Abstract
Background: Observational studies have demonstrated that variation in normal range thyroid function is associated with major cardiovascular risk factors, including dyslipidemia, hypertension, type 2 diabetes (T2D), and obesity. As observational studies are prone to residual confounding, reverse causality, and selection bias, we used a Mendelian randomization (MR) approach to investigate whether these associations are causal or not.
Methods: Two-sample MR analysis using data from the largest available genome-wide association studies on normal range thyrotropin (TSH) and free thyroxine (fT4) levels, serum lipid levels, blood pressure measurements, T2D, and obesity traits (body mass index [BMI] and waist/hip ratio).
Results: A one standard deviation (SD) increase in genetically predicted TSH levels was associated with a 0.037 SD increase in total cholesterol levels (p = 3.0 x 10(-4)). After excluding pleiotropic instruments, we also observed significant associations between TSH levels and low-density lipoprotein levels (beta = 0.026 SD, p = 1.9 x 10(-3)), pulse pressure (beta = -0.477 mmHg, p = 7.5 x 10(-10)), and T2D risk (odds ratio = 0.95, p = 2.5 x 10(-3)). While we found no evidence of causal associations between TSH or fT4 levels and obesity traits, we found that a one SD increase in genetically predicted BMI was associated with a 0.075 SD decrease in fT4 levels (p = 3.6 x 10(-4)).
Conclusions: Variation in normal range thyroid function affects serum cholesterol levels, blood pressure, and T2D risk.