Background
Graves’ disease (GD) is an autoimmune disease associated with an increased incidence of other autoimmune diseases. To investigate the causality between GD and Diabetes mellitus (DM), we designed bidirectional two-sample Mendelian randomization (MR) and multivariable MR (MVMR) studies.
Methods
Single-nucleotide polymorphisms (SNPs) associated with GD, thyroid peroxidase (TPO), thyroglobulin (Tg), thyroid-stimulating hormone (TSH), type 1 diabetes (T1D), and type 2 diabetes (T2D) were obtained from the IEU Open GWAS and FinnGen biobank databases. For the forward MR study, we used GD (sample size = 458,620) as the exposure and T1D (sample size = 520,580) and T2D (sample size = 211,766) as the outcomes. Next, T1D and T2D were used as exposure variables, and GD was used as the outcome variable for the reverse MR analysis. Finally, MVMR analysis was conducted to investigate the probable relationship between DM and indicators for thyroid function like TPO, Tg, and TSH. The inverse variance weighting (IVW) was used as the main method. Finally, the heterogeneity and sensitivity were assessed.
Results
There were 27, 88, and 55 SNPs associated with GD, T1D, and T2D, respectively. A significant causal connection between GD and T1D (odds ratio [OR] [95% confidence interval, CI] = 1.411 [1.077–1.848], P = 0.012) and T2D (OR [95% CI] = 1.059 [1.025–1.095], P = 5.53e-04) was found in the forward MR analysis. However, reverse MR suggested that there was a genetic susceptibility to T1D that increased the likelihood of developing GD (OR [95% CI] = 1.059 [1.025–1.095], P = 5.53e-04), while T2D did not (OR [95% CI] = 0.963 [0.870–1.066], P = 0.468). Furthermore, there was inadequate evidence to suggest that abnormal TSH, TPO, and Tg levels increase the risk of incident T1D or T2D in individuals with GD. MVMR revealed no causal relationship among Tg, TSH, TPO, T1D, or T2D.
Conclusion
Evidence of a bidirectional causative relationship between GD and T1D and a unidirectional causal relationship between GD and T2D was discovered using MR analyses. MVMR analysis showed no statistically relevant causality between TSH, TPO, or Tg and either T1D or T2D.