Serum lipid levels in relation to serum thyroid-stimulating hormone and the effect of thyroxine treatment on serum lipid levels in subjects with subclinical hypothyroidism: the Tromsø Study
Amjid Iqbal, Department of Cardiology, University Hospital of North Norway, 9038 Tromsø, Norway.
(fax: 47 776 26863; e-mail: email@example.com).
Objective. To evaluate the relation between serum thyroid-stimulating hormone (TSH) and lipids.
Design. Cross-sectional epidemiological study, nested case–control study, and a placebo-controlled double-blind intervention study.
Methods. In the 5th Tromsø study serum TSH, total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C) were measured. Subjects with subclinical hypothyroidism (SHT) and a matching control group were re-examined and apolipoprotein A1 (Apo A1) and apolipoprotein B (Apo B) were also measured. Subjects with SHT were included in an intervention study with thyroxine supplementation for 1 year.
Results. A total of 5143 subjects from the 5th Tromsø study were included. A significant and positive correlation between serum TSH levels and serum TC and LDL-C levels were found in both genders. However, in the females this did not reach statistical significance after adjusting for age and BMI. The serum LDL-C were significantly higher and the Apo A1 levels significantly lower in 84 SHT subjects compared with 145 controls, and in the SHT females the TC levels were also significantly elevated. In the intervention study (32 subjects given thyroxine and 32 subjects given placebo), we observed a significant reduction in the Apo B levels after thyroxine medication. In those that at the end of the study had serum TSH levels in the range 0.2–2.0 mIU L−1, the serum TC and LDL-C levels were also significantly reduced.
Conclusions. There is a positive association between serum TSH levels and TC and LDL-C levels. These lipid levels are reduced with thyroxine treatment in subjects with SHT.
Overt hypothyroidism is associated with hypercholesterolemia . However, whether subjects with subclinical hypothyroidism (SHT), defined as an elevated serum thyroid-stimulating hormone (TSH) level with serum free thyroxine (free T4) and free triiodothyronine (free T3) levels within the normal range, have hyperlipidaemia is more uncertain as is the effect of thyroxine treatment in these subjects. Thus, normal [2–4] as well as elevated [5, 6] lipid levels have been reported in SHT, and a lipid lowering [7, 8] as well as no effect [9–11] of thyroxine treatment have been found. Furthermore, in several studies on the relation between serum TSH levels and lipid profile no adjustment for confounders like age and BMI has been performed [5, 6]. This is a prerequisite as both serum TSH and serum lipid levels have considerable covariation with age and BMI [12, 13].
Since 1974, there have been five large epidemiological studies in Tromsø, Northern Norway. The main focus of these studies has been cardiovascular risk factors and diseases, and in the last one in 2001 with >8000 participants, measurement of serum TSH was also included. We, therefore, had the opportunity to test the relation between serum TSH and serum lipids in a large cohort. From this cohort, we also invited subjects with elevated serum TSH levels and matched controls to a follow-up study, and those with SHT were in addition invited to participate in a double-blind placebo-controlled intervention study with thyroxine for 1 year.
Patients and methods
The 5th Tromsø study was performed as a general health survey in 2001 . All men and women older than 29 years, living in the municipality of Tromsø and that participated in the second phase of the 4th Tromsø study  or became 30, 40, 45, 60, or 75 years old during 2001, were invited to participate. All subjects filled out a health questionnaire that included smoking status and the use of thyroxine and lipid lowering medication. Nonfasting blood samples were drawn and analysed for serum TSH.
Subjects with a serum TSH level between 3.50 and 10.0 mIU L−1 were invited to a follow-up examination at the Clinical Research Unit at the University Hospital of Tromsø. Those who reported a history of coronary infarction, angina pectoris or stroke in the questionnaire, those participating in other follow-up studies, those using thyroid medication, and those above the age of 80, were not invited. The hospital records were also checked to identify and exclude subjects with serious diseases not reported in the questionnaire. For each subject with serum TSH between 3.5 and 10 mIU L−1, an age- and sex-matched control subject with serum TSH in the range 0.5–3.49 mIU L−1 was also invited. The invitation letter did not disclose the subject's TSH status.
At the follow-up blood was drawn in the nonfasting state and serum samples analysed for TSH, free T4, free T3, and antibodies towards thyroid peroxidase (anti-TPO antibodies). A clinical examination was performed, height and weight were measured in light clothing wearing no shoes, and body mass index (BMI) was calculated as weight divided by squared height. Those that still had serum TSH between 3.5 and 10 mIU L−1 and free T4 and free T3 within the reference range and that did not have obvious clinical symptoms of hypothyroidism, were considered to have SHT. They were informed that their thyroid status were similar to that at the Tromsø study, and invited to further examinations. The control subjects that at the follow-up still had serum TSH within the range 0.5–3.49 mIU L−1 and serum free T4 and free T3 within the reference range were similarly informed that their thyroid status was unchanged and invited to further examinations. Subjects using lipid lowering medication were excluded. Those consenting had additional blood samples for lipid analysis drawn on a separate day in the nonfasting state.
The subjects were then informed about their thyroid status. Those with SHT were invited to a 12 months intervention study, but they were not informed that the effect on blood lipids was the main goal of the study. Those consenting were randomized to either placebo or thyroxine treatment. During the first 6 weeks all in the thyroxine group were given 50 μg daily, and for the following 6 weeks 100 μg daily. Thereafter, serum TSH and free T4 were measured every third month and the thyroxine dose adjusted in steps of 25–50 μg to achieve a TSH level between 0.5 and 1.5 mIU L−1. After 12 months height and weight were again measured and blood samples for lipid analyses were drawn.
Serum TSH from the 5th Tromsø study and at inclusion in the follow-up study were analysed with a Modular ETM automated clinical chemistry analyser (Roche Diagnostics, Mannheim, Germany) with 0.2–4.0 mIU L−1 as reference range according to the manufacturer. The laboratory changed analyser and inherent reagents to AxSYM (Abbott, IL, USA) at the beginning of the intervention study. As this was known in advance, serum samples from the SHT/control groups were kept frozen until analysis with the new assay. All serum TSH, free T4, and free T3 values presented from follow-up and intervention studies were therefore equally analysed, and the reference values, applied according to the manufacturer, were 0.20–4.3 mIU L−1, 9–22 pmol L−1, and 2.8–7.1 pmol L−1, respectively. Anti-TPO antibodies were measured with Immulite® 2000 automated analyzer (Diagnostic Products Corporation, Los Angeles, CA, USA) by an immunometric assay based on chemiluminescence with reference values <50.0 IU mL−1.
Serum total cholesterol (TC) was measured with an enzymatic colorimetric test using Modular PTM (Roche) automated clinical chemistry analyser. The reference values used were 30–39 years: 3.7–8.3 mmol L−1 and >40 years: 4.1–8.7 mmol L−1. Serum triglycerides (TG) and high-density lipoprotein cholesterol (HDL-C) were analysed likewise, and reference values applied were for serum TG 0.2–1.8 mmol L−1 (TG) and for HDL-C 0.9–2.2 mmol L−1 for females and 0.8–2.0 mmol L−1 for males. Serum low-density lipoprotein cholesterol (LDL-C) was calculated using the formula LDL-C =TC ÷ HDL-C ÷ (TG × 0.46), provided the serum TG value was <4.0 mmol L−1. Serum Apolipoprotein A1 (Apo A1) and apolipoprotein B (Apo B) were analysed with Modular PTM (Roche) using tests based on immunoturbidimetry, and reference values applied were 1.02–1.73 and 0.48–1.16 g L−1, respectively.
Normal distribution was evaluated with determination of skewness and kurtosis and visual inspection of histograms. Except for serum TG all dependent variables used in the regression analyses were considered normally distributed. After logarithmic transformation serum TG assumed normal distribution and was used as such when used as dependent variable. In the 5th Tromsø study, the subjects were divided in six groups according to serum TSH levels: <2.5 percentile, >97.5 percentile, and quartiles with the 2.5–97.5 percentile range. To test for interactions, factor analyses with TSH group, gender, and smoking status (current smoker/nonsmoker) as factors, and age and BMI as covariables, were performed. Regarding lipids, this revealed significant interactions between gender and smoking status in the 5th Tromsø study, and between gender and TSH group in the follow-up study. For these cohorts, males and females were therefore analysed separately. The serum lipids were analysed for linear trend across the six TSH groups using linear regression with age, smoking status and BMI as covariables.
In the follow-up study, the SHT and control groups were compared with Student's t-test for unpaired samples, and also with a general linear model with the parameter in question as dependent variable, TSH group (SHT or control), smoking status (and gender) as factors, and age and BMI as independent variables.
In the intervention study, the thyroxine and placebo groups were compared similarly, but with gender included as a factor in the general linear model. The associations between delta lipids (serum lipid level at start of the study minus serum lipid level at end of the study) and delta TSH values, were evaluated with a linear regression model with delta BMI, age and smoking status as covariables. Comparison between values at baseline and at the end of the study was performed with the Student's paired t-test. Correlations were evaluated with Spearman's rho coefficient.
Unless otherwise stated, all data are expressed as mean ± SD. All tests were done two-sided, and P < 0.05 was considered statistically significant. Statistical analyses were performed with SPSS version 11.0 (SPSS Inc., Chicago, IL, USA).
The Regional Ethics Committee approved the study, and all participants gave their written informed consent.
The 5th Tromsø study
A total of 10 419 men and women, aged 29–89, were invited to the 5th Tromsø study and 8128 persons attended. Serum TSH was successfully measured in 7954. We excluded present (n = 521) and previous (n = 49) thyroxine users, and those with missing data on thyroid medication (n =1220). In addition, 194 subjects with diabetes were excluded. In the remaining 5970 subjects, information on use of lipid lowering drugs were available in 5832, and among these, 5143 did not use or had not used lipid lowering drugs. Characteristics of these 5143 subjects are given in Table 1.
Table 1. Characteristics of the 2269 males and 2874 females from the 5th Tromsø study
|Age (years)||58.1 ± 14.4||56.6 ± 14.2|
|BMI (kg m−2)||26.6 ± 3.5||26.1 ± 4.4|
|Serum TSH (mIU L−1)||1.9 ± 2.4||1.9 ± 2.4|
|Serum TC (mmol L−1)||6.1 ± 1.1||6.3 ± 1.2|
|Serum TG (mmol L−1)||1.7 ± 1.0||1.4 ± 0.7|
|Serum HDL-C (mmol L−1)||1.4 ± 0.4||1.6 ± 0.4|
|Serum LDL-C (mmol L−1)a||4.0 ± 1.0||4.0 ± 1.1|
In both genders, there were significant correlations between serum TSH levels and age, BMI, serum TC, and serum LDL-C (Table 2). Accordingly, when dividing the cohort in the six TSH groups (<2.5 percentile, quartiles within the 2.5–97.5 percentile range, and >97.5 percentile) there was in both genders a gradual increase in age, BMI, serum TC, and serum LDL-C with increasing TSH level (Table 3). This linear trend remained statistically significant in the males after adjusting for age, BMI and smoking status. In the females, the linear trend for serum TC and LDL-C remained significant after adjusting for age alone or BMI alone, but not when both age and BMI were included in the linear regression model (Table 3). Judging separately those with serum TSH >10 mIU L−1 (eight males and 13 females), there was no further increase in the serum lipid levels with serum TC being 6.41 ± 1.13 and 6.66 ± 1.11 mmol L−1 in the males and females, respectively.
Table 2. Spearman's correlation coefficient (rho) between age, BMI, serum TSH and lipid levels in the 2269 males and 2874 females from the 5th Tromsø study
|Age (years)|| ||−0.06**||0.19**|| ||0.18**||0.21**|
|BMI (kg m−2)||−0.06**|| ||0.05*||0.18**|| ||0.13**|
|Serum TSH (mUI L−1)||0.19**||0.05*|| ||0.21**||0.13**|| |
|Serum TC (mmol L−1)||0.11**||0.17**||0.09**||0.50**||0.22**||0.13**|
|Serum TG (mmol L−1)||−0.17**||0.36**||0.01||0.19**||0.34**||0.08**|
|Serum HDL-C (mmol L−1)||0.22**||−0.33**||0.01||0.14**||−0.24**||0.02|
|Serum LDL-C (mmol L−1)||0.11**||0.16**||0.08**||0.46**||0.25**||0.12**|
Table 3. Characteristics of the 2269 males and 2874 females from the 5th Tromsø study in relation to serum TSH levels
| Serum TSH <0.55 mUI L−1||54||61.7 ± 13.1||26.7 ± 3.5||5.8 ± 0.8||5.8 ± 1.1||1.5 ± 0.9||1.3 ± 0.4||52||3.8 ± 0.7||3.8 ± 1.0|
| Serum TSH 0.56–1.20 mUI L−1||529||54.3 ± 15.0||26.3 ± 3.4||6.0 ± 1.0||6.1 ± 1.1||1.7 ± 1.0||1.3 ± 0.3||519||3.9 ± 0.9||4.0 ± 1.0|
| Serum TSH 1.21–1.60 mUI L−1||539||56.6 ± 14.0||26.6 ± 3.5||6.1 ± 1.1||6.2 ± 1.1||1.7 ± 0.9||1.3 ± 0.4||526||4.0 ± 1.0||4.1 ± 1.0|
| Serum TSH 1.61–2.15 mUI L−1||547||58.2 ± 13.9||26.7 ± 3.2||6.2 ± 1.1||6.2 ± 1.2||1.7 ± 1.1||1.4 ± 0.4||529||4.1 ± 1.0||4.1 ± 1.1|
| Serum TSH 2.16–4.46 mUI L−1||544||62.3 ± 13.6||26.8 ± 3.7||6.2 ± 1.1||6.2 ± 1.2||1.6 ± 1.0||1.4 ± 0.4||532||4.1 ± 1.0||4.1 ± 1.2|
| Serum TSH >4.46 mUI L−1||56||64.3 ± 11.6**||26.7 ± 4.3*||6.4 ± 1.2**||6.1 ± 1.4||1.7 ± 1.0||1.4 ± 0.4||55||4.3 ± 1.1**||4.0 ± 1.3|
| Serum TSH <0.48 mUI L−1||68||54.6 ± 16.6||24.5 ± 4.0||5.9 ± 1.1||6.1 ± 1.1||1.3 ± 0.8||1.6 ± 0.5||67||3.7 ± 0.9||3.9 ± 1.0|
| Serum TSH 0.48–1.11 mUI L−1||674||52.7 ± 14.5||25.5 ± 4.0||6.1 ± 1.2||6.3 ± 1.1||1.3 ± 0.8||1.6 ± 0.4||652||3.9 ± 1.1||4.1 ± 1.0|
| Serum TSH 1.12–1.53 mUI L−1||690||55.1 ± 14.2||26.0 ± 4.3||6.2 ± 1.2||6.3 ± 1.1||1.3 ± 0.7||1.6 ± 0.4||679||4.0 ± 1.0||4.1 ± 1.0|
| Serum TSH 1.54–2.14 mUI L−1||686||57.3 ± 14.0||26.3 ± 4.5||6.3 ± 1.2||6.3 ± 1.2||1.4 ± 0.7||1.6 ± 0.4||668||4.1 ± 1.1||4.1 ± 1.1|
| Serum TSH 2.15–4.52 mUI L−1||685||60.7 ± 12.8||26.9 ± 4.7||6.4 ± 1.2||6.4 ± 1.3||1.4 ± 0.8||1.6 ± 0.4||667||4.2 ± 1.1||4.1 ± 1.2|
| Serum TSH >4.52 mUI L−1||71||62.7 ± 12.3**||26.9 ± 4.7**||6.7 ± 1.2||6.6 ± 1.4||1.4 ± 0.8||1.7 ± 0.5||70||4.4 ± 1.0||4.3 ± 1.3|
Among the 7954 subjects with serum TSH measurements in the 5th Tromsø study, 1253 were excluded due to reports in the health questionnaire (heart disease, stroke or diabetes) and an additional 576 subjects refused to participate in follow-up studies. Among the remaining 6125 subjects, 363 had serum TSH 3.50–10.0 mIU L−1. After checking their hospital records, 114 were excluded because of thyroxine use, illness or participation in other studies. The remaining 249 subjects were invited to the follow-up study, 167 attended, 89 fulfilled the criteria for SHT among whom five were using lipid lowering drugs and were thus excluded. Among the remaining 84 subjects 40 (52.4%) were anti-TPO positive. A total of 249 age and sex-matched controls that in the 5th Tromsø study had serum TSH 0.5–3.49 mIU L−1 were invited, 162 attended, and 154 had normal serum free T4, free T3 and TSH in the same range as in the 5th Tromsø study at re-examination. Nine of them used lipid medication and were therefore excluded. Among the remaining 145 subjects 24 (16.6%) were anti-TPO positive.
The characteristics of the 84 SHT and 145 control subjects are given in Table 4. The age and sex ratio were similar in the SHT and control group, but there were significantly more smokers in the control group than in the SHT group. The SHT group had higher BMI than the control group, but the difference was not statistically significant after adjusting for smoking status. In the females, the serum TC and LDL-C levels were significantly higher in the SHT group than in the controls. With males and females analysed together, the serum Apo A1 levels were significantly lower in the SHT group than in the controls. These differences remained significant also after adjusting for age, BMI, and smoking status. The serum TG levels were significantly higher and the serum HDL-C significantly lower in male SHT than in control subjects, and in the females the serum Apo B levels were significantly higher in the SHT females than in the controls. However, these differences were not significant after adjusting for the other variables (Table 4).
Table 4. Characteristics of the subjects in the SHT and control groups
|Age (years)||62.5 ± 12.0||60.6 ± 12.8||62.3 ± 12.8||63.3 ± 11.8||62.6 ± 11.4||58.4 ± 13.2|
|BMI (kg m−2)a||28.0 ± 5.2*||26.3 ± 3.8||28.2 ± 3.2*||26.7 ± 3.4||27.9 ± 6.7||25.9 ± 4.1|
|Serum TSH (mIU L−1)|| 5.7 ± 1.7**,††|| 1.8 ± 0.7|| 5.4 ± 1.3**,††|| 1.8 ± 0.7|| 6.0 ± 2.0**,††|| 1.7 ± 0.7|
|Serum FT4 (pmol L−1)||12.7 ± 1.9**,††||14.1 ± 2.1||13.4 ± 1.8**,††||15.2 ± 2.3||12.0 ± 1.8**,††||13.2 ± 1.4|
|Serum FT3 (pmol L−1)|| 3.7 ± 0.7,†|| 3.9 ± 0.8|| 4.1 ± 0.6|| 4.3 ± 0.9|| 3.3 ± 0.6*|| 3.5 ± 0.5|
|Serum TC (mmol L−1)a|| 5.8 ± 1.0|| 5.6 ± 1.0|| 5.7 ± 1.0|| 5.6 ± 0.9|| 6.0 ± 1.11|| 5.6 ± 1.1|
|Serum TG (mmol L−1)|| 1.5 ± 0.8|| 1.4 ± 1.0|| 1.8 ± 1.0*|| 1.4 ± 0.7|| 1.3 ± 0.5|| 1.4 ± 1.2|
|Serum HDL-C (mmol L−1)|| 1.5 ± 0.5*|| 1.7 ± 0.4|| 1.3 ± 0.4*|| 1.5 ± 0.4|| 1.7 ± 0.5|| 1.8 ± 0.4|
|Serum LDL-C (mmol L−1)a,b|| 3.6 ± 0.9*,††||3.30 ± 0.9|| 3.6 ± 0.9|| 3.4 ± 0.8|| 3.7 ± 1.0*,††|| 3.2 ± 0.9|
|Serum Apo A1 (g L−1)|| 1.4 ± 0.3**,†|| 1.5 ± 0.3|| 1.3 ± 0.2**,†|| 1.4 ± 0.3|| 1.5 ± 0.3*|| 1.6 ± 0.3|
|Serum Apo B (g L−1)a|| 1.3 ± 0.3**|| 1.2 ± 0.3|| 1.3 ± 0.3|| 1.2 ± 0.2|| 1.2 ± 0.3*|| 1.1 ± 0.3|
Sixty-four subjects were included in the intervention study, 32 in the thyroxine group and 32 in the placebo group. In the thyroxine group, two subjects had at the end of the study serum TSH <0.2 mIU L−1, seven subjects serum TSH 0.2–0.49 mIU L−1, eight subjects serum TSH 0.50–1.50 mIU L−1, eight subjects serum TSH 1.51–1.99 mIU L−1, four subjects serum TSH 2.00–3.49 mIU L−1, and three subjects serum TSH 3.50–6.85 mIU L−1. A subanalysis was, therefore, done in the 23 subjects who at the end of the study had serum TSH levels within the range 0.2–2.0 mIU L−1 (thyroxine subgroup). Characteristics for these three groups are given in Table 5. The average thyroxine dose during the last 9 months in the thyroxine group and thyroxine subgroup were 97.1 ± 20.1 and 95.7 ± 20.2 μg, respectively, and the mean compliance rate in all three groups were 91%.
Table 5. Characteristics of the subjects in the intervention study at baseline and at the end of the study
|Males/females||16/16|| ||17/15|| ||13/10|| |
|Age (years)||62.0 ± 11.9|| ||62.7 ± 12.4|| ||63.0 ± 11.7|| |
|BMI (kg m−2)||28.7 ± 5.7||28.4 ± 5.8||27.1 ± 3.7||27.0 ± 4.1||29.2 ± 5.7||29.0 ± 6.0|
|Smokers/nonsmokers||6/36|| ||3/29|| ||3/20|| |
|Serum TSH (mIU L−1)||5.8 ± 1.8||1.5 ± 1.4**||5.4 ± 1.3||5.4 ± 2.0||5.7 ± 1.9||1.0 ± 0.6**|
|Serum free T4 (pmol L−1)||12.7 ± 1.8||17.7 ± 2.8**||12.9 ± 2.0||13.8 ± 2.1||12.7 ± 1.7||17.5 ± 2.2**|
|Serum free T3 (pmol L−1)||3.5 ± 0.6||4.4 ± 0.6**||3.8 ± 0.7||4.7 ± 0.7||3.6 ± 0.6||4.4 ± 0.4**|
|Serum TC (mmol L−1)||5.9 ± 1.1||5.7 ± 1.1||5.8 ± 0.8||5.8 ± 0.9||6.1 ± 1.1||5.8 ± 1.1*,†|
|Serum TG (mmol L−1)||1.5 ± 0.9||1.5 ± 1.0||1.6 ± 0.9||1.6 ± 0.7||1.7 ± 1.0||1.7 ± 1.1|
|Serum HDL-C (mmol L−1)||1.5 ± 0.4||1.5 ± 0.4||1.5 ± 0.6||1.5 ± 0.5||1.5 ± 0.4||1.5 ± 0.4|
|Serum LDL-C (mmol L−1)b||3.7 ± 0.9||3.6 ± 0.9||3.6 ± 0.8||3.6 ± 1.0||3.8 ± 1.0||3.5 ± 1.0**|
|Serum Apo A1 (g L−1)||1.4 ± 0.2||1.4 ± 0.2||1.4 ± 0.3||1.5 ± 0.3||1.4 ± 0.3||1.5 ± 0.3|
|Serum Apo B (g L−1)||1.3 ± 0.3||1.2 ± 0.3**||1.3 ± 0.3||1.2 ± 0.3||1.3 ± 0.3||1.2 ± 0.3**|
When comparing the values at the end of the study versus the baseline values, there was a significant reduction in serum Apo B levels in the thyroxine and the thyroxine subgroup. In the thyroxine subgroup, there was also a significant reduction in the serum TC and LDL-C levels at the end of the study (Table 5).
Regarding delta values (value at inclusion minus value at end of study) there were no significant differences between the thyroxine and placebo group. However, delta TC was significantly higher in the thyroxine subgroup when compared with the placebo group [−0.28 ± 0.52 and 0.02 ±0.59 mmol L−1, respectively (P < 0.01, general linear model with age, BMI, and smoking status as covariables)].
Furthermore, when considering all the 64 subjects in the intervention study together, there was a significant and positive correlation between delta serum TSH and delta serum cholesterol (ρ = 0.27, P < 0.05), an association that was also present in a linear regression model that included delta BMI, smoking status, gender, and age (R2 = 0.12, standardized beta coefficient = 0.29, t = 2.24, and P < 0.05). There were no other significant associations between delta serum TSH and the other serum lipids, and analysing the data from the intervention study sex-specific did not add further information.
In the 5143 subjects from the 5th Tromsø study, we found the serum TSH level to be significantly associated with the serum TC and LDL-C levels. Thus, after adjusting for age and BMI the difference between those in the lowest and highest serum TSH quartiles within the normal range was for serum TC and serum LDL-C only 0.18 and 0.12 mmol L−1 for the males and 0.06 and 0.05 mmol L−1 for the females, respectively. For those in the upper 2.5 percentile there was no further increase in serum TC and LDL-C in the males, whereas there was a modest increase in the females. Accordingly, serum TSH is not a major determinant of the serum TC and LDL-C levels. However, in our study, there were a few subjects with serum TSH >10 mIU L−1 and in profound hypothyroidism the relation between serum TSH and lipids may be different.
In the follow-up study, where we looked specifically at those with SHT, the serum TC and LDL-C levels were significantly higher in the females in the SHT group than in the controls. The serum TC and LDL-C levels were also higher in the males SHT subjects, but the difference did not reach statistical significance. The relation between SHT and serum lipids has been reported in numerous studies but the results differ. Thus, some have found no significant difference in TC between SHT and control subjects [3, 16], whereas other have found elevated TC levels in SHT subjects [6, 17, 18]. However, we have not been able to find any study where the TC or LDL-C levels were significantly lower in SHT subjects than controls, and we therefore consider it likely that our finding of a slightly increased TC and LDL-C levels in subjects with SHT is correct. This is also in line with the modest positive relation between serum TSH and lipid levels found in the epidemiological part of our study.
The association between serum TC and serum LDL-C and coronary heart disease is well established , but recently the importance of other lipoproteins than LDL-C has become apparent. Thus, low levels of Apo 1, which is the major apolipoprotein associated with HDL-C, and high Apo B, which is the major apolipoprotein associated with LDL-C, are linked to increased risk of coronary heart disease . In line with this, we found significantly lower Apo A1 levels and higher Apo B levels in the SHT group than in the controls. The difference regarding Apo A1 was also significant after adjusting for confounders, and has to our knowledge not been reported before. However, there are many studies where no significant difference between SHT subjects and controls regarding Apo A1 have been reported [2, 4, 6, 8, 17, 21, 22]. On the other hand, there are several reports on elevated Apo B levels in SHT subjects, as we also observed before adjusting for confounding factors [6, 8, 17, 21, 22]. However, unlike us, these investigators did not take into consideration differences in age and BMI.
The intervention study revealed at the end of the study a significant reduction in Apo B levels in those given thyroxine, but insignificant differences between the thyroxine and placebo groups regarding delta values (values at baseline minus value at end of the study). However, only eight subjects in the thyroxine group had serum TSH levels within the target range of 0.5–1.5 mIU L−1 at the end of the study, whereas 23 subjects had serum TSH levels in the 0.2–2.0 mIU L−1 range. We, therefore, did a subanalysis on these 23 subjects, and there was a significant reduction in serum TC, LDL-C, and Apo B after 1 year of thyroxine medication. Furthermore, the mean delta TC value was 0.30 mmol L−1 higher than in the control group, and this difference was statistically significant.
There are many previous reports on the effect of thyroxine treatment on serum TC, TG, HDL-C and LDL-C in SHT subjects, which has been summarized in three recent meta-analyses [23–25]. They all conclude that thyroxine substitution has no effect on the TG and HDL-C levels, whereas the TC and LDL-C cholesterol levels are reduced with the effect most pronounced in those with the highest pretreatment serum TSH  and serum TC levels [23, 24].
We observed no significant effect of thyroxine treatment on the Apo A1 levels, which is in accordance with most studies [6–8, 10, 21, 22] except the one by Efstatiadou et al.  who found a significant decrease after thyroxine treatment. For serum Apo B levels our finding of a significant reduction after thyroxine treatment is similar to that reported by others [7, 17, 21].
Thus, when considering the epidemiological data together with results from the intervention studies it is reasonable to assume that there is a causal relation between thyroid function and the cardiovascular risk factors TC, LDL-C, and Apo B. Furthermore, subjects with SHT appear to have an atherogenic lipid profile, and although the difference versus healthy controls in this respect is modest, thyroxine treatment in this regard seems favourable .
Our study has several limitations. First, we did not measure serum free T4 and free T3 in the 5th Tromsø study, which would have added more information on the subjects’ thyroid function. Secondly, when comparing the SHT group with the controls, and when comparing the thyroxine group with the placebo group in the intervention study, there were several differences that might have reached statistical significance had we included a larger number of subjects. And thirdly, to do analysis on a ‘thyroxine subgroup’ that only included subjects with serum TSH values at or close to the treatment target range at the end of the study, was not intended when the study was planned.
On the other hand, our study has considerable strength. We included a large group of subjects in the epidemiological part of the study, and the analyses were done with careful correction for confounding factors. Furthermore, the subjects in the SHT and control groups were recruited from an epidemiological study with strict selection criteria. Recruitment from clinical practice would favour inclusion of subjects with high pretreatment lipid levels as this in many cases are the reason for doing thyroid function tests, and could possibly bias the results. Our subjects also had a stable thyroid function with the serum TSH level elevated (or normal) on at least two occasions before inclusion in the study, which recently has been emphasized as an important inclusion criterion . We also used a rather low TSH range for inclusion in the SHT group, as these subjects are the ones where it is difficult to decide on starting treatment with thyroxine or not. The upper ‘normal’ TSH level is a matter of definition and depends on whether subjects with presence of thyroid autoantibodies and other risk factors for hypothyroidism are included. Thus, in the study by Bjøro et al.  the upper ‘normal’ serum TSH level was c. 3.5 mIU L−1 when excluding subjects with thyroid autoantibodies, and this level was therefore applied when selecting our study groups.
In conclusion, we have found modest but significant associations between serum TSH levels and serum TC and LDL-C levels. These associations are most likely causal, and thyroxine medication reduces the slightly elevated TC and LDL-C levels in subjects with SHT.
This study has been financed with the aid of EXTRA funds from the Norwegian Foundation for Health and Rehabilitation and by a grant from the Norwegian Research Council. The superb assistance by the staff at the Clinical Research Unit, University Hospital of North Norway, and by Inger Myrnes and Astrid Lindvall at the Department of Clinical Chemistry, University Hospital of North Norway is gratefully acknowledged. The thyroxine and placebo tablets were generously supplied by NycoMed Pharma.
Conflict of interest statement
The thyroxine and placebo tablets were supplied by NycoMed Pharma, but apart from that no other relationships exist that might lead to a conflict of interest.