Computed Axial Tomographic Scan Measurement of Abdominal Fat Distribution
and Its Correlation With Anthropometry and Insulin Secretion in
Healthy Asian Indians
C. Snehalatha, A. Ramachandran, K. Satyavani, M. Yezhisai Vallabi, and Vijay Viswanathan
Asian Indians have high insulin resistance, hyperinsulinemia, a high prevalence of diabetes, and a high waist to hip ratio
(WHR), although the rate of obesity is low. WHR and visceral fat (VF) are highly correlated, and both are associated with insulin
resistance. This study was performed to determine the normal ranges of abdominal fat distribution (subcutaneous [SF] and
VF) in nondiabetic South Indians and also to study its correlations with WHR, plasma insulin, and metabolic profiles. Fat areas
were measured by computed axial tomographic scan at the L4 to Ls level. Mean areas of SF and VF in men and women in this
study were similar to the values in white populations. Women had significantly less VF than men Gender differences were
observed in the contribution of fat areas to anthropometric, hormonal, and metabolic variables. In general, in men, total fat
(TF) area showed significant independent correlation with body mass index (BMI), WHR, and total cholesterol, and VF
correlated with insulin secretion. In women, TF and BMI were correlated and SF showed a correlation with total cholesterol.
Insulin secretion in women did not show a correlation with fat areas.
Copyright ~ 1997by W.B. Saunders Company
S
ERIAL EPIDEMIOLOGIC STUDIES in urban South Indi-
ans have shown that they have a high prevalence of
diabetes, and it is increasing. 1~4Although the study group has a
low prevalence of obesity measured as the body mass index
(BMI), an increase in BMI even in the non-obese range has been
noted to be a risk factor for diabetes, z3 Moreover. despite the
low BMI, they have high upper-body adiposity, measured as the
waist to hip ratio (WHR). WHR also was an additional risk
factor for diabetes. 2,3 Several reports have shown that for a
given BMI. Asian Indians have a higher WHR compared with
whites. 5-7 WHR and visceral fat (VF) were highly correlated,
and both were associated with insulin resistance: the risk
associated With high WHR was considered to be due to VF. 8-1°
Hitherto, there have been no studies on the direct measurement
of abdominal fat distribution in Indians In this study, we
measured the distribution of abdominal fat in the subcutaneous
(SF) and VF areas by computed axial tomographic scan in
normal healthy Indian men and women. The correlation of fat
areas with the insulin secretory response was also studied.
SUBJECTS AND METHODS
All persons in dais study underwent a standard oral glucose tolerance
test (OGTT) with 75 g glucose, and glucose tolerance was classified
according to World Health Organization criteria. 11 Forty healthy
controls (aged 36 -+ 10 years; male to female ratio, 21:19) with normal
glucose tolerance and no known family history of diabetes were
selected for the study. All of them provided written consent for the
study. All Women in this study were of premenopausal age.
AnthropOmetric measurements including height, weight, and waist
and hip girth were made by standard procedures described previously. 2,3
BMI and WHR were calculated. Systolic (SBP) and diastolic (DBP)
blood pressure were also measured.
The mean intake of calories in the study group was 1,400 to 2,250
kcalJd. All of them were habituated to a high-carbohydrate, cereal-based
diet with an approximate composition of 60% carbohydrate, 20%
protein, and 20% fat, and had no changes in the routine diet pattern at
From the Diabetes Research Centre, Royapuram, Madras, Indla.
Submitted February 14, 1997; accepted March 7, 199Z
Address reprint requests to C. Snehalatha, DSc, Diabetes Research
Centre, No 5, Main Road, Royapuram, Madras 600 013.
Copyright © 1997 by W.B. Saunders Company
0026-0495/97/4610-0023503.00/0
least 1 week before the OGTT. None of them had a change in bo0y
weight recently. During the OGTT. plasma samples were collected for
glucose, fasting lipid Itotal cholesterol, high-density lipoprotein choles-
terol [HDL-C], low-density Iipoprotein cholesterol [LDL-C], very-low-
density lipoprotein cholesterol, and triglycerides [TG]/. serum insulin.
and C-peptide (CP) measurements.
Plasma glucose level was measured by the glucose oxidase method
and lipid levels by enzymatic procedures using Boehringer Mannheim
(Mannheim. Germany) reagents and a Hitachi 704 autoanalyzer. Serum
IRI was determined by a radioimmunoassay procedure using a kit
supplied by the Bhabha Atomic Research Centre (Bombay, India). A
modified procedure of Herbert et a112 with a double antibody and PEG
precipitation was used. The lowest detection limit was 2 ~aU/mL. and
intraassay and interassay coefficients of variation (CV) were less than
5% and less than 7%. respectively. The CP level was measured Using
radioimmunoassay reagents from Diagnostic Systems Laboratories
(Texas). The intraassay CV was 5.5%. interassay CV less than 4.8%.
and lowest detection limit 0.003 pmol/mL. Areas under the curves were
calculated using Thai's formula. ~3 Ratios of insulin to glucose areas
were calculated and represented as I/G. The incremental area of insulin
at 30 minutes (30-minute value - fasting value) was divided by the
glucose Concentration at 30 minutes (IRI in picomoles and glucose in
millimoles~ and represented as AI/G for evaluating [3-cell secretion. 14-15
Fat measurements were made using a Hitachi W.450 x-ray CT
system. The details of test procedures were as follows: exposure time,
2.8 seconds; slice thickness, 10 mm; exposure factors, 90 mA, 120 kV,
Filter 2 °, attenuation values: mesenteric fat, -90 to -110; retroperito-
neal fat, -120 to -130; and SF, -130 to -145. Measurements were
taken at the L4 to L s level. Areas of fat were expressed as centimeters
squared. Mesenteric and retroperitoneal fat areas were added to obtain
the VF area. VF by SF ratios (V/S ratios) were calculated. 16VF and SF
areas were added to obtain total fat area (TF).
Statistical Analysis
Values are given as the mean _+ SD. Intergroup variations were tested
by ANOVA or t test as relevant. Multiple linear regression analyses
were performed to determine the correlation of various fat areas to
anthropometry and lipid and hormonal variables. Fat areas, insulin area,
and CP area were logarithmically transformed for regression analysis.
The SPSS package (Version 4.01; SPSS, Chicago, ILl was used for the
calculations. Pearson's correlations between the various parameters
were also determined. Data on fat measurements were tested separately
in men and women due to the differences in fat distribution.
1220 Metabolism, Vol 46, No 10 (October), 1997: pp 1220-1224