转录因子对结肠癌预后模型的构建  

Construction of a prognostic model of transcription factors for colon cancer

作　　者：屈超 陈子璐 徐正水 赵晨野 叶长春 林文浩 郑见宝[1] 余钧辉[1] 赵伟[1] 孙学军[1] Qu Chao;Chen Zilu;Xu Zhengshui;Zhao Chengye;Ye Changchun;Lin Wenhao;Zheng Jianbao;Yu Junhui;Zhao Wei;Sun Xuejun(Department of General Surgery,the First Affiliated Hospital of Xi’an Jiaotong University,Xi’an 710061,China)

机构地区：[1]西安交通大学第一附属医院普通外科,西安710061

出　　处：《中华内分泌外科杂志》2022年第3期303-308,共6页Chinese Journal of Endocrine Surgery

基　　金：国家自然科学基金(81972720)。

摘　　要：目的探索转录因子(transcription factors,TFs)和结肠癌预后的联系,通过TCGA和GEO双数据库构建预后模型,从而量化患者的风险并指导临床治疗决策。方法本研究运用TCGA和GEO数据库中结肠癌的转录组和临床数据,先将转录组数据进行基因注释并计算基因表达量,对TCGA和GEO中TFs行差异性分析(|log2FC|>1,P-Value(Fdr)<0.05)。取双数据交集的差异TFs行相关预后分析(P<0.01)。利用COX多因素分析计算出预后相关TFs的风险系数和其风险值,应用"survival"和"glmnet"包进行COX模型构建TFs预后模型。绘制出序列集和验证集的生存曲线(P<0.001)及ROC曲线(AUC>0.75),对风险值的分布进行可视化。按风险值分组后计算GSEA富集分析,构建基因集网格,进行靶基因预测,最后进行GO和KEGG的通路富集分析。结果取TCGA和GEO数据库两者交集的387个表达差异的TFs绘制热图,火山图及TFs相关的森林图,按照COX多因素分析构建出结肠癌预后模型=0.310×HSF4+0.137×IRX3-0.127×ATOH1+0.290×OVOL3+0.137×HOXC6+0.155×SIX2+0.092×ZNF556-0.444×CXXC5+0.429×TIGD1+0.413×TCF7L1。通过富集分析,结果显示这些预后因子可能直接或间接作用于癌通路,如基础细胞癌与癌症信号通路、局部组织与细胞黏附及细胞外基质等。结论构建的TFs对结肠癌预后模型可量化结肠癌预后风险,其高风险组是结肠癌预后的独立风险因素,该模型是一种评估结肠癌预后的新方式。Objective To investigate the relationship between transcription factors(TFs)and the prognosis of colon cancer,and to construct a prognosis model through TCGA and GEO dual databases,so as to quantify the risk of patients and guide clinical treatment decisions.Methods The transcriptome and clinical data of colon cancer in TCGA and GEO databases were used in this study.The transcriptome data were annotated and the gene expression was calculated.The difference analysis of TFs in TCGA and GEO(log2FC>1,P-value(Fdr)<0.05)was performed.The difference TFs of double data intersection were used for correlation prognosis analysis(P<0.01).The risk coefficient and risk value of prognosis-related TFs were calculated by COX multivariate analysis,and the prognosis model of TFs was constructed by COX model with"survival"and"glmnet"package.The survival curve(P<0.001)and ROC curve(AUC>0.75)of the sequence set and verification set were drawn,and the distribution of risk value was visualized.After grouping according to risk value,GSEA enrichment analysis was calculated,gene set grid was constructed,target genes were predicted,and finally,pathway enrichment analysis of GO and KEGG was carried out.Results 387 TFs with different expressions in TCGA and GEO databases were used to draw heat map,volcanic map and TFs-related forest map,and the prognosis model of colon cancer was constructed according to COX multivariate analysis=0.310×HSF4+0.137×IRX3-0.127×ATOH1+0.290×OVOL3+0.137×HOXC6+0.155×SIX2+0.092×ZNF556-0.444×CXXC5+0.429×TIGD1+0.413×TCF7L1.Through enrichment analysis,our results showed that these prognostic factors may directly or indirectly act on cancer pathways,such as basic cell carcinoma and cancer signaling pathway,local tissue-cell adhesion,and extracellular matrix.Conclusions The constructed TFs prognosis model of colon cancer can quantify the prognostic risk of colon cancer,and its high-risk group is an independent risk factor of colon cancer prognosis.This model is a new way to evaluate the prognosis of colon cancer

关 键 词：结肠癌 转录因子 预后模型 

分 类 号：R574.62[医药卫生—消化系统]