Thyroid Function and Obesity

The obesity epidemic is a major threat to health in most countries [1], although the increase in the prevalence of obesity seems to have stopped in some countries according to recent investigations [2,3]. The international focus on obesity has led to a steep increase in the number of studies dealing with possible interactions between obesity and other diseases as well as the relation between obesity and physiology and pathophysiology of the various organs and tissues of the body.

In thyroidology, there have been studies of weight and thyroid cancer showing a borderline positive association between the risk of cancer and body mass index (BMI) [4] similar to the association demonstrated for a number of other types of cancer [5], and maybe there is a more aggressive behaviour of thyroid cancer in obese people [6]. On the other hand, it has been suggested that obesity may protect against thyroid nodules [7].

However, the most straightforward scientific and practical clinical question in the thyroid and obesity field is the potential association between thyroid function and weight control, and the issue has been addressed in many studies. The present review discusses data suggesting that even rather small differences in thyroid function within a population are associated with differences in BMI and with the prevalence of obesity. The association is complex, because overeating with accumulation of fat and the development of obesity seems to activate the hypothalamic-pituitary-thyroid axis leading to changes in thyroid function tests overlapping with the abnormalities seen in primary thyroid disease.

A special area of practical concern is the common complaint from patients treated for thyroid disease that therapy has led to ‘irreversible’ gain of weight – or that the weight increase occurring in association with the disease did not reverse upon therapy.

Finally, a few comments will be added on new studies that may add a new role for L-T₃ replacement in control of body weight independent of energy expenditure.

Resting energy expenditure (REE), which comprises around 60% of total energy expenditure in adult man [8], is high in overt hyperthyroidism and low in overt hypothyroidism [9]. However, measurable differences in REE have been described also with smaller variation in thyroid function. al-Adsani et al. [10] found changes in REE of 7–8% when the dose of L-T₄ to hypothyroid patients was modified to change serum TSH by a factor of 10, corresponding to the difference from the lower to the upper limit of the laboratory reference range for serum TSH (0.4–4.0 mU/l). Based on data from many studies, Eisenberg and Distefano [11] developed a model on the association between the absorbed doses of L-T₄ given as supplement to a thyroidectomized patient and serum TSH. In this model, a change in serum TSH from 4.0 to 0.4 mU/l corresponds to a quite substantial (approx. 40%) increase in the absorbed dose of L-T₄.

A 7–8% difference in REE corresponds to a daily difference in metabolism of around 10 g of lipid. Thus, assuming unaltered energy intake and energy expenditure from exercise as well as unaltered food intake and thermoregulation, differences in thyroid function tests within the normal laboratory range might theoretically be associated with several kilogram loss or gain of weight per year.

Notably, results of a thyroid function test being within the ‘normal’ laboratory reference range may not be normal for an individual. Individual variation in thyroid function tests is considerably narrower than the broad laboratory reference range, and around 50% of this ‘normal’ range is not normal for the individual person [12].

Figure 1 illustrates the association between thyroid function tests and BMI in the Danish DanThyr 1997–1998 population cohort [13]. Small differences in thyroid function over the whole range of test results were associated with differences in BMI when studying this population. With higher levels of TSH, even within the laboratory reference range, BMI was higher, also after adjusting for many relevant confounding factors [14]. Moreover, the opposite association was observed between serum fT₄ and BMI, whereas no association was observed between fT₃ in serum and BMI [14]. Participants were selected to have fT₄ and fT₃ within laboratory reference ranges. In the group of participants with serum TSH >3.6 mU/l, the odds ratio for obesity with BMI >30 was 2.1 compared with the group with serum TSH of 1–1.99 mU/l. The estimated difference in weight between an average woman with a serum TSH of 0.28 mU/l and one with TSH of 4.5 mU/l was 5.5 kg, and a significant association was observed between retrospectively recorded weight gain over the last 5 years and serum TSH [14].

Higher serum TSH with lower fT₄ and less affected fT₃ is the typical picture of low-degree thyroid failure. The patterns observed in the DanThyr study suggest that the association between thyroid function variables and BMI is primarily driven by differences in thyroid hormone-dependent energy expenditure. A number of other studies have observed a negative association between serum fT₄ and BMI as well as markers of the metabolic syndrome [15,16,17,18].

However, as discussed below, other studies of the association between body weight and thyroid function parameters have come up with different main conclusions based on patterns that are to some degree overlapping with those shown above, but considerably different in some details.

Thyroid function tests in people who are morbidly obese may differ from those in a comparable group of lean people, with a serum TSH that is higher in the obese, but with no tendency of thyroid hormones in serum to be low in the obese. On the contrary, fT₃ and also in some studies fT₄ in serum tends to be higher in obese people [19,20,21]. The pattern has been most clearly observed in overweight children and adolescents (fig. 2) [22], where the frequency of underlying thyroid disease in the population is much lower than in adults. Characteristically, weight reduction by change of lifestyle and diet [22,23] or following gastric banding [24] or gastric bypass surgery [25] tends to ameliorate the small aberration in thyroid function tests in the majority of obese patients.

Thyroid hormones are important regulators of metabolism, and conversely metabolism modifies the system of hypothalamic-pituitary-thyroid-peripheral activation/inactivation effect of thyroid hormones. It is well known that such interaction may have effects on the thyroid function test used to evaluate patients for the presence of thyroid disease [26].

The exact mechanisms involved in the energy intake-dependent variations in the thyroid system are only partly known but probably quite complex [27]. Signs of low activity of the thyroid system are not only found during short-term fasting, but characteristically also in chronic caloric deprivation such as anorexia nervosa [28]. On the other hand, morbid obesity leads to signs of an elevated activity with slightly elevated TSH, and free thyroid hormones in serum (fig. 2). In the complex system of hypothalamic regulation, the factor favoured by most authors when discussing the cause for activation of the thyroid in overweight is activation of hypothalamic centres by leptin released from adipocytes in fat tissue [19,29,30,31]. There is a complex interaction between the thyroid hormones and adipose tissue where TSH and thyroid hormones may participate in adipocyte differentiation [32] and lipolysis regulation [33], whereas various adipocyte cytokines may interact with the hypothalamic-pituitary-thyroid axis [34,35,36].

Obesity is now very common in most populations with 32.2% of adult USA citizens being obese in the 2003–2004 National Health and Nutrition Examination Survey (NHANES) [37], and health care providers are seeing many more obese children and adults now than previously.

In a population with few thyroid abnormalities, but much obesity, it is to be expected that the predominant pattern of association between thyroid function tests and BMI will be activation of the thyroid axis with a positive association between body weight and elevated TSH and thyroid hormones in serum, as illustrated in the study of obese and lean children (fig. 2). On the other hand, in a study such as DanThyr with less obesity, but a more common occurrence of small thyroid function abnormalities, the findings will be those of a negative association between BMI and thyroid function as shown in figure 1.

In many population studies, only the positive association between serum TSH and BMI has been reported and thyroid hormones in serum were not measured [38,39,40]. From this, it is not possible to judge if the main factor involved is a change in body weight secondary to a change in thyroid function or an increase in TSH caused by obesity. Likely, both mechanisms would to some degree be involved in all studies.

A good example is a recent study of the 2003–2004 NHANES cohort [41] where serum TSH correlated positively with BMI as in many other studies. In addition, a significant positive association was observed between serum fT₃ and BMI pointing to an important role of thyroid axis activation by obesity in this cohort with >30% obese people [37]. On the other hand, serum fT₄ correlated negatively with BMI, which may suggest an additional role of higher BMI with a lower thyroid function. In the 1997–1998 DanThyr cohort illustrated in figure 1, the prevalence of obesity was considerably lower (12.8%) [3] than in the USA study.

A similar suggestion for the differences in results between studies was given in an Italian report investigating morbidly obese patients [20]. The authors found little evidence for thyroid failure being the cause for mildly elevated TSH in their group of severely obese patients.

Other methodological differences between studies might be involved. For example, the absolute differences in results of thyroid function tests between groups having different BMI are relatively small, around the magnitude of day-to-day variation in assays. In some studies, it is not indicated if serum samples from different groups of participants (e.g. obese patients and lean controls) were analysed in random order in the same batch. If not, this may have induced bias. Moreover, if samples have been kept at room temperature for more than short periods, free fatty acids from lipolysis may alter the results of free thyroid hormone assays [26], the magnitude being dependent on concentration of lipids in blood.

As discussed above, as well as in several recent reviews [19,29,30], obesity tends to increase serum TSH and nothing suggests that an isolated slightly elevated TSH in a morbidly obese person is more abnormal to the person than a low serum T₃ in a person with severe underweight. It is an adaptation and the appropriate therapy is adjustment of energy balance and body weight.

Patients should not be treated with thyroid hormone replacement unless there are other signs of thyroid disease. Therapy of obese patients with pharmacologic doses of thyroid hormones leading to hyperthyroidism might reduce their weight, but side effects of the hyperthyroidism may be severe, and currently there is no indication for such therapy [42].

Because of the low metabolic rate, it is generally assumed that patients with longstanding overt hypothyroidism are overweight and clearly, body weight tends to increase during the development of severe hypothyroidism. In a recent case-control study of patients with newly diagnosed overt autoimmune hypothyroidism identified in a population study versus age-, gender- and inhabitancy-matched euthyroid controls from the same background population, the patients weighed on average 7 kg more than the controls [43]. The main mechanism behind such increase in body weight seems not to be fat as might be anticipated, but an expanded water compartment.

Two studies have examined body composition using dual-energy X-ray absorptiometry (DXA) before and after L-T₄ replacement therapy of overtly hypothyroid patients [44,45]. The main result was the same: the loss of weight after therapy is caused by a decrease in lean body mass and not in fat mass, as illustrated in figure 3. Severe hypothyroidism leads to accumulation in skin and other tissues of water-binding glycosaminoglycans, which is a main factor in giving the myxoedematous character of hypothyroid patients. With therapy, tissue composition normalizes and the excess water is excreted. Thus, during the initial phase of therapy there may be excessive diuresis [46].

The notion that overt hypothyroidism does not cause morbid overweight and that most, if not all, of the gain in weight is caused by retention of water in tissues was already brought forward by Plummer [47] in 1940. Plummer published a number of patient photos illustrating the difference in myxoedematous appearance with excess body water before and after therapy (fig. 4). Based on the weights of the 200 hypothyroid patients (with an average basal metabolic rate of –32.3%) whom Plummer had seen, he judged that the average gain of weight in untreated patients was 4.6 kg, and that this was all water.

An interesting question is, if the difference in body weight observed with relatively small differences in thyroid function in studies of populations (fig. 1) might to some degree be caused by differences in body water content or if it is entirely caused by a difference in the amount of body fat.

When discussing weight gain in patients with hypothyroidism, it is of clinical importance to remember the large increase in risk of developing overt autoimmune hypothyroidism during the first 2 years after stop of tobacco smoking. Weight gain is common after smoking cessation [48]. In a 10-year study, weight gain attributable to smoking cessation was on average 4–5 kg. [49]. However, any patient complaining of weight gain after quitting smoking should have thyroid function tested to exclude that this is caused by hypothyroidism, because there is a 6- to 7-fold increase in the risk of developing autoimmune hypothyroidism after quitting smoking [43].

Hoogwerf and Nuttall [50] studied a group of 87 patients treated for hyperthyroidism with radioiodine and/or antithyroid drugs and who had been followed for a mean of 7.5 years after therapy. The setting was a USA military veteran clinic and 84% of the patients were men. Most of the patients had had their body weight recorded already before they became hyperthyroid, and body weight just before and during follow-up after therapy was systematically recorded. At the time of treatment, body weight was 16% below the baseline recorded before the disease, and only 1 patient had no loss of weight in association with the disease. After therapy, weight started to increase and at 24 months after initial therapy patients had on average the same weight as before they became hyperthyroid. Body weight continued to increase slightly and 8 years after therapy it was on average 1.7 kg higher.

In general, this study, which was published in 1984, suggests that there are few problems with weight gain after therapy of hyperthyroidism, and the authors concluded that in the absence of significant metabolic derangement, body weight is regulated within narrow limits over many years. The conclusion followed the concept of a set point for body weight regulation proposed by Nisbett in 1972 [51].

The idea of ‘body weight autoregulation’ has been severely challenged by the current obesity epidemic and a number of studies have subsequently indicated that a considerable proportion of patients experience excessive weight gain after therapy of hyperthyroidism [52,53,54,55,56,57,58,59]. In a questionnaire-based follow-up study of 235 Swedish patients previously treated for hyperthyroidism, Berg et al. [54] found that weight gain was a problem in 79% of the individuals, and O’Malley et al. [55] found 69% of patients experiencing more weight gain than previous weight loss after therapy of hyperthyroidism.

de la Rosa et al. [60] reported that a considerable part of the initial weight gain after therapy of hyperthyroidism consisted of lean body mass. This is in accordance with a longitudinal study performed by Lönn et al. [61]. Based on serial investigations using DXA and CT scan they concluded that initial gain of weight after therapy of hyperthyroidism was mostly caused by an increase in muscle mass, whereas the continued gain in weight was caused by accumulation of fat. On the other hand, Zimmermann-Belsing et al. [62] found that increases in lean mass and fat mass measured by DXA were nearly parallel over 12 months after therapy of hyperthyroidism.

In another follow-up study, Dale et al. [56] found that pre-existing obesity, Graves’ disease, prior weight loss, and development of hypothyroidism predicted excessive gain in weight. Similarly, Brunova et al. [57] found the main factors associated with weight gain to be the need for replacement therapy and poor control of thyroid function.

The exact reason for the excessive weight gain encountered by many patients after therapy of hyperthyroidism is not clear. Jansson et al. [53] hypothesized that hyperthyroidism may induce a long-lasting disturbance in the neurochemical regulation of appetite and weight. Common suggestions have been a psychological delay in adapting food intake to the fall in energy expenditure after therapy. Another suggested mechanism is low energy expenditure caused by overtreatment of hyperthyroidism with induction of hypothyroidism [63]. In a study on the effect of dietary advice in patients being treated for hyperthyroidism, Alton and O’Malley [64] found that patients with no dietary advice gained 16% in weight during the period of normalisation of thyroid function from carbimazole therapy. The group advised by a dietician gained 9.8% (p < 0.05).

An interesting speculation is on the role of T₃ in body weight regulation. T₃ has important effects in hypothalamic centres apart from the feedback effects on the pituitary-thyroid axis. A study performed in rats concluded that changes in energy balance in hyperthyroidism mostly occurred via T₃ regulation of lipid metabolism in the hypothalamus, leading via the sympathetic nervous system to induction of brown adipose tissue [65], and animals with profound circannual biological rhythms of food intake and reproduction show no cycles if hypothalamic T₃ content is kept stable [66]. Thus, the annual variations in food intake driven by changes in hours of daylight seem to be mediated by hypothalamic T₃ content. Moreover, in rats, direct T₃ stimulation of food intake via hypothalamic effects has been demonstrated [67]. However, when the same investigators performed an experiment in 11 healthy normal weight men, a short-term 45% increase in serum T₃ after oral intake of 10 µg L-T₃ did not affect subsequent food consumption [68].

A recent carefully controlled study may even suggest that T₃ has an appetite reducing effect in humans. Celi et al. [69] studied the metabolic effects of L-T₃ in hypothyroidism by performing a randomized double-blind crossover trial of full replacement with L-T₃ versus L-T₄. Medication was split over 3 daily doses to avoid the large fluctuation in serum T₃ associated with once daily administration, and doses of L-T₃ and L-T₄ were carefully adjusted to give the same normal level of serum TSH. Each treatment period was 4–5 months. Quality of life, REE, insulin sensitivity and cardiovascular function were studied in detail, but with no difference between groups. However, L-T₃ therapy was associated with a significant weight loss of 2.1 kg compared with L-T₄ therapy, and also with significant reductions in serum low-density lipoprotein cholesterol and apolipoprotein B. The weight loss despite unaltered REE may suggest a decrease in food intake during L-T₃ therapy. On the other hand, the fall in serum cholesterol and a concomitant increase in serum sex-hormone-binding hormone suggest an effect on liver metabolism.

Interestingly, a randomized Dutch 15-week study on the effect of L-T₄ + L-T₃ combination therapy in two ratios versus L-T₄ alone for hypothyroidism [70] observed an L-T₃ dose-dependent decrease in body weight during the combined therapy. Furthermore, such an effect was statistically significant (although of small absolute magnitude) in a meta-analysis of studies evaluating L-T₄ + L-T₃ therapy [71].

A relatively more profound effect of L-T₃ versus L-T₄ on body weight even if TSH in serum is the same, may be one cause for the observed increase in weight in patients kept euthyroid with L-T₄ alone after total thyroidectomy [72] even if this is not a universal finding [73]. It may also be a mechanism involved in the excessive weight loss in Graves’ disease where thyroidal T₃ production from T₄ is excessive [74] – and the subsequent excess in weight gain in patients who become hypothyroid and are replaced with L-T₄.

The effect of L-T₃ on body weight adds new fuel to the fire of discussion between experts who claim that pure L-T₄ replacement therapy of hypothyroidism is optimal, and the patients who insist that preparations with high T₃ content such as desiccated thyroid makes them feel healthier [75].

Clearly, more long-term intervention studies are needed to expand our knowledge on the association between body weight and different types of thyroid hormone replacement therapy.

The authors declare no conflicts of interest.