To our knowledge, this is the first prospective population-based cohort study describing the relation between thyroid function within the normal range and the risk of diabetes and progression from prediabetes and type 2 diabetes. Higher TSH levels and lower FT4 levels are associated with an increased risk of diabetes and progression from prediabetes to diabetes.
There are no other studies addressing the relation between diabetes and thyroid function in the euthyroid range or in individuals with prediabetes. Even though there are many cross-sectional reports studying the prevalence of diabetes and thyroid dysfunction, only few have investigated the association of thyroid function with the occurrence of diabetes and all were register-based studies. Our results are in contrast to a Danish nationwide registry study by Brandt et al. [17] that reported an increased risk of diabetes in hyperthyroid individuals, whereas we did not find an increased risk of diabetes with higher thyroid function. However, there are several factors that could explain these differences, including variance in the mean age and possible iodine status of the studied population. Most importantly, the study by Brandt et al. [17] did not include laboratory measurements of thyroid function and therefore misclassification of the diagnosis of hyperthyroidism could have occurred. Further, they did not provide estimates in the euthyroid range of thyroid function. Two other register-based studies report an increased risk of diabetes in hypothyroid individuals [18, 19] and our results are largely in line as we find an increased risk of diabetes in lower thyroid function.
There are several pathways that may explain the observed relation between low and low-normal thyroid function and the risk of diabetes. Overt and subclinical hypothyroidism are associated with a decreased insulin sensitivity and glucose tolerance, partially due to a decreased ability of insulin to increase glucose utilization mainly in muscle [14, 25]. Other mechanisms, such as downregulation of plasma membrane glucose transporters and direct effects on insulin degradation, have also been described [26–28]. Treatment of hypothyroidism has been shown to restore insulin sensitivity and the secretion of glucoregulatory hormones [15]. Furthermore, hypothyroidism is associated with several components of the metabolic syndrome and could therefore indirectly relate to the increased risk of diabetes [29]. However, in our analyses, adjusting for several cardiovascular risk factors and components of the metabolic syndrome did not shift risk estimates towards the null. Additionally, excluding participants using thyroid hormone replacement therapy at baseline only slightly altered the results. Even though overt hyperthyroidism is also associated with insulin resistance, our data show that high and high-normal thyroid function are protective against the development of or progression to diabetes. It could be that insulin resistance in hyperthyroid patients is counterbalanced by other mechanisms associated with prolonged thyroid hormone excess, such as improved beta-cell function and increased insulin secretion [6]. However, the exact pathophysiological mechanisms through which thyroid function could affect diabetes risk in the general population remain to be determined.
The clinical importance of these findings could be several. First of all, the association of thyroid function with development from prediabetes to diabetes is prominent. Thus, individuals with a low-normal thyroid function, which includes a large proportion of the population, are at an even higher risk of progression from prediabetes to diabetes. Secondly, with ageing and increasingly obese populations, there is need for better screening and prevention options for diabetes [30]. One could hypothesize that, in individuals with prediabetes with low or low-normal thyroid function (i.e., high TSH and low FT4), lifestyle interventions or diabetes treatment could be prompted in an earlier phase than those with normal or high thyroid function. Alternatively, having prediabetes could be an argument to start treatment of subclinical hypothyroidism to aim for prevention of overt diabetes. Current guidelines do not recommend or specifically address screening of thyroid function or treatment of thyroid dysfunction in individuals with type 2 diabetes [31, 32].
The relative risk increase of developing diabetes with thyroid function differences is modest. However, due to the high population risk of diabetes, the implications on the absolute risk are large. Despite this high occurrence of both conditions in the general population, the relation between thyroid dysfunction and diabetes had remained largely unexplored. Further research is needed to determine to what extend the association could be driven by thyroid hormone-related acceleration of development of diabetes or perhaps by other mechanisms such as a common genetic predisposition. If our results are confirmed, subsequent studies could focus on screening and prevention strategies as well as questions concerning treatment of subclinical hypothyroidism in patients at risk for diabetes.
Strengths of our study include the large number of individuals, the variety of available confounders adjusted for, and the long follow-up. Furthermore, we were able to investigate both diabetes risk as well as progression from prediabetes to diabetes. Limitations of our study should also be acknowledged. Residual confounding cannot be excluded in an observational study, even with the large number of potential confounders adjusted for in our analyses. Furthermore, the Rotterdam Study is predominantly composed of white participants aged 45 years and older and results may therefore not be generalizable to other populations.