Practice guidelines lack recommendations on subtle thyroid function alterations, such as subclinical thyroid disease or thyrotropin (TSH) near the reference limits. Several studies investigated the association of TSH or subclinical thyroid dysfunction with cognitive impairment in older adults. However, neuropathologic alterations begins a long time earlier than cognitive complaints, and little is known about the role of subtle thyroid alterations on cognitive impairment in middle-aged adults. This article comments on the recent work of Szlejf et al. that investigated the association of TSH and subclinical thyroid disease in a large sample of Brazilian middle-aged adults, free of dementia. The study found that lower TSH was associated with poorer performance on executive function.

Current guidelines recommend dosing serum thyrotropin (TSH) when evaluating individuals with cognitive complaints because overt thyroid dysfunctions are reversible causes of cognitive decline.1 However, dementia practice guidelines lack recommendations on subtler thyroid function alterations and practitioners may face uncertainty when dealing with patients with cognitive decline and subclinical thyroid disease or TSH near the reference limits. A better understanding of the role of TSH levels and subclinical thyroid disease on cognitive performance may help improve the management of these patients.

In a recent article, Szlejf et al. demonstrated that low serum TSH levels were associated with poorer cognitive performance in Brazilian adults without overt thyroid dysfunction.2 The aim of the study was to investigate the association of cognitive performance with serum TSH and subclinical thyroid disease in middle-aged adults from a middle-income country. It differed from previous studies in the field since most of them were conducted in older adults from developed countries. Using baseline data from the ELSA-Brasil study, a cohort of individuals aged 35-74 years at baseline, free of dementia, 10,362 participants were included in this complete case analysis. The main exclusion criteria were age ≥ 65 years, overt thyroid dysfunction, use of medications that can alter thyroid function and that indicate neurologic or psychiatric disease, and previous stroke. Participants with altered baseline TSH were screened for subclinical thyroid disease through free T4 (FT4) measurements. Cognition was evaluated by the CERAD´s delayed word recall test, semantic verbal fluency test, and the trail-making test version B. Sociodemographic characteristics, cardiovascular risk factors, and depression were considered as possible confounders and used for adjustment in the analysis. Participants in the first TSH tertile had worse performance in the trail-making test than participants in the middle TSH tertile. After restricting the analysis for the 9769 participants with TSH within the normal range, the results were similar, with a negative association between the first TSH tertile and performance on the trail-making test. Moreover, a one-unit increase in TSH had a discrete, although the significant, protective effect on performance on the trail-making test (OR 0.93, 95% CI 0.87-0.99). When participants were divided according to thyroid function, neither subclinical hypothyroidism nor subclinical hyperthyroidism was associated with performance on the cognitive tests.

The authors mentioned some limitations of the study, mainly the cross-sectional design, and not exploring the association between FT4 and cognitive performance. The authors also comment on possible pathophysiological pathways that could explain the findings, such as: (1) low TSH levels could reflect an excess of thyroid hormones, which have neurotoxic effects,3,4 (2) the brain disease that is leading to cognitive decline may also reduce thyrotropin-releasing hormone, leading to mild central hypothyroidism with lower TSH secretion and reduced thyroid hormone turnover,3,4 and (3) a non-causal association between TSH and cognitive impairment, due to non-addressed confounders. It is important to notice that the only significant association was with the trail making test, which evaluates executive function, processing speed, and visuospatial ability.5 These cognitive domains may be the first ones to alter the cognitive decline related to low TSH. 

Previous studies, most of them conducted in older adults, have found similar or correlated results with the study of Szlejf et al.2 Lower TSH and subclinical hyperthyroidism were associated with a higher risk of dementia in a cross-sectional analysis of a cohort of Brazilian older adults, although when restricting the sample to participants with TSH within the normal range, the association was no longer significant.6 In the Korean Longitudinal Study of Health and Aging, participants with incident mild cognitive impairment or dementia had lower baseline TSH than participants without these conditions.7 Chaker et al. found that higher TSH lowered the dementia risk in the Rotterdam Study, a prospective cohort of middle-aged and older adults.8 Higher TSH was also associated with better performance on cognitive tests and better Clinical Dementia Rating scores in community-dwelling older adults.9 In a systematic review including 23 studies, 14 cross-sectional and longitudinal well-designed and well-powered studies demonstrated association of low TSH within the reference interval or subclinical hyperthyroidism with dementia or cognitive impairment.4 A meta-analysis with cross-sectional and prospective studies did not find an association between subclinical hypothyroidism and cognitive function in older adults.10 Another recent meta-analysis of prospective studies also did not show that subclinical hypothyroidism was a risk factor for dementia or cognitive decline, although subclinical hyperthyroidism did increase the risk of dementia.11

On the other hand, some works showed results that are notably different. In an analysis of the Framingham original cohort, the lowest and highest tertiles of TSH were associated with dementia risk after a mean follow-up of 12.7 years. However, when restricting the analysis to participants with TSH within the normal range, these associations were no longer significant.12 A longitudinal study with very old adults from the Netherlands did not find an association between TSH levels and cognitive function at baseline and after 3 years of follow-up.13 In another longitudinal Dutch study, TSH level at baseline was not associated with risk of dementia or with brain atrophy on magnetic resonance imaging. However, non-demented subjects with higher FT4 had more hippocampal and amigdalar atrophy.14 In the Women´s Health and Aging Study, a cohort of community-dwelling older women with some degree of disability and TSH within the normal range, baseline TSH was not associated with baseline cognitive performance on the Mini-Mental State Examination and with cognitive decline after 3 years of follow-up. However, participants with total thyroxine in the lowest tertile had a greater decline in cognitive performance.15 In a cross-sectional analysis of the Healthy Aging Study in England and Wales, participants with higher TSH had lower scores on the Mini-Mental State Examination.16 Also, Beydoun et al. demonstrated that higher TSH was associated with poorer performance in cognitive tests in a cross-sectional analysis of data from two different American cohorts of adults.17,18

Although there are many studies in the field, controversy still remains regarding the association of cognitive performance with TSH and subclinical thyroid disease. The work of Szlejf et al. amplifies the scope of previous studies since it was conducted in a very large sample of middle-aged adults, free of dementia and overt thyroid disease. As the pathological alterations of dementia precede the clinical manifestations, it is pivotal to conduct further studies on the influence of thyroid functioning in cognitive impairment in middle-aged adults. Knowing that TSH near the lower limit of the reference range is associated with poorer performance in executive function, processing speed, and visuospatial ability opens a new window of opportunities for future studies.