子宫内膜异位症关键通路的鉴定和诊断模型的建立  

Identification of key pathways in endometriosis and development of a diagnostic model

作　　者：杨瑾 李圃[2] YANG Jin;LI Pu(Graduate School,Tianjin Medical University,Tianjin 300070;Department of Gynecology,Clinical School of Obstetrics and Gynecology Center,Tianjin Medical University,Tianjin 300199,China)

机构地区：[1]天津医科大学研究生院,天津300070 [2]天津医科大学中心妇产科临床学院妇科,天津300199

出　　处：《临床与病理杂志》2023年第6期1104-1118,共15页Journal of Clinical and Pathological Research

摘　　要：目的:子宫内膜异位症是一种发病率较高、诊断困难且目前病理机制不清的妇科疾病。通过基因集富集分析(gene set enrichment analysis,GSEA)和基因集变异分析(gene set variation analysis,GSVA)鉴定与子宫内膜异位症相关的关键通路和基因,并建立子宫内膜异位症的临床诊断模型。方法:利用基因表达综合数据库(Gene Expression Omnibus,GEO)获取子宫内膜异位症患者的基因芯片数据和临床信息。从子宫内膜异位症患者和非子宫内膜异位症患者的转录组数据中获取差异表达基因,并进行GSEA和GSVA。两者结果的交集就是子宫内膜异位症的关键表达途径,随后又构建这些关键通路基因的蛋白质-蛋白质相互作用(protein-protein interaction,PPI)网络,并利用Cytoscape对关键基因进行鉴定,应用最小绝对收缩和选择算子(least absolute shrinkage and selection operator,LASSO)和逻辑回归建立子宫内膜异位症的临床诊断模型,最后利用另一组GEO数据集验证和评估该模型。结果:共鉴定出394个上调基因和37个下调基因和2个关键通路(FOSTER_TOLERANT_MACROPHAGE_DN和TGGTGCT_MIR29A_MIR29B_MIR29C)。2条通路中分别鉴定出96个和142个关键基因。进一步筛选后,建立了包含4个基因和2个基因的2个预测模型,并根据其风险评分构建风险预测模型(列线图)。3个模型在训练集[曲线下面积(area under the curve,AUC)=0.860、0.805、0.883]和验证集(AUC=0.606、0.636、0.633)中均表现良好。结论:本研究运用GSEA和GSVA交集法初步筛选出子宫内膜异位症的关键通路,建立了诊断模型,这些模型具有良好的诊断效能。Objective:Endometriosis is a gynecological disease with high incidence,difficult diagnosis,and unclear pathological mechanism.We identified key pathways and genes associated with endometriosis by gene set enrichment analysis(GSEA)and gene set variation analysis(GSVA),and developed a clinical diagnostic model for endometriosis.Methods:Gene microarray data and clinical information of endometriosis patients were obtained from Gene Expression Omnibus(GEO)database.Differentially expressed genes were obtained from transcriptomic data of endometriosis patients and non-endometriosis patients.And GSEA and GSVA analysis were performed.The intersection of 2 results was considered to be key expression pathways of endometriosis.Then we constructed the protein-protein interaction(PPI)network of these key pathway genes and identified the hub genes using Cytoscape.Least absolute shrinkage and selection operator(LASSO)-logistic regression was then used to construct the clinical diagnosis model of endometriosis.The model was validated and evaluated using another GEO datasets.Results:A total of 394 up-regulated genes and 37 down-regulated genes were identified.Two key pathways were identified(FOSTER_TOLERANT_MACROPHAGE_DN and TGGTGCT_MIR29A_MIR29B_MIR29C).A total of 96 and 142 hub genes were identified from the 2 pathways,respectively.After further screening,2 prediction models containing 4 genes and 2 genes were established,and a risk prediction model(nomogram)was constructed based on their risk scores.The 3 models performed well on both the training set[area under the curve(AUC)=0.860,0.805,0.883]and the validation set(AUC=0.606,0.636,0.633).Conclusion:The key pathways of endometriosis are initially screened by GSEA and GSVA intersection method and a diagnostic model is established.These models have good diagnostic efficacy.

关 键 词：子宫内膜异位症 基因集富集分析 基因集变异分析 列线图 诊断 

分 类 号：R711.71[医药卫生—妇产科学]