Thyroid-Stimulating Hormone and Thyroid Hormones and Incidence of Diabetes: Prospective Results of the Brazilian Longitudinal Study of Adult Health (ELSA-BRASIL).

Background: There are conflicting data regarding the association of thyroid function with incident diabetes. We prospectively investigated thyrotropin (TSH), free thyroxine (fT4), free triiodothyronine (fT3), and its conversion ratio (fT3:fT4) with the risk of developing diabetes in euthyroid subjects and those with subclinical thyroid dysfunction. Our hypothesis is that this relationship is a U-shaped curve since both subclinical thyroid diseases may be associated with diabetes. Methods: ELSA-Brasil is a highly admixed cohort study of 35-74 years old at baseline (2008-2010). Levels of TSH, fT4, fT3, and fT3:fT4 ratio were evaluated at baseline and incident diabetes was estimated over an 8.2-year follow-up (2017-2019). Diabetes was identified based on medical diagnosis, prescriptions, and laboratory tests. The risk of diabetes was evaluated according to quintiles of TSH, fT4, fT3, and fT3:fT4 ratio using Poisson regression with robust variance presented as relative risk (RR) with confidence interval [CI] of 95% after multivariable adjustment for sociodemographic and cardiovascular risk factors (reference third quintile), and as continuous variables. Results: We included 7948 participants (mean age, 50.2 [standard deviation 8.6] years; 54.4% female): 7177 euthyroid, 726 with subclinical hypothyroidism, and 45 with subclinical hyperthyroidism. Incidence of diabetes was 14.8%. No association was found for TSH, fT4, fT3, and fT3:fT4 ratio quintiles with incident diabetes. Using continuous variables, the increase of 1-unit (1-U) of fT4 decreased the risk of diabetes (RR 0.94 [CI 0.91-0.99]), while the increase of 1-U of the fT3:fT4 ratio increased the diabetes risk (RR 1.37 [CI 1.15-1.63]). The increase of 1-U of fT3 was associated with an increased risk of diabetes, but without significance after multivariable adjustment. In body mass index-stratified analysis, people with overweight or obesity presented a modest significantly higher risk of diabetes in the lowest quintile of fT4 (RR 1.04 [CI 1.01-1.07]) and an inverse association with incident diabetes in the first quintile of fT3:fT4 ratio (RR, 0.95 [CI 0.93-0.98]). The analyses using continuous variables presented similar findings. Conclusion: These findings suggest that fT4 and fT3 levels and the conversion rate might be additional risk factors associated with incident diabetes, especially in the presence of overweight or obesity. However, they need to be confirmed in future studies. (ClinicalTrials.gov Identifier: NCT02320461).