Relationships Between Thyroid Function and Lipid Status or Insulin Resistance in a Pediatric Population

Abstract

Background:

In adults without thyroid disease, increasing levels of thyroid-stimulating hormone (TSH) within the range of that considered normal have been shown to be associated with increases in total cholesterol, low-density lipoprotein cholesterol, non-high-density lipoprotein cholesterol, and triglycerides, and with decreases in high-density lipoprotein cholesterol. Serum TSH has also been found to be positively associated with fasting and postload insulin concentrations and negatively associated with insulin sensitivity in euthyroid adults. We hypothesized that such relationships also exist in euthyroid children and adolescents.

Methods:

This was a retrospective record review of pediatric outpatients (ages 2–18 years) having measurements of TSH or free thyroxine (T4) and a concurrent lipid panel, fasting glucose, or fasting insulin. Pearson correlation coefficient was used to estimate the correlation between TSH or free T4 and logarithmic transformed lipid, plasma glucose, or insulin levels. Lipid levels, fasting plasma glucose, insulin, and homeostasis model assessment (HOMA) were also compared between subjects with TSH levels in the high normal range (2.5–5 mIU/L) and those with TSH in the low normal range (0.3–2.4 mIU/L).

Results:

TSH levels were positively correlated with triglyceride levels (r = 0.10, p = 0.001). Conversely, free T4 levels were inversely correlated with triglyceride levels (r = −0.10, p = 0.011). TSH levels were also positively correlated with fasting insulin (r = 0.26, p = 0.002) and with HOMA (r = 0.27, p = 0.001). These associations remained significant after adjustment for age, gender, and body mass index z-score. Children who had TSH levels between 2.5 and 5.0 mIU/L had higher triglycerides (p = 0.003), insulin levels (p = 0.040), and HOMA (p = 0.021) than those having TSH values between 0.3 and 2.4 mIU/L.

Conclusions:

In euthyroid children without a history of hypo- or hyperthyroidism, increasing levels of TSH and decreasing levels of free T4 are associated with higher triglyceride levels and elevated markers of insulin resistance. Whether these findings carry implications regarding optimal TSH levels in children at increased risk for cardiovascular disease awaits further study.

Introduction

H ypothyroidism is associated with an increased risk of coronary artery disease (1). This observation is partly related to the hyperlipidemia known to be present in patients with both primary and secondary hypothyroidism (2). Subclinical hypothyroidism, characterized by elevated thyroid-stimulating hormone (TSH) and normal thyroxine (T4) levels, has also been identified as a potential risk factor for cardiovascular disease and mortality in adults (3). Several cross-sectional studies have reported increased levels of total cholesterol and low-density lipoprotein (LDL) cholesterol in adults with subclinical hypothyroidism compared with euthyroid controls (4). In adults, increasing levels of TSH within the range of that is considered normal have been shown to be associated with increases in total cholesterol (5,6), LDL cholesterol (6,7), non-high-density lipoprotein (HDL) cholesterol, and triglycerides (6), and with decreases in HDL cholesterol (6,7).

In healthy euthyroid adults, both serum TSH and the freeT4*TSH product have been found to be positively associated with fasting and postload insulin concentrations and negatively associated with insulin sensitivity (8). It is unknown whether these same associations are present during childhood or develop later in life. We undertook this study to determine relationships between thyroid function and lipid concentrations or measures of insulin resistance in the pediatric population. We also wanted to determine whether there might be differences in lipid levels and insulin resistance in children with TSH in the upper normal range compared with those having TSH in the lower normal range.

Materials and Methods

Patients

All subjects were evaluated in the pediatric outpatient clinics at Mayo Clinic, Rochester, MN, between January 1, 2004, and September 30, 2007. Medical records of patients, aged 2–18 years, who had simultaneous measurement of TSH or free T4 and a fasting lipid panel (total cholesterol, HDL cholesterol and triglycerides), or fasting glucose, or fasting insulin level were reviewed. In patients with multiple measurements of TSH and/or free T4 during the study period, only the first measurement was included in the analysis (“index date”). Subjects found to have diabetes mellitus within 6 months before or after the index date, and those receiving thyroid hormone replacement therapy within 30 days before or after the index date were excluded. The closest body mass index (BMI) measurement within 30 days of the index date was utilized. Patients without a BMI measurement in their medical record were excluded. The Mayo Clinic Institutional Review Board approved the retrospective review of patients' medical records, and records were reviewed for research purposes in accordance with Minnesota statute 144.335.

Data were collected on patients, including age, sex, height, weight, and, in a subset of subjects, primary indication for thyroid function tests. Presenting signs and symptoms were quite variable. In a randomly selected sub-sample of 200 of our study subjects, the most common indications for thyroid function evaluation were multiple somatic complaints (33 subjects), psychiatric complaints (29 subjects), obesity or weight gain (27 subjects), and gastrointestinal complaints (25 subjects).

Age- and sex- specific BMI percentiles were determined using the 2000 Center for Disease Control growth charts (9). In addition, the BMI z-scores of the subjects were determined using the age-specific and sex-specific median BMI, generalized coefficient of variation (S), and the power of the Box-Cox transformation (L) by the following formula: {[(BMI/median BMI)^L] −1}/(L × S), based on U.S. Centers for Disease Control and Prevention growth curves (10).

Laboratory methods

All measurements were performed at Mayo Medical Laboratories. Serum concentrations of TSH were measured using an immunoenzymatic assay (Beckman Coulter Ireland Incorporated, Ireland). Inter-assay coefficients of variation (CVs) for TSH are 8% for 0.41 mIU/L and 3.3% for 17.1 mIU/L, respectively. Before January 9, 2006, TSH was measured by a two-site chemiluminescent sandwich immunoassay on the ACS-180 automated immunoassay system (Bayer Diagnostics Corp.). Inter-assay CVs were 7%, 3%, and 4% at 0.6, 4.6, and 15.5 mIU/L, respectively. In method comparison studies between the Beckman and Centaur the slope and intercept of DXI = 0.9567 × (CENTAUR) + 0.0599 was observed using linear regression analysis. The correlation coefficient between the two methods was R² = 0.9989. Free T4 levels were measured by chemiluminometric assay (Bayer Diagnostics). Intra-assay CVs for free T4 are 7.35% at 0.48 ng/dL, 4.7% at 1.01 ng/dL, and 2.05% at 3.52 ng/dL, respectively. The laboratory reference ranges were 0.3–5.0 mIU/L for TSH and 0.8–1.8 ng/dL for free T4.

Total cholesterol was measured on the Hitachi 912 chemistry analyzer using Technicon cholesterol reagent (Bayer Corp.). Intra-assay CVs are 0.9% and 0.8% at 121 and 196 mg/dL, respectively. Inter-assay CVs are 1.6%, 1.8%, and 1.7% at 159, 207, and 280 mg/dL, respectively. Triglycerides were measured on the Hitachi 912 chemistry analyzer using Technicon triglyceride reagent (Bayer Corp.). Intra-assay CVs are 1.4% and 1.0% at 97 and 215 mg/dL, respectively. Inter-assay CVs are 2.9%, 3.3%, and 2.9% at 78, 132, and 214 mg/dL, respectively. HDL cholesterol was measured on the Hitachi 912 chemistry analyzer using direct HDL-C plus reagent (Roche Diagnostics). HDL cholesterol intra-assay CVs are 0% and 0.4% at 25 and 84 mg/dL, respectively. Inter-assay CVs are 2.2%, 1.1%, and 4.3% at 41, 112, and 25 mg/dL, respectively. Non-HDL cholesterol was calculated using the following formula: Non-HDL cholesterol = Total cholesterol −HDL cholesterol.

Glucose was measured on the Cobas Mira using the hexokinase reagent from Roche Diagnostics. Intra-assay CVs are 0.6% at 85 mg/dL and 1.0% at 310 mg/dL. Inter-assay CVs are 4.7% at 64 mg/dL and 1.8% at 280 mg/dL. Insulin in the initial period of the study was measured by a two-site immunoenzymatic assay performed on the DxI automated immunoassay system (Beckman Instruments). Inter-assay CVs are 6.2% at 5.3 μU/mL, 6.5% at 46.1 μU/mL, and 7.7% at 120.4 μU/mL. Insulin was later performed by electochemiluminescence assay (Roche Diagnostic Corp.). In our method comparison studies between the Beckman and Roche, the slope and intercept of ROCHE = 1.08 × (DXI) +0.46, R² = 0.99 was observed using linear regression analysis. The correlation coefficient between the two method was R² = 0.99.

Reference ranges for the homeostasis model assessment (HOMA)-insulin resistance (IR) was calculated using the formula: HOMA-IR = fasting insulin concentration (μU/mL) × fasting glucose concentration (mg/dL)/405.

Statistical analyses

The distribution of each laboratory parameter (TSH, free T4, total cholesterol, HDL cholesterol, non-HDL cholesterol, triglycerides, fasting glucose, fasting insulin, and HOMA) was assessed graphically, and each of these measurements was then transformed using the natural logarithmic transformation to obtain a more normal distribution. Simple correlations were summarized using the Pearson correlation coefficient. The relationship between TSH and each of the laboratory parameters (dependent variable) was evaluated in a separate general linear model. Each model was assessed for two-way interaction effects with gender, age, and BMI z-score. In the absence of any statistically significant interaction effects, the models were adjusted for these three factors. Since the dependent variable was on a log scale, the parameter estimate for TSH is interpreted as the percentage change in the dependent variable per a 1 mIU/L increase in TSH. Similar analyses were performed for free T4. The distribution of each laboratory parameter (total cholesterol, triglycerides, HDL cholesterol, non-HDL cholesterol, glucose, insulin, and HOMA) was also compared using the Wilcoxon rank sum test between subjects with TSH levels between 0.30 and 2.4 versus 2.5 and 5.0 mIU/L. All calculated p-values were two-sided and p-values less than 0.05 were considered statistically significant. Analyses were performed using the version 9.1 SAS software package (SAS Institute, Inc.).

Results

TSH and lipids

TSH and lipids were measured concurrently in 1210 children and adolescents, of which 12 (1%) had TSH below the reference range, and 52 (4.3%) had TSH above the reference range. The 1146 (94.7%) subjects having TSH within the reference range (0.3–5.0 mIU/L) formed the basis for the analysis (Table 1). Increasing concentrations of TSH were correlated with increasing concentrations of triglycerides (r = 0.10, p = 0.001, Table 2). The relationship between TSH and triglycerides remained significant after adjustment for gender, age, and BMI z-score (i.e., nonsignificant interaction effects). After adjusting for these factors, the percent increase in triglycerides per a 1 mIU/L increase in TSH was 4.19 (95% CI, 1.25–7.13; p = 0.005). There were no significant relationships found between TSH and any of the other lipid parameters (total cholesterol, HDL cholesterol or non-HDL cholesterol, Table 2).

Table 1.

Summary of Characteristics of Patients with Normal Thyroid-Stimulating Hormone or Normal Thyroxine

	Patients with TSH and lipids (n = 1146)	Patients with TSH and fasting glucose (n = 2590)	Patients with T4 and lipids (n = 578)	Patients with T4 and fasting glucose (n = 1252)
Female, n (%)	659 (57.5)	1489 (57.5)	319 (55.2)	720 (57.5)
Age (years), mean (SD)	14.2 (4.0)	13.5 (4.4)	13.1 (4.0)	12.7 (4.5)
Age group, n (%)
2 to <10 years	179 (15.6)	564 (21.8)	128 (22.2)	334 (26.7)
10 to <19 years	967 (84.4)	2026 (78.2)	450 (77.8)	918 (73.3)
BMI z score, mean (SD)	0.92 (1.38)	0.58 (1.38)	1.09 (1.41)	0.58 (1.47)
interquartile range	0.05–2.04	−0.26–1.64	0.16–2.20	−0.32–1.68
range	−5.35–5.93	−6.09–5.92	−4.94–5.93	−5.08–6.61
BMI category, n (%)
<5th percentile	45 (3.9)	159 (6.1)	19 (3.3)	89 (7.1)
5th to 84th percentile	541 (47.2)	1432 (55.3)	243 (42.0)	665 (53.1)
85th to 94th percentile	140 (12.2)	355 (13.7)	71 (12.3)	174 (13.9)
≥95th percentile	420 (36.7)	644 (24.9)	245 (42.4)	324 (25.9)

BMI, body mass index; T4, thyroxine; TSH, thyroid-stimulating hormone.

Table 2.

Relationship Between Thyroid-Stimulating Hormone in the Normal Range and Lipid Levels and Measures of Insulin Resistance

	Total serum cholesterol (n = 1146)	HDL cholesterol (n = 1146)	Non-HDL cholesterol (n = 1146)	Triglycerides (n = 1146)	Fasting glucose (n = 2590)	Fasting insulin (n = 131)	Homeostatic model assessment (n = 131)
Pearson correlation coefficient	0.02	−0.03	0.04	0.10	0.04	0.26	0.27
Unadjusted percent change^a	0.46	−0.93	1.08	4.98	0.26	17.49	19.35
95% CI	(−0.81, 1.73)	(−2.51, 0.64)	(−0.68, 2.84)	(1.96, 7.99)	(−0.14, 0.67)	(6.36, 28.63)	(7.54, 31.15)
p-value for linear trend	0.48	0.24	0.23	0.001	0.20	0.002	0.001
Adjusted^b percent change	0.26	−0.15	0.44	4.19	0.21	14.54	15.70
95% CI	(−1.00, 1.52)	(−1.67, 1.36)	(−1.29, 2.17)	(1.25, 7.13)	(−0.20, 0.61)	(4.30, 24.78)	(5.01, 26.40)
p-value for linear trend	0.69	0.84	0.62	0.005	0.32	0.005	0.004

Percent change in laboratory parameter per each 1 mIU/L increase in TSH.

Adjusted for gender, age, and BMI z-score.

HDL, high-density lipoprotein.

TSH and markers of insulin resistance

A total of 2590 children and adolescents had simultaneous measurement of TSH within the reference range and fasting plasma glucose (Table 1). A subset of 131 subjects had simultaneous measurements of TSH, fasting insulin, and glucose. Increasing concentrations of TSH were significantly correlated with increasing concentrations of insulin (r = 0.26, p = 0.002) and HOMA (r = 0.27 and p = 0.001), Table 2). The relationship between TSH and these laboratory parameters did not vary significantly with gender, age, and BMI z-score. After adjusting for these factors, the percent increase in insulin per each 1 mIU/L increase in TSH was 14.54 (95% CI, 4.30–24.78; p = 0.005) and the percent increase in HOMA per each 1 mIU/L increase in TSH was 15.70 (95% CI, 5.01–26.40; p = 0.004).

TSH 0.30–2.4 versus 2.5–5.0 mIU/L

As expected, the BMI was lower in subjects with TSH levels between 0.3 and 2.4 mIU/L relative to those with TSH between 2.5 and 5.0 mIU/L. In the 1146 children with simultaneous measurement of TSH within the reference range and lipids, the median BMI was significantly lower in subjects with TSH levels between 0.30 and 2.4 mIU/L when compared to those with TSH levels between 2.5 and 5.0 mIU/L (BMI percentile 81.6 vs. 88.5, BMI z-score 0.90 vs. 1.20, Wilcoxon rank sum test p < 0.001). Similarly, in the 2590 children and adolescents with simultaneous measurement of TSH within the reference range and fasting plasma glucose, the median BMI was significantly lower in subjects with TSH levels between 0.30 and 2.4 mIU/L when compared to those with TSH levels between 2.5 and 5.0 mIU/L (BMI percentile 71.9 vs. 76.9, BMI z-score 0.58 vs. 0.74, Wilcoxon rank sum test p = 0.006).

The median triglyceride level was significantly lower in subjects with TSH levels between 0.30 and 2.4 mIU/L when compared to those with TSH levels between 2.5 and 5.0 mIU/L (87 mg/dl vs. 98 mg/dl; p = 0.003, Fig. 1). The median insulin level was significantly lower in subjects with TSH levels between 0.30 and 2.4 mIU/L (13.0 uIU/mL vs. 16.0 uIU/mL; p = 0.040, Fig. 2) when compared to those with TSH levels between 2.5 and 5.0 mIU/L. The median HOMA was also significantly lower in subjects with TSH levels between 0.30 and 2.4 mIU/L (2.6 vs. 3.7; p = 0.021, Fig. 2) when compared to those with TSH levels between 2.4 and 5.0 mIU/L.

FIG. 1.

Comparison of total cholesterol, high-density cholesterol, non-HDL cholesterol, and triglycerides between subjects with TSH 0.3–2.4 mIU/L and those with TSH 2.5–5.0 mIU/L) (*p = 0.003). HDL, high-density lipoprotein; TSH, thyroid-stimulating hormone.

FIG. 2.

Comparison of fasting glucose, fasting insulin, and HOMA between subjects with TSH values 0.3–2.4 mIU/L and those with TSH 2.5–5.0 mIU/L (*p = 0.040, **p = 0.02). HOMA, homeostatic model assessment.

Free T4 and lipids

Free T4 and lipids were measured concurrently in 590 children and adolescents, of which 6 (1.0%) had free T4 below the reference range, and 6 (1.0%) had free T4 above the reference range. The 578 (98.0%) subjects with free T4 within the reference range (0.8–1.8 ng/dL) form the basis for the analysis (Table 1). There was a significant inverse relationship between free T4 and triglycerides (r = −0.10, p = 0.011, Table 3). The relationship between free T4 and triglycerides remained significant after adjustment for gender, age, and BMI z-score. The relationship between free T4 and HDL cholesterol varied significantly with gender (p = 0.026) and age (p = 0.027). However, the relationships between free T4 and HDL within the subgroups (females/males or 2–10/10–19 years) were not statistically significant. Increasing concentrations of free T4 were not significantly correlated with increasing concentrations of total cholesterol (r = 0.01) or non-HDL cholesterol (r = −0.01), and these relationships did not vary significantly with gender, age, or BMI z-score.

Table 3.

Relationship Between Free Thyroxine in the Normal Range and Lipid Levels and Measures of Insulin Resistance

	Triglycerides	HDL cholesterol		HDL cholesterol		Fasting glucose
	All subjects (n = 578)	Females (n = 319)	Males (n = 259)	2 to <10 years (n = 128)	10 to <19 years (n = 450)	All subjects (n = 1252)
Pearson correlation coefficient	−0.10	−0.08	0.10	0.09	<−0.01	<−0.01
Unadjusted percent change^a	−32.47	−11.36	17.43	13.71	−0.40	−0.42
95% CI	(−57.60, −7.33)	(−27.36, 4.63)	(−2.72,37.57)	(−13.57, 40.99)	(−15.25, 14.44)	(−3.73, 2.90)
p-value for linear trend	0.011	0.16	0.09	0.32	0.96	0.81
Adjusted percent change	−26.77^b	−5.48^c	11.11^c	13.41^d	−0.09^d	1.17^b
95% CI	(51.18. −2.37)	(−20.45, 9.48)	(−7.89, 30.11)	(−11.75, 38.57)	(−13.78, 13.60)	(−2.19, 4.54)
p-value for linear trend	0.032	0.47	0.25	0.30	0.99	0.49

Percent change in laboratory parameter per 1 ng/dL increase in free T4.

Adjusted for gender, age, and BMI z-score.

Adjusted for age and BMI z-score.

Adjusted for gender and BMI z-score.

Free T4 and markers of insulin resistance

1252 subjects had simultaneous measurement of free T4 within the reference range and fasting glucose (Table 1). A subset of 95 subjects had simultaneous measurements of free T4 and fasting insulin and glucose. Overall, there was no significant relationship between free T4 and glucose, and this relationship did not vary significantly with gender, age, or BMI z-score (Table 3). The data suggested a significant interaction effect between free T4 and age on fasting insulin (p = 0.004). Among subjects between ages 2 and 10 years, the correlation between free T4 and fasting insulin was −0.42, whereas among subjects between 10 and 18 years of age, the correlation was nearly zero. Given that there were only 18 patients under 10 years of age with free T4 and insulin, further analysis could not be explored.

Discussion

In the current study, we found that increasing TSH values within the normal range were associated with increasing triglyceride levels. This relationship remained significant even after adjustment for age, gender, and adiposity. Conversely, free T4 levels within the reference range were inversely correlated with triglyceride levels. A recent study in obese adolescents found a positive correlation between TSH and triglyceride levels that remained significant even after correction for BMI SDS (11). Similarly, two studies in adults also reported a positive correlation between TSH and serum triglycerides (6,12). These relationships in adults were also found to be significant even after adjustment for age, sex, and BMI (12). Our study represents the first report to our knowledge of a similar relationship in a cohort, including normal-weight children. This relationship has been attributed to reduced activity of lipoprotein lipase (13,14), the presence of fewer cell surface receptors for LDL (1), and impaired clearance of lipoproteins in patients with higher TSH values (15).

Relationships between thyroid function and lipids have been examined in the pediatric population to a limited extent only. Paoli-Valeri et al. noted decreased HDL cholesterol in children with subclinical hypothyroidism defined as TSH >4.65 and normal free T4, relative to healthy controls (16). In a study of children with subclinical hypothyroidism from areas of iodine deficiency, iodine treatment was noted to result in significant decrease in total and LDL cholesterol levels (17). In contrast to these pediatric reports (16,17) and adult studies (5 –7), we did not find any relationship between TSH and total cholesterol, HDL cholesterol, or non-HDL cholesterol in our pediatric subjects. The absence of significant relationship of TSH with total or LDL cholesterol may be partly because we limited our study to subjects with normal thyroid function, whereas the previous studies evaluated levels in children with subclinical hypothyroidism. It is also possible that the physiological mechanisms responsible for the associations between TSH and these specific lipid parameters may develop slowly over time and therefore may not manifest during childhood.

We also found that increasing TSH values within the reference range are associated with increasing insulin and HOMA levels, suggesting an association between TSH levels and insulin resistance. Since TSH has been shown to be positively associated with BMI (18), we examined this relationship after adjustment for adiposity and found that it remained significant. We are not aware of any similar reports in normal weight or obese children and adolescents. An association between TSH and fasting insulin and insulin sensitivity has been previously reported in adults with type 2 diabetes (19), as well as in obese adults (20). Serum TSH and free T4·TSH product have also been found in lean adults to be associated linearly and positively with fasting and postglucose load insulin concentrations and negatively with a strong measure of insulin sensitivity (8).

Experimental studies suggest that thyroid hormones may impact insulin sensitivity by influencing expression or activation of uncoupling protein, β adrenergic receptor, and peroxisome proliferator-activated receptor-gamma (21 –23). Treatment of obese diabetic rodents with thyroid hormones increases insulin sensitivity and decreases hyperglycemia and hyperinsulinemia (24). In tissue cultures from newborn rats, exposure to thyroid hormone resulted in increased expression of GLUT 4 in precursor cells of brown adipocytes and increased the maximal transport rate of glucose (25). Increased expression of GLUT1, GLUT 3, and GLUT 4 transporters on the monocyte membrane has been demonstrated in human subjects with hyperthyroidism (26). It may be that even mild hypothyroidism leads to decreased transcription of glucose transporters such as GLUT 4 and thereby contributes to insulin resistance.

Serum TSH measurement is considered to be the most sensitive screening test for primary thyroid dysfunction. The laboratory reference range is generally chosen by determining the 95% confidence limits of a population of individuals free of known thyroid dysfunction (27). While using this method generally sets the upper limit of normal for TSH around 4.0–5.0 mIU/L for both adults and children, this remains a matter of controversy (28). It has been suggested that when subjects with thyroid antibodies and/or abnormal findings on thyroid ultrasound are excluded, the upper limit of the reference range may be closer to 2.5 mIU/L (29). Several professional organizations, including the American Association of Clinical Endocrinologists, have therefore recommended changing the upper limit of normal TSH values for adults to 3.0 mIU/L (30).

We found that subjects with TSH values in the upper portion of our reference range (2.5–5.0 mIU/L) had significantly higher triglyceride levels when compared with subjects having TSH in the lower portion of the normal range (0.3–2.4 mIU/L). Similar differences in triglycerides relative to TSH levels were also noted in a study of adult euthyroid subjects (12). We also found that euthyroid children with TSH <2.5 mIU/L had lower HOMA-IR and fasting insulin levels compared with children having TSH between 2.5 and 5.0 mIU/L. A recent study in women with polycystic ovary syndrome also noted higher degrees of insulin resistance as measured by HOMA-IR in those with TSH >2.0 mIU/L compared with those having TSH <2.0 mIU/L (31).

Our study suggests an association between thyroid function and lipid and glucose metabolism in children and adolescents. To our knowledge, this is the first study that has examined an association between TSH and free T4 levels and triglycerides and insulin/HOMA in a cohort of euthyroid children and adolescents. Since our study population included both normal weight and obese children and adolescents, we were able to conclude that these relationships were found to be independent of adiposity, age, and sex. The statistically significant differences in triglycerides and insulin that were noted between those with TSH in the lower half of normal range relative to those in the upper half of normal range may not be clinically significant. Therefore, further clinical studies are also needed to determine the reduction in triglycerides and insulin levels that may have a clinical impact. The retrospective and cross-sectional nature of our study design results in several limitations. We did not have any information on the status of thyroid autoimmunity in the study subjects. This may be an important potential confounding variable as the favorable effects of L-T4 therapy on lipids (total cholesterol and LDL cholesterol) have been found to be most pronounced in adults with high normal TSH (2.0–4.0 μIU/mL) and positive thyroid autoantibodies (32). Other limitations include lack of information concerning the ethnicity, physical activity, and dietary intake of the subjects. We also did not collect information on family history of thyroid disorder, diabetes, or hyperlipidemia. Finally, our findings do not, of course, prove causality between thyroid function status and any of the studied variables. The strengths of the associations found in this study of pediatric patients seeking outpatient pediatric care do however suggest that these relationships may exist in the general pediatric population also. There are currently no pediatric studies that have examined the effect of thyroid hormone treatment in those with TSH between 2.5 and 5 mIU/L on triglycerides or insulin levels.

Our findings suggest that further studies are required to determine the optimal TSH goal in patients with hypothyroidism who additionally have adverse cardiovascular risk factors, insulin resistance, and hypertriglyceridemia. Large and prospective randomized clinical trials are warranted to determine whether there is benefit in thyroid hormone treatment of subclinical hypothyroidism in these children. Similarly, prospective trials might be undertaken to determine whether decreasing the target TSH level to the lower half of the normal range in at risk children with overt primary hypothyroidism will improve triglyceride levels or markers of insulin resistance.

Footnotes

Disclosure Statement

The authors declare that no competing financial interests exist.

References

Cappola

, Ladenson

. 2003. Hypothyroidism and atherosclerosis. J Clin Endocr Metab, 88:2438–2444.

O'Brien

, Dinneen

, O'Brien

, Palumbo

. 1993. Hyperlipidemia in patients with primary and secondary hypothyroidism. Mayo Clin Proc, 68:860–866.

Ochs

, Auer

, Bauer

, Nanchen

, Gussekloo

, Cornuz

, Rodondi

. 2008. Meta-analysis: subclinical thyroid dysfunction and the risk for coronary heart disease and mortality. Ann Intern Med, 148:832–845.

Walsh

, Bremner

, Bulsara

, O'Leary

, Leedman

, Feddema

, Michelangeli

. 2005. Thyroid dysfunction and serum lipids: a community-based study. Clin Endocrinol, 63:670–675.

Pallas

, Koutras

, Adamopoulos

, Marafelia

, Souvatzoglou

, Piperingos

, Moulopoulos

. 1991. Increased mean serum thyrotropin in apparently euthyroid hypercholesterolemic patients: does it mean occult hypothyroidism? J Endocrinol Invest, 14:743–746.

Asvold

, Vatten

, Nilsen

TIL

, Bjoro

. 2007. The association between TSH within the reference range and serum lipid concentrations in a population-based study. The HUNT Study. Eur J Endocrinol, 156:181–186.

Bakker

, ter Maaten

, Popp-Snijders

, Slaets

, Heine

, Gans

. 2001. The relationship between thyrotropin and low density lipoprotein cholesterol is modified by insulin sensitivity in healthy euthyroid subjects. J Clin Endocr Metab, 86:1206–1211.

Fernandez-Real

J-M

, Lopez-Bermejo

, Castro

, Casamitjana

, Ricart

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

Kuczmarski

, Ogden

, Guo

, Grummer-Strawn

, Flegal

, Mei

, Wei

, Curtin

, Roche

, Johnson

. 2002. 2000 CDC Growth Charts for the United States: methods, development., Vital & Health Statistics—Series 11: Data From the National Health Survey May. 1–190.

10.

A SAS program for the CDC growth charts. www.cdc.gov/nccdphp/dnpa/growthcharts/resources/sas.htm. 2010 August 23.

11.

Shalitin

, Yackobovitch-Gavan

, Phillip

. 2009. Prevalence of thyroid dysfunction in obese children and adolescents before and after weight reduction and its relation to other metabolic parameters. Horm Res, 71:155–161.

12.

Ruhla

, Weickert

, Arafat

, Osterhoff

, Isken

, Spranger

, Schofl

, Pfeiffer

, Mohlig

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

13.

Lithell

, Boberg

, Hellsing

, Ljunghall

, Lundqvist

, Vessby

, Wide

. 1981. Serum lipoprotein and apolipoprotein concentrations and tissue lipoprotein-lipase activity in overt and subclinical hypothyroidism: the effect of substitution therapy. Eur J Clin Invest, 11:3–10.

14.

Pykalisto

, Goldberg

, Brunzell

. 1976. Reversal of decreased human adipose tissue lipoprotein lipase and hypertriglyceridemia after treatment of hypothyroidism. J Clin Endocr Metab, 43:591–600.

15.

Liu

, Rahman

, Bagdade

, Alaupovic

, Kannan

. 1998. Effect of thyroid hormone on plasma apolipoproteins and apoA- and apoB-containing lipoprotein particles. Eur J Clin Invest, 28:266–270.

16.

Paoli-Valeri

, Guzman

, Jimenez-Lopez

, Arias-Ferreira

, Briceno-Fernandez

, Arata-Bellabarba

. 2005. Atherogenic lipid profile in children with subclinical hypothyroidism. An Pediatr, 62:128–134.

17.

Zimmermann

, Aeberli

, Melse-Boonstra

, Grimci

, Bridson

, Chaouki

, Mbhenyane

, Jooste

. 2009. Iodine treatment in children with subclinical hypothyroidism due to chronic iodine deficiency decreases thyrotropin and C-peptide concentrations and improves the lipid profile. Thyroid, 19:1099–1104.

18.

Nyrnes

, Jorde

, Sundsfjord

. 2006. Serum TSH is positively associated with BMI. Int J Obes, 30:100–105.

19.

Chubb

SAP

, Davis

TME

. 2005. Interactions among thyroid function, insulin sensitivity, and serum lipid concentrations: the Fremantle diabetes study. J Clin Endocr Metab, 90:5317–5320.

20.

Michalaki

, Vagenakis

, Leonardou

, Argentou

, Habeos

, Makri

, Psyrogiannis

, Kalfarentzos

, Kyriazopoulou

. 2006. Thyroid function in humans with morbid obesity. Thyroid, 16:73–78.

21.

Dallongeville

, Helbecque

, Cottel

, Amouyel

, Meirhaeghe

. 2003. The Gly16—>Arg16 and Gln27—>Glu27 polymorphisms of beta2-adrenergic receptor are associated with metabolic syndrome in men. J Clin Endocr Metab, 88:4862–4866.

22.

Wang

, Chu

, Lu

, Hasstedt

, Kern

, Elbein

. 2004. Uncoupling protein-2 polymorphisms in type 2 diabetes, obesity, and insulin secretion. Am J Physiol Endocrinol Metab, 286:E1–E7.

23.

Frederiksen

, Brodbaek

, Fenger

, Jorgensen

, Borch-Johnsen

, Madsbad

, Urhammer

. 2002. Comment: studies of the Pro12Ala polymorphism of the PPAR-gamma gene in the Danish MONICA cohort: homozygosity of the Ala allele confers a decreased risk of the insulin resistance syndrome. J Clin Endocr Metab, 87:3989–3992.

24.

Koritschoner

, Alvarez-Dolado

, Kurz

, Heikenwalder

, Hacker

, Vogel

, Munoz

, Zenke

. 2002. Thyroid hormone regulates the obesity gene tub. EMBO Rep, 2:499–504.

25.

Shimizu

, Shimazu

. 2002. Thyroid hormone augments GLUT 4 expression and insulin-sensitive glucose transport system in differentiating rat brown adipocytes in culture. J Vet Med Sci, 64:677–681.

26.

Dimitriadis

GME

, Alevizaki

, Boutati

, Psara

, Papasteriades

, Raptis

. 2005. Thyroid hormone excess increases basal and insulin-stimulated recruitment of GLUT 3 glucose transporters on cell surface. Horm Metab Res, 37:15–20.

27.

Fatourechi

. 2007. Editorial: Upper limit of normal serum thyroid stimulating hormone: a moving and now an aging target? J Clin Endocr Metab, 92:4560–4562.

28.

Brabant

, Beck-Peccoz

, Jarzab

, Laurberg

, Orgiazzi

, Szabolcs

, Weetman

, Wiersinga

. 2006. Is there a need to redefine the upper normal limit of TSH? Eur J Endocrinol, 154:633–637.

29.

Baloch

, Carayon

, Conte-Devolx

, Demers

, Feldt-Rasmussen

, Henry

, LiVosli

, Niccoli-Sire

, John

, Ruf

, Smyth

, Spencer

, Stockigt

. 2003. Laboratory medicine practice guidelines. Laboratory support for the diagnosis and monitoring of thyroid disease. Thyroid, 13:3–126.

30.

Baskin

, Cobin

, Duick

, Gharib

, Guttler

, Kaplan

, Segal

. 2002. American Association of Clinical Endocrinologists medical guidelines for clinical practice for the evaluation and treatment of hyperthyroidism and hypothyroidism. Endocr Pract, 8:457–469.

31.

Mueller

, Schofl

, Dittrich

, Cupisti

, Oppelt

, Schild

, Beckmann

, Haberle

. 2009. Thyroid-stimulating hormone is associated with insulin resistance independently of body mass index and age in women with polycystic ovary syndrome. Hum Reprod, 24:2924–2930.

32.

Michalopoulou

, Alevizaki

, Piperingos

, Mitsibounas

, Mantzos

, Adamopoulos

, Koutras

. 1998. High serum cholesterol levels in persons with “high-normal” TSH levels: should one extend the definition of subclinical hypothyroidism? Eur J Endocrinol, 138:141–145.