The Relationship Between TSH Levels and Body Mass Index in Overweight Euthyroid Subjects : Role of Autoimmunity

Summary: Also if a positive correlation between TSH and BMI has been observed in the majority of the studies, data on this topic remain controversial, especially when TSH values are in normal range. However, it is reasonable to argue that the high-normal levels of TSH found in obesity can be partly attributed to a silent autoimmune thyroiditis. The aim of the present study was to evaluate the interconnection between TSH and autoimmunity in obese patients with TSH values within the normal range. Methods: We enrolled 891 overweight and obese patients (209 Males), with TSH values in normal range (mean age ± SD: 45.1 ± 15 years. Anti thyroid auto antibodies(ATAs: thyroid-peroxides auto antibodies (TPOAb) and/or thyroglobulin auto antibodies(TgAb) and TSH were measured in all patients. The entire sample was then subdivided according to the tertiles of TSH within the range between 0.4 and 4.1 μUI/L. Results: No significant correlation between TSH and BMI was found within these groups. According to the presence (ATA+) or absence (ATA-) of ATAs, we found 225 (25,4%) ATA+ (32 males) and 665 ATA(177 males) patients. A statistically significant correlation between TSH and BMI was found only in females with ATA positivity (Pearson: r = 0.1426, p= 0.04; Spearman: r = 0.1616, p = 0.02). Conclusion: In the initial assessment of obesity, the determination of TSH could be associate to ATAs, for identifying, among the patients with increased TSH, those affected by autoimmune thyroiditis. We suggest that in these patients, the beginning of LT4 therapy could be taken into account.


Introduction
Overweight and obesity are defined as "abnormal or excessive fat accumulation that may impair health" [1].This condition leads to adipocyte hypertrophy, resulting in dysfunction of white adipose tissue, development of hypoxia, oxidative stress and inflammation [2,3].Fat tissue dysfunction can lead to metabolic changes in several organs and systems, among which in the endocrine system where changes in plasma levels, secretion patterns and clearance of different hormones are observed [4,6].
As far as thyroid function is concerned, literature data suggest either a possible influence of obesity on the hypothalamicpituitary-thyroid axis (HPT) or vice versa of thyroid on the development and maintenance of obesity [7][8][9][10][11][12][13][14][15].It has also been documented that even minimal changes in the HPT axis and/or in thyroid function with hormone levels in the reference range can be positively associated with components of metabolic syndrome and an increase in the risk of thyroid cancer [16][17][18][19][20].
In addition, specific TSH receptors have recently been found on adipose tissue, suggesting a direct role of TSH on adipocyte function and energy expenditure [21].
Several clinical and population studies have shown positive correlation between TSH and BMI, but this is in contrast with other clinical and population studies, in which this correlation was not found .
Nevertheless, few studies have considered autoimmunity in the context of the complex relationship between excessive weight and thyroid disease [10,33,[41][42][43].This might partly explain the discordance of literature data on the relationship between BMI and TSH.With regard to this, the association between body mass and subclinical autoimmune hypothyroidism would seem to be conceivable in patients with less pronounced adiposity, since it was not observed in a study that selected patients with Class III obesity and elevated levels of TSH.Moreover, Rotondi et al [43].showed that thyroid ultrasound in obese patients is a confounding factor in the diagnosis of HT, so that only auto antibodies should be used to identify the underlying thyroid disease.Given that in Sardinia there is a high prevalence of autoimmune diseases and an increasing prevalence of obesity, the greater prevalence of both conditions, could allow a greater statistical power to the study of the relationship between TSH levels and autoimmunity in obese and overweight patients [67][68][69][70][71]72].In the present study we considered exclusively patients with TSH values within the conventional normal range in order to elucidate the relationship between excessive body weight and high-normal TSH levels and the influence of autoimmunity on this relationship.

Patient's Enrolment
From a total population of 1246 outpatients consecutively afferent at the Obesity Operating Unit (University of Cagliari; Italy) from 2014 to 2016; we excluded 13 patients with known history of hypothyroidism or hyperthyroidism; 184 patients with diabetes mellitus; [73]15 underweight and 22 normal weight patients.
Thus the study group consisted of 891 patients (209 Males); with TSH values between 0.4 and 4.1 μUI/L.None subjects was taking drugs affecting thyroid function [66] The age of males ranged between 19 and 78 years (mean ± SD 45.1 ± 15 years); the age of females ranged between 13 and 80 years (mean ± SD 44.8 ± 14.8 years).

Thyroid Evaluation
Venous blood samples for hormonal and antibodies determinations; as well as for the above mentioned blood chemical parameters; were taken from the antecubital vein of the arm in fasting condition between 08:00 and 09:00 a.m.; hormone values and antibodies were tested using commercial kits.

Statistical Analysis
Results for quantitative variables are expressed as mean ± SD for normally distributed data; and mean (Interquartile range) for non-parametric data.The regulation at normal distribution was tested with the Kolmogorov-Smirnov (K-S) test.
Data were analysed by using Student's t-test for unpaired data and relationship between quantitative variable by regression analysis with Pearson and Spearman correlation methods.
For all tests; a p-value < 0.05 was considered as a limit of significance.
The statistical analysis was performed using commercial software (Graph-Pad; Prism).

Results
TSH concentrations (Mean ± SD) among patients categorized in 4 groups on the basis of BMI were found to be overlapped in all groups (Table 1).(The entire sample was then subdivided according to the tertiles of TSH within the range between 0.4 and 4.1 μUI/L; the comparison of the BMI mean values in the three subgroups thus obtained did not show any significant statistical difference: 1) In the 1st tertile (TSH: 0.4-1.36μUI/L) including 251 patients (75 Males) the BMI was found to be 35.4 ± 6.68 Kg/m²; 2) In the 2nd tertile (TSH: 1.37-2.73μUI/L) including 477 patients (107 Males) the BMI was found to be 35.7 ± 6.71 Kg/m²; 3) In the 3rd tertile (TSH: 2.74-4.1 μUI/L) including 163 patients (27 Males) the BMI was found to be 35.7 ± 6.48 Kg/m².
No statistically significant correlation between TSH and BMI was found within these groups (data not shown).

Discussion
In our work we found that a significant; yet weak; positive correlation between TSH and BMI is limited to ATA positive female patients.
This result can be justified by the observations of Spencer et al. in NHANES study [74]: for each age and ethnic group; the inclusion of antibody positive subjects shifted the medians and upper limits of TSH (97.5 percentile) to the right; while it had no effect on the lower limits (2.5 percentile).Moreover; it is well known that thyroid autoimmunity is more frequent in females [64][65][66].
The close association between the presence of Hashimoto's thyroiditis and increased levels of TSH has been strangely neglected or underestimated by the majority of authors who studied the complex relationships between thyroid and excessive weight.
In fact; only five studies; out of which three clinical trials and two performed in general population; evaluated the presence of ATAs and consequently verified whether the correlation between TSH and BMI was affected.Regarding the clinical trials; in the study of Manji et al. on 267 British euthyroid patients; tested for ATA out of whom only 35 resulted positive; no correlation between TSH and BMI was found.Also Michalaki et al. in the same year; did not observe any correlation between TSH and BMI in 144 large obese people of Greek ethnicity; selected for euthyroidism and ATA negativity.Finally in a subsequent study conducted by Rotondi et al. on 350 obese Italian subjects; out of whom 40 were ATA positive; no correlation between TSH and BMI was observed; even though the average TSH values were higher in obese than in the normal-weight Individuals [41][42][43].However; the authors did not separate positive ATAs patients (numerically poorly represented) from negative ATAs patients.Furthermore; the prevalence of ATA in obese subjects overlapped that observed in the general population; while in patients with subclinical hypothyroidism the authors reported a higher prevalence of ATA in normal weight (66.1%) than in obese individuals (32.1%).They concluded that autoimmunity weakly influences the behaviour of TSH in obesity.
In partial accordance with our data; Diez et al. examined 778 Spanish obese euthyroid patients; enrolled in a "Thyroid clinic" [75].ATA were measured on 412 patients; out of whom 46 were positive.A significant correlation between TSH and BMI was found in the obese patients of both genders but no correlation was observed in negative ATA patients; although the number of positive patients was really small.In addition; the influence of thyroid autoimmunity on the relationship between BMI and TSH was underlined by the higher TSH values in patients with positive autoimmunity compared to patients with negative autoimmunity.
Inconsistent results on the relationship between serum TSH and obesity are also reported in population studies.In 2005; Knudsen et al. examined a population of 4082 aged 18-65 years Danish people.The sample was drawn among women in the age groups 18-22; 25-30; 40-45; and 60-65 years; with the aim to represent women before child bearing age; within childbearing age; after childbearing age but premenopausal; and postmenopausal [33].A group of men aged 60-65 years was included for comparison between genders.The preponderance of women chosen for this study was due to the higher prevalence of thyroid abnormalities expected in female sex; with the aim to increase the statistical power of the study at the lowest cost.Likewise; the group of men was chosen in the age group with the highest expected prevalence of thyroid abnormalities.
These authors found a positive correlation between TSH and BMI; but; although ATAs were assessed; the results are surprisingly not reported; and therefore it is not possible to highlight the potential influence of ATA.
In the study of Kitahara et al. [10] on 3114 Americans over the age of 20 years; a positive correlation between TSH and BMI was found; but excluding the 325 ATA positive subjects; it surprisingly decreased significantly in females; while remained unchanged in males.Finally; although Diez et al [75] opens a key reading of these conflicting results; arguing that autoimmunity can influence the correlation between TSH and BMI; he does not expose a sufficient demonstration of this assumption.In fact; the number of positive ATA subjects is too low to justify the great difference in the results obtained after the exclusion of the positive ATAs.
In our study; the presence of thyroid autoimmunity in obese and overweight patients; similar to that observed in our previous study performed in the general population [68]; influences the relationship between TSH and body mass.
The high prevalence of autoimmunity in Sardinia; allowed to highlight an association between excessive weight and autoimmune thyroid disease; which in other ethnicities with a lower prevalence was not observed and therefore denied [43].However; the observed association was weak.This can be partly explained by the range restriction of TSH values; considering that the value of r can be lower if the variability among the observations is small [68].A possible limitation of the present study is the lack of thyroid hormones assay in our patients.However; considering that we enrolled only subjects with normal TSH levels; and given that the central hypothyroidism is extremely rare in the population; we considered this analysis not relevant.Furthermore; a considerable variability of thyroid hormones; as a result of the nutritional status; has been reported in obesity [15].
In overweight and obese patients; the determination of TSH should be associated to ATAs; for identifying; among the patients with increased TSH; those affected by autoimmune thyroid disease.The low employment of this diagnostic tool and the resulting lack of this information could partly explain the discrepancy of the results obtained by the researchers with regard to the relationship between TSH and BMI in obese population.
Particularly in obese ATA positive women; high normal TSH levels could be attributed to a slight thyroid failure; while in all other cases obesity per sè could play a pivotal role [76]; confirming in a larger number of obese patients; the hypothesis of Diez et al (68) that the relationship between TSH and BMI could be due to the presence of underlying autoimmune thyroid disease.This consideration is supported by the reduction of TSH levels in obese patients undergoing bariatric surgery with exception of those affected by autoimmune thyroid disease [73;76;77] .

Conclusion
In conclusion our results suggest the importance of the systematic evaluation of thyroid autoimmunity in addition to that of TSH; in the analysis of the complex relationship between thyroid and body weight.

Table 1 :
TSH concentrations (M ± SD) among patients in the various