摘要
目的基于高通量基因表达(GEO)数据库结合CT影像预测肺癌临床分期的分子标志物,并建立相应的诊断模型及预测列线图。方法从GEO数据库中获取小结节肺癌患者包括训练队列和验证队列的基因表达谱数据,按临床分期将2个队列分别分为2组:早期(Ⅰ期)组和中晚期(Ⅱ、Ⅲ期和Ⅳ期)组;分析2组基因的差异表达及功能富集;通过单因素分析筛选共差异基因后进行Logistic回归分析和ROC分析,并在训练队列中绘制诊断列线图。结果对训练队列和验证队列的差异分析分别获得161和437个差异基因,并筛选获得7个共差异表达基因(SLC16A14、LHX2、PRAME、ZNF257、SOX2、KCNJ16、GSTA1)。Logistic回归分析显示,高表达的ZNF257和低表达的SOX2、KCNJ16、GSTA1与中晚期肺癌显著相关(均P<0.05),以此四基因构建的模型灵敏度为83.3%,特异度为92.9%,建立的诊断列线图在小结节肺癌的恶性诊断预测中显示出极好的潜力。结论成功预测肺癌临床分期的4个分子标志物(ZNF257、SOX2、KCNJ16、GSTA1);以这4个差异表达基因可建立小结节肺癌诊断模型及诊断预测列线图;该诊断模型具有较好的特异度和灵敏度,列线图具有预测CT筛查小结节肺癌的潜能。
Objective To predict the molecular markers for lung cancer staging based on GEO database combined with CT images,and to establish the corresponding diagnostic model and prediction alignment diagram.Methods The gene expression profile data of patient with small nodular lung cancer were obtained from GEO database,including training and validation cohorts.According to clinical stage,the two cohorts were divided into early stage group(stageⅠ)and middle-advanced stage group(stageⅡ,ⅢandⅣ).The differential expression and functional enrichment of genes were analyzed.After the overlapping differential genes were screened by univariate analysis,logistic regression and ROC analysis were performed and diagnostic alignment diagram was drawn in training cohort.Results A total of 161 and 437 differential genes were obtained by training cohort and validation cohort,including 7 overlapping differential genes(SLC16A14,LHX2,PRAME,ZNF257,SOX2,KCNJ16 and GSTA1).Logistic regression analysis showed that high expression of ZNF257 and low expression of SOX2,KCNJ16 and GSTA1 were significantly associated with advanced lung cancer(P<0.05).The sensitivity and specificity of the model stablished based on the four genes were 83.3%and 92.9%,respectively.Furthermore,the established diagnostic alignment diagram had a great potential in the prediction of small nodular lung cancer.Conclusion Four molecular markers for lung cancer staging(ZNF257,SOX2,KCNJ16 and GSTA1)were successfully predicted.The four differentially expressed genes can be used to establish diagnostic model and diagnostic prediction alignment diagram for small nodular lung cancer.The diagnostic model has good specificity and sensitivity,and the alignment diagram has potential to predict CT screening for small nodular lung cancer.
作者
欧阳锦
罗亭
余石群
黄邵鑫
汪鑫
OU YANG Jin;LUO Ting;YU Shi-qun;HUANG Shao-xin;WANG Xin(Jiangxi Provincial Key Laboratory of Preventive Medicine,Nanchang University,Nanchang 330006,China;Precision Preventive Medicine Laboratory of Basic Medical School,Jiujiang University,Jiujiang 332000,China)
出处
《南昌大学学报(医学版)》
2021年第5期1-7,共7页
Journal of Nanchang University:Medical Sciences
基金
国家自然科学基金(81660541)
江西自然科学基金(20181BAB20506)
中国博士后科学研究基金(2019M652334)。