欢迎访问《中国循证儿科杂志》官方网站，今天是
English Version
中国循证儿科杂志

中国循证儿科杂志 ›› 2024, Vol. 19 ›› Issue (2): 141-145.DOI: 10.3969/j.issn.1673-5501.2024.02.012

• 论著 • 上一篇    下一篇

基于两样本孟德尔随机化分析妊娠期糖尿病与子代儿童期哮喘风险的因果关系

戴雨晨,玛伊热·牙生,叶伟涛,张新光   

  1. 上海中医药大学附属市中医医院 上海，200071
  • 收稿日期:2023-09-26 修回日期:2023-11-02 出版日期:2024-04-25 发布日期:2024-04-25
  • 通讯作者: 张新光

Causal relationship between gestational diabetes mellitus and the risk of childhood asthma in offspring: A two-sample Mendelian randomized analysis

DAI Yuchen, MAYIRE·Yasheng, YE Weitao, ZHANG Xinguang   

  1. Shanghai Municipal Hospital of TCM, Shanghai University of Traditional Chinese Medicine, Shanghai 200071, China
  • Received:2023-09-26 Revised:2023-11-02 Online:2024-04-25 Published:2024-04-25
  • Contact: ZHANG Xinguang

PDF

31

摘要： 背景：在传统流行病学队列研究中，母体在妊娠期间发生的高血糖可能会影响胎儿的免疫系统和肺部发育，导致儿童更容易发生哮喘。然而，这一研究结果可能受到混杂因素的影响，有必要进一步探讨GDM与儿童哮喘之间的因果关联。 目的：采用两样本孟德尔随机化（TSMR）的方法来探究GDM与子代儿童期哮喘之间的因果关联，为临床防治提供建议。 设计：病例对照研究。 方法：通过全基因关联研究（GWAS）汇总数据库中选择一项年份最近、样本量最大的GDM和子代儿童期哮喘密切相关的遗传位点，以单核苷酸多态性（SNPs）作为工具变量，GDM作为暴露因素，儿童期哮喘为结局变量，分析GDM与子代儿童期哮喘的因果关系。在工具变量的筛选上设置P值（P<5.0×10-7），为避免连锁不平衡（LD）偏倚，与GDM相关的显著SNP的LD须满足条件：r2<0.001和遗传距离为10 000 kb，剔除统计量F＜10的SNP。使用最常用及较公认的逆方差加权法（IVW）的同时，也参考使用加权中位数法、MR-Egger回归法进行TSMR分析推断因果关系。并采用MR-Egger回归、Cochran Q检验和逐个剔除检验“留一法”行敏感性分析。统计分析采用R 4.3.1软件包TwoSampleMR进行，检验水准α取0.05。 主要结局指标：GDM与子代儿童期哮喘关联性。 结果：在GWAS汇总数据库的芬兰人种数据集的GDM数据集中，病例组6 033例，对照组110 330例，SNPs总计16 379 684个；哮喘数据集，病例组3 025例，对照组135 449例，SNPs总计16 379 865个。根据研究工具变量的筛选标准进行筛选，最终确定10个SNPs作为工具变量，所有的SNP位点对应的F统计量均＞10。IVW结果表明GDM和子代儿童期哮喘风险增加存在因果关系（OR=1.15，95%CI：1.01~1.32，P=0.039），Cochran Q检验表明纳入的SNPs不存在异质性（P均＞0.05）；MR-Egger回归分析，截距项为-0.013（P=0.68），差异无统计学意义，表明基因多效性影响的可能性较小；“留一法”对因果效应进行敏感性分析，逐步剔除各SNP后，未发现工具变量中具有对结果有强影响性的SNP位点，这表明因果关系具有一定的稳定性；漏斗图发现以单个SNP作为工具变量时，GDM与子代儿童期哮喘的因果效应基本呈对称分布。 结论：GDM与子代儿童期哮喘之间存在正向因果关联。

关键词: 妊娠期糖尿病, 儿童哮喘, 孟德尔随机化

Abstract: Background：Gestational diabetes mellitus (GDM) is one of the common metabolic disorders of pregnancy, and the hyperglycemia that occurs in the mother during pregnancy may affect the fetal immune system and lung development, leading to a greater susceptibility to asthma in children. However, in traditional epidemiologic cohort studies, the existing findings may be affected by confounding factors, and there is a need to further explore the causal association between GDM and childhood asthma and develop more effective prevention and management strategies. Objective：To investigate whether there is a causal association between gestational diabetes mellitus (GDM) and childhood asthma in offspring by using a two-sample Mendelian randomization (TSMR) subgroup, and to provide recommendations for clinical management. Design：Case-control study. Methods：Utilizing the most recent and largest sample size genome-wide association study (GWAS) summary data, we selected genetic loci closely related to GDM and childhood asthma, using single nucleotide polymorphisms (SNPs) as instrumental variables, GDM as the exposure factor and childhood asthma as the outcome variable, to analyze the causal relationship between GDM and childhood-onset asthma. A P value (P<5.0×10-7) was set for the screening of instrumental variables, and in order to avoid linkage disequilibrium (LD) bias, the LD of significant SNPs associated with GDM had to satisfy the conditions: r2<0.001 and a genetic distance of 10, 000 kb. SNPs with the F-statistics<10 were excluded. The inverse variance weighted(IVW) method, alongside weighted median and MR-Egger regression methods, were used for TSMR analysis to infer causality. Sensitivity analyses were performed in parallel using MR-Egger regression, Cochran's Q-test, and the "leave-one-out" method. Statistical analyses were performed using R 4.3.1 software package TwoSampleMR with a significance level of α=0.05. Main outcome measures：Association between GDM and childhood asthma. Results：In the Finnish GWAS summary data, the GDM dataset included 6,033 cases and 110,330 controls, totaling 16,379,684 SNPs. The asthma dataset included 3,025 cases and 135,449 controls, with a total of 16,379,865 SNPs. According to the screening criteria for research instrumental variables, 10 SNPs were selected as instrumental variables, all with F-statistics>10. IVW results indicated a causal relationship between GDM and increased risk of childhood asthma (OR=1.15, 95%CI: 1.01-1.32, P=0.039). Cochran's Q test showed no heterogeneity among the included SNPs (P> 0.05). MR-Egger regression's intercept was -0.013 (P=0.68), indicating a low likelihood of pleiotropic effects. Sensitivity analysis using the leave-one-out method did not reveal any SNPs with a strong influence on the results, suggesting stability in the causal relationship. Funnel plot analysis demonstrated that the causal effect of GDM on childhood asthma was symmetrically distributed when using individual SNPs as instrumental variables. Conclusion：There is a positive causal association between GDM and childhood-onset asthma.

Key words: Gestational diabetes mellitus, Childhood asthma, Mendelian randomization