Pediatr Cardiol (2013) 34:1166–1174

DOI 10.1007/s00246-013-0638-z

ORIGINAL ARTICLE

Subclinical Hypothyroidism as a Risk Factor for the Development

of Cardiovascular Disease in Obese Adolescents With
Nonalcoholic Fatty Liver Disease
Ahmet Sert • Ozgur Pirgon • Ebru Aypar •

Hakan Yilmaz • Dursun Odabas

Received: 20 September 2012 / Accepted: 8 January 2013 / Published online: 24 January 2013
Ó Springer Science+Business Media New York 2013

Abstract No data are available on the relationship such as total cholesterol (r = 0.606, p = 0.001), triglyc-
between subclinical hypothyroidism and risk factors for erides (r = 0.476, p = 0.016), low-density lipoprotein
the development of cardiovascular disease in obese ado- cholesterol (r = 0.461, p = 0.004), insulin (r = 0.607,
lescents with nonalcoholic fatty liver disease (NAFLD). p = 0.001), HOMA-IR (r = 0.596, p = 0.002), carotid
This study aimed to determine whether an association IMT (r = 0.894, p \ 0.0001), and LVM (r = 0.563,
exists between subclinical hypothyroidism and risk factors p = 0.003). The findings demonstrated that the obese
for the development of cardiovascular disease in obese adolescents with NAFLD and subclinical hypothyroidism
adolescents with NAFLD. The study enrolled 111 obese had a more adverse cardiovascular risk profile and a higher
adolescents and 42 lean subjects. The obese subjects were carotid IMT and LVM.
divided into two subgroups based on the presence or
absence of fatty liver with high transaminases: a NAFLD Keywords Nonalcoholic fatty liver disease 
group and a non-NAFLD group. Subclinical hypothyroid- Subclinical hypothyroidism
ism was defined as a thyroid-stimulating hormone (TSH)
level higher than 4 mIU/l and a normal free-thyroxine
level (0.6–1.8 ng/dl). Insulin resistance was calculated by Introduction
the homeostasis model assessment (HOMA-IR). Left
ventricular mass (LVM), LVM index measurements, car- Carotid intima media thickness (IMT), left ventricular mass
otid intima media thickness (IMT), and HOMA-IR values (LVM), and adolescent subclinical hypothyroidism are risk
were higher in the NAFLD obese group with TSH levels factors for the development of cardiovascular disease in obese
higher than 4 mIU/l than in the NAFLD obese group with adolescents with nonalcoholic fatty liver disease (NAFLD).
TSH levels lower than 4 mIU/l. Elevated TSH values in Obesity in childhood and adolescence has been associated
the NAFLD obese group were positively correlated with with adverse changes in cardiac geometry, including
most of the metabolic and cardiovascular risk parameters increased LVM [15]. Abnormal LVM and cardiac geometry
are well-established risk factors for cardiovascular events [4].
Due to the current epidemic of pediatric obesity, NAFLD
A. Sert (&)  E. Aypar  D. Odabas has become the most frequently diagnosed liver disease
Department of Pediatric Cardiology, Konya Training among children and adolescents [18]. Findings show NAFLD
and Research Hospital, 42080 Konya, Turkey to be strongly associated with obesity, insulin resistance,
e-mail: ahmetsert2@[Link]
hypertension, and dyslipidemia, and NAFLD currently is
O. Pirgon regarded as the liver manifestation of metabolic syndrome
Department of Pediatric Endocrinology and Diabetes, [24]. Increasing evidence supports an association between
Faculty of Medicine, S. Demirel University, Isparta, Turkey NAFLD and an increased risk of cardiovascular morbidity and
mortality [34]. The measurement of carotid artery IMT is
H. Yilmaz
Department of Radiology, Konya Training and Research increasingly used to evaluate the cardiovascular risk and target
Hospital, Konya, Turkey

123
Pediatr Cardiol (2013) 34:1166–1174 1167

organ damage in children and adolescents with metabolic were taking medications, or had a condition known to effect
abnormalities [17]. insulin action or insulin secretion (e.g., glucocorticoid ther-
Recently, increased interest has focused on the associ- apy, Cushing’s disease). We also excluded all patients who
ation between thyroid dysfunction and obesity. In 7–23 % had any previous diagnosis or treatment for thyroid
of obese children, moderately elevated thyroid-stimulating dysfunction.
hormone (TSH) levels are manifest in association with Subclinical hypothyroidism was defined as a TSH level
normal fT4 or fT3 levels, but the mechanisms underlying exceeding 4.0 mIU/l and a normal free thyroxine level
such thyroid hormonal changes in obese children are (0.6–1.8 ng/dl). According to data from the National
unclear [25]. Patients with primary hypothyroidism have a Health and Nutrition Examination Survey (NHANES), in a
threefold greater risk for early atherosclerosis, as shown disease-free population, the TSH values between percen-
independently for other risk factors such as atherogenic tiles 2.5 and 97.5 ranged from 0.45 to 4.17 mIU/l. Based on
lipid profile, hypertension, and impaired endothelial func- these data, we chose our cutoffs to be values less than
tion. Whether subclinical hypothyroidism has an influence 0.45 mIU/l and greater than 4.0 mIU/l [11].
on the same risk factors and is associated with athero- Anthropometric measurements of all the patients were
sclerosis still is debated [38]. Some studies show an asso- performed. Height and weight were measured with an
ciation [6, 10, 31], but others do not [28, 37]. empty bladder in postabsorptive conditions. The height was
However, no data are available on the relationship measured to the nearest 0.5 cm on a standard height board,
between subclinical hypothyroidism and risk factors for the and the weight was determined to the nearest 0.1 kg on a
development of cardiovascular disease in obese adoles- standard physician’s beam scale, with the subject dressed
cents with NAFLD. Therefore, we aimed to determine only in light underwear without shoes. The BMI was cal-
whether an association exists between subclinical hypo- culated as weight (kg) divided by height (m) squared.
thyroidism and risk factors for the development of car- Patients with a BMI at the 95th percentile or higher
diovascular disease in obese adolescents with NAFLD. according to reference curves for Turkish adolescents were
accepted as obese [23]. Pubertal development stage was
assessed by a single pediatric endocrinologist using the
Materials and Methods Tanner criteria, and the staging for sexual maturation
exceeded two for all the patients (Tanner stages 2–4).
Study Population After the child had rested for at least 5 min and was in a
sitting position, diastolic and systolic pressure (mmHg)
From April 2011 to April 2012, 111 obese adolescents (57 measurements were taken using a mercury-gravity manom-
girls and 54 boys) with mean age of 13.19 ± 1.3 years eter and a cuff appropriate for body size.
(range 12–17 years) and a mean body mass index (BMI) of
29.89 ± 3.30 kg/m2 were recruited from obese adolescents Laboratory Assessment
admitted to the Pediatric Endocrinology Unit. The obese
group was divided into two subgroups: NAFLD patients (25 Fasting blood samples were obtained by venipuncture in
girls and 33 boys; mean age, 13.13 ± 1.26 years; mean the morning at 8 a.m., after an overnight fast of at least
BMI, 30.36 ± 3.08 kg/m2) with high alanine aminotrans- 12 h, to measure serum glucose, insulin levels, and other
ferase (ALT) levels ([40 U/L) and ultrasound evidence of parameters. Glucose was determined by the glucose oxi-
fatty changes in the liver (group 1) and non-NAFLD patients dase method.
(32 girls and 21 boys; mean age, 13.25 ± 1.34 years; mean Serum concentrations of total cholesterol, high-density
BMI, 29.33 ± 3.47 kg/m2) with low ALT levels (\40 U/L) lipoprotein (HDL) cholesterol, and triglycerides were mea-
and no ultrasound evidence of fatty changes in the liver sured using routine enzymatic methods with the (Abbott
(group 2). Lean adolescents (17 girls and 25 boys; mean age, diagnostics C16000 chemistry analyzer, Illinois, USA). The
13.65 ± 1.4 years; range 12–17 years) with a mean BMI of value of low-density lipoprotein (LDL) cholesterol was
19.41 ± 2.21 kg/m2 were enrolled in the study from non- calculated using Friedwald’s equation.
obese healthy adolescents who attended the hospital for Serum insulin levels were measured by an immulite
minor illnesses such as common cold or conjunctivitis. immunoassay (Diagnostic Products, Los Angeles, CA,
The study protocols were approved by the institutional USA). The TSH level (reference range 0.45–4.0 mIU/l)
review board of the Konya Research Hospital Ethical was measured by means of a chemiluminescence immu-
Committee. Signed informed consent forms were obtained noassay (ADVIA Centaur XP, Immunoassay System; Sie-
from the parents of the adolescents. mens Healthcare Diagnostic Inc, Tarrytown, NY, USA).
Patients were excluded from the study if they had any Free thyroxine was determined using a chemiluminescence
systemic disease, including type 1 or 2 diabetes mellitus, immunoassay (reference range 0.6–1.8 ng/dl). Standard

123
1168 Pediatr Cardiol (2013) 34:1166–1174

liver function tests (ALT, aspartate aminotransferase Echocardiographic Examination

[AST]) were performed on the same day using an
autoanalyzer. All echocardiographic and Doppler assessments were per-
Insulin resistance was estimated using the homeostasis formed by the same expert pediatric cardiologist, who was
model assessment for insulin resistance (HOMA-IR) as blinded to the clinical and laboratory results of the study
follows: fasting insulin concentration (lU/mL) 9 fasting group. ProSound Alpha 7 echocardiography equipment
glucose concentration (mmol/l)/22.5 [20] (Aloka, Hitachi-Aloka Medical, Tokyo, Japan) with a
Insulin resistance was defined for adolescents as levels 3-MHz phased-array transducer was used for each study
of the HOMA-IR greater than 3.16 [13]. subject.
The patients were examined in the left lateral decubitus
Liver Ultrasonography position, and images were acquired at passive end expira-
tion to minimize global cardiac movement from standard
All the patients with abnormally high transaminases and an parasternal long-axis and apical planes. The M-mode
abnormal liver ultrasound were screened for other liver echocardiographic study of the left ventricle (LV) was
conditions (hepatitis B surface antigen, hepatitis C anti- performed under two-dimensional control. The ventricular
body, prothrombin time, iron, total iron-binding capacity, septal and posterior wall thicknesses at end diastole as well
ferritin, and antinuclear antibodies), which all showed as the dimensions of LV end-diastole (LVED) and end-
negative results. Liver ultrasound was performed by a systole were determined from M-mode echocardiography
trained operator blinded to all clinical and laboratory according to the American Society of Echocardiography
characteristics of the participants. Scans were performed recommendations. The calculations of LVM were per-
for all the subjects using a GE Healthcare Logic 7 machine formed using the following equation [7]:
(GE Healthcare Logic; Livonia, MI, USA) equipped with [Link]½ðLVED þ LV posterior wall thickness
7.5-MHz probes for younger adolescents and 5-MHz
þ interventricularseptal thicknessÞ3  LVED3 g
probes for larger or markedly obese adolescents.
þ 0:6 and indexed for height2:7 :
The presence of NAFLD was assessed via the scoring
system defined by Tominaga et al. [35] according to the Statistical Analysis
hyperechogenicity of liver tissue, the discrepancy between
the liver and the diaphragm, and the visibility of vascular Mean and standard deviations were used as descriptive
structures. The diagnosis of NAFLD usually is made from statistics. Obese and normal-weight groups were compared
mild elevations of liver enzymes during a routine blood test with respect to continuous variables using Student’s t-test.
and liver ultrasonography of an overweight or obese child. Bivariate associations of continuous variables were asses-
Although liver ultrasonography cannot estimate either sed with Pearson correlation coefficients. Normality
fibrosis or inflammation, it has a sensitivity of 89 % and a assumptions were assessed before parametric tests were
specificity of 93 % for detecting histologic steatosis [32]. used. Multivariable stepwise regression analysis was per-
Elevated ALT was defined as more than 40 IU/l in our formed to assess independent predictors of TSH levels. A
study [2]. In the current study, all the obese patients with p value lower than 0.05 was used to indicate statistical
NAFLD had high ALT levels. significance. Statistical analyses were performed with
SPSS for Windows, version 15 (SPSS, Chicago, IL, USA).
Carotid IMT Measurements

Carotid ultrasound studies were performed by a single

radiologist blinded to the clinical and laboratory status of Results
the patients using high-resolution B-mode ultrasonography
(GE Healthcare Logic 7) with a high-resolution linear array The characteristics of the study population are shown in
vascular transducer (14 MHz). An optimal two-dimen- Table 1. Gender, age, and BMI were similar in the two
sional image of the common carotid artery was obtained in groups. The control group (lean) included 42 sex-, age-,
which both the near and far wall intima/media complexes and pubertal stage-matched nonobese healthy subjects
were well visualized. without liver steatosis.
After the subject had rested 10 min and following The NAFLD obese group had significantly higher sys-
standard guidelines, the M-mode curser was placed 1 cm tolic and diastolic blood pressures than the non-NAFLD
proximal to the beginning of the carotid artery bulb during and lean groups. The NAFLD obese group also had sig-
end diastole. Carotid IMT was calculated by taking the nificantly higher levels of AST, ALT, total cholesterol,
mean value of three measurements. LDL cholesterol, triglycerides, fasting insulin, and TSH

123
Pediatr Cardiol (2013) 34:1166–1174 1169

Table 1 Characteristics of lean and obese adolescentsa

Lean Obese
Non-NAFLD NAFLD

N (Girl/boy ratio) 42 (17/25) 53 (32/21) 58 (25/33)

Age (years) 13.65 ± 1.4 (13) 13.25 ± 1.34 (13) 13.13 ± 1.26 (13)
BMI (kg/m2) 19.41 ± 2.21 (19.24) 29.33 ± 3.47 (28.53)a 30.36 ± 3.08 (30.35)b
a
Systolic blood pressure (mmHg) 102.9 ± 11.4 (100) 112.6 ± 11.6 (110) 119.3 ± 13.8 (120)b, c

a b, c
Diastolic blood pressure (mmHg) 65.2 ± 8.0 (70) 71.9 ± 9.8 (70) 76.9 ± 9.6 (80)
Total cholesterol (mg/dl) 138.2 ± 18.6 (137.5) 170.7 ± 23.8 (168)a 190.8 ± 29.1 (186)b, c

a b, c
Triglycerides (mg/dl) 75.2 ± 28.7 (70) 137.8 ± 65.9 (121) 180.8 ± 73.4(170.5)
LDL (mg/dl) 75.5 ± 17.3 (70) 100.4 ± 22.3 (101.4)a 117.2 ± 28.9 (113.4)b, c

HDL (mg/dl) 47.6 ± 8.2 (47) 43.3 ± 7.66 (43)a 39.3 ± 6.5 (38.5)b, c

a b
Fasting glucose (mg/dl) 76.7 ± 7.5 (75.5) 88.7 ± 9.6 (89) 91.7 ± 6.5 (92)
Fasting insulin (IU/ml) 3.9 ± 1.9 (4.0) 16.2 ± 10.3 (12.5)a 22.2 ± 9 (21)b, c

a
Free T4 (ng/dl) 1.25 ± 0.15 (1.27) 1.16 ± 0.13 (1.15) 1.20 ± 0.15 (1.18)
TSH (mIU/l) 1.77 ± 0.63 (1.64) 2.86 ± 1.13 (2.61)a 3.56 ± 1.34 (3.46)b, c

AST (IU/l) 16.7 ± 2.9 (16) 23.6 ± 5 (24)a 41.29 ± 3.3 (40.5)b, c

a b, c
ALT (IU/l) 14.2 ± 5.5 (13) 23.2 ± 8.6 (22) 45.3 ± 3.5 (45)
HOMA-IR 0.76 ± 0.29 (0.77) 3.59 ± 2.53 (2.85)a 5.14 ± 2.41 (4.66)b, c

a
AST/ALT ratio 1.31 ± 0.41 (1.27) 1.10 ± 0.31 (1) 0.911 ± 0.057 (0.912)b, c

a b, c
Carotid IMT (mm) 0.358 ± 0.011 (0.360) 0.380 ± 0.018 (0.380) 0.440 ± 0.025 (0.445)
LVM (g) 117.7 ± 35.2 (118.4) 156.4 ± 45.3 (147.8)a 177.9 ± 38.3 (176.5)b, c

2.7 a b
LVM index (g/m ) 33.3 ± 10.8 (34.7) 46.1 ± 10.9 (44.3) 49.5 ± 8.6 (48.9)
Data were expressed as the mean ± standard deviation with median given in parentheses
NAFLD nonalcoholic fatty liver disease; BMI body mass index; LDL low-density lipoprotein; HDL high-density lipoprotein; TSH thyroid-
stimulating hormone; AST aspartate transaminase; ALT alanine transaminase; HOMA-IR homeostasis model assessment for insulin resistance;
IMT intima-media thickness; LVM left ventricular mass
a
Lean versus obese without NAFLD (p \ 0.05)
b
Lean versus obese with NAFLD (p \ 0.05)
c
Obese without NAFLD versus obese with NAFLD (p \ 0.05)

than the non-NAFLD and lean groups, whereas the HDL The levels of AST, fasting glucose, and insulin also were
cholesterol and AST/ALT ratio values were lower in the significantly higher in the NAFLD obese group with a TSH
NAFLD obese group than in the non-NAFLD and lean level higher than 4 mIU/l than the NAFLD obese group with
groups. The non-NAFLD obese group had significantly a TSH level lower than 4 mIU/l, whereas the HDL choles-
higher BMI and systolic and diastolic blood pressures than terol levels were lower in the NAFLD obese group with a
the control group. The levels of AST, ALT, total choles- TSH level exceeding 4 mIU/l than in the NAFLD obese
terol, triglyceride, LDL cholesterol, fasting glucose, insulin, group with a TSH level lower than 4 mIU/l. The NAFLD
and TSH also were significantly higher in the non-NAFLD obese group with TSH level higher than 4 mIU/l had sig-
obese group than in the control group, whereas the HDL nificantly higher HOMA-IR values (6.24 ± 2.89 vs. 3.59 ±
cholesterol levels were lower in the non-NAFLD obese 2.53), carotid IMT (0.454 ± 0.026 vs 0.430 ± 0.018 mm),
group than in the control group. LVM (195.4 ± 38.4 vs 164.6 ± 32.9 g), and LVM index
The lean group had lower HOMA-IR values than the non- (52.5 ± 7.2 vs 47.2 ± 9.0 g/m2.7) than the NAFLD obese
NAFLD and NAFLD obese groups (respectively, 0.76 ± 0.29 group with a TSH level lower than 4 mIU/l (Table 2).
vs 3.59 ± 2.53 vs 5.14 ± 2.41). Moreover, the NAFLD obese In the NAFLD obese group with subclinical hypothyroid-
group compared with the non-NAFLD and lean groups had ism, high TSH values were positively correlated with most of
significantly higher TSH values (3.56 ± 1.34 vs 2.86 ± 1.13 the metabolic parameters such as total cholesterol (r = 0.606,
vs 1.77 ± 0.63 mIU/l) and carotid IMT (0.440 ± 0.025 vs p = 0.001), triglycerides (r = 0.476, p = 0.016), LDL cho-
0.380 ± 0.018 vs 0.358 ± 0.011 mm), as well as a higher LVM lesterol (r = 0.461, p = 0.004), insulin (r = 0.607,
(177.9 ± 38.3 vs 156.4 ± 45.3 vs 117.7 ± 35.2 g) (Table 1). p = 0.001), AST (r = 0.467, p = 0.019), HOMA-IR

123
1170 Pediatr Cardiol (2013) 34:1166–1174

Table 2 Characteristics obese adolescents with nonalcoholic fatty liver disease (NAFLD) according to serum thyroid-stimulating hormone
(TSH) levels
Obese with NAFLD
TSH (\4 mIU/l) TSH ([4 mIU/l) p Value

N 33 25
Age (years) 12.97 ± 1.15 (12.5) 13.35 ± 1.38 (13) 0.259
BMI (kg/m2) 30.26 ± 3.02 (30.26) 30.49 ± 3.22 (30.40) 0.777
Systolic blood pressure (mmHg) 118.9 ± 12.2 (120) 119.8 ± 15.97 (120) 0.817
Diastolic blood pressure (mmHg) 76.2 ± 8.2 (80) 77.8 ± 11.3 (80) 0.541
Total cholesterol (mg/dl) 186.7 ± 16 (184) 191.4 ± 29.3 (186) 0.222
Triglycerides (mg/dl) 177.4 ± 85.3 (164) 176.3 ± 73.1(172) 0.684
LDL (mg/dl) 113.7 ± 18.3 (113) 118.2 ± 29.1 (117.6) 0.293
HDL (mg/dl) 40.9 ± 6.3 (40) 39.9 ± 7.1 (35) 0.031
Fasting glucose (mg/dl) 89.8 ± 5.9 (91) 94.2 ± 6.5 (99) 0.010
Fasting insulin (IU/ml) 19.1 ± 6 (17.1) 26.3 ± 10.6 (25) 0.002
Free T4 (ng/dl) 1.21 ± 0.16 (1.26) 1.17 ± 0.14 (1.11) 0.344
TSH (mIU/l) 2.56 ± 0.72 (2.85) 4.89 ± 0.59 (4.94) \0.0001
AST (IU/l) 39.7 ± 3.1 (39) 43.2 ± 2.5 (44) \0.0001
ALT (IU/l) 44.8 ± 4 (43.8) 45.9 ± 2.8 (46) 0.267
HOMA-IR 3.59 ± 2.53 (2.85) 6.24 ± 2.89 (6.11) 0.002
AST/ALT ratio 0.888 ± 0.05 (0.902) 0.940 ± 0.037 (0.955) \0.0001
Carotid IMT (mm) 0.430 ± 0.018 (0.430) 0.454 ± 0.026 (0.455) \0.0001
LVM (g) 164.6 ± 32.9 (158.2) 195.4 ± 38.4 (200.4) 0.002
LVM index (g/m2.7) 47.2 ± 9.0 (46.4) 52.5 ± 7.2 (51.1) 0.019
Data are expressed as the mean ± standard deviation, with the median given in parentheses
BMI body mass index; LDL low-density lipoprotein; HDL high-density lipoprotein; AST aspartate transaminase; ALT alanine transaminase;
HOMA-IR homeostasis model assessment for insulin resistance; IMT intima-media thickness; LVM left ventricular mass

(r = 0.596, p = 0.002), carotid IMT (r = 0.894, p \ 0.0001), NAFLD [5, 16]. Liangpunsakul and Chalasani [16] repor-
and LVM (r = 0.563, p = 0.003) (Table 3; (Figs. 1, 2, 3). ted the prevalence of hypothyroidism in 174 patients with
In the multivariate stepwise regression analysis, the TSH nonalcoholic steatohepatitis (NASH) to be double that seen
level was positively correlated with increased carotid IMT in 442 matched control subjects. In a large population
(b = 0.243, p = 0.001) in the NAFLD obese group with TSH study involving 10,292 outpatient adults with a wide range
exceeding 4 mIU/l even after adjustment for age, sex, BMI, of age and thyroid function tests (TSH and fT4), serum
systolic and diastolic blood pressures, total cholesterol, tri- ALT and gamma-glutamyl transferase concentrations
glycerides, LDL cholesterol, HDL cholesterol, fasting glucose increased steadily across the increasing TSH categories.
and insulin, AST/ALT ratio, and HOMA-IR as cofactors, with Likewise, a negative and graded relationship was found
the total variance explained as 79.8 % (Table 4, 5). between serum ALT and gamma-glutamyl transferase
concentrations and fT4 categories [33].
In a recent study, Carulli et al. [5] found that euthyroid
Discussion patients with biopsy-proven NASH have high (although
still in the upper normal range) TSH values compared with
This study demonstrated that a significant relationship those who have simple steatosis. Although the association
exists between subclinical hypothyroidism and cardiac between thyroid hormonal derangements and fatty liver is
geometry in obese adolescents with NAFLD and obesity- established, which one causes the other still is unclear [25].
related cardiovascular risk factors. We suggest that the In our multivariate regression analysis, all the metabolic
presence of subclinical hypothyroidism may have an variables showed TSH levels positively correlated with
additional increasing impact on LVM and carotid IMT in carotid IMT and LVM in the NAFLD obese group with a
these patients with NAFLD. TSH level higher than 4 mIU/l.
Some studies support an association between overt or Thyroid hormones may have a negative effect on the
subclinical thyroid disease and the development of primary major metabolic pathways, and thyroid dysfunction may

123
Pediatr Cardiol (2013) 34:1166–1174 1171

Table 3 Relationship between carotid intima-media thickness (IMT)

and cardiovascular risk factors in obese adolescents with nonalcoholic
fatty liver disease (NAFLD)
Obese with NAFLD
r p Value

Age (years) 0.300 0.022

BMI (kg/m2) 0.319 0.018
Systolic blood pressure (mmHg) 0.445 \0.0001
Diastolic blood pressure (mmHg) 0.505 \0.0001
Total cholesterol (mg/dl) 0.583 \0.0001
Triglycerides (mg/dl) 0.120 0.371
LDL (mg/dl) 0.551 \0.0001
HDL (mg/dl) -0.328 0.012
Fasting glucose (mg/dl) 0.546 \0.0001
Fasting insulin (IU/ml) 0.724 \0.0001
Free T4 -0.256 0.052
TSH 0.598 \0.0001
AST (IU/l) 0.551 \0.0001
ALT (IU/l) 0.220 0.01 Fig. 2 Relationship between thyroid-stimulating hormone (TSH)
HOMA-IR 0.722 \0.0001 level and left ventricular mass in obese patients with nonalcoholic
AST/ALT ratio 0.495 \0.0001 fatty liver disease (NAFLD) whose TSH levels are higher than
4 mIU/l
LVM (g) 0.686 \0.0001
LVM index (g/m2.7) 0.469 \0.0001
BMI body mass index; LDL low-density lipoprotein; HDL high-den-
sity lipoprotein; TSH thyroid-stimulating hormone; AST aspartate
transaminase; ALT alanine transaminase; HOMA-IR homeostasis
model assessment for insulin resistance; LVM left ventricular mass

Fig. 3 Relationship between thyroid-stimulating hormone (TSH)

level and left ventricular mass index in obese patients with
nonalcoholic fatty liver disease (NAFLD) whose TSH levels are
higher than 4 mIU/l

play a role in the pathogenesis of NAFLD. Patients with

hypothyroidism not only have evidence of hyperlipidemia,
Fig. 1 Relationship between thyroid-stimulating hormone (TSH)
level and carotid intima media thickness in obese patients with
but also exhibit decreased fatty acid oxidation and hepatic
nonalcoholic fatty liver disease (NAFLD) whose TSH levels are output of triglycerides. Otherwise, hypothyroidism is
higher than 4 mIU/l associated with insulin resistance, which seems to be a

123
1172 Pediatr Cardiol (2013) 34:1166–1174

Table 4 Relationship between left ventricular mass (LVM) and (TSH [ 4.65 mIU/L and normal T4) and healthy control
cardiovascular risk factors in obese adolescents with nonalcoholic subjects. The mean HDL cholesterol was significantly
fatty liver disease (NAFLD)
lower in the children with subclinical hypothyroidism.
Obese with NAFLD In our study, the NAFLD obese group had significantly
r p Value higher levels of total cholesterol, LDL cholesterol, and
triglycerides than the non-NAFLD and lean groups. Fur-
Age (years) 0.319 0.015 thermore, the HDL cholesterol levels were lower in the
BMI (kg/m2) 0.326 0.013 NAFLD obese group with TSH levels higher than 4 mIU/l
Systolic blood pressure (mmHg) 0.457 \0.0001 than in the NAFLD obese group with TSH levels lower
Diastolic blood pressure (mmHg) 0.486 \0.0001 than 4 mIU/l. The findings showed TSH values positively
Total cholesterol (mg/dl) 0.454 \0.0001 correlated with most of the metabolic parameters such as
Triglycerides (mg/dl) 0.075 0.578 total cholesterol, triglycerides, LDL cholesterol in the
LDL (mg/dl) 0.441 0.001 NAFLD obese group with TSH levels higher than 4 mIU/l.
HDL (mg/dl) -0.261 0.047 An increased risk of hypertension also has been reported
Fasting glucose (mg/dl) 0.445 \0.0001 in some studies of patients with subclinical hypothyroid-
Fasting insulin (IU/ml) 0.709 \0.0001 ism. Because it is in overt disease, increased peripheral
Free T4 -0.241 0.069 vascular resistance, increased arterial stiffness, and endo-
TSH 0.476 \0.0001 thelial dysfunction can contribute to systemic hypertension
AST (IU/l) 0.387 0.003 in subclinical hypothyroidism [25]. In our study, the
ALT (IU/l) 0.124 0.353 NAFLD obese group had significantly higher systolic and
HOMA-IR 0.695 \0.0001 diastolic blood pressures than those in the non-NAFLD and
AST/ALT ratio 0.377 0.004 lean groups.
LVM index (g/m2.7) 0.719 \0.0001 Thyroid hormones are important determinants of glu-
cose homeostasis. Increases in plasma thyroid hormone
BMI body mass index; LDL low-density lipoprotein; HDL high-den-
sity lipoprotein; TSH thyroid-stimulating hormone; AST aspartate levels impair the ability of insulin to suppress hepatic
transaminase; ALT alanine transaminase; HOMA-IR homeostasis glucose production and to increase glucose uptake in
model assessment for insulin resistance muscle [25]. Maratou et al. [19] showed that patients with
subclinical hypothyroidism had insulin resistance.
Studies of children are limited. In euthyroid children
Table 5 Multiple regression analysis of determinants of carotid without a history of hypo- or hyperthyroidism, increasing
intima-media thickness (IMT) and left ventricular mass (LVM) in the levels of TSH and decreasing levels of free T4 are reported to
obese adolescents with nonalcoholic fatty liver disease (NAFLD) and
be associated with elevated markers of insulin resistance
subclinical hypothyroidism
[25]. A recent study of obese children showed that during
R2 R2 Change b 95 % CI p Value weight loss, independent of changes in body weight or body
Carotid IMT versus variable fat and lean tissue decreases in elevated serum TSH predict
TSH level 0.791 0.800 0.894 0.4–10.3 0.001 decreases in fasting insulin and insulin resistance [1].
LVM versus variable In our study, the NAFLD obese group had significantly
BMI 0.566 0.113 0.368 0.58–8.1 0.026
higher HOMA-IR values than the non-NAFLD and lean
HOMA-IR 0.523 2.6–11.1 0.003
groups. Moreover, the HOMA-IR values were significantly
higher in the NAFLD obese group with TSH levels higher
CI confidence interval; TSH thyroid-stimulating hormone; BMI body than 4 mIU/l than in the NAFLD obese group with TSH
mass index; HOMA-IR homeostasis model assessment for insulin
resistance levels lower than 4 mIU/l. The TSH values were positively
correlated with HOMA-IR in the NAFLD obese group with
TSH levels higher than 4 mIU/l. These findings could
justify the increased risk for insulin resistance–associated
hallmark of NASH. Finally, hypothyroidism is associated disorders such as the cardiovascular disease observed in
with lipid peroxidation, one of the leading candidates for NAFLD patients with subclinical hypothyroidism.
cellular injury in patients with NASH [25]. In the current study, we found marked abnormalities in
In obese children, a relationship between TSH levels, LV structure including increased LVM and LVM index in
triglycerides, total cholesterol, and LDL cholesterol have the NAFLD obese adolescents with TSH levels higher than
been reported in some but not all studies [25]. Paoli-Valeri 4 mIU/l. Monzani et al. [21] noted significantly higher
et al. [26] compared the lipid profile between 2- to 9-year- values for LVM in 20 patients with subclinical hypothy-
old Spanish children with subclinical hypothyroidism roidism than in control subjects, similar to the findings of

123
Pediatr Cardiol (2013) 34:1166–1174 1173

our study. However, other studies have reported no sig- References

nificant associations between subclinical hypothyroidism
and LVM [3, 27, 39]. Rodondi et al. [29] observed an 1. Aeberli I, Jung A, Murer SB, Wildhaber J, Wildhaber-Brooks J,
Knöpfli BH, Zimmermann MB (2010) During rapid weight loss
increase in LVM over 5 years in the subgroup of subclin-
in obese children, reductions in TSH predict improvements in
ical hypothyroidism with TSH levels of 10.0 mU/l or insulin sensitivity independent of changes in body weight or fat.
higher. However, an epidemiologic study performed by J Clin Endocrinol Metab 95:5412–5418
Iqbal et al. [12] that included more than 2,000 subjects 2. Alisi A, Manco M, Vania A, Nobili V (2009) Pediatric nonal-
coholic fatty liver disease in 2009. J Pediatr 155:469–474
could not find any significant association between serum
3. Biondi B, Fazio S, Palmieri A, Carella C, Panza N, Cittadini A,
TSH and LVM index after adjustment for age, BMI, and Bonè F, Lombardi G, Saccà L (1999) Left ventricular diastolic
systolic blood pressure, which is in accordance with the dysfunction in patients with subclinical hypothyroidism. J Clin
report by Dörr et al. [8]. In their study of 1,510 individuals, Endocrinol Metab 84:2064–2067
4. Bluemke DA, Kronmal RA, Lima JA, Liu K, Olson J, Burke GL,
the adjusted LVM index was almost identical in subjects
Folsom AR (2008) The relationship of left ventricular mass and
with elevated, normal, and decreased serum TSH levels, geometry to incident cardiovascular events: the MESA (multi-ethnic
whereas those with overt hyperthyroidism had an increased study of atherosclerosis) study. J Am Coll Cardiol 52:2148–2155
LVM index. 5. Carulli L, Ballestri S, Lonardo A, Lami F, Violi E, Losi L, Bonilauri L,
Verrone AM, Odoardi MR, Scaglioni F, Bertolotti M, Loria P (2011)
Carotid IMT is a reliable index of subclinical athero-
Is nonalcoholic steatohepatitis associated with a high-though-normal
sclerosis, and epidemiologic studies have demonstrated a thyroid stimulating hormone level and lower cholesterol levels? Intern
significant association between carotid IMT and cardiovas- Emerg Med. Epub ahead of print 11 May 2011
cular disease [30]. No study has assessed whether subclinical 6. Dean JW, Fowler PB (1985) Exaggerated responsiveness to
thyrotropin-releasing hormone: a risk factor in women with
hypothyroidism is associated with carotid atherosclerosis in
coronary artery disease. Br Med J 290:1555–1561
children or adolescents with subclinical hypothyroidism, but 7. Devereux RB, Alonso DR, Lutas EM, Gottlieb GJ, Campo E,
this association is shown in adult studies. Valentina et al. [36] Sachs I, Reichek N (1986) Echocardiographic assessment of left
showed that subclinical hypothyroidism is associated with an ventricular hypertrophy: comparison to necropsy findings. Am J
Cardiol 57:450–458
increase in carotid IMT, independently of classical risk
8. Dörr M, Wolff B, Robinson DM, John U, Lüdemann J, Meng W,
factors for atherosclerosis. Felix SB, Völzke H (2005) The association of thyroid function
Kim et al. [14] studied to determine whether subclinical with cardiac mass and left ventricular hypertrophy. J Clin
hypothyroidism is associated with an increase in the IMT Endocrinol Metab 90:673–677
9. Eckel RH, Grundy SM, Zimmet PZ (2005) The metabolic syn-
of the common carotid artery and whether thyroid hormone
drome. Lancet 365:1415–1428
replacement can reverse this change in the carotid IMT. 10. Hak AE, Pols HA, Visser T, Drexhage HA, Hofman A, Witteman
Recently, in a double-blind placebo-controlled study, JC (2000) Subclinical hypothyroidism is an independent risk
Monzani et al. [22] demonstrated that early carotid IMT factor for atherosclerosis and myocardial infarction in elderly
women: the Rotterdam study. Ann Intern Med 132:270–278
alterations also were present in patients with subclinical
11. Hollowell JG, Staehling NW, Flanders WD, Hannon WH, Gunter
hypothyroidism, mainly appearing during the fourth decade EW, Spencer CA, Braverman LE (2002) Serum TSH, T(4), and
of life and onward. thyroid antibodies in the United States population (1988–1994):
Furthermore, several factors may contribute to cardio- National health and nutrition examination survey (NHANES III).
J Clin Endocrinol Metab 87:489–499
vascular risk in subclinical hypothyroidism including BMI,
12. Iqbal A, Schirmer H, Lunde P, Figenschau Y, Rasmussen K,
fat distribution, lipid profile, and vascular dysfunction [9]. Jorde R (2007) Thyroid-stimulating hormone and left ventricular
A population-based survey concluded that subclinical function. J Clin Endocrinol Metab 92:3504–3510
hypothyroidism was associated with aortic atherosclerosis 13. Keskin M, Kurtoglu S, Kendirci M, Atabek ME, Yazici C (2005)
Homeostasis model assessment is more reliable than the fasting
and myocardial infarction in adults, independently of
glucose/insulin ratio and quantitative insulin sensitivity check
serum cholesterol levels [10]. index for assessing insulin resistance among obese children and
adolescents. Pediatrics 115:500–503
14. Kim SK, Kim SH, Park KS, Park SW, Cho YW (2009) Regres-
sion of the increased common carotid artery-intima media
Conclusion
thickness in subclinical hypothyroidism after thyroid hormone
replacement. Endocr J 56:753–758
The current study confirms that obese adolescents who 15. Li X, Li S, Ulusoy E, Chen W, Srinivasan SR, Berenson GS
have NAFLD with subclinical hypothyroidism demonstrate (2004) Childhood adiposity as a predictor of cardiac mass in
adulthood: the Bogalusa heart study. Circulation 110:3488–3492
a more adverse cardiovascular risk profile and higher
16. Liangpunsakul S, Chalasani N (2003) Is hypothyroidism a risk
LVM. Increased LVM and carotid IMT are significant factor for nonalcoholic steatohepatitis? J Clin Gastroenterol
predictors of cardiovascular-related death. Addition of 37:340–343
carotid IMT measurements to the echocardiographic 17. Litwin M, Niemirska A (2009) Intima-media thickness mea-
surements in children with cardiovascular risk factors. Pediatr
examination may assist in risk stratification of NAFLD
Nephrol 24:707–719
obese adolescents with subclinical hypothyroidism.

123
1174 Pediatr Cardiol (2013) 34:1166–1174

18. Manco M, Bottazzo G, DeVito R, Marcellini M, Mingrone G, 29. Rodondi N, Bauer DC, Cappola AR, Cornuz J, Robbins J, Fried
Nobili V (2008) Nonalcoholic fatty liver disease in children. LP, Ladenson PW, Vittinghoff E, Gottdiener JS, Newman AB
J Am Coll Nutr 27:667–676 (2008) Subclinical thyroid dysfunction, cardiac function, and the
19. Maratou E, Hadjidakis DJ, Kollias A, Tsegka K, Peppa M, risk of heart failure. The cardiovascular health study. J Am Coll
Alevizaki M, Mitrou P, Lambadiari V, Boutati E, Nikzas D, Cardiol 52:1152–1159
Tountas N, Economopoulos T, Raptis SA, Dimitriadis G (2009) 30. Simon A, Megnien JL, Chironi G (2010) The value of carotid
Studies of insulin resistance in patients with clinical and sub- intima-media thickness for predicting cardiovascular risk. Arte-
clinical hypothyroidism. Eur J Endocrinol 160:785–790 rioscler Thromb Vasc Biol 30:182–185
20. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher 31. Squizzato A, Gerdes VEA, Brandjes DPM, Büller HR, Stam J
DF, Turner RC (1985) Homeostasis model assessment: insulin (2005) Thyroid diseases and cerebrovascular disease. Stroke 36:
resistance and beta-cell function from fasting plasma glucose and 2302–2310
insulin concentrations in men. Diabetologia 28:412–429 32. Stephen CH, Tri HL, Stacey MA (2002) Nonalcoholic steato-
21. Monzani F, Di Bello V, Caraccio N, Bertini A, Giorgi D, Giusti hepatitis. In: Sciff’s disease of the liver, 9th edn. Lippincott
C, Ferrannini E (2001) Effect of levothyroxine on cardiac func- William & Wilkins, Philadelphia, p 1261–1289
tion and structure in subclinical hypothyroidism: a double-blind, 33. Targher G, Montagnana M, Salvagno G, Moghetti P, Zoppini G,
placebo-controlled study. J Clin Endocrinol Metab 86:1110–1115 Muggeo M, Lippi G (2008) Association between serum TSH, free
22. Monzani F, Caraccio N, Kozàkowà M, Dardano A, Vittone F, T4, and serum liver enzyme activities in a large cohort of unse-
Virdis A, Taddei S, Palombo C, Ferrannini E (2004) Effect of lected outpatients. Clin Endocrinol 68:481–484
levothyroxine replacement on lipid profile and intima-media 34. Targher G, Day CP, Bonora E (2010) Risk of cardiovascular
thickness in subclinical hypothyroidism: a double-blind, placebo- disease in patients with nonalcoholic fatty liver disease. N Engl J
controlled study. J Clin Endocrinol Metab 89:2099–2106 Med 363:1341–1350
23. Ozturk A, Mazicioglu MM, Hatipoglu N, Budak N, Keskin G, 35. Tominaga K, Kurata JH, Chen YK, Fujimoto E, Miyagawa S,
Yazlak Z, Balci N, Yildiz H, Yildiz K, Ustunbas HB, Kurtoglu S Abe I, Kusano Y (1995) Prevalence of fatty liver in Japanese
(2008) Reference body mass index curves for Turkish children 6 children and relationship to obesity: an epidemiological ultraso-
to 18 years of age. J Pediatr Endocrinol Metab 21:827–836 nographic survey. Dig Dis Sci 40:2002–2009
24. Pacifico L, Nobili V, Anania C, Verdecchia P, Chiesa C (2011) 36. Valentina VN, Marijan B, Chedo D, Branka K (2011) Subclinical
Pediatric nonalcoholic fatty liver disease, metabolic syndrome hypothyroidism and risk to carotid atherosclerosis. Arq Bras
and cardiovascular risk. World J Gastroenterol 17:3082–3091 Endocrinol Metabol 55:475–480
25. Pacifico L, Anania C, Ferraro F, Andreoli GM, Chiesa C (2012) 37. Vanderpump MP, Tunbridge WM, French JM, Appleton D, Bates
Thyroid function in childhood obesity and metabolic comorbid- D, Clark F, Grimley EJ, Rodgers H, Tunbridge F, Young ET
ity. Clin Chim Acta 413:396–405 (1996) The development of ischemic heart disease in relation to
26. Paoli-Valeri M, Guzmán M, Jiménez-López V, Arias-Ferreira A, autoimmune thyroid disease in a 20-year follow-up study of an
Briceño-Fernández M, Arata-Bellabarba G (2005) Atherogenic English community. Thyroid 6:155–160
lipid profile in children with subclinical hypothyroidism. An 38. Willeit J, Kiechl S, Oberhollenrer F, Rungger G, Egger G, Bo-
Pediatr 62:128–134 nora E, Mitterer M, Muggeo M (2000) Distinct risk profiles of
27. Pearce EN, Yang Q, Benjamin EJ, Aragam J, Vasan RS (2010) early and advanced atherosclerosis. Arterioscler Tromb Vasc Biol
Thyroid function and left ventricular structure and function in the 20:529–537
Framingham heart study. Thyroid 20:369–373 39. Yazici M, Gorgulu S, Sertbas Y, Erbilen E, Albayrak S, Yildiz O,
28. Rodondi N, Newman AB, Vittinghoff E, de Rekeneire N, Sat- Uyan C (2004) Effects of thyroxin therapy on cardiac function in
terfield S, Harris TB, Bauer DC (2005) Subclinical hypothy- patients with subclinical hypothyroidism: index of myocardial
roidism and the risk of heart failure, other cardiovascular events, performance in the evaluation of left ventricular function. Int J
and death. Arch Intern Med 165:2460–2466 Cardiol 95:135–143

123