Thyroid stimulating hormone, independent of thyroid hormone, can elevate the serum total cholesterol level in patients with coronary heart disease: a cross-sectional design.

Abstract

The relationship between TSH and the lipid profile is contradictory because few studies have excluded the potential influence of the thyroid hormones (TH). The aim of the present study was to evaluate the relationship between serum TSH levels and the lipid profile independent of TH.
1302 CHD patients diagnosed by coronary angiography were retrospectively studied. The prevalence and distribution of thyroid dysfunction were analyzed first. To assess the impact of TSH on serum lipids, Pearson's correlation analysis was performed after adjustments for classic factors and TH. To calculate the extent of the effect of TSH on the serum cholesterol level, the partial least squares method and additional statistical methods were used.
After the exclusions, a total of 568 patients (270 males and 298 females with a mean age of 63.56 ± 11.376 years) were selected. The prevalence of thyroid dysfunction among the patients was 18.66%, and the prevalence of hypothyroidism (15.32%) was higher than that of hyperthyroidism (3.34%). Even after adjusting for confounding factors, such as sex, age, smoking status, fasting plasma glucose levels and TH, a significant positive impact of TSH on the serum total cholesterol (TC) level was revealed (r = 0.095, p = 0.036). Each 1 mIU/L increase in the TSH level might be linked to a 0.015580712 mmol/L elevation of the serum TC value.
TSH can increase the TC level in CHD patients independent of TH. The present study suggests a potential physiological role of TSH and the importance of maintaining an appropriate TSH level in CHD patients.

Full text

Page 1

RESEARCHOpen Access
Thyroid stimulating hormone, independent of
thyroid hormone, can elevate the serum total
cholesterol level in patients with coronary heart
disease: a cross-sectional design
Chao Xu1,4, Xiaomei Yang1, Wenhui Liu2, Haitao Yuan3, Chunxiao Yu1,4, Ling Gao4,5and Jiajun Zhao1,4*
Abstract
Background: The relationship between TSH and the lipid profile is contradictory because few studies have
excluded the potential influence of the thyroid hormones (TH). The aim of the present study was to evaluate the
relationship between serum TSH levels and the lipid profile independent of TH.
Methods: 1302 CHD patients diagnosed by coronary angiography were retrospectively studied. The prevalence and
distribution of thyroid dysfunction were analyzed first. To assess the impact of TSH on serum lipids, Pearson’s
correlation analysis was performed after adjustments for classic factors and TH. To calculate the extent of the effect
of TSH on the serum cholesterol level, the partial least squares method and additional statistical methods were
used.
Results: After the exclusions, a total of 568 patients (270 males and 298 females with a mean age of
63.56±11.376 years) were selected. The prevalence of thyroid dysfunction among the patients was 18.66%, and the
prevalence of hypothyroidism (15.32%) was higher than that of hyperthyroidism (3.34%). Even after adjusting for
confounding factors, such as sex, age, smoking status, fasting plasma glucose levels and TH, a significant positive
impact of TSH on the serum total cholesterol (TC) level was revealed (r=0.095, p=0.036). Each 1 mIU/L increase in
the TSH level might be linked to a 0.015580712 mmol/L elevation of the serum TC value.
Conclusions: TSH can increase the TC level in CHD patients independent of TH. The present study suggests a
potential physiological role of TSH and the importance of maintaining an appropriate TSH level in CHD patients.
Keywords: TSH, Cholesterol, Coronary heart disease
Background
The thyroid hormones exert a wide range of functions in
several organs, including the heart [1]. Abnormal thyroid
hormone metabolism may lead to different forms of
heart disease and hypothyroidism, in particular, is a well-
known cause of accelerated coronary atherosclerosis
[2,3]. Moreover, similar consequences were found for
subclinical hypothyroidism (SCH), which is character-
ized by elevated serum thyroid stimulating hormone
(TSH) levels and normal thyroxine (T4) levels [4]. Ele-
vated TSH levels have recently aroused interest due to
the potential for TSH to induce injury, especially in
patients with coronary heart disease (CHD). A series of
studies reported that a high level of TSH was associated
with a deleterious change of serum lipids, with an in-
crease of lipid abnormalities [5-8]; however, this issue
has been the subject of considerable debate [9], and sev-
eral studies have not observed such an association
[10,11]. The differences in the studies have been ascribed
to the influence of some confounding factors, such as
age, gender and body mass index (BMI). Existing evi-
dence has demonstrated that the relationship of TSH
and lipid levels was different between overweight and
* Correspondence: jjzhao@medmail.com.cn.
1Department of Endocrinology, Provincial Hospital affiliated to Shandong
University, Jinan, China
4Institute of Endocrinology, Shandong Academy of Clinical Medicine, Jinan,
China
Full list of author information is available at the end of the article
© 2012 Xu et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
Xu et al. Nutrition & Metabolism 2012, 9:44
http://www.nutritionandmetabolism.com/content/9/1/44

Page 2

normal weight populations and between men and
women [12]. Furthermore, the thyroid hormones play an
important role in regulating lipid metabolism. Numerous
studies have confirmed the presence of an inverse rela-
tionship between serum thyroxin and cholesterol levels
[2]. Even within the reference range, serum free thyrox-
ine (FT4) levels near the upper limit have been asso-
ciated with different metabolic markers in euthyroid
subjects and patients with coronary artery disease
[13,14]. Therefore, to evaluate the essential relationship
between TSH and the lipid status, it is necessary to ad-
just for age, gender, BMI and thyroid hormone levels.
Regrettably, few studies have excluded the potential
influences of the thyroid hormones when assessing the
relationship between TSH and the lipid status.
Interestingly, in vivo and in vitro research by our la-
boratory on the function of TSH has shown that TSH,
independent of thyroid hormones, can upregulate the
expression of hepatic 3-hydroxy-3-methyl-glutaryl coen-
zyme A reductase (HMGCR), which is the rate-limiting
enzyme in cholesterol synthesis, and increase the choles-
terol content in the liver [15]. Therefore, we hypothe-
sized that TSH, independent of thyroid hormones,
would be positively associated with the serum choles-
terol level.
The present study evaluated the relationship between
TSH and the lipid status after adjusting for classic con-
founding factors and the thyroid hormones. We also
analyzed the extent to which TSH can affect serum lipid
parameters. The present study yielded insights into po-
tentially novel effects of TSH on serum lipids and sug-
gested that it is necessary to routinely test thyroid
function in CHD patients. Maintaining serum TSH
levels in an appropriate range will achieve homeostasis
of the lipid levels and slow the progression of athero-
sclerosis in CHD patients.
Materials and methods
Patients
A total of 1302 patients who were hospitalized in either
the Provincial Hospital or the Qianfushan Hospital,
which are affiliated with Shandong University (Jinan,
China), from2004to2010
reviewed. All of the patients were diagnosed with CHD
by coronary angiography according to the international
criteria. Information on medication and a history of pre-
vious medical or surgical diseases for each patient was
obtained. The smoking histories of the patients were also
recorded. The blood pressure values were obtained from
the medical records and presented as the mean of two
measures taken in the sitting position according to a
standardized protocol.
The following criteria were used for exclusion: (1) Eu-
thyroid sick syndrome, being characterized by low serum
were retrospectively
triiodothyronine (T3); (2) Acute myocardial infarction at
the moment of hospitalization; (3) Reduced (<50%) left
ventricular ejection fraction at echocardiography; (4)
Hypothalamus and/or pituitary gland diseases, diabetes
mellitus or other endocrine diseases; (5) Intake of drugs
that influence serum lipids or thyroid function within
the past 3 months; (6) cerebral vascular disease, a malig-
nant tumor, hereditary hyperlipidemia, or serious liver
or renal dysfunctions; (7) History of myocardial infarc-
tion or revascularization prior to hospitalization; and (8)
pregnancy. Generally, the patients were clinically stable
at the moment of hospitalization and those with serious
condition or in intensive care unit were excluded. In the
end, 568 patients (270 males and 298 females with a
mean age of 63.56±11.376 years) were selected and en-
rolled in the present study.
The local ethics committee approved the retrospective
review of the patients’ medical records and licensed the
records for research purposes only.
Laboratory analysis
All of the measurements were performed in the clinical
laboratory that is affiliated with Shandong University.
Blood samples were collected from all of the patients be-
tween 8:00 A.M. and 10:00 A.M. after a minimum of a
10-h fast. Chemiluminescent procedures (Cobas E610;
Roche, Basel, Switzerland) were employed to determine
the thyroid function of the patients, TSH, free triiodo-
thyronine (FT3), FT4 and reverse T3 (rT3). The labora-
tory reference ranges were 0.27-4.2 mIU/L for TSH, 3.1-
6.8 pmol/L for FT3, 12–22 pmol/L for FT4 and 0.54-
1.46 nmol/L for rT3. Thyroid function of the patients
was measured twice, before hospitalization and the sec-
ond day after hospitalization. The patients were
excluded when distinct variations between the two
results were found.
The levels of plasma glucose, total cholesterol (TC),
triglycerides (TG), low-density lipoprotein (LDL) choles-
terol, high-density lipoprotein (HDL) cholesterol and
uric acid (UA) were determined using an Auto Biochem-
ical Analyzer (MODULAR-000GS; Roche, Basel, Switz-
erland). Hypercholesterolemia was defined as a TC value
over 6.21 mmol/L, which is in accordance with the Na-
tional Cholesterol Education Program Adult Treatment
Panel III criteria (NCEP/ATPIII) [16].
All of the CHD patients were initially divided into two
groups based on their thyroid function: euthyroid and
thyroid dysfunction. Euthyroidism was defined by circu-
lating levels of FT3, FT4 and TSH that were within the
reference ranges. The patients with thyroid dysfunction
were further divided into the following 3 subgroups: (1)
hyperthyroidism, which was classified as a TSH level less
than 0.27 mIU/L and/or FT3 and FT4 levels above the
reference ranges; (2) subclinical hypothyroidism, which
Xu et al. Nutrition & Metabolism 2012, 9:44
http://www.nutritionandmetabolism.com/content/9/1/44
Page 2 of 8

Page 3

was classified as a TSH level above 4.2 mIU/L and FT3
and FT4 levels in the reference ranges; and (3)
hypothyroidism, which was classified as FT3 and FT4
levels less than the reference ranges with a TSH level
above 4.2 mIU/L.
Statistical analyses
Statistical tests were performed in a blinded fashion by
two statisticians using SPSS version 17.0 for Windows
(Chicago, IL, USA). Parametric and nonparametric data
are given as the mean±SD and the percentage. Groups
were compared using a one-way analysis of variance
(ANOVA) or Chi-squared test. Variables with a skewed
distribution were transformed to their natural logarithm
to optimize the models. The relationships between thy-
roid function and serum lipid parameters were evaluated
with Pearson’s correlation analysis. To adjust for the thy-
roid hormones and several risk factors, a partial correl-
ation analysis and a factorial analysis were performed
while evaluating the relationship between TSH and TC.
To examine the relationship between TSH and TC, we
used four different statistical methods, each with its own
advantages. We used linear regression analysis, principal
component analysis, the partial least squares method
and path analysis. These analyses are often used to
analyze similar issues and using them together can com-
pensate for the weaknesses of the individual methods.
All of the calculated p-values are two-sided, and p-values
less than 0.05 are considered statistically significant.
Results
Characteristics of the 568 CHD patients according to
serum TSH concentrations
The composition and general characteristics of the
patients are summarized in Table 1. The normal TSH
levels were stratified into three groups (0.27–1.57 mIU/L,
1.58–2.88 mIU/L and 2.89–4.19 mIU/L), and the patients
with CHD were divided into 5 groups according to the
serum TSH concentrations. Significant differences were
detected among the five groups with respect to the
levels of FPG (p=0.001), TC (p=0.013), TG (p=0.003),
FT3 (p<0.001) and FT4 (p<0.001). No significant dif-
ferences were found among the five groups with regard to
age, SBP, DBP, LDL, HDL or UA.
The prevalence and distribution of thyroid dysfunction in
the CHD patients
Among all of the CHD patients, the prevalence of thy-
roid dysfunction was 18.66% (Table 2). Interestingly, the
prevalence of hypothyroidism (15.32%), including overt
Table 1 The characteristics of 568 CHD patients who were stratified according to serum TSH concentration
Serum TSH concentration (mIU/L)
P
value
Total
<0.270.27-1.571.58-2.882.89-4.19
≥4.2
Age (y)64.84(11.64)65.47(11.27)61.92(11.54)63.69(10.57)64.46(10.82)0.05263.76(11.28)
SBP (mm Hg)137.47(17.01)134.77(22.72)132.87(19.45) 136.51(21.17)136.94(20.35)0.481 134.74(20.79)
DBP (mm Hg)80.42(7.77)79.02(12.54)79.22(11.96)78.51(11.83)80.43(10.54)0.84779.27(11.78)
FPG (mmol/L)6.1328(1.24)5.4976(1.04)5.3802(1.30)5.0863(0.55)5.2059(0.83)0.0015.3733(1.093)
TC (mmol/L)4.6163 (1.3305)4.8699(1.03)4.9598(1.15)4.9685(1.03)5.3518(1.34)0.0134.9829(1.15)
TG (mmol/L)1.3226(0.69)1.3669(0.82)1.7644(1.62)1.6951(0.85)1.9766(1.66)0.0031.6538(1.32)
LDL (mmol/L)3.00(0.851)2.91(0.77)2.88(0.841)2.91 (0.74)3.16 (0.91)0.1112.94(0.82)
HDL (mmol/L)1.1506(0.29)1.2891(0.35)1.3229(0.37)1.2762(0.32)1.3557(0.29)0.1411.3060(0.34)
UA (μmol/L)297.9688 (107.94)313.9742(90.85)321.0955(101.73)311.7162(70.90) 321.63(109.47)0.817317.0161(95.98)
TSH (mIU/L)0.1498(0.11)1.0532(0.34)2.1371(0.35)3.5424(0.56)19.760(33.33)0.000 4.6504(14.52)
FT3 (pmol/L)8.1885(6.20)4.5007(0.79) 4.5055(0.71) 4.5020(0.83) 3.5891(1.24)0.000 4.4864(1.58)
FT4 (pmol/L) 31.2682(17.85) 17.5750(2.50)17.3971(2.84) 16.7438(2.44) 13.4475(4.76)0.00017.2147(5.30)
Smoking statusA17 121 159 72 800.0000
B023 2635 0.0008
C223 21610.0009
D025110.2222
1. The data are presented as the mean±standard deviation (SD). The SD is shown in parentheses.
2. The groups were compared using a Chi-squared test. P-values less than 0.05 were considered statistically significant. 3. Abbreviations: SBP, systolic blood pressure;
DBP, diastolic blood pressure; FPG, free plasma glucose; TC, total cholesterol; TG, triglycerides; LDL, low-density lipoprotein cholesterol; HDL, high-density lipoprotein
cholesterol; and UA, uric acid. 4. The following reference ranges were based on our laboratory standards: Plasma glucose (3.9-6.1 mmol/L), total cholesterol
(3.6-6.2 mmol/L), triglycerides (<1.71 mmol/L), low-density lipoprotein cholesterol (0–3.36 mmol/L), high-density lipoprotein cholesterol (>1.15 mmol/L), TSH
(0.27–4.2 mIU/L), FT3 (3.1-6.8 pmol/L), FT4 (12–22 pmol/L) and uric acid (142–339 μmol/L). 4. The smoking status categories are as follows: A, B, C and D represent no
smoking, smoking 1–339, 440–1099, and above 1100 cigarettes per year, respectively. The data are presented as the number of people in each category.
Xu et al. Nutrition & Metabolism 2012, 9:44
http://www.nutritionandmetabolism.com/content/9/1/44
Page 3 of 8

Page 4

and subclinical hypothyroidism, was higher than the
prevalence of hyperthyroidism (3.34%). The prevalence
of thyroid dysfunction in the females (26.17%) was
higher than the prevalence in the males (10.37%).
The prevalence of hypercholesterolemia in the CHD
patients increases with the serum TSH levels
In the entire study population, there were 76 cases with
hypercholesterolemia, which was approximately 13.5%.
The prevalence of hypercholesterolemia showed a linear
and significant increase with elevations of the serum
TSH levels (Pearson’s Chi-squared test, linear trend
0.010, p<0.05; Figure 1). As expected, the patients with
a TSH level below 0.27 mIU/L did not suffer from
hypercholesterolemia. However, the prevalence of hyper-
cholesterolemia increased significantly in patients who
showed an increased TSH level within the reference
range. The prevalence of hypercholesterolemia was also
significantly higher in the patients with the serum TSH
levels between 2.89 and 4.19 mIU/L compared with the
patients with serum TSH levels between 0.27 and 1.57
mIU/L (14.6% vs. 10.1%; p<0.05). The patients with
values of TSH greater than or equal to 4.2 mIU/L
showed the highest prevalence of hypercholesterolemia
(20.7%).
The relationship between thyroid function and the serum
lipid parameters
Because the serum lipid parameters did not fit a normal
distribution, log transformations [17] were applied to the
values of TC, TG, LDL and HDL (Table 3).
In total patients, significant negative correlations were
identified between the levels of the thyroid hormones
(FT3 and FT4) and the serum TC or HDL levels
(r=−0.125, p=0.003 for FT3 and LogTC; r=−0.132,
p=0.002 for FT3 and LogHDL; r=−0.190, p<0.001 for
FT4 and LogTC; and r=−0.184, p<0.001 for FT4 and
LogHDL). The levels of FT4 were also negatively corre-
lated with the serum LDL levels (r=−0.084, p=0.047).
Interestingly, the correlations remained significant after
we adjusted for age, gender, smoking status and FPG.
There was no significant relationship between the thy-
roid hormones (FT3 and FT4) and the serum TG level.
We observed a significant positive impact of TSH on
the serum TC, TG and HDL levels (r=0.160, p<0.001
for TSH and LogTC; r=0.085, p=0.044 for TSH and
LogTG; and r=0.102, p=0.015 for TSH and LogHDL).
Interestingly, the correlations remained significant for
the TC level after we adjusted for age, gender, smoking
status and FPG (r=0.146, p=0.001). After an additional
adjustment for the thyroid hormone levels (FT3 and
Table 2 The prevalence and distribution of thyroid dysfunction in CHD patients
CasesEuthyroidThyroid dysfunction
HyperthyroidismSubclinical hypothyroidismHypothyroidism
Male270242 (89.63%)4 (1.48%)19 (7.04%)5 (1.85%)
Female298220 (73.83%)15 (5.03%)38 (12.75%)25 (8.39%)
Total568462 (81.34%)19 (3.34%)57 (10.04%)30 (5.28%)
p value
<0.001
<0.01
<0.01
<0.05
The data are presented as the number (percentage) of individuals in each category. The groups were compared using a Chi-squared test. P-values less than 0.05
were considered statistically significant.
Figure 1 The prevalence of hypercholesterolemia in CHD patients who were stratified according to the serum TSH concentrations. The
data are presented as the number (percentage) of individuals in each category. The prevalence of hypercholesterolemia showed a linear and
significant increase with elevations of the serum TSH levels (Pearson’s Chi-squared test, linear trend 0.010, p<0.05; Figure 1).
Xu et al. Nutrition & Metabolism 2012, 9:44
http://www.nutritionandmetabolism.com/content/9/1/44
Page 4 of 8

Page 5

FT4), the significance still remained for the TC level
(r=0.095, p=0.036 for TSH and LogTC) but not for the
levels of TG, LDL or HDL.
We also conducted statistics in euthyroidic patients
separately and the results were shown in Table 3. In gen-
eral, the results were similar to those conducted in total
patients.
The response relationship between TSH and TC
All of the four different statistical methods revealed that
TSH could increase the serum TC level to a certain extent
(Data are available upon request). Interestingly, the partial
least squares method had advantage on overcoming the
adverse effects caused by correlation between the inde-
pendent variables. Table 4 shows the results of this
method. The partial least squares method revealed that
each 1 mIU/L increase in TSH was estimated to elevate
theTC level by 0.015580712 mmol/L (p<0.001). However,
the thyroid hormones (FT3 and FT4) could negatively
affect the serum TC level (coefficients=−0.001921199 and
−0.011725050 for FT3 and FT4, respectively; p <0.001),
which was consistent with previous observations [2].
Discussion
The present study reported the relationship between
TSH levels and the lipid status after adjustments for the
thyroid hormones and/or other potentially confounding
factors in patients with CHD. The principal finding is
that the prevalence of hypercholesterolemia increased as
the serum TSH level increased and TSH per se was posi-
tive correlated with TC. Although the TSH-induced in-
crease in TC is weak, it may be physiologically relevant
and clinically significant. Thyroid function has an im-
portant role in the risk stratification of these patients
with suspected CHD and should be routinely tested in
the patients at risk of CHD. Maintaining the serum TSH
levels in an appropriate range will achieve homeostasis
of the lipid levels and slow the progression of athero-
sclerosis in CHD patients.
To ensure our study to have considerable strength, we
used more stringent exclusion criteria and the impact of
potential confounders was minimized. Firstly, we ruled
out the individuals with euthyroid sick syndrome (ESS)
as possible as we can. ESS, also known as low triiodo-
thyronine syndrome, is often seen in starvation, critical
illness or patients in intensive care unit. The prevalence
of ESS was as higher as 29.2% in the patients with car-
diovascular diseases [18]. The most prominent altera-
tions in this condition are low serum T3 and elevated
rT3 levels [19]. Individuals with such characteristics
were excluded. Secondly, the patients enrolled our study
were clinically stable. Acute myocardial infarction and
any other acute or serious diseases were not included.
Third and most important, individuals using drugs that
influence serumlipidsor
excluded. This may sound weird because nearly all CHD
patients should take lipid-lowering drugs known as
thyroidfunction were
Table 3 Correlation analysis of thyroid function and
serum lipid parameters in total patients and euthyroidic
patients
total patientsLogTCLogTGLogLDLLogHDL
FT3Unadjustedr=-0.125r=-0.010r=-0.053r=-0.132
P=0.003P=0.810 P=0.212P=0.002
Adjusted*r=-0.122r=-0.021r=-0.057r=-0.098
P=0.007P=0.646 P=0.210 P=0.030
FT4Unadjustedr=-0.190r=-0.046r=-0.084r=-0.184
P=0.000P=0.269P=0.047P=0.000
Adjusted*r=-0.197r=-0.035r=-0.105 r=-0.176
P=0.000 P=0.442P=0.021P=0.000
TSHUnadjusted r=0.160r=0.085r=0.047r=0.102
P=0.000P=0.044P=0.269 P=0.015
Adjusted*r=0.146 r=0.051r=0.074 r=0.069
P=0.001 P=0.261P=0.103P=0.131
Adjusted**r=0.095r=0.044r=0.048r=0.020
P=0.036P=0.337P=0.295P=0.658
euthyroidic patients LogTCLogTGLogLDLLogHDL
FT3 Unadjustedr=-0.150r=-0.113 r=-0.102r=0.052
P=0.001 P=0.015P=0.029P=0.269
Adjusted*r=-0.190r=-0.072r=-0.141 r=-0.031
P=0.000 P=0.153 P=0.005P=0.537
FT4 Unadjustedr=-0.012 r=-0.043r=-0.008 r=0.140
P=0.004P=0.355 P=0.859 P=0.003
Adjusted*r=-0.011r=-0.057 r=-0.020r=0.120
P=0.003 P=0.255P=0.690P=0.017
TSHUnadjusted r=0.055 r=0.199r=0.003r=0.002
P=0.040P=0.000 P=0.955P=0.974
Adjusted*r=0.050 r=0.185r=0.012r=-0.010
P=0.018P=0.001P=0.815P=0.847
Adjusted**r=0.049r=0.190 r=0.011r=-0.019
P=0.027P=0.007P=0.821P=0.707
TC, TG, LDL, and HDL were transformed to their natural logarithm to optimize
the models. P-values less than 0.05 were considered statistically significant.
* Adjusted for sex, age, smoking status and FPG.
** Adjusted for sex, age, smoking status, FPG and thyroid hormones.
Table 4 The response relationship between thyroid
function and the serum TC level
VariableCoefficientsSignificant frequencies P value
TSH0.015580712 89
<0.001
FT3 -0.00192119956
<0.001
FT4-0.011725050 97
<0.001
The data are presented as the coefficients and significant frequencies. P-values
less than 0.05 were considered statistically significant.
Xu et al. Nutrition & Metabolism 2012, 9:44
http://www.nutritionandmetabolism.com/content/9/1/44
Page 5 of 8