Thyrotropin Levels,Insulin Resistance,and Metabolic Syndrome: A Cross-Sectional Analysis in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil)

Abstract

Background:

Previous studies have described an association with thyrotropin (TSH) levels, insulin resistance, and metabolic syndrome. We performed a cross-sectional analysis to investigate the relationship between TSH levels, insulin resistance, and metabolic syndrome using baseline data from the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil).

Methods:

Diabetics and individuals using medications that interfere in thyroid function were excluded, leaving 10,935 participants (54.3% women) for current analyses. Homeostasis Model Assessment of Insulin Resistance (HOMA-IR) values above the 75th percentile was considered as indicative of presence of insulin resistance. Logistic regression models were built using HOMA-IR and metabolic syndrome as the dependent variable, and quintiles of TSH as the independent variable (first quintile as reference). Odds ratios (OR) were presented with multivariate adjustment for socioeconomic/cardiovascular risk factors for insulin resistance, and adjustment only for socioeconomic factors and smoking for metabolic syndrome.

Results:

Age, body mass index, waist measurement, fasting glucose and fasting and post load insulin and HOMA-IR increased according to TSH quintiles. Subjects in the fifth TSH quintile presented an OR of association with insulin resistance of 1.86 [95% confidence interval (95% CI) 1.26–2.75], regardless of gender. For the metabolic syndrome, subjects in the fifth quintile presented an OR of 1.21 (95% CI 1.01–1.45) and remained positive only for men (OR 1.37; 95% CI 1.07–1.76). Restricting the analysis to quintiles of TSH in the normal range did not change the results.

Conclusions:

In this cross-sectional evaluation, high TSH quintiles were associated to insulin resistance/metabolic syndrome.

Introduction

Several studies have investigated the association between thyroid dysfunction and cardiovascular diseases. Recent prospective studies showed that subclinical hypothyroidism has been associated with increased risks of nonfatal and fatal coronary heart disease,¹ heart failure,^2,3 all-cause,⁴ and cardiovascular mortality.^1,4,5 One possible explanation for this association may be a higher frequency of aortic atherosclerosis,⁶ an atherogenic lipid profile,^7
–9 alterations in cardiac function,^10,11 or less studied factors, such as insulin resistance, among individuals with subclinical hypothyroidism than in those without this condition. Few studies with a small number of subjects, mainly women, have evaluated the relationship between subclinical hypothyroidism and insulin resistance, with conflicting results.^12
–14 The association of subclinical hypothyroidism and metabolic syndrome was also evaluated, especially in Asian populations, also with conflicting results.^15

–18 In regard to the association between insulin resistance, as for the association with metabolic syndrome, most studies categorize thyroid function as clinical or subclinical hypothyroidism. Few studies have specifically evaluated thyrotropin (TSH) levels and insulin resistance^19,20 or metabolic syndrome,^21

–24 and most of those also were performed in Asians.^21
–23

The Brazilian Longitudinal Study of Adult Health (ELSA-Brasil) is an ongoing prospective, multicenter cohort study in a Brazilian sample of civil servants. The ethnic diversity of the ELSA-Brasil sample, the high number of participants, the information at baseline examination of TSH levels, insulin resistance (homeostasis model assessment for insulin resistance, HOMA-IR), and detailed use of medicines makes ELSA-Brasil an ideal setting to evaluate the relationship of TSH (all range) and only TSH in the normal range, with insulin resistance and metabolic syndrome. Therefore, we performed a cross-sectional analysis to investigate the relationship between TSH levels, insulin resistance, and metabolic syndrome, using baseline data of ELSA-Brasil.

Methods

Subjects

The ELSA-Brasil study has been described previously.^25
–27 Briefly, 15,105 civil servants aged 35–74 years, from six cities in Brazil, were enrolled between August 2008 and December 2010. The aim of the study was to determine the incidence of cardiovascular diseases and diabetes and their associated risk factors. The ELSA-Brasil protocol was approved at all six centers by the Institutional Review Boards addressing research in human participants, according to the Declaration of Helsinki. All the participants signed an informed consent form.

Data collection

Each participant was interviewed in the workplace, and visited the Trial Center for clinical exams; according to standard protocols.²⁸ The interview and examination at each site were performed by trained personal, with strict quality control. The questionnaires included questions on age, sex, self-reported skin color/race (white, brown, black, Asian, and Indigenous), formal education (<9, 9–11, and >11 years), mean monthly income, smoking status (never, former and current), alcohol consumption, and physical activity during leisure time (low, mild, or vigorous), using the International Physical Activity Questionnaire—long form. All prescription and over-the-counter pill bottles were reviewed for medications taken in the last 15 days.²⁹

Blood pressure (BP) was taken using the validated Omron HEM 705CP INT oscillometric device. Three measurements were taken, at one-minute intervals. The mean of the two latest BP measurements was considered as the value for definition of hypertension. Body mass index (BMI) was calculated by dividing weight in kilograms by height in square meters. Hypertension was defined as previous use of medication to treat hypertension, and systolic BP ≥140 mm Hg, or diastolic BP ≥90 mm Hg. Dyslipidemia was defined as low density lipoprotein (LDL) cholesterol ≥130 mg/dL or use of cholesterol-lowering medications. All diabetics defined as using medication to treat diabetes or presenting a fasting plasma glucose ≥126 mg/dL, a 2-hour plasma glucose ≥200 mg/dL, or an HbA1C ≥6.5% were excluded from the analysis.

Definition of thyroid function, insulin resistance, and metabolic syndrome

Venous blood samples were obtained after fasting overnight. TSH levels were measured by a third generation immunoenzymatic assay (Siemens). Free thyroxine (FT4) levels were only evaluated in participants who presented with altered TSH levels. Therefore, if TSH levels were normal, FT4 levels were not obtained. We excluded from the analysis participants using thyroid hormones or antithyroid medications, such as propylthiouracil or methimazole, as well as other drugs that can interfere with thyroid function, such as amiodarone, carbamazepine, carbidopa, phenytoin, furosemide, haloperidol, heparin, levodopa, lithium, metoclopramide, propranolol, primidone, rifampicin, and valproic acid.^30,31 Cutoff levels for TSH were <0.4 mIU/L for hyperthyroidism and >4.0 mIU/L for hypothyroidism. Cutoff levels for FT4 were <0.8 ng/dL for hypothyroidism and >1.9 ng/dL for hyperthyroidism. Cutoff values for TSH and FT4 were similar to those used in the National Health and Nutritional Examination Survey³² and recommended by Surks et al.³³

Insulin resistance was estimated using the homeostasis model assessment for insulin resistance (HOMA-IR) based on the formula: fasting glucose (mg/dL)×insulin (μIU/mL)/405.³⁴ HOMA-IR values above the 75th percentile were considered elevated. For measurement of fasting and post-load glucose we used the hexokinase method (ADVIA 1200, Siemens); for fasting and post-load insulin, the immunoenzymatic assay; for glycated hemoglobin, high pressure liquid chromatography; for total, high density lipoprotein (HDL) cholesterol; and for triglycerides, the enzymatic colorimetric assay (ADVIA 1200, Siemens). LDL cholesterol was calculated using the Friedewald equation, except for cases with elevated triglyceride levels, when an enzymatic colorimetric assay was used (ADVIA 1200, Siemens). High-sensitivity C-reactive protein (hsCRP) was measured by immunochemistry (nephelometry, Siemens).

Metabolic syndrome was defined, according to the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III) criteria, as the presence of three of the following items: waist measurement >88 cm for women or 102 cm for men, HDL cholesterol <50 mg/dL for women or <40 mg/dL for men, systolic blood pressure ≥130 m Hg or ≥85 mm Hg, serum triglyceride levels ≥150 mg/dL, and fasting plasma glucose ≥100 mg/dL.

Statistical analysis

Categorical variables were expressed as proportions and compared using chi-square. Continuous variables were expressed as mean (standard deviation) and compared using analysis of variance. HOMA-IR and hsCRP were log transformed for comparison, but the values in the tables are back-transformed to their natural scales. For the main analyses, TSH levels were categorized in quintiles using all range distribution. Additional analysis restricted to quintiles of TSH within the normal range. Logistic regression models were built using HOMA-IR above 75th percentile or metabolic syndrome (MS) as the dependent variables, and the first quintile of the TSH was considered as the reference level. For insulin resistance, odds ratio (OR) and 95% confidence interval (95% CIs) are presented crude, age–sex adjusted, and with multivariate adjustment for age, sex, race, education, monthly income, BMI, waist measurement, hypertension, dyslipidemia, smoking, fasting plasma glucose ≥100 mg/dL, triglycerides ≥150 mg/dL, low HDL-cholesterol (<40 mg/dL for men and <50 mg/dL for women), and systolic blood pressure ≥130 mm Hg or diastolic blood pressure ≥85 mm Hg. For metabolic syndrome, OR and 95% CI are presented crude, age–sex adjusted and with multivariate adjustment for age, sex, race, education, monthly income, and smoking, as the other variables were represented by the inclusion of MS diagnosis in the model, and their inclusion would mask the putative association under study.

The analyses were performed using Statistical Packages for Social Sciences, version 22.0. A p-value<0.05 was considered as significant.

Results

Of the 15,105 participants, 49 had missing values for TSH or information on use of medication, leaving 15,056 eligible participants. Of these, 466 used medications that could alter thyroid function or interfere with TSH and FT4 tests, and 1,118 were using l-thyroxine or medication to treat hyperthyroidism and were also excluded from the analysis, resulting in 13,472 participants. We further excluded 2,537 participants with diabetes, leaving 10,935 participants with available information that were included in the analyses. Table 1 shows the general characteristics of participants according to quintiles of TSH. There was a statistically significant increase in age, BMI, waist circumference, fasting plasma glucose, glycated hemoglobin, fasting, and post-load insulin and HOMA-IR according to TSH quintiles. Individuals in the highest TSH quintiles tended to be of white race and current smokers. The proportion of participants with a HOMA-IR value above percentile 75 increased in the highest TSH quintiles, as well as, the proportion of participants with metabolic syndrome in the fourth and fifth TSH quintiles were higher than in the first, second and third TSH quintiles.

Table 1.

General Characteristics of the Sample According to the Presence of Subclinical Hypothyroidism

	Quintiles
	First n=2087	Second n=2287	Third n=2224	Fourth n=2267	Fifth n=2070	p	p for trend
Age (years), mean (SD)	50.4 (8.7)^a	50.1 (8.8)^a	50.4 (8.9)^a	50.6 (8.6)^a	52.1 (9.1)^a	<0.001	<0.001
Women, n (%)	1151 (55.2)	1213 (53)	1184 (53.2)	1216 (53.6)	1170 (56.5)	0.11
Race (%)						<0.001
White	925 (45)	1094 (48.4)	1178 (53.5)	1319 (58.8)	1259 (61.3)
Brown	638 (31)	705 (31.2)	628 (28.5)	590 (26.3)	535 (26)
Black	437 (21.2)	382 (16.9)	311 (14.1)	267 (11.9)	198 (9.6)
Asian	39 (1.9)	54 (2.4)	64 (2.9)	52 (2.3)	44 (2.1)
Indigenous	18 (0.9)	26 (1.1)	19 (0.9)	17 (0.8)	19 (0.9)
Education, n (%)						<0.001
Below high school	238 (11.4)	230 (10.1)	225 (10.1)	218 (9.6)	242 (11.7)
High school	780 (37.4)	824 (36)	736 (33.1)	716 (31.6)	692 (33.4)
College	1069 (51.2)	1233 (53.9)	1263 (56.8)	1333 (58.8)	1136 (54.9)
Monthly family income, n (%)						<0.001
US$
Up to 1245	596 (28.7)	602 (26.4)	538 (24.2)	532 (23.5)	516 (25)
1246–3319	799 (38.5)	875 (38.4)	818 (36.9)	845 (37.4)	808 (39.2)
≥3320	681 (32.8)	800 (35.1)	863 (38.9)	883 (39.1)	738 (35.8)
BMI (kg/m²), mean (SD)	26.2 (4.3)^b	26.3 (4.3)^b	26.4 (4.5)	26.5 (4.6)	26.7 (4.6)^b	0.002	<0.001
Waist circumference (cm), mean (SD)	88.9 (12.1)	89.2 (11.9)	89.6 (12.3)	89.5 (12.4)	89.9 (12.3)	0.07	0.004
Systolic blood pressure ≥130 or diastolic blood pressure ≥85 mm Hg, n (%)	508 (24.9)	611 (27.2)	579 (26.5)	577 (25.9)	538 (26.6)	0.51
Hypertension, n (%)	580 (27.8)	673 (29.4)	622 (28)	629 (27.7)	573 (27.7)	0.66
Dyslipidemia, n (%)	1115 (53.8)	1243 (54.7)	1238 (55.9)	1254 (55.5)	1194 (58.1)	0.07
Smoking, n (%)						<0.001
Never	1149 (55.1)	1352 (59.1)	1336 (60.1)	1384 (61)	1241 (60)
Past	563 (27)	609 (26.6)	592 (26.6)	631 (27.8)	640 (30.9)
Current	375 (18)	326 (14.3)	296 (13.3)	252 (11.1)	189 (9.1)
Fasting glucose ≥100 mg/dL, n (%)	1393 (66.7)	1565 (68.5)	1459 (65.6)	1440 (63.5)	1311 (63.3)	0.001
Low HDL-cholesterol, n (%)	105 (33)	148 (38.5)	128 (36.5)	149 (38.1)	123 (36.5)	0.60
Triglycerides ≥150 mg/dL, n (%)	484 (23.2)	604 (26.4)	608 (27.3)	656 (28.9)	630 (30.4)	<0.001
Metabolic syndrome,^c n (%)	281 (13.5)	341 (14.9)	330 (14.8)	378 (16.7)	332 (16)	0.04
Physical activity at leisure, n (%)						0.92
Low	1596 (77.1)	1727 (76.5)	1645 (75.4)	1683 (75.7)	1561 (76.7)
Mild	278 (13.4)	316 (14)	311 (14.3)	308 (13.8)	277 (13.6)
Vigorous	196 (9.5)	215 (9.5)	225 (10.3)	233 (10.5)	196 (9.6)
TSH (IU/mL), mean (SD)	0.70 (0.20)^d	1.14 (0.11)^d	1.56 (0.13)^d	2.21 (0.24)^d	5.16 (11.8)^d	<0.001	<0.001
Fasting plasma glucose, (mg/dL), mean (SD)	103.6 (8.4)^b	103.8 (8.3)^b	103.2 (8.7)	103.2 (8.6)	102.7 (8.6)^b	0.001	<0.001
Post-load blood glucose, mean (SD)	123.4 (28)	122.4 (28.6)	121.4 (27.8)	122.1 (28.1)	122 (27.8)	0.22	0.11
Glycated hemoglobin (mg/dL), mean (SD)	5.24 (0.53)^e	5.23 (0.55)	5.20 (0.52)	5.18 (0.51^e)	5.20 (0.51)	0.005	0.001
Fasting insulin (mcIU/mL), mean (SD)	6.2 (5.6)^f	6.8 (5.9)^f	7.4 (6.1)^f	8.2 (6.7)^f	8.3 (6.8)^f	<0.001	<0.001
Post-load insulin (μIU/mL), mean (SD)	50.7 (43.8)^g	55.1 (48.4)^g	56.8 (47.7)^g	61.2 (55)^g	62.4 (55.6)^g	<0.001	<0.001
HOMA-IR, mean (SD)	1.60 (1.49)	1.77 (1.59)	1.92 (1.64)	2.12 (1.81)	2.15 (1.83)	<0.001	<0.001
HOMA-IR ≥P75, n (%)	255 (12.2)	354 (15.5)	415 (18.7)	483 (21.3)	456 (22)	<0.001	<0.001
hs-CRP (mg/dL), mean (SD)	2.58 (4.80)^f	2.45 (3.48)^f	2.69 (5.05)^f	2.53 (3.86)^f	2.58 (3.52)^f	0.43	0.76

Data presented as mean (±standard deviation) or n (%), as indicated.

HDL cholesterol <40 mg/dL in men and <50 mg/dL in women.

First through fourth quintiles are≠fifth quintile, p<0.0001.

First and second quintiles are≠fifth quintile, p<0.02.

Metabolic syndrome defined according to NCEP ATP III.

All quintiles are≠from the others except for the second quintile, which is not different from the third quintile, p≤0.04.

First quintile≠fourth quintile, p=0.007.

All quintiles are≠from the others (p<0.03) except for the fourth quintile, which is not different from the fifth quintile.

All quintiles are different from the others (p≤0.04), except for the second quintile, which is not different form the third.

BMI, body mass index; HDL, high density lipoprotein; HOMA-IR, homeostasis model assessment for insulin resistance; hs-CRP, high-sensitivity C-reactive protein; P75, 75th percentile; TSH, thyrotropin; NCEP ATP III, National Cholesterol Education Program Adult Treatment Panel III.

Table 2 shows the results of logistic regression models using first quintile of TSH as reference. Subjects in the fifth quintile of TSH presented an OR of association with insulin resistance of 1.86 (95% CI 1.26–2.75) for all sample, of 2.44 (95% CI 1.86–3.21) for men, and of 1.81 (95% CI 1.37–2.38) for women. For metabolic syndrome, the OR was 1.21 (95% CI 1.01–1.45) considering all sample, and 1.37 (95% CI 1.07–1.76) for men. This association was not significant for women (OR 1.09; 95% CI 0.84–1.40). Restricting the analysis to TSH levels within the normal range, we found similar results for insulin resistance. The OR of association between the fifth quintile of TSH with insulin resistance was 1.67 (95% CI 1.08–2.58) for both sexes, 2.22 (95% CI 1.64–3.01) for men, and 1.90 (95% CI 140–2.59) for women. Regarding metabolic syndrome, the only association that remained significant was for men in the fifth quintile (OR 1.43; 95% CI 1.43–1.90). Neither association was found for women (OR 1.05; 95% CI 0.79–1.40) or for the whole sample (OR 1.22; 95% CI 0.99–1.49) (Table 3).

Table 2.

Odds Ratio and 95% Confidence Interval of an Association of Insulin Resistance (HOMA–IR) According to Quintiles of TSH Levels in Non–diabetic Subjects for All Sample and According to Sex

	Crude OR (95% CI)	Model 1 OR (95% CI)	Model 2 OR (95% CI)
Insulin resistance
All (10,935)
First quintile TSH	1.0 (reference)	1.0 (reference)	1.0 (reference)
Second	1.32 (1.11–1.57)	1.31 (1.10–1.56)	1.16 (0.79–1.70)
Third	1.65 (1.39–1.95)	1.64 (1.38–1.94)	1.43 (0.97–2.10)
Fourth	1.95 (1.65–2.30)	1.94 (1.65–2.29)	1.58 (1.08–2.32)
Fifth	2.03 (1.72–2.40)	2.03 (1.71–2.40)	1.86 (1.26–2.75)
Men (5001)
First quintile TSH	1.0 (reference)	1.0 (reference)	1.0 (reference)
Second	1.28 (1.01–1.63)	1.30 (1.02–1.64)	1.36 (1.03–1.79)
Third	1.63 (1.29–2.06)	1.64 (1.30–2.07)	1.66 (1.27–2.18)
Fourth	2.16 (1.72–2.71)	2.17 (1.73–2.71)	2.31 (1.77–3.01)
Fifth	2.30 (1.82–2.90)	2.28 (1.81–2.87)	2.44 (1.86–3.21)
Women (5934)
First quintile TSH	1.0 (reference)	1.0 (reference)	1.0 (reference)
Second	1.34 (1.04–1.72)	1.33 (1.03–1.72)	1.32 (0.99–1.74)
Third	1.65 (1.29–2.11)	1.64 (1.28–2.10)	1.68 (1.28–2.22)
Fourth	1.71 (1.34–2.18)	1.70 (1.33–2.18)	1.76 (1.33–2.32)
Fifth	1.82 (1.43–2.32)	1.78 (1.39–2.28)	1.81 (1.37–2.38)
Metabolic syndrome
All (10935)
First quintile TSH	1.0 (reference)	1.0 (reference)	1.0 (reference)
Second	1.13 (0.95–1.34)	1.14 (0.96–1.35)	1.18 (0.99–1.41)
Third	1.12 (0.94–1.33)	1.11 (0.94–1.32)	1.18 (0.99–1.40)
Fourth	1.29 (1.09–1.52)	1.28 (1.08–1.52)	1.35 (1.13–1.60)
Fifth	1.23 (1.03–1.46)	1.17 (0.98–1.39)	1.21 (1.01–1.45)
Men (5,001)
First quintile TSH	1.0 (reference)	1.0 (reference)	1.0 (reference)
Second	1.06 (0.83–1.35)	1.08 (0.85–1.38)	1.13 (0.88–1.45)
Third	1.19 (0.94–1.51)	1.20 (0.94–1.53)	1.26 (0.99–1.61)
Fourth	1.40 (1.11–1.78)	1.41 (1.11–1.79)	1.46 (1.14–1.85)
Fifth	1.35 (1.06–1.73)	1.31 (1.03–1.68)	1.37 (1.07–1.76)
Women (5934)
First quintile TSH	1.0 (reference)	1.0 (reference)	1.0 (reference)
Second	1.19 (0.94–1.51)	1.18 (0.93–1.50)	1.27 (0.99–1.62)
Third	1.04 (0.81–1.33)	1.03 (0.80–1.31)	1.13 (0.88–1.46)
Fourth	1.16 (0.92–1.48)	1.15 (0.90–1.46)	1.28 (0.99–1.64)
Fifth	1.13 (0.88–1.44)	1.02 (0.80–1.31)	1.09 (0.84–1.40)

Insulin resistance: Model 1, adjusted for age and sex; Model 2-, adjusted for age, sex (only for the model with both sexes), race/skin color, education, monthly family income, body mass index, waist measurement, hypertension, dyslipidemia, smoking, low HDL cholesterol, fasting plasma glucose ≥100 mg/dL, systolic blood pressure ≥130 mm Hg or diastolic blood pressure ≥85 mm Hg, and triglycerides ≥150 mg/dL.

Metabolic syndrome: Model 1, adjusted for age and sex; Model 2, adjusted for age, sex (only for the model with both sexes), race/skin color, education, monthly family income, and smoking.

Table 3.

Odds Ratio and 95% Confidence Interval of an Association of Insulin Resistance (HOMA-IR) According to Quintiles of TSH Levels in the Normal Range in Nondiabetic Subjects for All Sample and According to Sex

	Crude OR (95% CI)	Model 1 OR (95% CI)	Model 2 OR (95% CI)
Insulin resistance
All (10,935)
First quintile TSH	1.0 (reference)	1.0 (reference)	1.0 (reference)
Second	1.32 (1.11–1.58)	1.33 (1.11–1.58)	1.11 (0.75–1.64)
Third	1.65 (1.39–1.96)	1.65 (1.39–1.96)	1.36 (0.92–2.03)
Fourth	1.95 (1.65–2.31)	1.95 (1.64–2.32)	1.53 (1.04–2.26)
Fifth	1.95 (1.62–2.35)	1.92 (1.005–1.02)	1.67 (1.08–2.58)
Men (5001)
First quintile TSH	1.0 (reference)	1.0 (reference)	1.0 (reference)
Second	1.26 (0.99–1.61)	1.28 (1.002–1.63)	1.33 (1.003–1.76)
Third	1.61 (1.27–2.04)	1.61 (1.27–2.04)	1.64 (1.25–2.16)
Fourth	2.13 (1.69–2.68)	2.13 (1.69–2.68)	2.27 (1.73–2.97)
Fifth	2.13 (1.64–2.75)	2.11 (1.63–2.73)	2.22 (1.64–3.01)
Women (5934)
First quintile TSH	1.0 (reference)	1.0 (reference)	1.0 (reference)
Second	1.38 (1.06–1.79)	1.37 (1.05–1.78)	1.34 (1.00–1.79)
Third	1.70 (1.31–2.19)	1.69 (1.31–2.18)	1.74 (1.30–2.31)
Fourth	1.76 (1.37–2.27)	1.75 (1.36–2.26)	1.81 (1.36–2.41)
Fifth	1.83 (1.40–2.42)	1.80 (1.37–2.37)	1.90 (1.40–2.59)
Metabolic syndrome
All (10,935)
First quintile TSH	1.0 (reference)	1.0 (reference)	1.0 (reference)
Second	1.16 (0.98–1.39)	1.18 (0.99–1.40)	1.21 (1.01–1.45)
Third	1.15 (0.97–1.38)	1.15 (0.96–1.37)	1.20 (1.01–1.44)
Fourth	1.33 (1.12–1.57)	1.32 (1.11–1.57)	1.38 (1.15–1.64)
Fifth	1.22 (1.002–1.48)	1.17 (0.96–1.42)	1.22 (0.99–1.49)
Men (5001)
First quintile TSH	1.0 (reference)	1.0 (reference)	1.0 (reference)
Second	1.11 (0.86–1.42)	1.14 (0.89–1.47)	1.17 (0.91–1.51)
Third	1.25 (0.97–1.60)	1.26 (0.98–1.61)	1.31 (1.01–1.68)
Fourth	1.47 (1.15–1.87)	1.48 (1.16–1.89)	1.51 (1.18–1.93)
Fifth	1.42 (1.08–1.87)	1.40 (1.06–1.84)	1.43 (1.08–1.90)
Women (5934)
First quintile TSH	1.0 (reference)	1.0 (reference)	1.0 (reference)
Second	1.21 (0.95–1.55)	1.19 (0.93–1.52)	1.27 (0.99–1.63)
Third	1.06 (0.82–1.36)	1.03 (0.80–1.33)	1.14 (0.88–1.47)
Fourth	1.18 (0.93–1.51)	1.15 (0.90–1.48)	1.28 (0.99–1.65)
Fifth	1.07 (0.81–1.41)	0.98 (0.74–1.30)	1.05 (0.79–1.40)

Insulin resistance: Model 1, adjusted for age and sexModel 2, adjusted for age, sex (only for the model with both sexes), race/skin color, education, monthly family income, body-mass index, waist measurement, hypertension, dyslipidemia, smoking, low HDL cholesterol, fasting plasma glucose ≥100 mg/dL, systolic blood pressure ≥130 mm Hg or diastolic blood pressure ≥85 mm Hg, and triglycerides ≥150 mg/dL.

Metabolic syndrome: Model 1, adjusted for age and sex; Model 2, adjusted for age, sex (only for the model with both sexes), race/skin color, education, monthly family income, and smoking.

Discussion

Our results showed an association of insulin resistance with quintiles of TSH considering the entire range of TSH values and restricting the analysis to normal TSH values for men, women, and all sample. For metabolic syndrome, there was also an association with TSH quintiles, although weaker than with insulin resistance, and in the analysis by sex, it was positive only for men. Further restriction of the analyses to quintiles of TSH in the normal range did not alter the results for insulin resistance, but the association with metabolic syndrome remained positive only for men. All of the analyses were done excluding diabetics and participants using medication to treat thyroid disorders or medications that interfere with TSH tests.

Few studies have analyzed the association of thyroid function with insulin resistance and metabolic syndrome. For insulin resistance, most of them investigated small samples, mainly consisting of women, selected from tertiary care facilities or specialized clinics, presenting conflicting positive^12
–14 and negative^35,36 results. Furthermore, all these previous analyses categorized thyroid function as subclinical or clinical hypothyroidism. Compared with these small studies, ELSA-Brasil included a high number of apparently healthy men and women, enabling an evaluation of the association between TSH and insulin resistance in a large sample of nondiabetics, more similar to general population and using all range of TSH values. Therefore, this is a good opportunity to test this association in a large sample, in which the probability of a negative result consequently to an underpowered analysis is unlikely. Only two previous studies evaluate insulin resistance according to TSH levels. Consistent with our findings, Garduño-Garcia reported on a sample of 3148 Hispanics in whom, after adjusting for age and sex, TSH values had a positive correlation with HOMA-IR with borderline significance.¹⁹ In addition, Fernández-Real et al. evaluated the relationship between TSH levels and insulin sensitivity in 221 men using an intravenous glucose tolerance test with minimal model analysis. Baseline levels of FT4 and TSH were obtained before the glucose infusion. Serum TSH was positively associated with fasting and post-load insulin concentrations, especially in lean men with TSH levels within the normal range.²⁰ Our findings are complementary to these previous data from Mexican and Spanish samples but also highlight important differences since Brazil has mainly Portuguese ancestry.

However, our data contrasts with those of Lai et al., who studied 1534 participants in China and did not find any association between TSH levels and HOMA-IR considering only TSH in the normal range.²¹ Differences in study samples, such as the presence of individuals with diabetes and the analysis strategy used, could help explain some differences among the studies. It is interesting to observe that we found a positive association not only for all the TSH range, but also restricting the analysis to normal TSH values, even excluding diabetics from the analysis. Another important point is that in our study, the association between HOMA-IR and TSH levels remained significant for men and women, even after multivariate adjustment for all components of metabolic syndrome using the NCEP ATP III definition.

For metabolic syndrome, most studies found a positive association with subclinical hypothyroidism for both sexes^37,38 or only for women.¹⁸ However, some other studies did not find any association.^16,19 Although studies that evaluated the association between thyroid function and metabolic syndrome included a high number of participants, most of them also categorized thyroid function according to TSH and free thyroxine values, and restricted the evaluation to individuals with subclinical hypothyroidism compared with the present analysis that evaluated all range of TSH values in quintiles. Except for the study of Waring et al.,¹⁷ which included older subjects, and that of Garduño-Garcia,¹⁹, which included subjects from 18 to 70 years of age, the other four studies^15,16,18,37 included subjects with similar ages to those of ELSA-Brasil. Regarding ethnicity, three studies were performed on Asian populations, one in Turkey, one in the United States, and one in Mexico. Therefore, we evaluated this association in a population with different characteristics compared with previous studies. Our results for MS are not as clear as for insulin resistance. We only found an association of TSH quintiles with metabolic syndrome in the overall sample and in the men only sample. These results contrast with the study of Nakajima et al., who evaluated the presence of metabolic syndrome and subclinical hypothyroidism in women but not in men.¹⁸ However, our results for men are consistent with the findings of Lee et al., who found an association between high-normal TSH concentration (still within the normal range) and metabolic syndrome.²² Recently, Gyawali et al. have also found a higher frequency of thyroid dysfunction in patients with metabolic syndrome.³⁹

Our results could be compared with those of the study by Roos et al. in relation to TSH levels.⁴⁰ They studied 2703 adults without diabetes and with normal thyroid function in a mid-size town in the Netherlands to evaluate the association with HOMA-IR, fasting insulin, post-load insulin, and metabolic syndrome (NCEP III). They found an association between HOMA-IR and TSH levels, but the significance did not persist after adjustment for obesity. In our sample, the association between HOMA-IR and TSH levels remained significant after multivariate adjustment for BMI (obesity) and waist measurement (central obesity), and even when considering only individuals with TSH levels within the normal range.

Our study has some strengths. Although there have been few prevalence studies on the frequency of thyroid disorders, two population-based studies show a high frequency of thyroid disorders in Brazil^41,42 compared with other countries. ELSA-Brasil is also a cohort study with a high number of overweight and obese participants (2/3 of the sample). Therefore, we have a good setting to evaluate the association of TSH values with insulin resistance and metabolic syndrome in ELSA-Brasil. We excluded from the analyses subjects with a diagnosis of diabetes and we found a positive association between TSH levels and insulin resistance in nondiabetic participants, even after restricting the analysis to TSH levels within the normal range. HOMA-IR is a good measure of insulin resistance in epidemiologic studies, and there is a strong linear correlation with glucose clamp estimates in nondiabetics.⁴³ Although not representative of the whole Brazilian population, ELSA-Brasil is a multicenter study carried out in six Brazilian cities, in a different population compared with previous studies, and characterized by a high level of racial blending. However, this is a cross-sectional analysis using baseline data that do not enable evaluation of causality. Also, we only have information about FT4 levels in participants with altered TSH levels. Therefore, it is impossible to evaluate a possible association of FT4 with insulin resistance and metabolic syndrome in this sample.

Concluding, we found a positive association among TSH levels with insulin resistance in the overall sample, as well as for men and women separately. This finding remained significant after restricting the analysis to individuals within the normal TSH range. We also found a significant association between higher TSH quintiles and metabolic syndrome in men, but not in women.

Footnotes

Acknowledgments

The authors would also like to acknowledge the 15,105 individuals recruited for this study for their participation, without which the ELSA-Brasil cohort would not have been possible.

The ELSA-Brasil baseline study was supported by the Brazilian Ministry of Health (Science and Technology Department) and the Brazilian Ministry of Science and Technology (Financiadora de Estudos e Projetos and CNPq National Research Council) (grants 01 06 0010.00 RS, 01 06 0212.00 BA, 01 06 0300.00 ES, 01 06 0278.00 MG, 01 06 0115.00 SP, and 01 06 0071.00 RJ).

Author Disclosure Statement

No conflicting financial interests exist.

References

Rodondi

, den Elzen

, Gussekloo

. Subclinical hypothyroidism and the risk of coronary heart disease and mortality. JAMA, 2010; 304:1365–1374.

Rodondi

, Bauer

, Cappola

, et al. Subclinical thyroid dysfunction, cardiac function, and the risk of heart failure. JACC, 2008; 52:1152–1159.

Gencer

, Collet

T-H

, Virgini

, et al. Subclinical thyroid dysfunction and the risk of heart failure events: an individual participant data analysis from 6 prospective cohorts. Circulation, 2012; 126:1040–1049.

Sgarbi

, Matsumura

, Kasamatsu

, et al. Subclinical thyroid dysfunctions are independent risk factors for mortality in 7.5-year follow-up: the Japanese Brazilian Thyroid Study. Eur J Endocrinol, 2010; 162:568–577.

Asvold

, Bjoro

, Platou

, et al. Thyroid function and the risk of coronary heart disease: 12-year follow-up of the HUNT study in Norway. Clin Endocrinol, 2012; 77:911–917.

Hak

, Pols

, Visser

, et al. Subclinical hypothyroidism is an independent risk factor for atherosclerosis and myocardial infarction in elderly women: the Rotterdam Study. Ann Intern Med, 2000; 132:270–278.

Kuusi

, Taskinen

, Nikkila

. Lipoproteins, lipolytic enzymes and hormonal status in hypothyroid women at different levels of substitution. J Clin Endocrinol Metab, 1988; 66:51–56.

Arikan

, Bahceci

, Tuzcu

, et al. Postprandial hyperlipidemia in overt and subclinical hypothyroidism. Eur J Intern Med, 2012; 23:e141–145.

Hernandez-Mijares

, Jover

, Bellod

, et al. Relation between lipoprotein subfractions and TSH levels in the cardiovascular risk among women with subclinical hypothyroidism. Clin Endocrinol, 2013; 78:777–782.

10.

Kahaly

, Dillmann

. Thyroid hormone action in the heart. Endocr Rev, 2005; 26:704–708.

11.

Roef

, Taes

, Kaufman

, et al. Thyroid hormone levels within reference range are associated with heart rate, cardiac structure, and function in middle-aged men and women. Thyroid, 2013; 23:947–954.

12.

Singh

, Goswami

, Mallika

. Association between insulin resistance and hypothyroidism in females attending tertiary care hospital. Indian J Clin Biochem, 2010; 25:141–145.

13.

Guzel

, Seven

, Guzel

, et al. Visfatin, leptin, and TNF-α: interrelated adipokines in insulin-resistant clinical and subclinical hypothyroidism. Endocr Res, 2013; 38:184–194.

14.

Maratou

, Hadjidakis

, Kollias

, et al. Studies of insulin resistance in patients with clinical and subclinical hypothyroidism. Eur J Endocrinol, 2009; 160:785–790.

15.

Liu

, Scherbaym

, Schott

, et al. Subclinical hypothyroidism and the prevalence of the metabolic syndrome. Horm Metab Res, 2011; 43:417–421.

16.

Wang

C-Y

, Chang

T-C

, Chen

M-F

. Associations between subclinical thyroid disease and metabolic syndrome. Endocr J, 2012; 59:911–917.

17.

Waring

, Rodondi

, Harrison

, et al. Thyroid function and prevalent and incident metabolic syndrome in older adults: the Health, Ageing and Body Composition Study. Clin Endocrinol, 2012; 76:911–918.

18.

Nakajima

, Yamada

, Azuzawa

, et al. Subclinical hypothyroidism and indices for metabolic syndrome in Japanese women: the one-year follow-up study. J Clin Metab, 2013; 98:3280–3287.

19.

Garduño-Garcia J de

, Alvirde-Garcia

, López-Carrasco

, et al. TSH and free thyroxine concentrations are associated with differing metabolic markers in euthyroid subjects. Eur J Endocrinol, 2010; 163:273–278.

20.

Fernández-Real

, López-Bermejo

, Castro

, et al. Thyroid function is intrinsically linked to insulin sensitivity and endothelium-dependent vasodilation in healthy euthyroid subjects. J Clin Endocrinol Metab, 2006; 91:3337–3343.

21.

Lai

, Jiang

, Wang

, et al. The relationship between serum thyrotropin and components of metabolic syndrome. Endocr J, 2001; 58:23–30.

22.

Lee

, Oh

, Park

, et al. Serum TSH levels in healthy Koreans and the association of TSH with serum lipid concentration and metabolic syndrome. Serum TSH level in healthy Koreans and the association of TSH with serum lipid concentration and metabolic syndrome. Korean J Int Med, 2011; 26:432–439.

23.

Jee-Young

, Sung

Y-A

, Lee

. Elevated thyroid stimulating hormone levels are associated with metabolic syndrome in euthyroid young women. Korean J Intern Med, 2013; 28:180–186.

24.

Ruhla

, Weickertt

, Arafat

, et al. A high normal TSH is associated with the metabolic syndrome. Clin Endocrinol, 2010; 72:696–701.

25.

Aquino

, Barreto

, Bensenor

, et al. Brazilian Longitudinal Study of Adult Health (ELSA-Brasil): objectives and design. Am J Epidemiol, 2012; 175:315–324.

26.

Lotufo

. [Setting up the longitudinal study for adult health (ELSA-Brasil)]. Rev Saude Publica, 2013; 47(Suppl 2):3–9.

27.

Schmidt

, Duncan

, Mill

, et al. Cohort profile: Longitudinal Study of Adult Health (ELSA-Brasil). Int J Epidemiol, 2014, 44:68–75.

28.

Bensenor

, Griep

, Pinto

, et al. [Routines of organization of clinical tests and interviews in the ELSA-Brasil investigation center]. Rev Saude Publica, 2013; 47(Suppl 2):37–47.

29.

Chor

, Alves

, Giatti

, et al. [Questionnaire development in ELSA-Brasil: challenges of a multidimensional instrument]. Rev Saude Publica, 2013; 47(Suppl 2):27–36.

30.

Lai

, Yang

, Lin

, et al. Use of antiepileptic drugs and risk of hypothyroidism. Pharmacoepidemiol Drug Saf, 2013; 22:1071–1079.

31.

Dong

. How medications affect thyroid function. West J M, 2000; 172:102–106.

32.

Hollowell

, Staehling

, Dana Flanders

, et al. Serum TSH, T(4), and thyroid antibodies in the United States population (1988–1994): National Health and Nutritional Examination Survey (NHANES III). J Clin Endocrinol Metab, 2002; 87:489–499.

33.

Surks

, Ortiz

, Daniels

, et al. Subclinical thyroid disease—scientific review and guidelines for diagnosis and management. JAMA, 2000; 291:228–238.

34.

Matthews

, Hosker

, Rudenski

, Naylor

, Treacher

, Turner

. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia, 1985; 28:412–9.

35.

Tuzcu

, Bahceci

, Gokalp

, Tuzun

, Gunes

. Subclinical hypothyroidism may be associated with elevated high-sensitive c-reactive protein (low grade inflammation) and fasting hyperinsulinemia. Endocr J, 2005; 52:89–94.

36.

Al Sayed

, Al Ali

, Abbas

, et al. Subclinical hypothyroidism is associated with early insulin resistance in Kwaiti women. Endocr J, 2006; 53:653–657.

37.

Liu

, Scherbaym

, Schott

, et al. Subclinical hypothyroidism and the prevalence of the metabolic syndrome. Horm Metab Res, 2011; 43:417–421.

38.

Uzunlulu

, Yorulmaz

, Oguz

. Prevalence of subclinical hypothyroidism in patients with metabolic syndrome. Endocr J, 2007; 54:71–76.

39.

Gyawali

, Takanche

, Shrestha

, et al. Pattern of thyroid dysfunction in patients with metabolic syndrome and its relationship with components of metabolic syndrome. Diabetes Metab J, 2015; 39:66–73.

40.

Roos

, Bakker

, Links

, et al. Thyroid function is associated with components of the metabolic syndrome in euthyroid subjects. J Clin Endocrinol Metab, 2007; 92:491–496.

41.

Benseñor

, Goulart

, Lotufo

, et al. Prevalence of thyroid disorders among older people: results from the São Paulo Ageing & Health Study. Cad Saude Publica, 2011; 27:155–161.

42.

Sichieri

, Baima

, Marante

, et al. Low prevalence of hypothyroidism among black and Mulatto people in a population-based study of Brazilian women. Clin Endocrinol (Oxf), 2007; 66:803–807.

43.

Muniyappa

, Lee

, Chen

, et al. Current approaches for assessing insulin sensitivity and resistance in vivo: advantages, limitations, and appropriate usage. Am J Physiol Endocrinol Metab, 2008; 294:E261–E270.