差异表达基因。利用String数据库结合Cytoscape 3.8.2软件构建差异表达基因蛋白质相互作用网络,并分析核心基因。运用ROC方法验证肺癌差异基因、核心基因对肺癌诊断的预测作用。通过TIMER数据库分析GPM6A基因表达及拷贝数变异与免疫细胞浸润的关系。结果:基于GEO数据库GSE89047、GSE108055与GSE116959肺癌表达数据集多芯片联合分析,筛选得到938个肺癌组织与正常肺组织间差异表达基因,以矫正的P值排序,TOP 10差异基因为GPM6A、WNT3A、SLC6A4、TMEM100、TCF21、BTNL9、HSPA12B、LIMS2、VGLL3和ITLN2。String数据库结合Cytoscape 3.8.2软件分析所得10个核心基因为CCNA2、CCNB1、CENPE、FOXM1、ITGAM、KIF11、KIF20A、KIF23、KIF2C和MMP9。ROC分析显示GPM6A的AUC(95%CI)为0.948(0.874~0.986);TOP10差异基因的AUC(95%CI)为0.961(0.886~0.992);10个核心基因的AUC(95%CI)为0.830(0.722~0.895),表明这些标志基因具有较好的肺癌预测能力。TIMER分析结果显示:肺腺癌及肺鳞癌中GPM6A表达均与巨噬细胞浸润相关性最高(肺腺癌:r=0.347,P<0.001;肺鳞癌:r=0.425,P<0.001),GPM6A基因拷贝数变异在肺腺癌中与B细胞、CD4+T细胞、巨噬细胞、中性粒细胞和树突状细胞的免疫浸润相关(均P<0.05),GPM6A基因拷贝数变异在肺鳞癌中与B细胞、CD8+T细胞、CD4+T细胞、巨噬细胞、中性粒细胞和树突状细胞的免疫浸润均具有较高的相关性(均P<0.05)。结论:通过多芯片联合分析初步开发、验证了对肺癌诊断具有较好预测能力的标志基因,并发现差异最显著的标志基因GPM6A与免疫细胞浸润关系密切。Objective:To screen out a group of genes closely related to the occurrence of lung cancer through GEO multichip combined analysis,as a key marker gene for predicting lung cancer,and conduct preliminary verification.Methods:Download the GSE89047,GSE108055 and GSE116959 lung cancer expression datasets from GEO database and merge them.The sva program package ComBat in the R language corrects the batch effect,and the limma program package performs gene differential expression analysis to screen out lung cancer differentially expressed genes.String database combined with Cytoscape 3.8.2 software to construct a differentially expressed gene protein-protein interaction network and analyze core genes.The ROC method was used to verify the predictive effect of lung cancer differential genes and core genes on the diagnosis of lung cancer.TIMER database was used to analyze the relationship between GPM6A gene expression and copy number variation and immune cell infiltration.Results:Based on the multi-chip combined analysis of the GEO database GSE89047,GSE108055 and GSE116959 lung cancer expression datasets,938 differentially expressed genes between lung cancer tissues and normal lung tissues were screened and sorted by the corrected P value.The TOP 10 differential genes were GPM6A,WNT3A,SLC6A4,TMEM100,TCF21,BTNL9,HSPA12B,LIMS2,VGLL3 and ITLN2.The 10 core genes analyzed by String database combined with Cytoscape 3.8.2 software are CCNA2,CCNB1,CENPE,FOXM1,ITGAM,KIF11,KIF20A,KIF23,KIF2C and MMP9.ROC analysis showed that the AUC(95%CI)of GPM6A was 0.948(0.874-0.986);the AUC(95%CI)of the TOP10 differential genes was 0.961(0.886-0.992);the AUC(95%CI)of the 10 core genes was 0.830(0.722-0.895),indicating that the marker genes selected in this study have good lung cancer prediction ability.TIMER analysis showed that GPM6A expression correlated highest with macrophage infiltration in both lung adenocarcinoma and lung squamous carcinoma(lung adenocarcinoma:r=0.347,P<0.001;lung squamous carcinoma:r=0.425,P<0.001),GPM6A gene copy number

关 键 词：肺癌 多芯片联合 预测 开发 验证 免疫浸润 

分 类 号：R734.2[医药卫生—肿瘤]