International Journal of Obesity (2001) 25, 1196±1205
ß 2001 Nature Publishing Group All rights reserved 0307±0565/01 $15.00
www.nature.com/ijo
PAPER
Weight-related differences in glucose metabolism and
free fatty acid production in two South African
population groups
C Punyadeera1, M-T van der Merwe2, NJ Crowther1*, M Toman1, AR Immelman1, GP Schlaphoff3
and IP Gray1
1
Department of Chemical Pathology, University of the Witwatersrand Faculty of Health Sciences, Johannesburg, South Africa;
Department of Endocrinology, Johannesburg General Hospital, Johannesburg, South Africa; and 3Department of Radiology,
University of the Witwatersrand Faculty of Health Sciences, Johannesburg, South Africa
2
OBJECTIVE: The effects of free fatty acids (FFA), leptin, tumour necrosis factor (TNF) a and body fat distribution on in vivo
oxidation of a glucose load were studied in two South African ethnic groups.
DESIGN AND MEASUREMENTS: Anthropometric and various metabolic indices were measured at fasting and during a 7 h oral
glucose tolerance test (OGTT). Body composition was measured using bioelectrical impedance analysis and subcutaneous and
visceral fat mass was assessed using a ®ve- and two-level CT-scan respectively. Glucose oxidation was evaluated by measuring
the ratio of 13CO2 to 12CO2 in breath following ingestion of 1-13C-labelled glucose.
SUBJECTS: Ten lean black women (LBW), ten obese black women (OBW), nine lean white women (LWW) and nine obese white
women (OWW) were investigated after an overnight fast.
RESULTS: Visceral fat levels were signi®cantly higher (P < 0.01) in obese white than black women, despite similar body mass
indexes (BMIs). There were no ethnic differences in glucose oxidation however; in the lean subjects of both ethnic groups the
area under the curve (AUC) was higher than in obese subjects (P < 0.05 for both) and was found to correlate negatively with
weight (r ˆ 70.69, P < 0.01) after correcting for age. Basal TNFa concentrations were similar in all groups. Percentage
suppression of FFAs at 30 min of the OGTT was 24 12% in OWW and 738 23% (P < 0.05) in OBW, ie the 30 min FFA
level was higher than the fasting level in the latter group. AUC for FFAs during the late postprandial period (120 ± 420 min) was
signi®cantly higher in OWW than OBW (P < 0.01) and LWW (P < 0.01) and correlated positively with visceral fat mass
independent of age (r ˆ 0.78, P < 0.05) in the OWW only. Leptin levels were higher (P < 0.01) both at fasting and during the
course of the OGTT in obese women from both ethnic groups compared to the lean women.
CONCLUSIONS: Glucose oxidation is reduced in obese subjects of both ethnic groups; inter- and intra-ethnic differences were
observed in visceral fat mass and FFA production and it is possible that such differences may play a role in the differing
prevalences of obesity-related disorders that have been reported in these two populations.
International Journal of Obesity (2001) 25, 1196 ± 1205
Keywords: glucose oxidation; free fatty acids; leptin; obesity; visceral fat; ethnic differences
Introduction
Type 2 diabetes occurs most commonly in individuals over
the age of 40 and is closely associated with obesity.1 The
prevalence of both diseases is increasing within the black
population of South Africa in conjunction with increased
*Correspondence: NJ Crowther, Department of Chemical Pathology,
South African Institute for Medical Research, University of the
Witwatersrand Faculty of Health Sciences, 7 York Road, Parktown 2193,
Johannesburg, South Africa.
E-mail: [email protected]
Received 1 June 2000; revised 31 January 2001;
accepted 2 February 2001
industrialisation and acculturation in Africa in general.2,3
Epidemiological studies have shown that the prevalence of
certain obesity-related disorders differs between black and
white urban women in South Africa: mortality from ischaemic heart disease is very rare in obese black South Africans
(8=100 000 vs 55=100 000),4 whilst hypertension (30 vs 15%)5
and type 2 diabetes (7.0 vs 3.6%)6 are more common. Therefore, a greater understanding is required of the biochemical
mechanisms mediating the relationship between obesity and
associated disorders within these two populations. This may
allow the development of methods to limit the metabolic
consequences of obesity, particularly insulin resistance.
Ethnic differences in glucose and lipid metabolism
C Punyadeera et al
A study using the euglycaemic hyperinsulinaemic clamp
technique showed obese black South African women to be
more insulin resistant in comparison with obese white
women (OWW).7 Other studies reported leptin8 and free
fatty acid (FFA)9 levels to be higher in obese black women
(OBW). Elevated levels of FFA,10 tumour necrosis factor a
(TNFa)11 and leptin12 have been suggested as key mediators
of obesity-linked insulin resistance. Both leptin13 and TNFa11
have been shown to interfere with insulin receptor kinase
activity and elevated FFA levels may cause hyperglycaemia
by inhibiting skeletal muscle glucose uptake14 and=or by
interfering with the ability of insulin to suppress hepatic
glucose output.15 FFAs reduce glucose utilisation in muscle
tissue by inhibiting insulin-dependent glucose uptake and
therefore decreasing glucose oxidation.16 ± 18
The present study was therefore designed to investigate
the effect of plasma FFA, TNFa and leptin and body fat
distribution on in vivo oxidation of a glucose load using
mass spectrometry. Metabolic differences between black
and white South African populations have only been
observed in the postabsorptive and early postprandial
phases7 ± 9 and therefore in the present study metabolic
pro®les were also investigated during the late postprandial
phase.
Methods
Subjects
Subjects were selected from an urban population of women
residing in the greater Johannesburg area according to four
selection criteria which included: age, BMI, absence of any
metabolic disorder and premenopausal status. The subjects
were subdivided into lean (body mass index (BMI) < 25) and
obese (BMI > 30) groups according to WHO criteria.19 The
study group comprised of 38 women: 19 obese subjects (10
black women (OBW), nine white women (OWW)) and 19
lean subjects (10 black women (LBW), nine white women
(LWW)). All subjects had normal liver and kidney function, a
parity of less than ®ve children and none of the women were
on oral contraceptives. Subjects with metabolic or endocrine
disorders were excluded from the study. All the subjects had
normal glucose tolerance according to WHO criteria20 and
the women were studied during the ®rst 10 days of the
follicular phase of the menstrual cycle.
The study protocol was approved by the University of
Witwatersrand, Faculty of Health Sciences Human Ethics
Committee. Each subject gave written consent after being
informed of the nature, purpose and the possible risks of the
study.
CT scan and anthropometric measurements
After a 10 ± 12 h overnight fast, the subjects were weighed
and their height measured. Lean mass (kg, %) and fat mass
(kg, %) were measured using the Bodystat 1500 bioelectrical
impedance analyser (Bodystat Ltd, Douglas, British Isles).9
The bioimpedance equations used were based on the South
African
populations
as
a
whole
(fat-free
mass
(kg) ˆ aH2=z ‡ bH2=wt ‡ cH ‡ d age ‡ e; where H is height;
wt is body mass, z is whole body impedance, and a ± e are
regression coef®cients, validated by the Dunn Nutritional
Centre, University of Cambridge, Cambridge, UK). Waist-tohip ratio (WHR) was measured by taking waist circumference
as the midpoint between the lower rib margin and the iliac
crest and hip circumference as the widest circumference of
the buttock. A ®ve-level CT-scan was performed using computerised axial tomography (CT scan, Philips SR 7000, the
Netherlands). The scan parameters were: (1) 10 mm slice
thickness; (2) 120 kV; (3) 250 mA; (4) 2 s; and (5) 34 ±
480 mm ®eld-of-view (FOV), depending on the size of the
patient. Photographic images were taken in the resting
expiration. Subcutaneous adiposity was measured using a
®ve-level CT-scan ranging from the diaphragm, umbilicus,
L4 ± 5 lumbar disc, widest diameter of the pelvis and midthigh (the total distance from the iliac crest to acetabulae
and from acetabulae to the knee joint, divided by 2). The ®rst
scanogram included the diaphragm, the umbilicus (marked
with a metallic coin), iliac crest and symphysis pubis, with a
maximal length of 500 mm from the umbilicus. The second
scanogram extended from the symphysis pubis to knees. The
widest parasagital diameter was measured at the level of L4 ±
5. The fat areas were calculated with a region of interest
seeding programme on a Philips Gyroview workstation (fat
values were chosen between 730 and 7130 HU). The areas
of the subcutaneous and visceral fat were calculated separately, and the anatomical boundaries were as described by
Kvist et al.21,22 Abdominal visceral fat was measured at the
top three levels.9
1197
Oral glucose tolerance test (OGTT) and measurement of
glucose oxidation and insulin resistance
Each subject was given a glucose load of 75 g containing 0.5 g
1-13C-glucose with 99 atom% 13C (Sigma Chemical Company, St Louis) in 200 ml of water over a 2 min period. With
the subject in the supine position, an intravenous needle was
inserted into the antecubital vein of the forearm, which was
covered with a heating blanket to ensure arterialisation of
venous blood. Blood samples were obtained at baseline, 30,
60, 120, 180, 240, 300, 360 and 420 min post-glucose load for
determination of plasma glucose, lactate and FFAs and serum
insulin, TNFa (only at basal level) and leptin. All the samples
were centrifuged at 4 C and the supernatants frozen and
stored at 770 C until analysis as a single batch to avoid
interassay variation.
Stable isotopes were used in this work to determine in vivo
oxidation of an orally administered glucose load and to
assess the differences between lean and obese subjects. The
use of 13C-labelled compounds is widespread in biomedical
research and diagnosis.23 A number of studies have used
13
C-labelled glucose to investigate carbohydrate metabolism
and shown that it is a reliable method for measuring the
International Journal of Obesity
Ethnic differences in glucose and lipid metabolism
C Punyadeera et al
1198
oxidation of glucose.24 ± 26 In addition, this method has also
been applied to demonstrate a lower glucose metabolism in
type 2 diabetic patients when compared to weight-matched
non-diabetic patients.24
Baseline 13CO2=12CO2 values were subtracted from postprandial levels to account for differences in dietary intake of
13
CO2-containing foods. Breath tests were executed over an
extended period of 7 h to accommodate the exogenous
glucose oxidation peak between 3 and 5 h postprandially.
The methodology used in this study was based on the work
by Schoeller and Klein.27 Breath samples were collected into
1.8 l Douglas bags (to minimise the diffusion of gases) before
glucose ingestion and 30, 60, 90, 120, 150, 180, 210, 240,
300, 360 and 420 min after the glucose load. The CO2 and
water were separated from the exhaled breath samples by
trapping in liquid nitrogen. The CO2 was then evaporated
into the apparatus while water remained trapped in a mixture of methanol and dry ice. The 13C=12C ratios of the
isolated CO2 were determined using an AEI MS-20 double
collector isotopic ratio mass spectrometer. The 13C=12C ratio
of sample was compared to that of commercially available
CaCO3 standard (Analar Reagent) using the formula of
Craig:28
13
13
12
C …del per mil† ˆ …… C= C of sample†=
…13 C=12 C of standard†† ÿ 1† 1000
All the results were expressed according to an international
Solenhofen standard29 and were used as a measure of in vivo
oxidation of the glucose load.
Biochemical analyses
Glucose, FFA and lactate levels were measured using commercially available enzymatic colorimetric methods (Boehringer Mannheim, Mannheim, Germany). Inter- and intraassay coef®cients of variation (CV) for glucose were 0.9 and
1.8%, respectively. Inter- and intra-assay CV for both FFA
and lactate were < 5%. Insulin levels were measured by
enzyme-ampli®ed immunoassay (Mercodia, Uppsala,
Sweden); the lower limit of sensitivity for the ELISA was
7.3 pmol=l and the intra- and inter-assay coef®cients of
variations were 4 and 3.6%, respectively. Serum TNFa was
measured using an enzyme immunometric assay (Chiron
Diagnostics, Los Angeles, USA) and leptin by a radioimmunoassay method (Linco, St Charles, Missouri, USA). The
intra- and inter-assay coef®cients of variation for TNFa
were 3.6 and 5.3%, respectively and for leptin were 4.7 and
3.6%, respectively.
Statistical analysis
Statistical analyses were carried out using Statistica version 5
and SPSS version 6.1 for Windows (Chicago, IL). A two-way
analysis of covariance (ANCOVA) model was used to invesInternational Journal of Obesity
tigate the effects of BMI and ethnicity on the variables
measured and to test for interactions between these factors;
age was found to be signi®cantly different between obese
and lean groups and was therefore included as a covariate.
One-way ANCOVA was used to measure differences between
means for the four study groups correcting for age in all
comparisons and correcting for age and weight when comparing OBW with OWW. Partial correlation analyses were
used to test whether correlation existed in any of the subgroups and in the full combined group. Differences in
correlations between groups were examined using multiple
regression analysis with age and weight included as covariates. Total levels of FFAs, leptin, glucose, insulin and glucose
oxidation during the 7 h OGTT were calculated by measuring
area under the curve (AUC) using the trapezoid rule.
The number of subjects in each study group was suf®cient
to allow us to detect statistically and clinically signi®cant
differences (see Results section) in the metabolic variables
under investigation.
Results
Anthropometric measurements
The body composition of the two lean groups was similar
(Table 1). In the obese subjects WHR (P < 0.05), fat mass
(P < 0.05) and weight (P ˆ 0.05) were higher in OWW than
OBW. Visceral fat area was also higher in OWW than OBW
(P < 0.01) even after adjusting for total body fat in a one-way
ANCOVA. Two-way ANCOVA showed that visceral fat levels
were higher in white than black females (F ˆ 30.0;
P < 0.0001) and that there was a signi®cant interaction
between race and BMI (F ˆ 20.5; P < 0.0001). Obese subjects
were older than lean in both ethnic groups, and therefore
age was included as an independent variable in all the
relevant statistical analyses. In addition, the two-way
ANCOVA showed the white women to be heavier than the
black women (P < 0.05) and therefore weight was corrected
for in the multiple regression analyses. However, the major
indices of obesity, ie BMI and percentage body fat were
similar between the lean and obese women from both
ethnic groups (Table 1).
Glucose, insulin and lactate levels
Glucose levels in LBW were signi®cantly lower than in OBW
at fasting (P < 0.05) and 60 min (P < 0.05) and lower than
LWW at 30 (P < 0.05) and 60 min (P < 0.05; Figure 1A).
Fasting glucose levels of LWW were lower than those of
OWW (P < 0.05). Total glucose levels (AUC) were lower in
LBW compared to OBW (P < 0.01) and LWW (P < 0.01). Total
glucose levels in OWW were not different from LWW (Table
2A). Two-way ANCOVA showed that there was an effect of
ethnicity on total glucose levels (P < 0.05) and that there
was an interaction between BMI and ethnicity (P < 0.05;
Table 2B).
Ethnic differences in glucose and lipid metabolism
C Punyadeera et al
Table 1 The anthropometric data for the four groups of women who participated in the study using age as a covariate
Lean white women
(n ˆ 9)
a
Age (y)
Weight (kg)
BMI (kg=m2)
Waist circumference (cm)
Hip circumference (cm)
WHR
Fat (kg)
Percentage fat (%)
Lean (kg)
Lean percentage (%)
2
Subcutaneous fat (cm )
2
Visceral fat (cm )
31.4 1.4
(33.0)
a
58.0 2.1
(57.3)
21.5 1.0a
(20.1)
a
65.6 1.2
(65)
a
94.4 1.8
(94.0)
0.69 0.01a
(0.69)
15.7 1.5a
(16.0)
a
26.7 1.9
(26.2)
42.3 1.3a
(42.2)
73.3 1.9a
(73.8)
a
171.4 23.4
(164.7)
a
24.1 7.9
(14.9)
Lean black women
(n ˆ 10)
b
32.1 2.6
(28.5)
b
53.1 2.8
(54.8)
20.7 0.7b
(20.6)
b
63.5 1.2
(65)
b
93.7 2.2
(95.5)
0.68 0.01b
(0.68)
16.8 1.3b
(15.5)
b
31.25 1.2
(30.5)
36.7 1.9b
(37.1)
68.8 1.2b
(69.5)
b
146.2 15.6
(140.5)
b
18 3.8
(16.2)
a
b
Obese white women
(n ˆ 9)
Obese black women
(n ˆ 10)
44.1 3.4
(42.0)
101.8 4.3
(103.5)
39.4 1.7
(39.6)
105.2 2.8
(104)
126.6 3.8
(126.0)
0.83 0.01c
(0.84)
51.5 3.4c
(51.8)
49.7 1.8
(51.0)
51.5 1.4
(51.2)
50.3 21.8
(49.0)
549.0 42.6
(475.7)
b
139.7 10.7
(128.1)
44.2 2.8
(42.5)
90.5 3.2
(86.6)
36.5 1.6
(36.1)
93.2 3.7
(94)
123.3 3.1
(124.5)
0.76 0.02
(0.76)
41.8 2.6
(41.2)
47.2 2.1
(46.1)
47.3 1.46
(47.0)
52.6 2.1
(53.0)
512.4 32.0
(499.5)
72.3 3.9
(70.6)
1199
c
Mean values s.e.m.: median values are given in parentheses; P < 0.01 vs OWW; P < 0.01 vs OBW; P < 0.05 vs OBW.
Fasting insulin levels were higher in OBW compared to
LBW (P < 0.05) and were also higher at 420 min (P < 0.05;
Figure 1B). The insulin levels in OWW were higher than
LWW at 30 min (P < 0.05) but lower at 300 min (P < 0.05).
LWW had higher insulin levels compared to LBW at 420 min
(P < 0.01). Two-way ANCOVA demonstrated a BMI effect
(P < 0.05) on total insulin (AUC) levels (Table 2A).
Fasting lactate levels were not different between groups
but postprandial concentrations (Figure 1C) were higher in
lean than obese women with signi®cant differences at 180,
240 and 300 min (P < 0.05 for all) in white women and at 180
and 240 min in black women (P < 0.05 for both). Two-way
ANCOVA con®rmed these results with a signi®cant BMI
effect (P < 0.05) on total (AUC) lactate levels (Table 2C).
The percentage lactate increase from the basal to the peak
value at 60 min in the four groups were as follows: LWW,
245 28%; LBW, 138 12%; OBW, 156 24%; and OWW,
144 19%. Signi®cant differences in percentage lactate
increase were observed between LWW and LBW (P < 0.05)
and LWW and OWW (P < 0.05).
FFA levels
There was a 24 12% suppression of FFA levels after 30 min
of the OGTT in the OWW which was not different from that
observed in the LWW (16 13%) but was signi®cantly higher
(P < 0.05) than in the OBW (738 23%; Figure 1D) in whom
there was no suppression. The percentage suppression of FFA
levels in the LBW (35.8 12%) was signi®cantly higher than
that observed in the OBW (P < 0.01). Two-way ANCOVA
demonstrated a signi®cant interaction between ethnicity
and BMI (F ˆ 6.3; P < 0.05) on percentage suppression of
FFA levels at 30 min.
The OWW had higher postprandial FFA levels at 240
(P < 0.05), 300 (P < 0.01) and 360 (P < 0.05) min than OBW
and higher levels at 240 (P < 0.05) and 300 (P < 0.01) min
than LWW (Figure 1D). The LWW and LBW had similar FFA
concentrations throughout the study. The OBW showed
signi®cantly higher FFA values at 30 (P < 0.01) and 60
(P < 0.01) min compared to the LBW.
Total (AUC) FFA levels between 0 ± 420 min of the OGTT
were signi®cantly higher in OWW than OBW and LWW
(P < 0.01 for both; Table 2D) and a similar relationship
existed for FFA levels (AUC) between 120 and 420 min
(Table 2E). Furthermore for total FFA levels between 0 and
420 and between 120 and 420 min two-way ANCOVA
demonstrated signi®cant ethnicity effects (P < 0.01 for
both) and an interaction between BMI and ethnicity
(P < 0.05 for both) and for the former variable there was
also a BMI effect (P < 0.05; see Table 2D, E). Total (AUC) FFA
levels between 0 and 120 min of the OGTT were signi®cantly
higher in OBW (32.2 5.9 mmol=l min) than LBW
(17.0 3.9 mmol=l min; P < 0.05), but there were no differences between OWW (35.0 4.6 mmol=l min) and LWW
(23.8 4.3 mmol=l min); two-way ANCOVA demonstrated
a BMI effect only (F ˆ 7.0; P < 0.05).
Total FFA (0 ± 420 min) correlated with visceral fat
(r ˆ 0.82, P < 0.05, n ˆ 9) independent of age in OWW only.
International Journal of Obesity
Ethnic differences in glucose and lipid metabolism
C Punyadeera et al
1200
Figure 1 Levels of (A) glucose (B) insulin (C) lactate and (D) FFAs during a 7 h OGTT in: (black triangle) lean black women; (white triangle) lean white women;
(black square) obese black women; (white square) obese white women. *P < 0.05 and **P < 0.01 for obese black vs lean black women; ‡ P < 0.05 and
‡ ‡ P < 0.01 for obese white vs lean white women; §P < 0.05 and §§P < 0.01 for obese white vs obese black women; #P < 0.05 and ##P < 0.01 for lean white vs
Late postprandial FFA levels (ie AUC for FFAs between 120
and 420 min) were the main contributors to this relationship
(r ˆ 0.78, P < 0.05) with no correlation found between early
postprandial FFA levels (ie AUC for FFAs between 0 and
120 min) and visceral fat area.
In vivo oxidation of a glucose load
There were no interethnic differences in glucose oxidation
levels (Figure 2). Glucose oxidation in the lean subjects of
both ethnic groups was higher than in the obese subjects
with signi®cant differences from 90 min onwards in both
ethnic groups. Total glucose oxidation levels (AUC) were
signi®cantly higher in LWW (12066 336 del per mil
International Journal of Obesity
(d) min) than OWW (8948 702 d min; P < 0.05) and in
LBW (12457 644 d min) than OBW (9183 507 d min;
P < 0.05). Two-way ANCOVA further demonstrated the signi®cant effect of BMI on total glucose oxidation level
(F ˆ 15.5; P < 0.0005). When all subject groups were combined, negative correlations were observed between glucose
oxidation (AUC) and weight (r ˆ 70.69, P < 0.01, n ˆ 38),
BMI (r ˆ 70.59, P < 0.01, n ˆ 38), percentage body fat
(r ˆ 70.52, P < 0.01, n ˆ 37), subcutaneous fat (r ˆ 70.63,
P < 0.01, n ˆ 30) and visceral fat (r ˆ 70.51, P < 0.01,
n ˆ 30), all corrected for age. In addition AUC glucose oxidation correlated negatively with both AUC leptin (r ˆ 70.50,
P < 0.01, n ˆ 34) and AUC FFA (r ˆ 70.40, P < 0.05, n ˆ 36)
correcting for age in a multiple linear regression analysis but
Ethnic differences in glucose and lipid metabolism
C Punyadeera et al
Table 2 Two-way ANCOVA of total levels of (A) glucose (B) insulin (C) lactate (D) FFA between 0 and 420 min of an OGTT and (E) FFA between 120 and
420 minutes of an OGTT
Lean
Obese
Overall
2358 66
a
2342 49
2350 40
2360 38
2224 48
99 7
82 12
90 7
113 20
115 16
114 12
C. Lactate (mmol=l min)
White women
Black women
Overall
355 28
376 26b
370 19
D. FFA between 0 and 420 min of the
OGTT (mmol=l min)
White women
Black women
Overall
E. FFA between 120 and 420 min of
the OGTT (mmol=l min)
White women
Black women
Overall
A. Total glucose (mmol=l min)
White women
Black women
Overall
B. Total insulin (nmol=l min)
White women
Black women
Overall
F-value
P-value
Ethnicity
BMI
Interaction
5.7
1.7
4.4
0.02
0.20
0.04
106 10
98 11
Ethnicity
BMI
Interaction
0.2
6.5
0.5
0.66
0.02
0.50
286 11
313 20
301 12
309 15
345 17
Ethnicity
BMI
Interaction
1.01
4.8
0.01
0.32
0.04
0.91
160 11c
140 11
150 8
237 24
160 7c
198 15
198 15
150 7
Ethnicity
BMI
Interaction
12.4
4.4
4.3
0.001
0.04
0.04
136 9c
123 10
130 7
201 20
128 7c
164 14
169 13
126 6
Ethnicity
BMI
Interaction
12.5
2.1
6.1
0.001
0.15
0.02
2362 37
2106 65
2234 49
a
Factors
1201
Mean values s.e.m.; aP < 0.01 vs LBW; bP < 0.05 vs OBW; cP < 0.01 vs OWW.
Leptin levels
Fasting and all postprandial leptin levels (Figure 3) were
signi®cantly higher (P < 0.01 for all) in both obese groups
compared to the two lean groups. Total (AUC) leptin levels
were signi®cantly higher in OBW (17.6 2.0 mg=ml min)
than LBW (5.7 1.9 mg=ml min; P < 0.01) and in OWW
(15.5 1.5 mg=ml min) than LWW (4.7 0.9 mg=ml min;
P < 0.01). Two-way ANCOVA also demonstrated a very signi®cant BMI effect (F ˆ 22.4; P < 0.0001). When all subject
groups were combined total leptin (AUC) correlated with
percentage body fat (r ˆ 0.70, P < 0.01, n ˆ 33), weight
(r ˆ 0.72, P < 0.01, n ˆ 34), BMI (r ˆ 0.68, P < 0.01, n ˆ 34),
subcutaneous fat (r ˆ 0.75, P < 0.01, n ˆ 30) and visceral fat
levels (r ˆ 0.39, P < 0.05, n ˆ 30).
Figure 2 Glucose oxidation levels during course of 7 h OGTT in: (black
triangle) lean black women; (white triangle) lean white women; (black
square) obese black women; (white square) obese white women.
*P < 0.05 and **P < 0.01 for lean black vs obese black women;
‡ P < 0.05 and ‡ ‡ P < 0.01 for lean white vs obese white women.
TNFa levels
There were no inter- or intraethnic differences in fasting
TNFa levels, which were as follows: LWW, 12.3 1.3; OWW,
18.9 4.2; LBW, 13.5 0.84; OBW, 12.3 1.7 ng=l. TNFa
levels correlated with weight (r ˆ 0.53, P < 0.01, n ˆ 36) and
BMI (r ˆ 0.44, P < 0.01, n ˆ 36), subcutaneous fat (r ˆ 0.44,
P < 0.01, n ˆ 36) and visceral fat area (r ˆ 0.43, P < 0.05,
n ˆ 36) in the combined subject groups, correcting for age.
signi®cance was lost in both relationships when weight was
included as an independent variable. In addition, AUC
insulin correlated positively with AUC glucose oxidation in
lean (r ˆ 0.66, P < 0.01, n ˆ 19) but not obese women when
corrected for age and weight.
Discussion
The present study demonstrates that oxidation of a glucose
load is greater in lean than obese women of both ethnic
International Journal of Obesity
Ethnic differences in glucose and lipid metabolism
C Punyadeera et al
1202
Figure 3 Leptin levels during course of 7 h OGTT in: (black triangle)
lean black women; (white triangle) lean white women; (black square)
obese black women; (white square) obese white women. **P < 0.01 for
obese black vs lean black women; ‡ ‡ P < 0.01 for obese white vs lean
groups, and correlates negatively with BMI and percentage
body fat. One possible explanation for these results is that
the obese subjects have a larger bicarbonate pool than the
lean subjects and therefore retain more of the 13CO2 in this
pool. However, this is unlikely because a study has shown
that following ingestion of 1-14C-glucose, after accounting
for bicarbonate pool size, 14CO2 production in obese subjects
was half that of non-obese subjects.30 Furthermore, two
previous studies have shown that after an oral dose of
1-13C-glucose 13CO2 breath levels were signi®cantly lower
in obese type 2 diabetic patients than weight-matched nondiabetic obese subjects.24,31 These results cannot be
explained by differences in bicarbonate pool size because
the subjects were weight matched, and therefore must be due
to reduced insulin sensitivity in the diabetics. This may
explain the negative association between weight and glucose
oxidation observed in the present study since there is a welldocumented association between obesity and reduced insulin sensitivity.1 This is demonstrated by the ®nding that total
(AUC) levels of insulin correlated positively and signi®cantly
with total glucose oxidation in lean but not obese women.
One interpretation of this association is that adequate glucose oxidation is maintained by insulin in subjects who are
lean, but the ability of insulin to stimulate glucose uptake
and hence increase glucose oxidation is reduced in subjects
who are obese.
This interpretation is supported by the postprandial lactate levels which were signi®cantly lower in both groups of
obese than in lean women. This may be due to higher insulin
resistance in the obese subjects, leading to reduced entry of
glucose into the glycolytic pathway and hence lower lactate
production.32 Similar results were observed by Lovejoy
International Journal of Obesity
et al,33 who demonstrated that diminished postprandial
lactate appearance correlated with degree of insulin resistance. The lower postprandial lactate increase of 138% in the
LBW, compared to the 245% increase in LWW, would suggest
a greater degree of insulin resistance in LBW and yet glucose
levels tend to be lower in LBW than LWW. Since adipose
tissue is a signi®cant source of lactate in the postprandial
period32,34,35 whereas muscle is not,36,37 it may be that LBW
have greater adipose tissue insulin resistance. These hypotheses obviously need to be tested but a differential degree of
tissue sensitivity to insulin has previously been observed by
Hansen et al (University of Maryland, personal communication) in a primate model of obesity and in studies showing
that in obese and type 2 diabetic subjects the levels of
glucose transporter 4 (GLUT-4) fall in adipocytes but
remain the same in skeletal muscle.38,39 In addition, insulin
dose ± response curves for stimulation of glucose uptake,
suppression of glucose production and suppression of lipolysis has been shown before.38,39 Lipolysis is the most insulin
sensitive process, followed by glucose production and only
then by glucose uptake (with EC50's in the physiological
range only for lipolysis and glucose production).41,42 Since
failure to adequately turn off lipolysis will directly affect liver
and muscle metabolism, whilst the opposite does not hold, it
may be that adipose tissue will be the primary site for the
defect, leading to insulin resistance elsewhere.40 Plasma
glucose levels (AUC) in LBW were lower than those in
LWW, yet insulin levels and glucose oxidation rates were
very similar. This may imply that, for a given amount of
insulin, the LBW utilise glucose more ef®ciently at the level
of muscle and liver compared to the LWW. The higher
plasma glucose levels in OBW compared to LBW may be
partly explained by lower glucose oxidation rates in the
OBW and greater tissue insulin resistance caused by higher
FFA levels, particularly during the early postprandial phase of
the OGTT.
The immediate suppression of FFAs after administration
of a glucose load, is primarily due to insulin's inhibitory
action on hormone-sensitive lipase.43,44 This insulin effect
was signi®cantly reduced in the OBW compared to OWW
and a reduced sensitivity of isolated adipocytes to insulininhibition of lipolysis has been demonstrated previously in
OBW.45 During the late postprandial period of the extended
OGTT, FFA concentrations were higher in the OWW than
OBW and LWW. This may arise as a result of increased
intravascular triglyceride hydrolysis by lipoprotein lipase
activity (LPL) or net outward movement of FFA from adipose
tissue due to increased hormone-sensitive lipase (HSL) activity or both. The possible in¯uence of acylation stimulating
protein46 in this context has not been studied in OBW.
Insulin levels were similar in OWW and OBW during the
late postprandial phase and as regards counterregulation,
fasting catecholamine concentrations have been studied in
these ethnic groups before and were found to be similar.9
Whether there are differences in late postprandial concentrations of catecholamines is not known. Regardless of the
Ethnic differences in glucose and lipid metabolism
C Punyadeera et al
mechanism, higher late postprandial FFA concentrations
could predispose the OWW to ischaemic cardiovascular
disease.47
If the movement of FFA into and out of adipose tissue is
governed by concentration gradients that can change in
direction, it is likely that FFA delivery from adipose tissue
will partly be governed by blood ¯ow through this tissue.43 A
recent study has shown that blood ¯ow in subcutaneous
abdominal and femoral fat depots is higher in black than
white obese women.9 Therefore, FFAs may be cleared more
slowly from the peripheral circulation in the OWW and this
may contribute to the higher late postprandial FFA levels. If
this leads to the lower glucose oxidation rates observed in
obese subjects then suppression of glucose oxidation should
be higher in the white compared to OBW. This was not the
case. It is therefore possible that OBW are more sensitive to
FFA-induced inhibition of glucose uptake and glycolysis than
OWW. This needs to be investigated by performing in vivo
13
C-palmitate infusion experiments in both ethnic groups.
Studies have demonstrated that high FFA levels can reduce
glucose uptake and oxidation.16 ± 18 In the present investigation total (AUC) FFA levels did correlate negatively with
glucose oxidation but not independently of weight. This
®nding is not unexpected, as FFA concentrations are largely
in¯uenced by the amount of adipose tissue and not only
insulin resistance per se.10
Visceral fat accumulation was greater in the OWW than
the OBW. This is not simply a result of the higher fat mass of
the OWW since one-way ANCOVA shows a signi®cant difference even after correcting for fat mass. Furthermore, in a
previous study we have shown that in subjects matched for
fat mass visceral fat area is greater in OWW than OBW.8 This
fat depot is known to be more lipolytically active and less
sensitive to the action of insulin than subcutaneous fat48 and
total (AUC) late postprandial FFA levels in the OWW correlated with visceral but not subcutaneous fat mass. This
con®rms an earlier study showing postprandial FFA levels
were increased in men with high visceral fat mass.49
Even though leptin levels tended to be higher in OBW
compared to OWW, differences were not signi®cant; a previous study has shown higher postabsorptive leptin levels in
OBW than OWW.8 Leptin levels correlated negatively with
glucose oxidation in both ethnic groups but not independently of weight. The data from other studies investigating
the effects of leptin on glucose tolerance are contradictory.
In rodents, leptin can increase both glucose utilisation50 and
insulin sensitivity;51 other studies have shown that leptin
can interfere with insulin receptor signalling13 and does not
directly affect glucose transport in adipocytes or muscle
cells.52 Leptin levels were found to correlate more strongly
with subcutaneous than visceral fat, supporting other studies
that showed that leptin mRNA levels are higher in subcutaneous compared to visceral fat depots.53,54 TNFa levels of
both ethnic groups were similar with no difference between
obese and lean women. This study does not rule out the
possibility that TNFa may act locally to reduce insulin
sensitivity.
Analysis of data from the present study by two-way
ANCOVA demonstrated that levels of glucose, FFA and
visceral fat area were signi®cantly in¯uenced by ethnicity
and that for each of these variables interaction existed
between ethnicity and BMI. Thus, the greater increase in
visceral fat area and total FFA levels observed in the OWW
may predispose these subjects to cardiovascular disease4
whilst the more pronounced decrease in glucose tolerance
in the OBW may predispose this population to type 2
diabetes.6
In summary, this study demonstrates a negative association between adiposity and glucose oxidation rates and
furthermore, OBW may experience suppression of glucose
oxidation at a lower level of plasma FFA than OWW. The
former group also have poorer suppression of early postprandial FFA concentrations than OWW whilst the OWW
have higher late postprandial FFA levels than the OBW. The
high visceral fat and postprandial FFA levels in conjunction
with the elevated triglyceride concentrations of the postabsorptive55,56 and the late postprandial57 periods may explain
the higher incidence of ischeamic heart disease in the white
than black population.4 It therefore seems that in a country
undergoing rapid industrialisation, both glucose and lipid
toxicity play a role in the pathogenesis of the co-morbid
diseases of obesity. However, different ethnic groups are not
necessarily affected by both to the same degree.
1203
Acknowledgements
This research was funded by the South African Institute for
Medical Research (SAIMR), the South African Sugar Association, the University of the Witwatersrand Research Committee, Servier and Roche Pharmaceuticals. The authors wish to
thank Sisters Nomsa Ramela and Nancy Holden for carrying
out the OGTTs. We would also like to thank the SAIMR
Chemical Pathology routine laboratories for their excellent
technical assistance and Professor John Kalk for his much
appreciated help.
References
1 Walker M. Obesity, insulin resistance, and its link to non-insulindependent diabetes mellitus. Metabolism 1995; 44(Suppl 3): 18 ±
20.
2 A SEMDSA Consensus Document, 1997, in association with
DESSA, SADA, ADSA. Type II diabetes mellitus clinical guidelines
at primary health care level. S Afr Med J 1997; 87: 493 ± 512.
3 Rossi-Espagnet A, Goldstein GB, Tabibzadah I. Urbanisation and
health in developing countries: a challenge for health for all.
World Health Stat Q 1991; 44: 185 ± 244.
4 Isles CG, Milne FJ. Low mortality from ischaemic heart disease
among urban Blacks in South Africa. J Clin Hypertens 1987; 3:
749 ± 775.
5 Seedat YK. Race, environment and blood pressure: the South
African experience. J Hypertens 1997; 1: 7 ± 12.
International Journal of Obesity
Ethnic differences in glucose and lipid metabolism
C Punyadeera et al
1204
6 Joffe BI, Seftel HC. Diabetes mellitus in the Black communities of
Southern Africa. J Intern Med 1994; 235: 137 ± 142.
7 van der Merwe M-T, Crowther NJ, Schlaphoff G, Gray IP, Joffe BI,
È nnroth PN. Evidence for insulin resistance in Black women
Lo
from South Africa. Int J Obes Relat Metab Disord 2000; 24: 1340 ±
1346.
8 van der Merwe M-T, Panz VR, Crowther NJ, Schlaphoff GP, Gray
È nnroth PN. Free fatty acids and insulin
IP, Froguel P, Joffe BI, Lo
levels Ð relationship to leptin levels and body composition in
various patient groups from South Africa. Int J Obes Relat Metab
Disord 1999; 23: 1 ± 9.
9 van der Merwe M-T, Crowther NJ, Schlaphoff G, Boyd IH, Gray IP,
È nnroth PN. Lactate and glycerol release from the
Joffe BI, Lo
subcutaneous adipose tissue of obese urban women from South
Africa, important metabolic implications. J Clin Endocrinol Metab
1998; 83: 4084 ± 4091.
10 Boden G. Role of fatty acids in the pathogenesis of insulin
resistance and NIDDM. Diabetes 1997; 46: 3 ± 10.
11 Hotamisligil GS, Peraldi P, Spiegelman BM. The molecular link
between obesity and diabetes. Curr Opin Endocrinol Diab 1996; 3:
16 ± 23.
12 Muller G, Ertl J, Gerl M, Preibbisch G. Leptin impairs metabolic
actions of insulin in isolated rat adipocytes. J Biol Chem 1997;
272: 10585 ± 10593.
13 Cohen B, Novick D, Rubinstein M. Modulation of insulin activities by leptin. Science 1996; 274: 1185 ± 1188.
14 Kelley DE, Mokan M, Simoneau JA, Mandarino LJ. Interaction
between glucose and free fatty acid metabolism in human skeletal muscle. J Clin Invest 1993; 92: 91 ± 98.
15 Fanelli C, Calderone S, Epifano L, DeVincenzo A, Modarelli F,
Pampanelli S, Perriello G, DeFeo P, Brunetti P, Gerich JE, Bolli GB.
Demonstration of a critical role for free fatty acids in mediating
counterregulatory stimulation of gluconeogenesis and suppression of glucose utilisation in humans. J Clin Invest 1993; 92:
1617 ± 1622.
16 Randle PJ, Garland PB, Hales CN, Newsholme EA. The glucosefatty acid cycle. Its role in insulin sensitivity and the metabolic
disturbances of diabetes mellitus. Lancet 1963; i: 785 ± 789.
17 Boden G, Chen X, Ruiz J, White JV, Rossetti L. Mechanisms of
fatty acid-induced inhibition of glucose uptake. J Clin Invest 1994;
93: 2438 ± 2446.
18 Thiebaud D, DeFronzo RA, Jacot E, Golay A, Acheson K, Maeder
E, Jequier E, Felber J-P. Effect of long chain triglyceride infusion
on glucose oxidation in man. Metab Clin Exp 1982; 31: 1128 ±
1136.
19 World Health Organization Expert Committee. Physical status: the
use and interpretation of anthropometry. WHO Technical Report
Series no. 854. WHO: Geneva; 1995.
20 World Health Organization. Diabetes mellitus. Report of a WHO
study group. WHO Technical Report Series no. 727. WHO:
Geneva; 1985.
È strom L, Tylen U. Adipose tissue volume determina21 Kvist H, Sjo
tions in women by computerised tomography. Int J Obes 1986;
10: 53 ± 67.
È strom L. Total and
22 Kvist H, Chowbury B, Grangard U, Tylen U, Sjo
visceral adipose tissue volumes derived from measurements with
computed tomography in adult men and women: predictive
equations. Am J Clin Nutr 1988; 48: 1351 ± 1361.
23 Matwiyoff NA, Ott DG. Stable isotope tracers in the life sciences
and medicine. Science 1973; 181: 1125 ± 1133.
24 Lefebvre PL, Mosora F, Lacroix M, Luyckx A, Lopez-Habib
G, Duchesne J. Naturally labelled 131-C-glucose: metabolic
studies in human diabetes and obesity. Diabetes 1975; 24: 185 ±
189.
25 Mosora F, Lefebvre P, Pirany F, Lacroix, Luycks A, Duchesne J.
Quantitative evaluation of the oxidation of an exogenous glucose
load using naturally labelled 13C-glucose. Metabolism 1976; 25:
1575 ± 1582.
26 Rating D, Langhans CD. Breath test; concepts, application and
limitation. Eur J Pediatr 1997; 156(Suppl 1): S18 ± S23.
International Journal of Obesity
27 Schoeller DA and Klein PD. A simpli®ed technique for collecting
breath CO2 for isotopic ratio mass spectrometry. Biomed Mass
Spectrom 1978; 5: 29 ± 31.
28 Craig H. Isotopic standards for carbon and oxygen and correction
factors for mass spectrometric analysis of carbon dioxide. Geo
phys Cosmo Acta 1957; 12: 133 ± 149.
29 Schoeller DA, Schneider JF, Solomons NW, Watkins JD, Klein PD.
Clinical diagnostics with the stable isotope 13C in CO2 breath
tests: methodology and fundamental considerations. J Lab Clin
Med 1977; 90: 412 ± 420.
30 Shreeve WW. Transfer of carbon-14 and tritium from substrates to
CO2, water and lipids in obese and diabetic subjects in vivo. Ann
NY Acad Sci 1965; 131: 464 ± 469.
31 Punyadeera C, Crowther NJ, van der Merwe M-T, Galpin J, Gray
IP. Comparison of glucose metabolism in South African obese and
diabetic women. (Abstract.) J Endocrinol Metab Diab S Afr 2000; 5:
42.
32 Di Girolamo M, Newby FD, Lovejoy J. Lactate production in
adipose tissue: a regulated function with extra-adipose implications. FASEB J 1992; 6: 2405 ± 2412.
33 Lovejoy J, Newby FD, Gebhart SS, Di Girolamo M. Insulin
resistance in obesity is associated with elevated basal lactate
levels and diminished lactate appearance following intravenous
glucose and insulin. Metabolism 1992; 41: 22 ± 27.
È rntorp P. Glucose uptake
34 MaÊrin P, RebuffeÂ-Scrive M, Smith U, Bjo
in human adipose tissue. Metabolism 1987; 36: 1154 ± 1160.
È m E, Arner P, Ungerstedt U, Bolinder J. Subcutaneous
35 Hagstro
adipose tissue: a source of lactate production after glucose ingestion in humans. Am J Physiol 1990; 258: E888 ± E893.
36 Jackson RA, Hamling JB, Sim B, Hawa MI, Blix PM, Nabarro JD.
Peripheral lactate and oxygen metabolism in man: the in¯uence
of oral glucose loading. Metabolism 1987; 36: 144 ± 150.
37 Yki-Jarvinen H, Bogardus C, Foley JE. Regulation of plasma
lactate concentration in resting human subjects. Metabolism
1990; 39: 859 ± 864.
38 Kahn BB. Facilitative glucose transporters: regulatory mechanisms and dysregulation in diabetes. J Clin Invest 1992; 89: 1367 ±
1374.
39 Abel ED, Shepherd PR, Kahn BB. Glucose transporters and pathological states. In: Le Roith D, Taylor SI, Olefsky JM (eds). Diabetes
mellitus: a fundamental and clinical text. Lippincott-Raven: Philadelphia, PA; 1996. pp 530 ± 543.
40 Campbell PJ, Carlson MG, Hill JO, Nurjhan N. Regulation of free
fatty acid metabolism by insulin in humans: role of lipolysis and
re-estri®cation. Am J Physiol 1992; 263: E1063 ± E1069.
41 Bonadonna RC, Groop L, Kramer N, Ferrannini E, Del Prato S, Del
Fronzo RA. Obesity and insulin resistance in humans: a dose
response study. Metabolism 1990; 39: 452 ± 459.
42 Stumval M, Jacob S. Multiple sites of insulin resistance: muscle,
liver and adipose tissue. Exp Clin Endocrinol Diabetes 1999; 107:
107 ± 110.
43 Frayn KN. Regulation of fatty acid delivery in vivo. In: Richter T
(ed). Skeletal muscle metabolism in exercise and diabetes. Plenum
Press: New York; 1998. pp 171 ± 179.
44 Coppack SW, Evans RD, Fisher RM, Frayn KN, Gibbons GF,
Humphreys SM, Kirk ML, Potts JL, Hockaday TDR. Adipose
tissue metabolism in obesity: lipase action in vivo before and
after a mixed meal. Metabolism 1992; 41: 264 ± 272.
È nnroth PN, Gray IP,
45 Buthelezi EP, van der Merwe M-T, Lo
Crowther NJ. Ethnic differences in the responsiveness of adipocyte lipolytic activity to insulin. Obes Res 2000; 8: 171 ± 178.
46 Sniderman AD, Cian¯one K, Arner P, Summers LKM, Frayn KN.
The adipocyte, fatty acid trapping and atherogenesis. Arterioscler
Thromb Vasc Biol 1998; 18: 147 ± 151.
47 Patsch JR, Miesenbock G, Hopferwieser T, Muhlberger V, Knapp E,
Dunn JK, Gotto AM, Patsch W. Relation of triglyceride metabolism and coronary artery disease. Studies in the postprandial
state. Arterioscler Thromb Vasc Biol 1992; 12: 1336 ± 1345.
48 Arner P. Difference in lipolysis between human subcutaneous and
omental adipose tissue. Ann Med 1995; 27: 435 ± 438.
Ethnic differences in glucose and lipid metabolism
C Punyadeera et al
49 Couillard C, Bergeron N, Prud'homme D, Bergeron J, Tremblay
A, Bouchard C, MaurieÁge P, DespreÂs J-P. Postprandial triglyceride response in visceral obesity in men. Diabetes 1998; 47:
953 ± 960.
50 Kamohara S, Burcelin R, Halaas JL, Friedman JM, Charron MJ.
Acute stimulation of glucose oxidation in mice by leptin treatment. Nature 1997; 389: 374 ± 377.
51 Sivitz WI, Walsh SA, Morgan DA, Thomas MJ, Haynes WG. Effects
of leptin on insulin sensitivity in normal rats. Endocrinology 1997;
138: 3395 ± 3401.
52 Zierath JR, Frevert EU, Ryder JW, Berggren PO, Kahn BB. Evidence
against a direct effect of leptin on glucose transport in skeletal
muscle and adipocytes. Diabetes 1998; 47: 1 ± 4.
53 Hube F, Lietz U, Igel M, Jensen PB, Tornqvist H, Joost HG, Hauner
H. Difference in leptin mRNA levels between omental and subcutaneous abdominal adipose tissue from obese humans. Horm
Metab Res 1996; 28: 690 ± 693.
54 Montague CT, Prins JB, Sanders L, Digby JE, O'Rahilly S. Depot
and sex speci®c differences in human leptin mRNA expression:
implications for the control of regional fat distribution. Diabetes
1997; 46: 342 ± 347.
55 Seftel HC, Asvat MS, Joffe BI, Raal FJ, Panz VR, Vermaark WJH,
Loock ME, Rajput MC, Omar MA, Jeenah MS. Selected risk factors
for coronary heart disease in male scholars from the major South
African population groups.
S Afr Med J 1993; 84: 891 ± 897.
È nn L, Joffe
56 van der Merwe M-T, Wing JR, Celgow LH, Gray IP, Lo
È nnroth PN. Metabolic indices in relation to body composiBI, Lo
tion changes during weight loss on Dexfen¯uramine in obese
women from two South African ethnic groups. Int J Obes Relat
Metab Disord 1996; 20: 768 ± 776.
57 Punyadeera C, Crowther NJ, van der Merwe M-T, Gray IP. Lipid
metabolism in two South African populations. (Abstract.) J Endocrinol Metab Diab S Afr 2000; 5: 52.
1205
International Journal of Obesity