See corresponding editorial on page 3.
See corresponding CME exam on page 183.
Clustering of metabolic abnormalities in adolescents with the
hypertriglyceridemic waist phenotype1–3
Ahmad Esmaillzadeh, Parvin Mirmiran, and Fereidoun Azizi
ABSTRACT
Background: It remains unknown whether the hypertriglyceride-
mic waist (HW) phenotype, an appropriate screening tool in adults,
can also be used to screen for metabolic abnormalities in adolescents.
Objective: We aimed to evaluate metabolic risk factors identified by
the HW phenotype in adolescents.
Design: Anthropometric and biochemical measurements were as-
sessed in a population-based cross-sectional study of 1413 male and
1623 female Iranian adolescents aged 10–19 y. The HW phenotype
was defined as serum triacylglycerol concentrations �110 mg/dL
and concurrent waist circumference �90th percentile for age and
sex. Elevated fasting glucose (�110 mg/dL), high LDL (�130 mg/
dL) and low HDL (�40 mg/dL) cholesterol, hypercholesterolemia
(total cholesterol �200 mg/dL), and hypertension (systolic or dia-
stolic blood pressure� 95th percentile for age, sex, and height) were
considered as risk factors.
Results: Adolescents with the HW phenotype had significantly
higher prevalences of all metabolic risk factors except elevated fast-
ing glucose than did those without the HW phenotype. After control
for potential confounding variables, adolescents with the HW phe-
notype were significantly more likely to have high LDL cholesterol
(odds ratio: 1.8; 95% CI: 1.3, 2.7), low HDL cholesterol (1.6; 1.3,
2.0), hypercholesterolemia (2.9; 2.0, 4.2), and�1 (1.4; 1.1, 1.7) and
�2 (2.2; 1.6, 3.0) risk factors than were those without the HW
phenotype. The HW phenotype had a significantly higher percentage
of correct prediction of metabolic abnormalities than did over-
weight, elevated triacylglycerol concentration, or enlarged waist
circumference.
Conclusion: This study shows a clustering of metabolic abnormal-
ities in adolescents with the HW phenotype and suggests this phe-
notype as a simple marker for identifying adolescents at risk of
metabolic syndrome and other metabolic abnormalities. Am J
Clin Nutr 2006;83:36–46.
KEY WORDS Hypertriglyceridemic waist, metabolic syn-
drome, adolescents, cardiovascular risk factors, overweight
INTRODUCTION
Cardiovascular diseases (CVDs) are one of the major causes of
mortality in Iran (1), and the prevalence of these disturbances
continues to rise (2). Persons with metabolic syndrome are at
greater risk of CVD. The metabolic syndrome is defined as a
constellation of laboratory and physical findings such as central
obesity, insulin resistance, hyperglycemia, dyslipidemia, and hy-
pertension (3). The third report of the National Cholesterol Ed-
ucation Program Adult Treatment Panel III recognized the met-
abolic syndrome as a secondary target of risk-reduction therapy
(4). A recent study in Tehran showed that the metabolic syn-
drome is highly prevalent in Tehranian adults; the estimated
prevalence is �30% (5), which is higher than that in most de-
veloped countries, such as the United States (6).
Although studies indicated that the process of atherosclerosis
starts at an early age and is linked even in childhood to obesity
and other components of the metabolic syndrome (7), few studies
have reported the prevalence of the metabolic syndrome pheno-
type in children and adolescents (8–11). Moreover, at present,
there is no globally accepted definition for the metabolic syn-
drome in children and adolescents. Research is needed to ascer-
tain whether the precursors of the metabolic syndrome in chil-
dren and adolescents differ from those in adults, but many
investigators assume that insulin resistance is the fundamental
metabolic defect underlying the metabolic syndrome (3, 4, 12,
13). The concurrent presence of insulin resistance, increased
serum apolipoprotein B concentrations, and high serum concen-
trations of small, dense LDL cholesterol has been referred to as
a metabolic triad (14) that could be identified by using the inex-
pensive screening tool of the hypertriglyceridemic waist (HW)
phenotype (15).
Some investigators have reported that HW, or waist-
triacylglycerol syndrome, predicts the presence of the metabolic
syndrome (16). Others recommended this index for the identifi-
cation of a syndrome of lipid overaccumulation (17). Subjects
with the HW phenotype were nearly 4 times as likely to have
angiographically defined coronary artery disease (CAD) as were
subjects who did not have the HW phenotype (15). Our previous
studies showed that the HW phenotype is highly prevalent in the
adult population of Tehran; the estimated prevalence is 19% in
1 From the Endocrine Research Center, Shaheed Beheshti University of
Medical Sciences, Tehran, Iran.
2 Supported by grant no. 121 from the National Research Council of the
Islamic Republic of Iran and by the combined support of the National Re-
search Council of Islamic Republic of Iran and the Endocrine Research
Center of Shaheed Beheshti University of Medical Sciences.
3 Reprints not available. Address correspondence to F Azizi, Endocrine
Research Center, Shaheed Beheshti University of Medical Sciences, PO Box
19395-4763, Tehran, Iran. E-mail: azizi@erc.ac.ir.
Received March 14, 2005.
Accepted for publication August 25, 2005.
36 Am J Clin Nutr 2006;83:36–46. Printed in USA. © 2006 American Society for Nutrition
by guest on April 5, 2011
w
w
w
.ajcn.org
D
ow
nloaded from
men (18) and 32% in women (19). Although the prevalence
(15–20) and correlates (21–23) of the HW phenotype have been
investigated in adults, comparatively little emphasis was placed
on its prevalence and correlates in children and adolescents, and
there is no report in the current literature regarding the prevalence
of this phenotype in adolescents. It also remains unknown
whether, as seen in adults, the HW phenotype is associated with
a similar clustering of metabolic risk factors in adolescents. We
found in our previous investigation (24) that the HW phenotype
is highly prevalent in Tehranian adolescents; ie, the estimated
prevalence is 6.5%. In the current study, we aimed to evaluate
whether the HW phenotype could be used as a screening tool for
metabolic abnormalities in adolescents.
SUBJECTS AND METHODS
Subjects
This study was conducted within the framework of the Tehran
Lipid and Glucose Study (TLGS), a prospective study performed
in a representative sample of residents of one district of Tehran
with the aims of ascertaining the prevalence of noncommunica-
ble disease risk factors and developing a healthy lifestyle to
curtail these risk factors (25). The TLGS provides a representa-
tive sample of the population of the city of Tehran (25):�15 000
people aged�3 y who were living in a district of Tehran and were
selected by multistage cluster random-sampling method, includ-
ing 3265 adolescents aged 10–19 y. In the current population-
based cross-sectional study, after the exclusion of subjects taking
medications that affect serum lipids, blood pressure, and carbo-
hydrate metabolism (n� 23), 3036 adolescents (1413 male and
1623 female) with full relevant data (206 adolescents had incom-
plete relevant data) were included in the study.
Written informed consent was obtained from each subject.
This study was approved by the research council of the Endocrine
Research Center of the Shaheed Beheshti University of Medical
Sciences.
Methods
Details of the TLGS protocol and all laboratory procedures
were published elsewhere (2, 26, 27). In brief, weight was mea-
sured with the use of digital scales while subjects were minimally
clothed and not wearing shoes, and it was recorded to the nearest
100 g. Height was measured with the use of a tape measure while
subjects were in a standing position and not wearing shoes and
while the shoulders were in a normal position. Body mass index
(BMI) was calculated as weight (in kg) divided by height (in m2).
Waist circumference (WC) was measured at the narrowest level,
and hip circumference was measured at the maximum level over
light clothing with the use of an unstretched tape measure and
without any pressure to body surface; measurements were re-
corded to the nearest 0.1 cm. Because the measurements were to
be made over light clothing, participants were asked to remove
belts and any tight or loose garments that may alter the shape of
the body, and the person measuring was asked to ensure that the
tape had the proper tension around the subject’s body—ie, nei-
ther too loose nor too tight. Although the narrowest waist is easy
to identify in most subjects, for some subjects there is no single
narrowest waist because of either a large amount of abdominal fat
or extreme thinness (28). In the current study, when the narrowest
point of waist was difficult to identify (particularly in obese
subjects), we measured the WC immediately below the end of the
lowest rib, because in most subjects the narrowest waist is at the
lowest rib (28). To reduce subjective error, all measurements in
all males were taken by the same male technician, and those in all
females were taken by the same female technician. There were
significant correlations between both the test and retest measure-
ments taken by male and female technicians (r� 0.75, P� 0.01
for both). Data on family history of diabetes were collected as the
subjects’ oral responses to the previously tested questionnaire.
The criterion for family history of diabetes was having�1 first-
degree relative with a diagnosis of diabetes after 30 y of age. Data
on physical activity, which were reported previously (29), were
obtained by using subjects’ oral responses to a previously tested
questionnaire, and subjects were categorized into 3 physical ac-
tivity groups—light, moderate, and heavy.
Between 0700 and 0900, a blood sample was drawn into evac-
uated tubes from all study participants after overnight (ie,�10 h)
fasting. Blood samples were taken while the subjects were in a
sitting position, according to the standard protocol, and were
centrifuged within 30–45 min. All blood lipid analyses were
done at the TLGS research laboratory on the day of blood col-
lection. The analysis of samples was performed by using a Se-
lectra 2 autoanalyzer (Vital Scientific, Spankeren, Netherlands).
Fasting plasma glucose was measured on the day of blood col-
lection by using the enzymatic colorimetric method with glucose
oxidase. Serum total cholesterol (TC) and triacylglycerol con-
centrations were measured by commercially available enzymatic
reagents (Pars Azmoon Inc, Tehran, Iran) adapted to the Selectra
autoanalyzer. HDL cholesterol was measured after precipitation
of the apolipoprotein B–containing lipoproteins with phospho-
tungistic acid. LDL cholesterol was calculated from serum TC,
triacylglycerol, and HDL cholesterol concentrations, except
when the triacylglycerol concentration was� 400 mg/dL. Assay
performance was monitored once every 20 tests by using the lipid
control serum Percinorm (normal range) and by using Percipath
(pathologic range) whenever applicable [Cat. no. 1446070
(Percinorm) and 171778 (Percipath); Boehringer Mannheim,
Mannheim, Germany]. Lipid standard (calibrated for automated
systems) (Cat. No. 759350; Boehringer Mannheim) was used to
calibrate the Selectra 2 autoanalyzer on each day of laboratory
analyses. All samples were analyzed when internal quality con-
trol met the acceptable criteria. Interassay and intraassay CVs
were 2% and 0.5% for TC and 1.6% and 0.6% for TAG, respec-
tively (26).
Participants were made to rest for 15 min before their blood
pressure was measured. A qualified physician then measured the
blood pressure of the seated subject twice by using a standard
mercury sphygmomanometer, and thereafter the mean of 2 mea-
surements was considered as the participant’s blood pressure.
The systolic blood pressure was defined as the appearance of the
first sound (Korotkoff phase 1), and the diastolic blood pressure
was defined as the disappearance of the sound (Korotkoff phase
5) during deflation of the cuff at a decrement rate of 2–3 mm/s of
the mercury column (30).
Definition of terms
Because no reference values for WC exist for children and
adolescents, we established the enlarged WC criteria by analyz-
ing all adolescents in the current dataset whose WCs were re-
corded. Subjects with a value �90th percentile for age and sex
from this sample population were classified as having enlarged
HYPERTRIGLYCERIDEMIC WAIST IN ADOLESCENTS 37
by guest on April 5, 2011
w
w
w
.ajcn.org
D
ow
nloaded from
WC. The choice of the 90th percentile was based on the associ-
ation between truncal fat and WC according to Taylor et al (31).
Other investigators also showed that clustering of risk factors
was significantly higher in subjects with a WC� 90th percentile
than in subjects with a WC� 90th percentile (32).
In developing a definition for elevated serum triacylglycerol
concentration in adolescents, we considered reference values
from the National Cholesterol Education Program’s Pediatric
Panel Report (32). In children aged 10–19 y, a borderline high
range for triacylglycerol concentrations is given as 90–129
mg/dL (1.02–1.46 mmol/L). Therefore, the midpoint value for
triacylglycerol concentrations (�110 mg/dL � 1.24 mmol/L)
was taken as the 90th percentile value for age. Subjects were
categorized in 4 phenotype groups on the basis of the mentioned
cutoffs: normal WC (�90th percentile for age and sex) and
normal serum triacylglycerol concentrations (�110 mg/dL);
normal WC (�90th percentile for age and sex) and elevated
serum triacylglycerol concentrations (�110 mg/dL); enlarged
WC (�90th percentile for age and sex) and normal serum tri-
acylglycerol concentrations (�110 mg/dL); and enlarged WC
(�90th percentile for age and sex) and elevated serum triacyl-
glycerol concentrations (�110 mg/dL).
Subjects with �3 of the 5 characteristics described in this
paragraph were categorized as having the metabolic syndrome.
The first characteristic was an enlarged WC (�90th percentile
for age and sex), and the second characteristic was an elevated
systolic or diastolic blood pressure, which was defined as a
value �90th percentile for age, sex, and height based on pub-
lished reference data (34). Because the updated report of the Task
Force on the Diagnosis and Management of Hypertension does
not include adolescents aged 18 and 19 y, we used cutoffs
of�130 and�85 mm Hg for systolic and diastolic blood pres-
sure, respectively, in those age categories, on the basis of the
recommendations of the Joint National Committee on Preven-
tion, Detection, Evaluation and Treatment of High Blood Pres-
sure (35). The third characteristic was a borderline low HDL
cholesterol concentration, in the range of 35–45 mg/dL (0.91–
1.16 mmol/L), as ascertained for all sexes and ages by the Na-
tional Cholesterol Education Program’s Report of the Expert
Panel on Blood Cholesterol Levels in Children and Adolescents
(33). Therefore, the midpoint value for HDL cholesterol (�40
mg/dL� 1.03 mmol/L) was used as a 10th percentile value. The
fourth characteristic was an elevated serum triacylglycerol con-
centration (�110 mg/dL). The fifth characteristic was a refer-
ence value of�110 mg/dL (�6.1 mmol/L) for elevated fasting
plasma glucose, which was taken from the American Diabetes
Association guideline (36).
Hypercholestrolemia was defined as TC �200 mg/dL and
high LDL cholesterol as �130mg/dL (33). Hypertension was
defined as systolic or diastolic blood pressure of�95th percen-
tile for sex, age, and height (34) and, in adolescents aged 18 and
19 y, as a systolic blood pressure �140 mm Hg or a diastolic
blood pressure� 90 mm Hg (35). The presence of�1 and�2 of
the 5 major risk factors for CVD (ie, elevated fasting glucose,
high LDL cholesterol, low HDL cholesterol, hypercholesterol-
emia, and hypertension) was also evaluated.
On the basis of the standardized percentile curves of BMI
suggested for Iranian children and adolescents, overweight was
defined as�95th percentile of BMI for age and sex (37). Risk of
overweight was defined as�85th to�95th percentile of BMI for
age and sex, and normal weight was defined as�85th percentile
of BMI for age and sex.
Statistical analysis
We used SPSS software (version 9.05; SPSS Inc, Chicago IL)
for all statistical analyses. The prevalence of the HW phenotype
and its 95% CI was calculated overall, by sex, by sex and family
history of diabetes, and by sex and BMI status. There was a
significant interaction between sex and age with regard to the
prevalence of HW (P� 0.05). Therefore, the prevalences of the
HW phenotype in boys and girls were examined by age. Preva-
lence values were compared by using the chi-square test. Signif-
icant differences in general characteristics across different phe-
notypes of serum triacylglycerol concentrations and WC were
searched by using 2-factor (enlarged WC versus normal WC and
elevated triacylglycerol versus normal triacylglycerol) analysis
of variance (ANOVA), and the main effects of WC and triacyl-
glycerol concentrations and the WC � triacylglycerol concen-
tration interaction were tested. We also used 2-factor ANOVA to
detect significant differences in the distribution of subjects
across different phenotypes of serum triacylglycerol concentra-
tion and WC with regard to the prevalence of metabolic abnor-
malities. We ascertained age- and sex-adjusted means of meta-
bolic risk factors across different phenotypes of serum
triacylglycerol concentrations and WC by using general linear
models. These models were further adjusted for BMI. Analysis of
covariance was used to compare these means. To determine the
association of the HW phenotype with metabolic abnormalities,
we used logistic regression models. Because the ORs estimated
from logistic regression models in cross-sectional studies are not
valid estimators of the rate ratios when the binary outcome vari-
able has a higher prevalence (38, 39), we used the formula sug-
gested by Zhang and Yu (39) to correct the adjusted ORs obtained
from logistic regression so that we could derive an estimate of
association that better represents the true relative risk. All models
were controlled for age (y), sex, and physical activity level (light,
moderate, and heavy). When a significant association with the
HW phenotype was observed, we further examined whether
overweight associated with the HW phenotype would explain the
associations by adjusting for BMI. In all multivariate models,
subjects with normal serum triacylglycerol concentrations and
normal WCs were considered as a reference. McNemar’s chi-
square statistic was used to ascertain whether significant differ-
ences were present between the HW phenotype and overweight
with respect to accuracy in classifying individuals according to
the presence or absence of risk factors. Such analyses were also
repeated by comparing the HW phenotype with the metabolic
syndrome, enlarged WC, or elevated triacylglycerol concentra-
tion alone. The frequency of subjects with and without CVD risk
factors in established cutoffs of the screening measures (HW
phenotype, overweight, and the metabolic syndrome) was ascer-
tained by cross-tabulation, and sensitivity and specificity were
calculated. The sensitivity of each screening measure was de-
fined as the proportion of the total number of subjects with a
given risk factor who were identified correctly by the same in-
dicator. The specificity of each screening measure was defined as
the proportion of the total subjects without the risk factor who
were correctly identified by the same measure.
38 ESMAILLZADEH ET AL
by guest on April 5, 2011
w
w
w
.ajcn.org
D
ow
nloaded from
RESULTS
Of the 3036 adolescents, 358 (11.8%) had a family history of
diabetes, 382 (12.5%) were overweight, and 399 (13.1%) were at
risk of overweight. The prevalence of the HW phen