实用肿瘤学杂志 ›› 2025, Vol. 39 ›› Issue (3): 208-215.doi: 10.11904/j.issn.1002-3070.2025.03.006

• 临床研究 • 上一篇    下一篇

基于生物学先验信息和基因调控网络筛选食管癌诊断及预后标志物

王心雨1, 张奇2   

  1. 1.国家儿童医学中心 首都医科大学附属北京儿童医院大数据中心(北京 100045);
    2.哈尔滨医科大学公共卫生学院卫生统计学教研室
  • 收稿日期:2024-11-26 修回日期:2025-02-13 出版日期:2025-06-28 发布日期:2025-07-02
  • 通讯作者: 张奇,E-mail:hmutjzq@163.com
  • 作者简介:王心雨,女,(1989—),博士,助理研究员,从事大数据挖掘和儿童慢性病管理的研究。
  • 基金资助:
    哈尔滨医科大学创新科学研究基金(编号:2022-KYYWF-0252)

Screening of diagnostic and prognostic markers for esophageal cancer based on biological prior information and gene regulatory network

WANG Xinyu1, ZHANG Qi2   

  1. 1. Big Data Center Beijing Children′s Hospital,Capital Medical University,National Center for Children′s Health,Beijing 100045,China;
    2. Department of Health Statistics,School of Public Health,Harbin Medical University
  • Received:2024-11-26 Revised:2025-02-13 Online:2025-06-28 Published:2025-07-02

摘要: 目的 筛选食管癌诊断和预后标志物,探究其潜在作用机制,为食管癌的早期诊断和精准治疗提供新思路。方法 提取并分析癌症基因组图谱(The Cancer Genome Atlas,TCGA)食管癌转录组测序数据(RNA-seq)中的蛋白编码基因,将差异表达基因所富集的KEGG通路和蛋白质-蛋白质相互作用(protein-protein interaction,PPI)网络关系作为先验信息,使用先验信息驱动的混合图模型(prior incorporation mixed graphical model,piMGM)构建综合调控网络,提取其中与疾病状态、生存时间和生存结局有关的基因作为诊断和预后标志物,利用多因素回归分析构建预测模型并计算风险评分。结果 共获得180个肿瘤组织与正常组织之间的差异表达基因,主要富集在细胞周期、细胞衰老、胃酸分泌、p53信号通路、IL-17信号通路等KEGG通路中;通过基因调控网络分析确定MT1M、SLC9A4、GPER1、MT1A、CCL20、MDFI为关键基因,与临床变量一起构建预后预测模型并计算风险评分,根据最佳截断值将患者分为预后显著不同的高低风险组:食管癌诊断模型的AUC为0.978(95% CI:0.935~0.996),1年、3年总生存率预测模型的AUC分别为0.783(95% CI:0.646~0.896)和0.779(95% CI:0.598~0.999),1年、3年无病生存率预测模型的AUC分别为0.787(95% CI:0.646~0.848)和0.762(95% CI:0.575~0.900)。结论 本研究确定的6个标志物能够有效地预测食管癌患者的发病和预后,为开发食管癌高效的诊断工具和精准治疗方案奠定坚实基础。

关键词: 生物学先验信息, 基因调控网络, 食管癌, 生物标志物

Abstract: Objective The aim of this study was to screen diagnostic and prognostic markers for esophageal cancer,explore their potential mechanisms of action,and provide new insights for early diagnosis and precision treatment of esophageal cancer. Methods The protein-coding genes in transcriptome sequencing data(RNA-seq)of esophageal cancer from The Cancer Genome Atlas(TCGA)were extracted and analyzed.The KEGG pathways and protein-protein interaction(PPI)network relationships enriched in differentially expressed genes(DEGs)were used as prior information.The prior incorporation mixed graphical model(piMGM)was employed to construct an integrative regulatory network.Genes related to disease status,survival time,and survival outcomes were identified as diagnostic and prognostic markers.The prediction models and calculate the risk score were constructed using multivariate Cox regression analysis. Results A total of 180 DEGs between tumor and normal tissues were obtained,which were mainly enriched in KEGG pathways such as cell cycle,cellular senescence,gastric acid secretion,p53 signaling pathway,and IL-17 signaling pathway.MT1M,SLC9A4,GPER1,MT1A,CCL20,and MDFI were identified as key genes through gene regulatory network analysis.Together with clinical variables,a prognostic prediction model was constructed and the risk score was calculated.According to the optimal cutoff value,the patients were divided into the high-and low-risk groups with significantly different prognoses:the area under the curve(AUC)of the esophageal cancer diagnosis model was 0.978(95% CI:0.935-0.996),the AUCs of the 1-year and 3-years overall survival prediction models were 0.783(95% CI:0.646-0.896)and 0.779(95% CI:0.598-0.999),respectively,and the AUCs of the 1-year and 3-years disease-free survival prediction models were 0.787(95% CI:0.664-0.848)and 0.762(95% CI:0.575-0.900),respectively. Conclusion The six markers identified in this study can effectively predict the incidence and prognosis of patients with esophageal cancer,laying a solid foundation for the development of efficient diagnostic tools and precise treatment regimens for esophageal cancer.

Key words: Biological priori information, Gene regulatory network, Esophageal cancer, Biomarker

中图分类号: