| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
BRIEF REPORT |
Department of Biochemistry (S.A.P.C.), Fremantle Hospital, Fremantle, Western Australia; and School of Medicine and Pharmacology (W.A.D., T.M.E.D.), University of Western Australia, Fremantle Hospital, Fremantle, Western Australia 6959, Australia
Address all correspondence and requests for reprints to: Dr. S. A. P. Chubb, Department of Biochemistry, Fremantle Hospital, P.O. Box 480, Fremantle, Western Australia 6959, Australia. E-mail: paul.chubb{at}health.wa.gov.au.
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsDiscussionReferences |
|---|
Objective: The aim of the study was to determine whether insulin sensitivity modifies the association between thyroid dysfunction and lipid parameters in diabetic patients.
Design: This is a cross-sectional study.
Setting: This is a community-based observational study.
Patients: One hundred seventeen females with type 2 diabetes who were not taking oral hypoglycemic therapy, insulin, or lipid-lowering therapy participated in the study.
Intervention: Serum TSH, insulin, total and high-density lipoprotein cholesterol, and triglycerides were measured.
Main Outcome Measures: Age-adjusted multiple linear regression analysis of serum lipid concentrations and derived parameters, as functions of serum TSH and homeostasis model assessment-derived insulin sensitivity (HOMA-S), were measured.
Results: The relationship among serum lipid concentrations, serum TSH, and HOMA-S was significantly modified by an interaction term ln(TSH)*ln(HOMA-S). In three-dimensional graphs, there were strong positive associations between TSH and lipid parameters with adverse cardiac risks at low insulin sensitivity that were absent at higher insulin sensitivity. The effect was strongest for lipid risk factors associated with insulin resistance.
Conclusions: The interaction between thyroid function and insulin sensitivity is an important contributor to diabetic dyslipidemia and may justify T4 replacement in some patients.
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
The Fremantle Diabetes Study (FDS) was a prospective observational study of diabetes in a postal code-defined population of 120,000 people in Western Australia (7). Of 437 unselected women with type 2 diabetes from the FDS who had participated in a substudy of thyroid disease (6), we included all 117 who were not taking oral hypoglycemic agents, insulin, or lipid-lowering therapy in the present study. All gave informed consent to participate in the FDS, which was approved by the Human Rights Committee, Fremantle Hospital.
Clinical methods
FDS patients were recruited between 1993 and 1996 and had comprehensive assessments at baseline and annually thereafter. At each visit, demographic and clinical information, including details of other illnesses, were documented, clinical examinations were carried out, and samples for biochemical tests were obtained (7). The present data were from baseline assessments.
Assay methodology
Fasting venous blood samples were collected between 0730 and 1200 h, centrifuged promptly, and separated sera stored at –80 C. Assays for insulin and TSH were carried out using an Elecsys 2010 analyzer (Roche Diagnostics, Castle Hill, New South Wales, Australia) and standardized against the World Health Organization first international reference preparation 66/304 for insulin and second international reference preparation 80/558 for TSH. Between-day coefficients of variation were 5.2% at 11.8 mU/liter insulin and 2.6% at 2.6 mU/liter TSH. Serum glucose, total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), and triglycerides (TGs) were analyzed using a Hitachi 911 analyzer (Roche Diagnostics). Low-density lipoprotein cholesterol (LDL-C) was calculated using the Friedewald equation (8) for those specimens with TGs less than 400 mg/dl (<4.5 mmol/liter). Non-HDL-C was calculated as the difference between TC and HDL-C.
Data analysis
Insulin sensitivity was estimated using homeostasis model assessment (HOMA-S) from fasting serum glucose and insulin using the Oxford HOMA calculator (http://www.dtu.ox.ac.uk/homa/index.html). In this model, insulin sensitivity is expressed relative to the mean of healthy adults. For the purposes of the present study, relative rather than absolute values were important. Statistical analyses were performed using SPSS for Windows (version 11; SPSS, Inc., Chicago, IL). Data are presented as proportions, mean ± SD or geometric mean (SD range). Two-sample comparisons were by Fisher’s exact test, Student’s t test, or Mann-Whitney U test. Multiple linear regression analysis was performed to identify relationships between serum lipid parameters as dependent variables and thyroid function and insulin sensitivity, after adjusting for age. Logarithmic (ln) transformation was applied to right-skewed data before analysis. Scientific Notebook (version 3.00; TCI Software Research, Boston, MA) was used to generate three-dimensional graphs illustrating these relationships. Coronary heart disease (CHD) risk was estimated using the United Kingdom Prospective Diabetes Study risk engine (9).
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
Sera from 17 patients (1.5%) were either unavailable or unsuitable for biochemical testing due to hemolysis. Compared with the other 563 type 2 FDS females at study entry, the remaining 100 patients were younger (61.6 ± 13.9 vs. 64.7 ± 11.3 yr; P = 0.034) and had shorter diabetes duration [1.0 (0.4–3.0) vs. 4.6 (1.8–10.0) yr; P < 0.001]. The proportions in each group taking T4 were not significantly different (10.0 vs. 7.1%; P = 0.31).
The mean serum TSH, TC, HDL-C, and TGs were 1.7 (0.7–4.3) mU/liter, 220 ± 46 mg/dl (5.7 ± 1.2 mmol/liter), 42.5 ± 12.4 mg/dl (1.10 ± 0.32 mmol/liter), and 159 (97–266) mg/dl (1.8 (1.1–3.0) mmol/liter), respectively, in the present 100 patients. The mean HOMA-S was 32% (16–62%). Eight patients had TSH results over 5.1 mU/liter (the upper limit of the reference range) (range 5.6–10.3 mU/liter), three of whom were taking T4.
Associations among serum TSH, insulin sensitivity, and serum lipids
The age-adjusted simple linear regression coefficients for the relationships between serum lipid parameters and ln(TSH) (model 1) and ln(HOMA-S) (model 2) are shown in Table 1
. ln(TSH) was positively associated with TC, LDL-C, and non-HDL-C. ln(HOMA-S) was associated negatively with TC, ln(TG), non-HDL-C, and ln(TC/HDL-C) and positively with HDL-C.
|
. Model 3 included ln(TSH) and ln(HOMA-S), whereas model 4 also included the interaction term ln(TSH)*ln(HOMA-S). In model 4, ln(TSH) was positively associated with all lipid parameters except HDL-C, whereas ln(HOMA-S) was associated positively with HDL-C and negatively with ln(TG) and non-HDL-C. The interaction terms were significantly associated with each lipid parameter except LDL-C, which just failed to reach significance. In each case, the adjusted coefficient of multiple determination (R2) values indicated that model 4 explained more of the variability in the serum lipid parameters than model 3. The multiple regression analyses were repeated after excluding 12 patients with history of thyroid disease (11 on thyroid replacement, one with thyroidectomy). The interaction terms remained significant for all parameters (P 
0.038) except LDL-C (P = 0.07).
Three-dimensional representations of the relationships of TC, HDL-C, TC/HDL-C, and TGs to TSH and HOMA-S in model 4 are shown in Fig. 1. For HDL-C, the shape of the surface was inverted, compared with that for other lipid parameters. For a 61.7-yr-old patient (the mean age in the present series) with HOMA-S of 10%, model 4 predicts that, at serum TSH of 1 and 7 mU/liter, TC would be 213 mg/dl (5.5 mmol/liter) and 290 mg/dl (7.5 mmol/liter), and HDL-C 37.9 mg/dl (0.98 mmol/liter) and 29.0 mg/dl (0.75 mmol/liter), respectively. In the same example, but at HOMA-S of 90%, the differences between TC [205 mg/dl (5.3 mmol/liter) and 186 mg/dl (4.8 mmol/liter)] and HDL-C [46.4 mg/dl (1.2 mmol/liter) and 61.9 mg/dl (1.6 mmol/liter)] at TSH values of 1 and 7 mU/liter were smaller and in the opposite direction. Using the above lipid data for HOMA-S of 10%, average values for age, diabetes duration (3.0 yr), systolic blood pressure (146 mm Hg), and hemoglobin A1c (6.8%) and assuming no atrial fibrillation or smoking, the 10-yr CHD risks at a serum TSH of 1 and 7 mU/liter were estimated at 14.8 and 26.5%, respectively.
|
|
Discussion |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
Bakker et al. (5) demonstrated significant interactions between insulin resistance and thyroid function in the prediction of serum TC, LDL-C, and non-HDL-C in euthyroid nondiabetic adults. Our observations are similar, but in the case of LDL-C, model 4 showed only a trend toward significance for the interaction term. A stronger association may have been present if we had been able to include patients with serum TGs greater than 400 mg/dl (4.5 mmol/liter). For HDL-C, however, our data differ conspicuously from those of Bakker et al. (5). Whereas TSH was the only predictor of HDL-C in euthyroid nondiabetic subjects (5), we found that TSH, HOMA-S, and the interaction term were all significant determinants of HDL-C in our subjects. This was also the case for TGs, a variable not assessed by Bakker et al. (5). Our observations reflect the association of insulin resistance with low serum HDL-C and increased TGs (13) but also reveal marked effects of small variations in TSH on these lipid parameters at levels of insulin sensitivity not examined previously.
The interaction between insulin resistance and high TSH in our data may have clinical implications. For the insulin-resistant patient (HOMA-S, 10%) in the example given above, the estimated 10-yr CHD risk at a TSH of 1 mU/liter was almost half that at a TSH of 7 mU/liter. Thus, at low insulin sensitivity, relatively minor differences in TSH are associated with marked changes in lipid risk factors and thus cardiovascular risk. By comparison, cardiovascular risk reduction in one trial of T4 replacement in nondiabetic patients with subclinical hypothyroidism was only 17% (14).
We recently reported a prevalence of approximately one in 12 for subclinical hypothyroidism in a community-based sample of female type 2 diabetic patients (6). Current guidelines recommend against routine treatment of this condition in patients with serum TSH less than 10 mU/liter (11). Although this remains controversial (12), T4 replacement trials in nondiabetic subjects found that the greatest improvements in serum lipids were in patients with higher serum TC and LDL-C and those with TSH greater than 10 mU/liter (14, 15, 16). T4 replacement did not influence serum HDL-C or TGs (14, 16). In the present study, there were too few patients with subclinical hypothyroidism to analyze separately, but the data suggest that insulin-resistant patients with TSH levels less than 10 mU/liter may benefit from adjunctive T4 treatment, especially those intolerant of statin therapy or those with therapy-resistant dyslipidemia.
Our study has limitations. First, the relatively simple multivariate regression models we used explained only a limited amount of the variation in serum lipid parameters, with the highest adjusted R2 value of 0.34 suggesting that other factors are also important. Second, we studied only diet-controlled females not taking hypolipidemic therapy, a group selected because it has a relatively high prevalence of subclinical hypothyroidism and in which pharmacological influences on serum lipid status were minimized. Third, our study was cross-sectional, so its ability to infer causality is limited. Despite these limitations, the associations we found were highly significant and consistent with those in studies of nondiabetic subjects (5).
The measurement of fasting serum lipids is an integral part of the routine assessment of the type 2 diabetic patient. The present data suggest that simultaneous assessment of thyroid status through serum TSH might also be justified as part of regular review, even in younger patients. However, placebo-controlled T4 replacement trials in more diverse groups of diabetic patients with subclinical hypothyroidism monitored for cardiovascular end points are needed before the value of correcting a mildly raised serum TSH is known.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
First Published Online June 28, 2005
Abbreviations: CHD, Coronary heart disease; FDS, Fremantle Diabetes Study; HDL-C, high-density lipoprotein cholesterol; HOMA-S, homeostasis model assessment-derived insulin sensitivity; LDL-C, low-density lipoprotein cholesterol; ln, natural logarithm; R2, coefficient of multiple determination; TC, total cholesterol; TG, triglyceride.
Received February 14, 2005.
Accepted June 22, 2005.
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  A. Roos, S. J. L. Bakker, T. P. Links, R. O. B. Gans, and B. H. R. Wolffenbuttel Thyroid Function Is Associated with Components of the Metabolic Syndrome in Euthyroid Subjects J. Clin. Endocrinol. Metab., February 1, 2007; 92(2): 491 - 496. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J.-M. Fernandez-Real, A. Lopez-Bermejo, A. Castro, R. Casamitjana, and W. Ricart Thyroid Function Is Intrinsically Linked to Insulin Sensitivity and Endothelium-Dependent Vasodilation in Healthy Euthyroid Subjects J. Clin. Endocrinol. Metab., September 1, 2006; 91(9): 3337 - 3343. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
