Visceral fat thickness measured by ultrasonography can estimate not only visceral obesity but also risks of cardiovascular and metabolic diseases

Visceral fat thickness measured by ultrasonography can estimate
not only visceral obesity but also risks of cardiovascular and
metabolic diseases1–3
Soo Kyung Kim, Hae Jin Kim, Kyu Yeon Hur, Sung Hee Choi, Chul Woo Ahn, Sung Kil Lim, Kyung Rae Kim,
Hyun Chul Lee, Kap Bum Huh, and Bong Soo Cha
ABSTRACT
disease (CVD; 1). In particular, the accumulation of adipose
Background: Visceral obesity is closely associated with cardiovas-
tissue predominantly in the visceral cavity plays a major role in
cular disease and the metabolic syndrome. Estimating the amount of
the development of metabolic syndrome, CVD, or both (2).
visceral fat is important and requires a straightforward, reliable, and
Therefore, estimating the visceral fat accumulation is impor-
practical method.
tant in terms of evaluating patients with a higher risk of CVD.
Downloaded from
Objective: We investigated whether visceral fat thickness (VFT)
Currently, computed tomography (CT) at the abdominal level is
measured by ultrasonography can adequately assess visceral fat ac-
recognized as the standard method (3). However, exposure to
cumulation and predict cardiovascular or metabolic diseases.
ionizing radiation, high cost, and low availability prevent the
Design: Diabetic patients (240 men and 106 women) underwent
wide use of CT in clinical and epidemiologic studies. Therefore,
ultrasonography to estimate visceral fat accumulation.
alternative, simple, noninvasive methods of assessing visceral
Results: The visceral adipose tissue area had the best correlation
fat accumulation are needed. Such methods would include the
www.ajcn.org
with VFT (r
0.799, P  0.001). VFT correlated with HDL-
anthropometric indexes, such as the body mass index (BMI; in
cholesterol, triacylglycerol, and high-sensitivity C-reactive protein
kg/m2), waist circumference, and the waist-to-hip circumference
concentrations, the homeostasis model assessment for insulin resis-
ratio (WHR; 4); dual-energy X-ray absorptiometry (5); and ul-
tance, and the intima-media thickness at the common carotid artery
trasonography (6 –9). Ultrasonography is a reliable and conve-
by guest on April 29, 2011
(r
0.30, 0.39, 0.34, 0.31, and 0.33, respectively; P  0.05) in
nient way of quantifying the amount of visceral fat, and the
men and with triacylglycerol and high-sensitivity C-reactive protein
diverse ultrasonographic values were reported to be useful (6 –
concentrations and the homeostasis model assessment for insulin
11). Nevertheless, the absence of a useful measurement with
resistance (r
0.33, 0.44, and 0.30, respectively; P  0.05) in
established simplicity and reliability prevents the widespread use
women. Men in the middle and high VFT tertiles had a higher odds
of ultrasonography. In addition, whether the ultrasonographi-
ratio (OR) of coronary artery disease [ORs: 4.48 (95% CI: 1.29, 5.51)
cally determined amount of visceral fat directly reflects the risk
and 2.04 (1.06, 3.94), respectively; P
0.016], hypertriacylglycer-
of CVD, other metabolic diseases (eg, hypertension, dyslipide-
olemia [ORs: 2.87 (1.41, 5.86) and 1.91 (1.24, 2.95), respectively;
mia, and metabolic syndrome), or both remains to be proven.
P
0.003], and the metabolic syndrome [ORs: 3.38 (1.61, 7.10) and
To assess the usefulness of ultrasonographic measurement for
1.95 (1.16, 3.27), respectively; P
0.003] than did those in the low
visceral fat, we investigated the correlation between the values
tertile, after adjustment for age, waist circumference, and body mass
measured with CT and with ultrasonography and the correlation
index.
between several ultrasonographic values and the risk factors of
Conclusion: VFT might be a reliable index for assessing the amount
CVD. This study particularly focused on whether the visceral fat
of visceral fat and for identifying diabetic patients, particularly men,
thickness (VFT) measured by ultrasonography could be an al-
who are at high risk of cardiovascular disease.
Am J Clin Nutr
ternative index for estimating the risk of CVD and whether it
2004;79:593–9.
could predict the presence of coronary artery disease (CAD) or
other metabolic diseases.
KEY WORDS
Abdominal obesity, cardiovascular disease,
coronary artery disease, ultrasonography, visceral fat, waist circum-
1 From the Department of Internal Medicine (SKK, HJK, KYH, SHC,
ference
CWA, SKL, KRK, HCL, KBH, and BSC) and Brain Korea 21 Project for
Medical Science (SKK, CWA, SKL, KRK, HCL and BSC), Yonsei Univer-
sity College of Medicine, Seoul, Korea.
2 Supported by grant no. R13-2002-054-01001-0 (2002) from the Basic
INTRODUCTION
Research Program of the Korea Science & Engineering Foundation.
3
Obesity has induced many public health problems related to
Address reprint requests to BS Cha, Department of Internal Medicine,
Yonsei University College of Medicine, 134 Shinchon-Dong, Seodaemoon-
metabolic diseases, including glucose intolerance, hypertension,
Ku, PO Box 120-749, Seoul, Korea. E-mail: bscha@yumc.yonsei.ac.kr.
dyslipidemia, hyperinsulinemia, and atherosclerosis. Moreover,
Received April 14, 2003.
these complexes are known to increase the risk of cardiovascular
Accepted for publication October 13, 2003.
Am J Clin Nutr 2004;79:593–9. Printed in USA. © 2004 American Society for Clinical Nutrition
593

---

594
KIM ET AL
SUBJECTS AND METHODS
the internal face of the rectoabdominal muscle perpendicular to
the aorta. The ratio of PFT
to SFT
was calculated and is
max
min
Subjects
referred to as the abdominal wall fat index (AFI). The ratio of
Three hundred forty-six diabetic patients (240 men and 106
VFT to SFT
was calculated as the value corresponding to the
max
women) who visited the Diabetes Center at the Severance Hos-
CT-determined ratio of visceral fat area (VFA) to subcutaneous
pital, Yonsei University, for glycemia control were enrolled in
fat area (SFA) (VSR).
the study. The exclusion criteria included pregnancy; endocri-
CT scanning was carried out on the same day as ultrasonog-
nopathies other than diabetes; a history of treatment with anti-
raphy. The CT scans were performed (Tomoscan 350; Philips,
obesity drugs, thiazolidinediones, or corticosteroid; abnormal
Mahway, NJ) to measure the VFA and SFA at L4 –L5 and the
renal function as determined from the age-adjusted creatinine-
muscle and fat areas at midthigh (the midpoint of the upper
clearance values; symptoms that are indicative of CAD within
border of the patella and the greater trochanter). Adipose tissue
the previous 6 mo; abnormal electrocardiogram findings without
was defined as having a density of
150 to
50 Hounsfield
a confirmation by angiography; and weight loss of Π3 kg during
units, and muscle tissue was defined as having a density of
49
the past 3 mo.
to 100 Hounsfield units (12). The VSR and the ratio of the VFA
All the participants underwent the standard examination and
to the muscle area (VMR) were calculated.
testing, which included measurement of the concentrations of
The bilateral common carotid arteries (CCAs) were scanned
fasting glucose, total cholesterol, HDL cholesterol, triacylglyc-
by using the SA9900 system (Medison) with a 7.5-MHz linear
erol, high-sensitivity C-reactive protein (hsCRP), and insulin;
transducer. Scanning was performed at the far wall of the middle
blood pressure, height, weight, and waist and hip circumference
and distal CCAs by using a lateral longitudinal projection. The
measurements (by the same investigator); a resting electrocar-
intima-media thickness (IMT) at the CCA, which is defined as
diogram; and completion of a questionnaire on CVD and other
the distance between the media-adventitia interface and the
Downloaded from
diseases. As an indicator of insulin resistance, this study used an
lumen-intima interface, was measured at the point of the greatest
index for a homeostasis model assessment for insulin resistance
thickness and at 1 cm upstream and 1 cm downstream of that
(HOMA-IR), which was calculated as follows:
point with the use of a Digimatic electronic caliper (Mitutoyo,
Kawasaki, Japan). The mean IMT was defined as the mean of 6
HOMA-IR
[fasting plasma glucose (mmol/L)
measurements, and the maximum IMT was defined as the largest
of 6 measurements. The same investigator performed all the
fasting serum insulin ( U/mL)]/22.5
(1)
www.ajcn.org
ultrasonographic procedures used to estimate the fat distribution
The body fat distribution for all participants was estimated by
and the IMT.
ultrasonography, and CT scan was performed on only 75 of the
subjects (50 men and 25 women). The study was approved by the
Definition of terms
by guest on April 29, 2011
Ethics Committee of Yonsei University College of Medicine,
As detailed in the report of the Adult Treatment Panel III (13),
and written informed consent was obtained from each subject.
metabolic syndrome was defined the presence of
2 of the con-
ditions hypertriacylglycerolemia, low-HDL cholesterolemia,
Anthropometric data
and hypertension. Dyslipidemia was defined as hypertriacyl-
glycerolemia (
1.7 mmol/L, or
150 mg/dL), low-HDL cho-
The height and weight were measured to the nearest 0.1 cm and
lesterolemia ( 1.0 mmol/L, or  40 mg/dL, in men;  1.3
0.1 kg, respectively, while the subjects were wearing light cloth-
mmol/L, or  50 mg/dL, in women), or receipt of hypolipidemic
ing and not wearing shoes. The waist circumference was mea-
treatment. Hypertension was defined as a systolic blood pressure
sured at the midpoint between the lateral iliac crest and the lowest
of
130 mm Hg and a diastolic blood pressure of
85 mm Hg,
rib, and the hip circumference was measured at the maximal
and the use of antihypertensive agents was also considered to
protrusion of the greater trochanter.
indicate hypertension. In this study, abdominal obesity was not
Ultrasonography was performed by using a high-resolution
included in the component of metabolic syndrome because sub-
ultrassonographic system (SA 9900; Medison, Seoul, Korea) as
jects with a larger waist circumference have a longer VTF, and
described by Suzuki et al (10) and Armellini et al (11). Briefly,
that fact might bias the results.
the subjects were examined in the supine position. All frozen
images were obtained immediately after respiration to avoid the
Analytic methods
influence of the respiratory status or abdominal wall tension. The
maximum thickness of the preperitoneal fat (PFT
) and the
The serum glucose concentrations were determined by using
max
minimum thickness of the subcutaneous fat (SFT
) were mea-
the glucose oxidase method. The plasma insulin concentrations
min
sured by longitudinal scanning with the use of a 7.5-MHz linear
were measured by using a radioimmunoassay according to a
probe from the xiphoid process to the umbilicus along the linea
double-antibody method and with a commercially available ra-
alba. The preperitoneal fat thickness was defined as the thickness
dioimmunoassay kit (Linco Research Inc, St Charles, MO). The
of the fat tissue between the liver surface and the linea alba, and
serum cholesterol and triacylglycerol concentrations were mea-
the subcutaneous fat thickness was defined as the thickness of the
sured enzymatically. The hsCRP concentration was quantified
fat tissue between the skin-fat interface and the linea alba. There-
by using a Nephelometer II (Dade Behring Diagnostics, Mar-
after, transverse scanning was performed to measure the maxi-
burg, Germany; 14).
mum subcutaneous fat thickness (SFT
) by using a 7.5- MHz
max
probe and to measure VFT by using a 3.5-MHz probe; both
Statistical analysis
measures were obtained 1 cm above the umbilicus. VFT was
Statistical analyses were performed by using SPSS software
defined as the distance between the anterior wall of the aorta and
(version 10.0; SPSS Inc, Chicago) and were carried out sepa-

---

ULTRASONOGRAPHIC ASSESSMENT OF VISCERAL FAT
595
TABLE 1
Clinical characteristics of the subjects by tertile of visceral fat thickness1
Men
Women
Low tertile
Middle tertile
High tertile
P for
Low tertile
Middle tertile
High tertile
P for
(n
78)
(n
81)
(n
81)
trend
(n
35)
(n
34)
(n
37)
trend
Age (y)
47.9
12.8
49.9
11.8
51.5
11.0
NS
52.2
11.3
53.7
7.5
55.0
5.9
NS
Weight (kg)
66.6
9.8
71.8
8.82
76.5
9.53,4
 0.001
58.1
8.3
59.6
7.1
66.0
9.53
 0.05
Waist (cm)
83.3
6.6
87.7
5.73
92.7
6.83,5
 0.001
79.8
10.5
83.7
7.1
90.4
7.63
 0.001
BMI (kg/m2)
23.2
2.5
24.7
2.62
27.0
3.03,5
 0.001
23.3
3.1
24.4
2.2
26.8
3.53
 0.001
SFT
(mm)
20.8
7.5
20.9
7.0
19.5
7.9
NS
25.4
8.4
25.4
8.4
26.6
6.4
NS
max
Glucose (mmol/L)
7.94
3.39
7.88
2.61
8.10
2.50
NS
8.60
2.94
7.78
3.39
8.94
2.50
NS
Hb A
(%)
7.9
1.9
7.7
1.5
7.8
1.6
NS
7.7
1.5
7.5
1.2
8.2
1.7
NS
1c
Total cholesterol (mmol/L)
4.73
0.83
4.71
0.93
4.66
0.85
NS
4.55
0.75
4.66
0.80
4.86
0.70
 0.05
HDL cholesterol (mmol/L)
1.15
0.18
1.11
0.19
1.08
0.19
NS
1.31
0.22
1.15
0.20
1.11
0.212
 0.05
Triacylglycerol (mmol/L)
3.46
1.50
4.96
1.982 5.25
2.113
 0.05
3.10
1.19
4.42
2.172
4.86
2.062
 0.05
Triacylglycerol:HDL cholesterol
3.32
1.92
5.15
4.262 6.13
4.452
 0.001
2.7
1.8
4.7
4.82
4.9
2.52
 0.05
hsCRP (mg/L)
1.50
1.12
1.57
0.96
2.1
1.852,4
 0.05
0.71
1.59
0.93
0.60
1.52
1.12,4
 0.05
Insulin (pmol/L)
46.6
32.3
50.2
29.4
71.8
342,4
 0.05
57.4
44.4
61.7
30.8
68.2
36.6
NS
HOMA-IR
2.4
1.3
2.5
1.4
3.5
2.12,4
 0.05
2.9
2.7
3.1
2.1
3.5
1.7
NS
IMT
(mm)
0.68
0.14
0.74
0.12
0.84
0.202
 0.05
0.70
0.21
0.69
0.11
0.71
0.10
NS
mean
IMT
(mm)
0.88
0.23
0.94
0.20
1.06
0.21
NS
0.85
0.29
0.89
0.26
0.86
0.10
NS
max
Downloaded from
1 All values are x
SD. Mean values for tertiles of visceral fat thickness were 36.2 mm (95% CI: 10.9, 44.0 mm) and 31.8 mm (15.0, 36.0 mm) (low tertile),
53.3 mm (44.1, 57.9 mm) and 44.7 mm (37.1, 47.0 mm) (middle tertile), and 75.8 mm (58.5, 104.0 mm) and 63.6 mm (47.4, 98.0 mm) (high tertile) in the men
and the women, respectively. SFT
, maximum subcutaneous fat thickness; Hb A , glycated hemoglobin; hsCRP, high-sensitivity C-reactive protein;
max
1c
HOMA-IR, homeostasis model assessment for insulin resistance; IMT
and IMT
, mean and maximum intima-media thickness at the common carotid
mean
max
artery, respectively.
2,3 Significantly different from the low tertile (one-way ANOVA followed by Scheffe’s post hoc test): 2P  0.05, 3 P  0.001.
www.ajcn.org
4,5 Significantly different from the middle tertile (one-way ANOVA followed by Scheffe’s post hoc test): 4P  0.05, 5 P  0.001.
rately in men and women because of the significant difference in
tertile to the middle tertile and the high tertile. In the men, the
body fat distribution between the sexes. The associations be-
fasting serum concentrations of triacylglycerol, hsCRP, and in-
by guest on April 29, 2011
tween the continuous variables were determined by using Pear-
sulin; the ratio of serum triacylglycerol to HDL cholesterol;
son’s correlation coefficients. The intergroup comparisons were
HOMA-IR; and the mean IMT in the high tertile group were
performed by using an independent-sample t test and a one-way
higher than those in the low tertile group. Women in the high VFT
analysis of variance followed by Scheffe’s post hoc test. The
tertile had higher fasting serum triacylglycerol and hsCRP con-
prevalence and risk relation of CAD or the other metabolic dis-
centrations, higher triacylglycerol:HDL, and a lower HDL-
eases among the groups were compared by using a chi-square
cholesterol concentration than did those in the low tertile.
test. Logistic regression, which was adjusted for age, waist cir-
cumference, and BMI, was used to analyze the associations be-
tween the tertiles of the VFT and the presence of CAD or other
Correlations of values measured by CT and
metabolic diseases. A low tertile of VFT was used as the refer-
ultrasonography in 75 subjects
ence category [odds ratio (OR): 1.00]. Receiver operating char-
To test the usefulness of ultrasonography for evaluating ab-
acteristic (ROC) curve analyses were performed to determine the
dominal adiposity, this study first compared the values obtained
sensitivity and specificity of using VFT as a forecaster of meta-
by ultrasonography with those obtained by CT scan in 75 subjects
bolic syndrome. P values  0.05 were considered significant.
selected from the study population. The 75 subjects did not differ
significantly from the entire population in anthropometric and
RESULTS
biochemical characteristics (data not shown).
The anthropometric characteristics of the 75 subjects who
Clinical characteristics of the subjects
underwent CT scanning are listed in Table 2. No differences in
The subjects were classified by the tertile on the basis of their
age, BMI, waist circumference, WHR, or the total abdominal fat
VFT; the mean values are  44.1 mm in men and  36 mm in
area were observed between the men and the women. Neverthe-
women for the low tertile, 44.1–58.0 mm in men and 36.0 – 47.0
less, the men had less subcutaneous fat deposition and a higher
mm in women for the middle tertile, and Π58.0 mm in men and
VSR than did the women. These differences identified by CT
Œ 47.0 mm in women for the high tertile. The patients’ charac-
scan were also observed in the ultrasonographic assessment.
teristics according to VFT tertiles are shown in Table 1. No
This study investigated the correlations between values ob-
differences in terms of age, SFT
, concentrations of fasting
tained by using CT scanning (VFA, VSR, and VMR) and those
max
glucose or glycated hemoglobin (Table 1), duration of diabetes,
obtained by using ultrasonography (VFT, PFT
, AFI, and
max
or systolic and diastolic blood pressures (data not reported) were
VFT:SFT
; Table 3). The VFT correlated with the VFA (r
max
observed between the tertiles in the men or the women. The
0.799, P  0.001), VSR, and VMR. The VFT:SFT
was well
max
weight, waist circumference, and BMI increased from the low
correlated with the VFA (r
0.473, P  0.05), VSR, and VMR.

---

596
KIM ET AL
TABLE 2
Correlations between several ultrasonographic values and
Anthropometric indexes in the 75 patients who underwent computed
clinical and laboratory data
tomography1
The correlation coefficients between the ultrasonographic val-
Men (n
50)
Women (n
25)
ues and the clinical data are shown in Table 4. In the men, VFT
Age (y)
45.7
9.8
46.7
10.3
correlated with BMI; waist circumference; concentrations of se-
Weight (kg)
73.7
12.1
61.1
14.12
rum HDL cholesterol (negatively), triacylglycerol, hsCRP, and
BMI (kg/m2)
25.3
3.8
24.1
3.8
insulin; triacylglycerol:HDL cholesterol; HOMA-IR; and the
Waist (cm)
88.0
7.8
84.2
11.2
mean and maximum IMT. In addition, VFT correlated with BMI,
WHR
0.93
0.04
0.93
0.07
waist circumference, triacylglycerol and hsCRP concentrations,
Computed tomography
triacylglycerol:HDL cholesterol, and HOMA-IR in the women.
Total fat area (cm2)
225.1
66.2
258.5
81.7
Waist circumference correlated with triacylglycerol (r
0.24,
VFA (cm2)
95.4
29.7
82.5
28.1
P
SFA (cm2)
129.7
48.7
175.9
66.32
 0.05), HDL-cholesterol (r
0.17, P  0.05), and insulin
VSR
0.79
0.28
0.51
0.182
(r
0.27, P  0.05) concentrations; triacylglycerol:HDL choles-
Ultrasonography
terol (r
0.24, P  0.001); HOMA-IR (r
0.27, P  0.05); and
VFT (mm)
51.7
15.4
37.1
18.22
mean IMT (r
0.20, P  0.05) in men. However, the correlation
SFT
(mm)
22.4
8.0
30.1
9.82
max
coefficients between waist circumference and the above risk factors
AFI
1.91
1.19
0.84
0.312
were lower than those between VFT and the above risk factors.
VFT:SFT
2.56
1.00
1.31
0.752
max
1 All values are x
SD. WHR, waist-to-hip circumference ratio; VFA,
Association between VFT and the prevalence of CAD and
visceral fat area; SFA, subcutaneous fat area; VSR, visceral-to-subcutaneous
fat area ratio; VFT, visceral fat thickness; SFT
, the maximum subcuta-
metabolic diseases
Downloaded from
max
neous fat thickness; AFI, abdominal wall fat index (ratio of maximum pre-
Next, this study compared the prevalence of CAD, hyperten-
peritoneal fat thickness to minimum subcutaneous fat thickness).
2
sion, dyslipidemia (hypertriacylglycerolemia, low-HDL choles-
Significantly different from the men, P  0.05 (independent-sample t
terolemia, or both), and metabolic syndrome in the men and the
test).
women according to VFT tertile (Table 5). Men in the middle
and high VFT tertiles had a greater prevalence of CAD, hyper-
www.ajcn.org
triacylglycerolemia, low-HDL cholesterolemia, and metabolic
In contrast, the correlation of waist circumference with VFA (r
syndrome than did those in the low tertile. In women, the prev-
0.543, P  0.05) was weaker than that with the VFT.
alence of dyslipidemia and metabolic syndrome was higher in the
middle and high tertiles than in the low tertile. However, the
by guest on April 29, 2011
prevalence of CAD did not differ significantly among tertiles.
Coefficients of variation of the ultrasonographic values
The logistic regression results showing the ORs for the various
The intraobservational reproducibility of the ultrasonographic
metabolic diseases in the VFT tertiles are listed in Table 6. The
estimations was 1.5–2.0% for the VFT, 1.8 –3.2% for the SFTmax
ORs for CAD, hypertension, hypertriacylglycerolemia, low-
and SFT
, and 3.0 –5.2% for the PFT
. The reproducibility
min
max
HDL cholesterolemia, and metabolic syndrome were higher for
between the 2 operators was 1.8 –2.8% for the VFT and 3.5–
men in the middle and high tertiles than for those in the low
8.1% for the other indexes. Therefore, the absolute value of the
tertile. The ORs for dyslipidemia and metabolic syndrome were
VFT might be a more useful marker of visceral fat deposition
higher for women in the middle and high tertiles than for those in
than might either the PFT
, AFI, or VFT:SFT
.
max
max
the low tertile. With the exception of hypertension and low-HDL
cholesterolemia in the men and hypertriacylglycerolemia in the
women in the high VFT tertile, these observations remained
significant after adjustment for age, waist circumference, and
TABLE 3
BMI.
Pearson’s correlation coefficients between the values measured by
As shown in Figure 1, the area under the ROC curve for VFT
computed tomography (CT) and by ultrasonography1
as a predictor of the presence of metabolic syndrome was 0.764
CT scan
(95% CI: 0.700, 0.827; P  0.001) in men and 0.730 (95% CI:
0.614, 0.845; P
0.001) in the women. A VFT of 47.6 and 35.5
Ultrasonography
VFA
VSR
VMR
mm in the men and the women, respectively, was found to be the
VFT
0.7992
0.3863
0.6113
discriminating cutoff for the metabolic syndrome (specificity of
PFT
0.3283
0.136
0.117
max
74% and 78% and sensitivity of 71% and 69% in the men and in
AFI
0.101
0.5023
0.303
the women, respectively). The use of ROC analyses to predict the
VFT:SFT
0.4733
0.5712
0.5842
max
presence or absence of CAD, other metabolic diseases, or both
1 75 patients underwent CT. VFA, visceral fat area; VSR, visceral-to-
showed that the area under the curves for VFT were 0.621 (95%
subcutaneous fat area ratio; VMR, visceral fat-to-thigh muscle area ratio;
CI: 0.527, 0.714) for CAD, 0.649 (95% CI: 0.572, 0.725) for
VFT, visceral fat thickness; PFT
, maximum preperitoneal fat thickness;
max
hypertension, 0.725 (95% CI: 0.657, 0.793) for hypertriacyl-
AFI, abdominal wall fat index (ratio of maximum preperitoneal fat thickness
glycerolemia, and 0.583 (95% CI: 0.507, 0.659) for low-HDL
to minimum subcutaneous fat thickness); VFT:S
, ratio of VFT to maxi-
max
cholesterolemia in the men and 0.687 (95% CI: 0.577, 0.799) for
mum subcutaneous fat thickness.
2 P
hypertriacylglycerolemia and 0.695 (95% CI: 0.544, 0.803) for
 0.001.
3 P  0.05.
low-HDL cholesterolemia in the women.

---

ULTRASONOGRAPHIC ASSESSMENT OF VISCERAL FAT
597
TABLE 4
Pearson’s correlation coefficients between several ultrasonographic values and the clinical or laboratory data1
Men (n
240)
Women (n
106)
VFT
P
AFI
VFT:SFT
VFT
P
AFI
VFT:SFT
max
max
max
max
BMI
0.612
0.332
0.08
0.143
0.602
0.25
0.273
0.05
Waist circumference
0.622
0.292
0.06
0.203
0.562
0.18
0.263
0.12
Total cholesterol
0.03
0.07
0.04
0.05
0.25
0.10
0.18
0.11
HDL cholesterol
0.303
0.15
0.07
0.153
0.17
0.20
0.11
0.09
Triacylglycerol
0.392
0.03
0.09
0.13
0.333
0.09
0.07
0.08
Triacylglycerol:HDL cholesterol
0.402
0.09
0.05
0.14
0.313
0.05
0.12
0.06
hsCRP
0.342
0.14
0.14
0.02
0.442
0.22
0.13
0.12
Fasting insulin
0.333
0.01
0.12
0.08
0.21
0.11
0.03
0.10
HOMA-IR
0.312
0.10
0.07
0.10
0.303
0.11
0.08
0.08
IMT
0.333
0.17
0.02
0.303
0.28
0.16
0.24
0.33
mean
IMT
0.243
0.11
0.09
0.233
0.15
0.27
0.453
0.15
max
1 VFT, visceral fat thickness; PFT
, the maximum preperitoneal fat thickness; AFI, abdominal wall fat index (ratio of maximum preperitoneal fat
max
thickness to minimum subcutaneous fat thickness); VFT:SFT
, ratio of VFT to SFT; hsCRP, high-sensitivity C-reactive protein; HOMA-IR, homeostasis
max
model assessment for insulin resistance; IMT
and IMT
, mean and maximum intima-media thickness at the common carotid artery, respectively.
mean
max
2 P  0.001.
3 P  0.05.
Downloaded from
DISCUSSION
abdominal sagittal diameter (4), or neck circumference (17).
Previous studies reported that ultrasonography could be used
These values provide a fast, easy, and noninvasive method of
to estimate visceral adiposity (8 –10). However, no definite val-
assessing regional adiposity, particularly in epidemiologic stud-
ues were established. This study showed that VFT had a better
ies (18, 19). However, it is possible that substantial variations in
correlation with the values measured by CT than did PFT
,
the visceral fat content may be observed among persons with a
www.ajcn.org
max
AFI, or VFT:SFT
and that VFT might be very useful in as-
similar waist circumference or WHR value because these in-
max
sessing the amount of visceral fat accumulation. Furthermore,
dexes are not the direct methods of quantifying the amount of fat
this study found that a high VFT was closely related to an in-
or of discriminating between visceral and subcutaneous fat. Ac-
creased risk of CVD and to a greater prevalence of CAD, other
cordingly, alternative and reliable methods are needed to over-
by guest on April 29, 2011
metabolic diseases, or both in men. Therefore, this study empha-
come these pitfalls of anthropometric indexes.
sizes that an assessment of visceral fat amounts via ultrasonog-
Although ultrasonography is not a method normally used to
raphy can be used to evaluate the risk of CVD and predict the
quantify the amount of fat, this study suggests that it may be an
presence of various metabolic diseases.
alternative method of assessing visceral adiposity. Furthermore,
Body fat distribution varies considerably, even among people
the ORs for CAD [ORs: 4.48 (95% CI: 1.29, 5.51) and 2.04 (95%
with similar total fat amounts. It is well known that visceral
CI: 1.06, 3.94); P
0.016], hypertriacylglycerolemia [ORs: 2.87
obesity may be a better predictor of metabolic syndrome—
(95% CI: 1.41, 5.86) and 1.91 (95% CI: 1.24, 2.95); P
0.003],
including hyperinsulinemia, glucose intolerance, dyslipidemia,
and metabolic syndrome [ORs: 3.38 (95% CI: 1.61, 7.10) and
and CVD—and cancer than may general obesity (1, 15). In ad-
1.95 (95% CI: 1.16, 3.27); P
0.003] in the men in the middle
dition, subjects judged by the BMI criteria to be lean may be
and high VFT tertiles, respectively, were significantly higher
insulin resistant if they have centrally located body fat (16).
than those in the men in the low tertile, even after adjustment for
Several methods of assessing the amount of visceral fat accu-
age, waist circumference, and BMI. In addition, the VFT corre-
mulation have been investigated. The simplest way is to use an
lated with several risk factors for CVD, such as the fasting plasma
anthropometric index such as BMI, waist circumference, WHR,
triacylglycerol, hsCRP, and insulin concentrations; triacylglyc-
TABLE 5
Prevalence of coronary artery disease (CAD) and metabolic diseases within the visceral fat thickness tertiles in the male and female subjects1
Men
Women
Low tertile
Middle tertile
High tertile
Low tertile
Middle tertile
High tertile
Prevalence
(n
78)
(n
81)
(n
81)
(n
35)
(n
34)
(n
37)
CAD (%)
5.4
17.92
21.62
2.9
12.9
9.1
Hypertension (%)
50.0
65.4
79.73
64.7
67.7
60.6
Hypertriacylglycerolemia (%)
35.1
66.73
77.03
35.3
67.72
69.72
Low-HDL cholesterolemia (%)
40.5
60.32
60.82
52.9
87.12
90.93
Metabolic syndrome (%)
35.9
72.83
86.43
57.1
82.42
94.63
1 The definitions of hypertension, hypertriacylglycerolemia, low-HDL cholesterolemia, and metabolic syndrome are given in the Definition of terms
subsection of Subjects and Methods. CAD confirmed by angiography.
2,3 Significantly different from the low tertile (chi-square analysis): 2P  0.05, 3 P  0.001.

---

598
KIM ET AL
TABLE 6
Odds ratios and 95% CIs of coronary artery disease (CAD) and the metabolic diseases by tertile of visceral fat thickness1
Unadjusted
Adjusted2
Low tertile
Middle tertile
High tertile
P for trend
Middle tertile
High tertile
P for trend
Men (n
240)
CAD
1.00
3.83 (1.20, 12.23)3
4.83 (1.53, 15.24)4
0.006
4.48 (1.29, 5.51)3
2.04 (1.06, 3.94)3
0.016
Hypertension
1.00
1.89 (0.98, 3.63)
3.93 (1.90, 8.14)4
 0.001 1.46 (0.70, 3.06)
1.40 (0.88, 2.23)
0.096
Hypertriacylglycerolemia
1.00
3.69 (1.89, 7.21)4
6.19 (3.01, 12.74)4
 0.001 2.87 (1.41, 5.86)3 1.91 (1.24, 2.95)3
0.003
Low-HDL cholesterolemia
1.00
2.22 (1.16, 4.26)3
2.28 (1.18, 4.40)3
0.014
1.77 (0.88, 3.53)
1.36 (0.90, 2.05)
0.18
Metabolic syndrome
1.00
4.43 (2.24, 8.78)4
11.14 (4.92, 25.23)4
 0.001 3.38 (1.61, 7.10)3 1.95 (1.16, 3.27)3
0.003
Women (n
106)
CAD
1.00
4.89 (0.52, 46.36)
3.30 (0.33, 33.46)
0.19
4.68 (0.48, 45.14)
1.57 (0.42, 5.88)
0.21
Hypertension
1.00
1.45 (0.41, 3.21)
0.84 (0.31, 2.26)
0.30
0.95 (0.30, 3.03)
0.67 (0.36, 41.28)
0.35
Hypertriacylglycerolemia
1.00
3.85 (1.37, 10.79)3
4.22 (1.52, 11.73)3
0.003
3.91 (1.33, 11.48)3 1.71 (0.95, 3.09)
 0.05
Low-HDL cholesterolemia
1.00
6.00 (1.72, 20.89)4
8.89 (2.27, 34.79)4
0.014
5.32 (1.48, 19.09)3 2.81 (1.33, 5.97)3
 0.50
Metabolic syndrome
1.00
4.73 (1.35, 16.54)3
22.4 (2.73, 98.92)4
 0.001 3.25 (1.19, 11.03)3 10.74 (1.12, 67.83)3
 0.05
1 The definitions of hypertension, hypertriacylglycerolemia, low-HDL cholesterolemia, and metabolic syndrome are given in the Definition of terms
subsection of Subjects and Methods. CAD confirmed by angiography.
2 Adjusted for age, waist circumference, and BMI.
3,4
Downloaded from
Significantly different from the low tertile (logistic regression analysis): 3P  0.05, 4 P  0.001.
erol:HDL cholesterol; and HOMA-IR. These correlations with
good method of predicting CAD, metabolic diseases, or both and
VFT were relatively stronger than were those with the waist
of evaluating the risk of CVD without the limitations of the
circumference or other ultrasonographic values. Moreover, the
anthropometric indexes.
mean and maximum IMTs at the CCA, which are a marker of
Although the PFT
measured by ultrasonography was pre-
max
www.ajcn.org
early atherosclerosis, correlated with VFT (r
0.33 and 0.24,
viously reported to correlate positively with the VFA measured
respectively; P  0.05) and VFT:SFT
(r
0.30 and 0.23,
by CT (10), we did not observe this correlation. This discrepancy
max
respectively; P  0.05) in men, but did not correlate with the
may result from the irreproducibility of the PFT
measure-
max
other anthropometric and imaging values, except for waist cir-
ments or from differences in the measuring sites. The sites were
by guest on April 29, 2011
cumference, which showed a weak correlation with the mean
not fixed because PFT
was measured at the thickest point in
max
IMT (r
0.20, P  0.05).
the preperitoneal fat. Therefore, it is likely that PFT
has a high
max
In addition, this study suggested that a VFT of 47.6 and 35.5
degree of irreproducibility and that it may be an unadvisable
mm in the men and the women, respectively, was a cutoff for
measure in clinical or epidemiologic studies. In the present study,
predicting the presence of CAD and various metabolic diseases.
the intraobservational and interobservational CVs for VFT were
In the case of metabolic syndrome, these cutoff values for VFT
lower (1.5–2.0% and 1.8 –3.0%, respectively) than those for the
were shown to have high specificity and sensitivity in both sexes.
other ultrasonographic measurements (1.8 –5.2% and 3.5– 8.1%,
Therefore, the ultrasonographic measurement of VFT might be a
respectively). In addition, VFT had a better correlation with VFA
FIGURE 1. Receiver operating characteristic (ROC) analyses of visceral fat thickness as a predictor of the presence or absence of the metabolic syndrome
in men (left) and women (right). The higher the area under the curve—ie, the greater the curvature away from the 50% line—the greater the predictive power.
A visceral fat thickness of 47.6 and 35.5 mm in men and women, respectively, was chosen as the discriminator value to predict the presence of the metabolic
syndrome (specificity of 74% and 78% and sensitivity of 71% and 69% in the men and the women, respectively). The area under the ROC curve was 0.764 in
the men and 0.730 in the women.

---

ULTRASONOGRAPHIC ASSESSMENT OF VISCERAL FAT
599
(r
0.799, P  0.001) than with the other ultrasonographic
2. Fujioka S, Matsuzawa Y, Tokunaga K, Tauri S. Contribution of intra-
values and with waist circumference. Therefore, VFT appears to
abdominal fat accumulation to the impairment of glucose and lipid
metabolism in human obesity. Metabolism 1987;36:54 –9.
be more reliable in estimating the amount of visceral fat accu-
3. Rossner S, Bo WJ, Hiltbrandt E, et al. Adipose tissue determinations in
mulation than are the other ultrasonographic indexes.
cadavers: a comparison between cross-sectional planimetry and com-
This study had several limitations. First, it was performed only
puted tomography. Int J Obes Relat Metab Disord 1990;14:893–902.
in a Korean population. Accordingly, the VFT cutoffs used for
4. Seidell JC, Cigolini M, Charzewska J, Ellsinger BM, Deslypere JP, Cruz
predicting the presence of CAD, metabolic diseases, or both may
A. Fat distribution in European men: a comparison of anthropometric
differ from those in other ethnic groups. Second, because all the
measurements in relation to cardiovascular risk factors. Int J Obes Relat
Metab Disord 1992;16:17–22.
participants were diabetic and almost every patient was being
5. Bertin E, Marcus C, Ruiz JC, Eschard JP, Leutenegger M. Measurement
treated with oral hypoglycemic, antihypertensive, or hypolipi-
of visceral adipose tissue by DXA combined with anthropometry in
demic agents, all of which are capable of affecting the serum lipid
obese humans. Int J Obes Relat Metab Disord 2000;24:263–70.
concentration and insulin sensitivity, it is possible that the inter-
6. Tornaghi G, Raiteri R, Pozzato C, et al. Anthropometric or ultrasonic
relations between VFT and the metabolic values related to the
measurements in assessment of visceral fat? A comparative study. Int J
risks of CVD were weakened. We believe that, had healthy sub-
Obes Relat Metab Disord 1994;18:771–5.
7. Armellini F, Zamboni M, Robbi R, et al. Total and intra-abdominal fat
jects been used in this study, a more conspicuous relation be-
measurements by ultrasound and computerized tomography. Int J Obes
tween VFT and the CVD risk would have been found. Third,
Relat Metab Disord 1993;17:209 –14.
whereas VFT in the men reasonably reflected CVD risk, it did not
8. Ribeiro-Filho FF, Faria AN, Kohlmann O Jr, et al. Ultrasonography for
do so in the women. The divergent significances of VFT accord-
the evaluation of visceral fat and cardiovascular risk. Hypertension
ing to the sex appear to be due to the lower prevalence of CAD in
2001;38:713–7.
9. Stolk RP, Meijer R, Mali WP, Grobbee DE, van der Graaf Y. Ultrasound
Korean women than in men. In addition, this study did not ana-
measurements of intraabdominal fat estimate the metabolic syndrome
lyze the role of VFT in the women after grouping them according
better than do measurements of waist circumference. Am J Clin Nutr
Downloaded from
to premenopausal and postmenopausal states. Accordingly, ad-
2003;77:857– 60.
ditional work will be needed to validate the value of using VFT
10. Suzuki R, Watanabe S, Hirai Y, et al. Abdominal wall fat index, esti-
in women.
mated by ultrasonography, for assessment of the ratio of visceral fat to
In conclusion, VFT measured by ultrasonography was
subcutaneous fat in the abdomen. Am J Med 1993;95:309 –14.
11. Armellini F, Zamboni M, Rigo L, et al. Sonography detection of small
strongly correlated with the amount of visceral fat and the risk
intra-abdominal fat variations. Int J Obes Relat Metab Disord 1991;15:
factors associated with CVD. A higher VFT was found to inde-
847–52.
www.ajcn.org
pendently predict CAD and metabolic diseases in the men.
12. Ashwell M, Cole TJ, Dixon AK. Obesity: new insight into the anthro-
Therefore, we believe that VFT is a more useful index than are the
pometric classification of fat distribution shown by computed tomogra-
other ultrasonographic values or anthropometric indexes, not
phy. Br Med J 1985;292:1692– 4.
only for assessing visceral obesity but also for screening patients,
13. Executive summary of the third report of the National Cholesterol Ed-
ucation Program (NCEP) expert panel on detection, evaluation, and
by guest on April 29, 2011
particularly men, with a higher risk of CAD, metabolic diseases,
treatment of high blood cholesterol in adults (Adult Treatment Panel III).
or both.
JAMA 2001;285:2486 –97.
14. van der Meer IM, de Maat MP, Kiliaan AJ, van der Kuip DA, Hofman A,
We thank the study subjects for their participation in this study.
Witteman JC. The value of C-reactive protein in cardiovascular risk
SKK performed the study design, data collection, ultrasonographic mea-
prediction: the Rotterdam Study. Arch Intern Med 2003;163:1323– 8.
surements, and statistical data analysis, and he wrote the manuscript. HJK
15. Bjorntorp P. Portal adipose tissue as a generator of risk factors for
and KYH performed the data collection and were responsible for quality
cardiovascular disease and diabetes. Arteriosclerosis 1990;10:493– 6.
control of the ultrasonographic measurements. SHC and CWA supervised
16. Kahn SE, Prigeon RL, Schwartz RS, et al. Obesity, body fat distribution,
the data collection. SKL, KRK, and HCL provided significant expertise and
insulin sensitivity and islet beta-cell function as explanations for meta-
reviewed the manuscript. KBH provided the conception for the study. BSC
bolic diversity. J Nutr 2001;131:354S– 60S.
contributed to the design and conduct of this study and provided critical
17. Laakso M, Matilainen V, Keinanen-Kiukaanniemi S. Association of
revision of the manuscript. None of the authors had any financial, personal,
neck circumference with insulin resistance-related factors. Int J Obes
or professional conflict of interest.
Relat Metab Disord 2002;26:873–5.
18. Pouliot MC, Despres JP, Lemieux S, et al. Waist circumference and
abdominal sagittal diameter; best simple anthropometric indexes of ab-
REFERENCES
dominal visceral adipose tissue accumulation and related cardiovascular
1. Larsson B, Svardsudd K, Welin L, Wilhelmsen L, Bjorntorp P, Tibblin
risk in men and women. Am J Cardiol 1994;73:460 – 8.
G. Abdominal adipose tissue distribution, obesity, and risk of cardio-
19. Lemieux S, Prud’homme D, Tremblay A, Bouchard C, Despress JP.
vascular disease and death: 13-year follow-up of participants in the study
Anthropometric correlates to changes in visceral adipose tissue over 7
of men born in 1913. Br Med J 1984;288:1401– 4.
years in women. Int J Obes Relat Metab Disord 1996;20:618 –24.

---