Department of Pulmonary and Critical Care Medicine, The affiliated hospital of Qingdao University, Qingdao University, Qingdao, China.
Medical Department of Qingdao University, Qingdao, China.
BMC Pulm Med. 2024 Mar 20;24(1):142. doi: 10.1186/s12890-024-02961-7.
The question as to whether or not diabetes mellitus increases the risk of idiopathic pulmonary fibrosis (IPF) remains controversial. This study aimed to investigate the causal association between type 1 diabetes (T1D), type 2 diabetes (T2D), and IPF using Mendelian randomization (MR) analysis.
We used two-sample univariate and multivariate MR (MVMR) analyses to investigate the causal relationship between T1D or T2D and IPF. We obtained genome-wide association study (GWAS) data for T1D and T2D from the European Bioinformatics Institute, comprising 29,652 T1D samples and 101,101 T1D single nucleotide polymorphisms (SNPs) and 655,666 T2D samples and 5,030,727 T2D SNPs. We also used IPF GWAS data from the FinnGen Biobank comprising 198,014 IPF samples and 16,380,413 IPF SNPs. All cases and controls in these datasets were derived exclusively from European populations. In the univariate MR analysis, we employed inverse variance-weighted (IVW), weighted median (WM), and MR-Egger regression methods. For the MVMR analysis, we used the multivariate IVW method primarily, and supplemented it with multivariate MR-Egger and multivariate MR- least absolute shrinkage and selection operator methods. Heterogeneity tests were conducted using the MR-IVW and MR-Egger regression methods, whereas pleiotropic effects were assessed using the MR-Egger intercept. The results of MR and sensitivity analyses were visualized using forest, scatter, leave-one-out, and funnel plots.
Univariate MR revealed a significant causal relationship between T1D and IPF (OR = 1.118, 95% CI = 1.021-1.225, P = 0.016); however, no significant causal relationship was found between T2D and IPF (OR = 0.911, 95% CI = 0.796-1.043, P = 0.178). MVMR analysis further confirmed a causal association between T1D and IPF (OR = 1.133, 95% CI = 1.011-1.270, P = 0.032), but no causal relationship between T2D and IPF (OR = 1.009, 95% CI = 0.790-1.288, P = 0.950). Sensitivity analysis results validated the stability and reliability of our findings.
Univariate and multivariate analyses demonstrated a causal relationship between T1D and IPF, whereas no evidence was found to support a causal relationship between T2D and IPF. Therefore, in clinical practice, patients with T1D should undergo lung imaging for early detection of IPF.
糖尿病是否会增加特发性肺纤维化（IPF）的风险仍存在争议。本研究旨在通过孟德尔随机化（MR）分析来研究 1 型糖尿病（T1D）、2 型糖尿病（T2D）与 IPF 之间的因果关联。
我们使用两样本单变量和多变量 MR（MVMR）分析来研究 T1D 或 T2D 与 IPF 之间的因果关系。我们从欧洲生物信息研究所获得了 T1D 和 T2D 的全基因组关联研究（GWAS）数据，包括 29652 例 T1D 样本和 101101 个 T1D 单核苷酸多态性（SNP），以及 655666 例 T2D 样本和 5030727 个 T2D SNP。我们还使用了来自芬兰基因生物银行的 IPF GWAS 数据，其中包括 198014 例 IPF 样本和 16380413 个 IPF SNP。这些数据集中的所有病例和对照均来自欧洲人群。在单变量 MR 分析中，我们采用了逆方差加权（IVW）、加权中位数（WM）和 MR-Egger 回归方法。对于 MVMR 分析，我们主要使用多变量 IVW 方法，并辅以多变量 MR-Egger 和多变量 MR-最小绝对收缩和选择算子方法。我们使用 MR-IVW 和 MR-Egger 回归方法进行异质性检验，使用 MR-Egger 截距评估多效性效应。MR 和敏感性分析的结果通过森林图、散点图、留一法和漏斗图进行可视化。
单变量 MR 显示 T1D 与 IPF 之间存在显著的因果关系（OR=1.118，95%CI=1.021-1.225，P=0.016）；然而，T2D 与 IPF 之间不存在显著的因果关系（OR=0.911，95%CI=0.796-1.043，P=0.178）。MVMR 分析进一步证实了 T1D 与 IPF 之间存在因果关系（OR=1.133，95%CI=1.011-1.270，P=0.032），但 T2D 与 IPF 之间不存在因果关系（OR=1.009，95%CI=0.790-1.288，P=0.950）。敏感性分析结果验证了我们研究结果的稳定性和可靠性。
单变量和多变量分析均表明 T1D 与 IPF 之间存在因果关系，而没有证据表明 T2D 与 IPF 之间存在因果关系。因此，在临床实践中，T1D 患者应进行肺部成像以早期发现 IPF。
