Numerous studies have examined whether vitamin D is associated with gestational diabetes mellitus (GDM). Nevertheless, it is still challenging to determine the causality, due to a number of shortcomings in observational research and randomized controlled trials. Mendelian randomization (MR) with two samples was conducted to investigate the potential causative association between 25-hydroxyvitamin D (25(OH)D), vitamin D binding protein (VDBP) and GDM risk. Publicly accessible summary data from independent cohorts were used for two-sample MR. For 25(OH)D, we obtained data from UK Biobank, IEU and EBI, then performed a meta-analysis to enhance the statistical power (via METAL); for VDBP, data were obtained from the INTERVAL study; for GDM, data were obtained from FinnGen. The inverse variance weighted (IVW) approach was performed as the main analysis, together with several sensitivity analyses, such as MR-Egger, maximum likelihood, weighted median, and weighted mode. The IVW results revealed a weak negative causal connection between 25(OH)D and GDM risk [OR (95% CI) = 0.71 (0.50, 0.99), p = 0.046]. However, the causal association was unstable according to sensitivity analyses, and Cochran's Q test revealed significant heterogeneity. After removing BMI-related IVs, the causal association between 25(OH)D and GDM disappeared [OR (95% CI) = 0.76 (0.55, 1.06), p = 0.101]. In addition, our study found no proof to support the assumption that VDBP level was related to GDM risk causally [OR (95% CI) = 0.98 (0.93, 1.03), p = 0.408]. According to this study, a weak negative causal association between 25(OH)D and GDM risk was found, while we had little proof to support the link between VDBP and GDM. To further explore whether total or free 25(OH)D levels and GDM are causally related, GWAS data with an emphasis on women of reproductive age and other ethnic groups are required.
Read full abstract