MATERIALS AND METHODS

We examined all children with non-syndromal obesity, defined by a body mass index (BMI) above the 97th centile who attended the intervention programme “Obeldicks”⁸ for obese children between 1999 and 2000. Reference data from Rolland-Cachera and colleagues⁹ were used, as proposed by the European childhood obesity group.

Thyroid hormone parameters (T3, T4, TSH), leptin, degree of overweight as a z score from the BMI according to the LMS method,^9,10 which includes an age dependent factor controlling for variation and skewness of the BMI distribution, as well as age, sex, and stage of puberty according to Tanner were established. After one year the same parameters were determined for all those taking part in the intervention programme “Obeldicks”. These data were compared to thyroid function (T3, T4, TSH) of 107 non-obese healthy children with constitutional or familial short or tall stature and a BMI below the 90th centile.⁹

TSH, total T3, and total T4 were determined by solid phase technique—chemiluminescence—immunoassays (Immulite). The norm values for the respective age range were between 0.2–0.4 and 5.4–6.1 mIU/l for TSH, between 0.6 and 3.1–3.5 nmol/l for T3, and between 46–54 and 160–178 nmol/l for T4. Leptin was measured with an enzyme immunoassay (DRG). The intraassay coefficients of variation were 3.9% for TSH, 5.4% for T3, and 6.3% for T4. The interassay coefficients of variation were 8.0% for TSH, 7.1% for T3, and 6.7% for T4.

In all children, autoimmune thyroiditis was excluded by measuring antimicrosomal and thyroglobulin antibodies for TSH concentrations above 5 mU/l or above the twofold standard deviation based on the 107 normal weight children (see table 1) and for goitre. Children with endocrine or metabolic disorders were excluded from the study. Smokers and children taking any medication including oral contraceptives were also excluded. Growth hormone deficiency was excluded in all children with short stature.

The reduction in weight was achieved through the intervention programme “Obeldicks” for obese children based on a one year therapy with behavioural components, physical exercise, and normal energy diet.⁹

Statistical analysis was performed by the Statistical Package for Social Sciences (SPSS). Stastically significant differences were tested by the non-parametric Mann-Whitney U test, paired observation by the non-parametric Wilcoxon test, correlations by the Spearman rank test, and dependent variables by partial correlation.

RESULTS

A total of 119 obese children presented for participation in the intervention programme “Obeldicks”. One child was excluded because of autoimmune thyroiditis. In the last three months before schooling no child had had a change in weight.

Obese children showed significantly higher serum concentrations of TSH, T3, and T4 than the non-obese children (see table 1). The group of obese children did not differ with respect to age, sex, or stage of puberty from the group of non-obese children. TSH serum concentrations were above the twofold standard deviation based on the 107 normal weight children in 14 (12%) of the 118 obese children, T3 serum concentrations in 18 (15%), and T4 serum concentrations in 13 (11%) obese children. There was no correlation between TSH, T3, and T4 serum concentrations, and age, sex, or stage of puberty. No child had a goitre. The thyroid hormones correlated with the degree of overweight (z score BMI; see table 2). The thyroid hormones did not correlate significantly with leptin (T3: r = 0.07, p = 0.46; T4: r = −0.05, p = 0.60; TSH: r = 0.16, p = 0.11). As both leptin¹¹ and the thyroid hormones correlated with the degree of overweight, partial correlation adjusted to the degree of overweight was used.

Two obese children without goitre had TSH values above 5 mIU/l (child 1: 10.5 mIU/l; child 2: 7.8 mIU/l) and normal autoantibodies (microsomal (MAK), thyroglobulin (TRAK), and thyroid activating autoantibodies (TAK)). T3 of these children was 3.8 (child 1) and 3.1 nmol/l (child 2); T4 was 130 nmol/l (child 1) and 102 nmol/l (child 2). They had no clinical signs of hyperthyroidism or hypothyroidism.

Sixty eight obese children took part in the intervention programme “Obeldicks”. Fifty five (81%) finished the training programme successfully with a reduction of their z scores BMI by at least 0.1 (median loss 0.50, range 0.20–1.90; see table 3). The group of successful children did not differ from that of the unsuccessful children as to age (median 11.3 years, range 4.5–15 years compared to median 11.9 years, range 6–15 years), sex (44% compared to 54% girls), or degree of overweight (z score BMI). Compared to their initial concentrations the successful children showed a significant decrease in serum T3, T4, and leptin (see table 3) but not in TSH, even if patients with TSH above twofold standard deviation based on the 107 normal weight children are excluded (TSH: p = 0.761; T3: p = 0.002; T4: p < 0.0001). The thyroid hormones of the 13 children who did not lose weight successfully after a year did not change significantly compared to their initial values (see table 4).

DISCUSSION

In obese children, serum concentrations of thyroid hormones (TSH, T3, T4) were on average above those of the children of normal weight, and decreased (T3, T4) after weight loss from a normal energy diet; this points to a reversible increase of thyroid hormones in obesity.

These changes of TSH and T3 are not caused by other influencing factors, as without weight loss in a collective of obese children of the same age, gender, and degree of overweight, there was no change in thyroid status over the same time period.

As both TSH and the peripheral thyroid hormones (T3 and T4) are increased in obesity, there is no necessity for treatment of the increased TSH serum concentrations,¹³ as T3 is the biologically active thyroid hormone and is crucial for metabolism.

A moderate rise in total and free T3 and TSH serum concentrations of obese children, as in our group, has been shown in a number of other studies.^12–15 The cause of the increased thyroid hormone concentrations in obesity is unclear. In a study in children, the increase in TSH serum concentrations was not accounted for by iodine deficiency or autoimmune thyroiditis.¹³ Increased thyroid hormone concentrations could point to hormone resistance, similar to insulin resistance in obesity.¹⁶ In support of this theory is the fact that in obesity T3 receptors are decreased,¹⁷ and the negative feedback between TSH and the peripheral thyroid hormones (T3, T4) is decreased, as both TSH and the peripheral thyroid hormones are increased in obesity. The patients with the highest serum TSH concentrations had high serum T3 concentrations. To determine whether pituitary responsivity is altered in obese children, a TRH test would be useful.

As the thyroid hormones—especially T3—regulate both the resting metabolic rate and thermogenesis and lead to lipolysis,^12,18,19 changes in thyroid hormones could also point to an adaptation process in obesity. When the peripheral thyroid hormones decrease after a reduction of overweight, we can also expect a reduction in resting metabolic rate and consequently a reduction in energy expenditure. This decrease in energy expenditure as a result of weight loss has been shown in several studies in children.^12,20

Production of TSH is also regulated by transmitters and hormones which regulate body weight and satiation, such as neuropeptide Y, α melanocyte stimulating hormone, and the agouti related peptide innervating hypophysiotropic TRH neurones.²¹ These transmitters are also influenced by leptin,²² which correlates with the degree of adipose tissue.¹¹ Some studies have described a correlation between leptin and TSH^2,6,7; indeed leptin has even been considered the mediator responsible for the increase in the production of TSH.^5,7 However, in our study as well as in one further study,⁴ no significant correlation between TSH and leptin could be identified, although in our analysis we used partial correlation adjusted to the degree of overweight, as both leptin and TSH correlated with the degree of overweight. As such there seems to be no direct connection between leptin and the thyroid hormones.

Surprisingly in our sample there was no significant decrease in TSH serum concentrations through weight loss as there was in most other studies.^12,23,24 The non-decrease of TSH despite weight loss could point to a neuroendocrine dysfunction as a cause of obesity, producing a bioinactive TSH²⁵ owing to a derangement in the hypothalamic-hypophysial axis, which also regulates body weight.¹⁵ Further causes for the non-decrease of TSH may be attributed to the small samples, to insufficient weight loss, to an improvement in the down regulated negative feedback from TSH to the peripheral thyroid hormones in obesity, or to the observations having taken place over too short a period of time. As the capacity of binding proteins of thyroid hormones changes with the state of nutrition,²⁶ the decrease in total concentration of T4 and T3 with unchanged TSH serum concentrations could therefore be a result of changes in the transport proteins, leaving the free amounts of peripheral thyroid hormones unchanged. In order to clarify this point, the free amounts of peripheral thyroid hormones and the transport proteins throughout long term weight reduction need to be determined.

Some studies have described a decrease in serum T3 concentrations a few weeks after weight loss,^12,27,28 while others reported constant T4 serum concentrations after weight loss.^12,23,24 All these studies dealt merely with short term changes after a few weeks on a low energy diet, indicating that a low energy diet leads to an increase in the production of biologically inactive rT3.^24,27,28 As both T3 and T4 decreased in our sample after weight loss on a normal energy diet, increased production of rT3 cannot be assumed. On the other hand, in our study total T4 decreased—but not significantly—even in the group of children not achieving weight loss. This points to other influencing factors of total T4. For example, the individuals in our study progressed into puberty during the course of the year, influencing thyroid binding.

Children taking any medication or having a medical condition which can affect binding proteins were excluded from the study. Measurement of total T3 and total T4 concentrations is therefore a good method to assess thyroid hormone status as measurement of free T3 in the obese showed the same results.^4,13

In summary, moderately increased TSH, total T3, and T4 serum concentrations are frequently found in obese children. There is no necessity for treatment of the increased thyroid hormones. A normal energy diet induces a long term decrease in the peripheral thyroid hormones as opposed to TSH. As in our study leptin showed no correlation with the thyroid hormones, it is doubtful that leptin is important as a mediator in thyroid changes. There is still a need for further research into the causes of the changes in thyroid hormones (especially TRH test, TSH, free T3, and T4) in obesity and weight loss.