基于机器学习方法的甲状腺癌预后标志物探索和免疫细胞浸润分析  

Exploration of prognostic biomarkers and immune cell infiltration analysis in thyroid cancer based on machine learning methods

作　　者：王子康 方艳 刘超 WANG Zi-kang;FANG Yan;LIU Chao(Department of General Surgery,the Second Affiliated Hospital of Zhengzhou University,Zhengzhou 450001,China;Department of Hepatobiliary Surgery,the Fifth Affiliated Hospital of Zhengzhou University,Zhengzhou 450001,China;Department of Pharmacy,the Second Affiliated Hospital of Zhengzhou University,Zhengzhou 450001,China)

机构地区：[1]郑州大学第二附属医院普通外科,郑州450001 [2]郑州大学第二附属医院药学部,郑州450001 [3]郑州大学第五附属医院肝胆外科,郑州450001

出　　处：《医药论坛杂志》2025年第5期459-465,共7页Journal of Medical Forum

基　　金：国家自然科学基金(82304615);河南省重点科技攻关项目(242102310239);河南省医学科技攻关计划联合共建项目(LHGJ20230344);河南省医学科技攻关计划省部共建项目(SBGJ202402071)。

摘　　要：目的旨在确定甲状腺癌(thyroid cancer,TC)中潜在的生物标志物,并探讨相关差异基因和免疫微环境在甲状腺癌中的作用和机制。方法正常和TC基因表达谱微阵列芯片来自基因表达综合(gene expression omnibus,GEO)数据库。为了识别具有高度相关甲状腺癌特征的差异表达基因(highly differentially expressed genes,HDEGs),使用了LASSO和机器学习-随机森林方法取交集。TCGA数据库来验证HDEGs表达及预后。应用单样本基因集富集分析(single-sample gene set enrichment analysis,ssGSEA)算法来观察TC的免疫浸润特性及其与目标基因的关系。结果在使用机器学习、LASSO等方法比较甲状腺癌样本与正常甲状腺样本的转录组数据后,发现CFD,RNF150,CCL21,RPS6KA5,TMEM139和DPT在甲状腺癌中均显著下调,CD93则在甲状腺癌中显著上调。RNF150、RPS6KA5、TMEM139和CD93的高表达组预后明显优于低表达组。使用ssGSEA算法,发现TC中7种类型的免疫细胞浸润显著差异,并且与HDEGs具有良好的相关性。结论可能通过诱导免疫炎症来促进甲状腺癌的进展。CFD,RNF150,CCL21,RPS6KA5,TMEM139,CD93和DPT可用作TC的新型诊断生物分子标志物和潜在的治疗靶点。Objective To identify potential biomarkers in thyroid cancer(TC)and investigate the role and mechanisms of differentially expressed genes(DEGs)and the immune microenvironment in thyroid cancer.Methods Gene expression microarray data from normal and TC tissues were obtained from the gene expression omnibus(GEO)database.To identify highly TC-relevant differentially expressed genes(HDEGs),the intersection of genes identified using LASSO regression and the machine learning-based random forest method was analyzed.The expression and prognostic significance of HDEGs were validated using data from the cancer genome atlas(TCGA).Single-sample gene set enrichment analysis(ssGSEA)was applied to assess immune infiltration characteristics in TC and their relationship with the target genes.Results Transcriptomic data comparing TC and normal thyroid samples using machine learning and LASSO methods revealed significant down regulation of CFD,RNF150,CCL21,RPS6KA5,TMEM139,andDPT,aswell as significant upregulation of CD93 in TC.Higher expression levels of RNF150,RPS6KA5,TMEM139,and CD93 were associated with better prognosis compared to lower expression levels.The ssGSEA algorithm identified significant differences in the infiltration of seven types of immune cells in TC,which correlated strongly with the HDEGs.Conclusion Thyroid cancer progression may be promoted by inducing immune inflammation.CFD,RNF150,CCL21,RPS6KA5,TMEM139,CD93,and DPT represent novel diagnostic biomarkers and potential therapeutic targets for TC.

关 键 词：甲状腺癌 免疫浸润 机器学习 生物标志物 

分 类 号：R736.1[医药卫生—肿瘤]