Content uploaded by Antony Karelis
Author content
All content in this area was uploaded by Antony Karelis on Mar 31, 2016
Content may be subject to copyright.
Relationship between insulin sensitivity and the
triglyceride–HDL-C ratio in overweight and obese
postmenopausal women: a MONET study
Antony D. Karelis, Stephanie M. Pasternyk, Lyne Messier, David H. St-Pierre,
Jean-Marc Lavoie, Dominique Garrel, and Re
´
mi Rabasa-Lhoret
Abstract: The objective of this cross-sectional study was to examine the relationship between the triglyceride–HDL-
cholesterol ratio (TG:HDL-C) and insulin sensitivity in overweight and obese sedentary postmenopausal women. The study
population consisted of 131 non-diabetic overweight and obese sedentary postmenopausal women (age; 57.7 ± 5.0 y; body
mass index (BMI), 32.2 ± 4.3 kg/m
2
). Subjects were characterized by dividing the entire cohort into tertiles based on the
TG:HDL-C (T1 < 0.86 vs. T2= 0.86 to 1.35 vs. T3 > 1.35, respectively). We measured (i) insulin sensitivity (using the
hyperinsulinenic–euglycemic clamp and homeostasis model assessment (HOMA)), (ii) body composition (using dual-
energy X-ray absorptiometry), (iii) visceral fat (using computed tomography), (iv) plasma lipids, C-reactive protein, 2 h
glucose concentration during an oral glucose tolerance test (2 h glucose), as well as fasting glucose and insulin, (v) peak
oxygen consumption, and (vi) lower-body muscle strength (using weight training equipment). Significant correlations were
observed between the TG:HDL-C and the hyperinsulinemic–euglycemic clamp (r = –0.45; p < 0.0001), as well as with
HOMA (r = 0.42; p < 0.0001). Moreover, the TG:HDL-C significantly correlated with lean body mass, visceral fat, 2 h
glucose, C-reactive protein, and muscle strength. Stepwise regression analysis showed that the TG:HDL-C explained
16.4% of the variation in glucose disposal in our cohort, which accounted for the greatest source of unique variance. Other
independent predictors of glucose disposal were 2 h glucose (10.1%), C-reactive protein (CRP; 7.6%), and peak oxygen
consumption (5.8%), collectively (including the TG:HDL-C) explaining 39.9% of the unique variance. In addition, the
TG:HDL-C was the second predictor for HOMA, accounting for 11.7% of the variation. High levels of insulin sensitivity
were associated with low levels of the TG:HDL-C. In addition, the TG:HDL-C was a predictor for glucose disposal rates
and HOMA values in our cohort of overweight and obese postmenopausal women.
Key words: hyperinsulinemic–euglycemic clamp, HOMA, blood lipids, muscle strength, visceral fat.
Re
´
sume
´
: Le but de cette e
´
tude transversale est d’analyser la relation entre le ratio de triglyce
´
rides/HDL-choleste
´
rol
(TG:HDL-C) et la sensibilite
´
a
`
l’insuline chez des femmes se
´
dentaires, postme
´
nopause
´
es et pre
´
sentant un surpoids
ou de l’obe
´
site
´
. Cent trente et une femmes non diabe
´
tiques, postme
´
nopause
´
es, se
´
dentaires, obe
`
ses ou pre
´
sentant un sur-
poids (a
ˆ
ge : 57,7 ± 5,0 ans, IMC : 32,2 ± 4,3 kg/m
2
) participent a
`
cette e
´
tude. Les femmes sont re
´
parties dans trois tertiles
selon leur ratio de triglyce
´
rides/HDL-choleste
´
rol (TG:HDL-C) : (T1 : < 0,86, T2 : entre 0,86 et 1,35, T3 : > 1,35. Les me-
sures suivantes sont prises : (i) la sensibilite
´
a
`
l’insuline au moyen d’un clamp hyperinsuline
´
mique-euglyce
´
mique et du
mode
`
le home
´
ostatique (HOMA), (ii) la composition corporelle par absorptiome
´
trie a
`
rayons-X en double e
´
nergie, (iii)le
contenu de gras abdominal par tomographie assiste
´
e par ordinateur, ( iv) la concentration plasmatique de lipides, la prote
´
ine
Cre
´
active, la concentration de glucose durant 2 h apre
`
slede
´
but de l’e
´
preuve d’hyperglyce
´
mie provoque
´
e par voie orale
de me
ˆ
me que les taux d’insuline et de glucose a
`
jeun, (v) la consommation d’oxyge
`
ne de cre
ˆ
te et (vi) la force musculaire
des membres infe
´
rieurs au moyen d’un appareil de musculation. On note des corre
´
lations significatives entre le ratio
TG:HDL-C et les valeurs observe
´
es au clamp hyperinsuline
´
mique–euglyce
´
mique (r = –0,45; p < 0,0001) et dans le mode
`
le
home
´
ostatique (r = 0,42; p < 0,0001). De plus, on note des corre
´
lations significatives entre le ratio TG:HDL-C et la masse
maigre, le contenu de gras abdominal, la concentration de glucose a
`
l’e
´
preuve de tole
´
rance, la concentration de la prote
´
ine
Cre
´
active de me
ˆ
me que la force musculaire. L’analyse de re
´
gression multiple pas a
`
pas re
´
ve
`
le que le ratio TG:HDL-C ex-
plique 16,4 % de la variance au niveau de l’e
´
limination du glucose, ce qui constitue la plus grande source de variance
unique. Les autres valeurs pre
´
dictives inde
´
pendantes de l’e
´
limination du glucose sont le taux de glucose lors de l’e
´
preuve
d’hyperglyce
´
mie provoque
´
e d’une dure
´
e de 2 h (10,1 %), la prote
´
ine C re
´
active (7,6 %) et la consommation d’oxyge
`
ne de
cre
ˆ
te (5,8 %), toutes ces variables, y compris le ratio TG:HDL-C, expliquant 39,9 % de la variance unique. En outre, le ra-
tio TG:HDL-C est la deuxie
`
me variable pre
´
dictive du mode
`
le home
´
ostatique et il compte pour 11,7 % de la variation ex-
Received 20 November 2006. Accepted 22 May 2007. Published on the NRC Research Press Web site at apnm.nrc.ca on 2 November
2007.
A.D. Karelis. Department of Kinanthropology, Universite
´
du Que
´
bec a
`
Montre
´
al, Montreal, QC H2X 3R9, Canada.
S.M. Pasternyk, L. Messier, D.H. St-Pierre, D. Garrel,
1
and R. Rabasa-Lhoret. Department of Nutrition, University of Montreal,
Montreal, QC H3T 1A8, Canada.
J. Lavoie. Department of Kinesiology, University of Montreal, Montreal, QC H3T 1A8, Canada.
1
Corresponding author (e-mail: dominique.garrel@umontreal.ca).
1089
Appl. Physiol. Nutr. Metab. 32: 1089–1096 (2007) doi:10.1139/H07-095
#
2007 NRC Canada
plique
´
e. Les hauts niveaux de sensibilite
´
a
`
l’insuline sont associe
´
s aux valeurs faibles du ratio TG:HDL-C; de plus, ce der-
nier est une variable pre
´
dictive du taux d’e
´
limination du glucose ainsi qu’un facteur du mode
`
le home
´
ostatique chez un
groupe de femmes postme
´
nopause
´
es, obe
`
ses ou pre
´
sentant un surpoids.
Mots-cle
´
s:clamp hyperinsuline
´
mique–euglyce
´
mique, HOMA, lipides sanguins, force musculaire, gras visce
´
ral.
[Traduit par la Re
´
daction]
________________________________________________________________________ ______________
Introduction
Insulin resistance has been shown to be associated with
the development of type 2 diabetes and cardiovascular dis-
ease (Reaven 1988; Abbasi et al. 2002; Ingelsson et al.
2005). Furthermore, high levels of triglycerides have been
associated with an increased risk of cardiovascular disease,
whereas high levels of high-density lipoprotein cholesterol
(HDL-C) have been reported to decrease the risk of cardio-
vascular disease (Gaziano et al. 1997). In addition, low
HDL-C concentrations combined with high triglyceride
(TG) levels have been reported to be associated wit h a
higher risk for type 2 diabetes and cardiovascular disease
(Haffner et al. 1990; Jeppesen et al. 1998).
The triglyceride–HDL-cholesterol (TG:HDL-C) ratio has
been proposed as a simple tool to identify subjects with in-
sulin resistance (McLaughlin et al. 2003, 2005). Interest-
ingly, a recent study observed a significant negative
relationship between TG:HDL-C and insulin sensitivity us-
ing an oral glucose insulin sensitivity index (r = –0.33) and
the quantitative insulin sensitivity check index (QUICKI)
(r = –0.37) in severely obese non-diabetic men and women
(Brehm et al. 2004). Thus, a better understanding of blood
lipid ratios and their relationship with metabolic risk factors
compared with single lipid markers may be of particular
clinical importance.
It should be noted that surrogate indices of insulin sensi-
tivity measurements were used in the previous studies,
which are related to the fasted state instead of direct meas-
urements of insulin sensitivity such as the hyperinsulinemic–
euglycemic clamp technique, which is related to the
postprandial state with elevated glucose metabolism and hy-
perinsulinemia. To our knowledge, the possible association
between insulin sensitivity usin g the hyperinsulinemic–
euglycemic clamp technique and TG:HDL-C has not been
examined in a well-characterized group of subjects. In ad-
dition, more research on the relationship between
TG:HDL-C and metabolic risk factors such as insulin
resistance has been advocated by the American Heart
Association and the National Heart, Lung, and Blood Insti-
tute, since triglycerides and HDL-C are key components of
the metabolic syndrome (Grundy et al. 2005). Therefore,
the purpose of the present study was to investigate the re-
lationship between TG:HDL-C and insulin sensitivity in
overweight and obese postmenopausal women, using two
measurements:thegoldstandardhyperinsulinemic–euglycemic
clamp and a widely used fasting-based index, the homeo-
stasis model assessment (HOMA). We hypothesized that
higher levels of insulin sensitivity would be associated
with lower levels of TG:HDL-C after controlling for poten-
tial confounding factors such as visceral fat, peak oxygen
consumption, and muscle strength. The rationale was as
follows: high visceral fat content may be an indicator of
defective free fatty acid trapping by the adipose tissue,
which may lead to metabolic complications such as dysli-
pidemia and insulin resistance (Lewis et al. 2002). Further-
more, high levels of peak oxygen uptake and muscle
strength could be indicators of physical fitness and
functional capacity, which may translate to higher levels
of energy expenditure and in turn improve metabolic com-
plications such as insulin sensitivity and dyslipidemia
(Young 1995). Moreover, a physiological link may exist
between insulin sensitivity and muscle strength, since skel-
etal muscle is one of the major sites of insulin-stimulated
glucose disposal.
Materials and methods
Subjects
The study population consisted of 131 non-diabetic over-
weight (n = 51) and obese (n = 80) postmenopausal women
aged between 44 and 73 years. It should be noted that the
cohort was normally distri buted for age and that the major-
ity of women were aged between 55 and 64 (n = 56).
Women were included in the study if they met the fol-
lowing criteria: (i) body mass index (BMI) ‡ 27 kg/m
2
,
(ii) cessation of menstruation for more than 1 year and
a follicle-stimulating hormone level ‡ 30 U/L, (iii) sed-
entary (<2 h/week of structured exercise), (iv) non-smoker,
(v) low to moderate alcohol consumption (<2 drinks/d),
(vi) free of known inflammatory disease, and (vii) did not
use hormone replacement therapy. On physical examination
or biological testing, all participants had no history or
evidence of cardiovascular disease, peripheral vascular disease
or stroke, diabetes (fasting plasma glucose <7.0 mmol/L
and (or) 2 h post 75 g oral glucose tolerance
test <11.1 mmol/L), orthopaedic limitations, body mass
fluctuation ±2 kg in the last 6 months, thyroid or pituitary
disease, infection (based on a medical examination along
with a complete blood count), or medica tions that could af-
fect cardiovascular function and (or) metabolism. The
study was approved by the University of Montreal ethics
committee. After reading and signing the consent form,
each participant was submitted to a series of tests.
Diet stabilization period
Before the study, subjects were asked to maintain normal
activity level and eating habits prior to the testing sequence
and to maintain their body mass within a ±2 kg range to re-
duce the acute effects of caloric restriction on outcome vari-
ables.
Hyperinsulinemic-euglycemic clamp
The study began at 07h30 after a 12 h overnight fast fol-
lowing the procedure described by DeFronzo et al. (1979).
1090 Appl. Physiol. Nutr. Metab. Vol. 32, 2007
#
2007 NRC Canada
An antecubital vein was cannulated for the infusion of 20%
dextrose and insulin (Actrapid
1
, Novo-Nordisk, Toronto,
Ont.). The other arm was cannulated so that bloo dsamples
could be drawn. Three basal samples of plasma glucose and
insulin were taken over 40 min. Then, insulin infusion was
initiated at the rate of 75 mUm
–2
min
–1
for 180 min. Plasma
glucose was measured every 10 min with a glucose analyzer
(Beckman Instruments, Fullerton, Calif.) and maintained at
fasting level with a variable infusion rate of 20% dextrose.
Glucose disposal was calculated as the mean rate of glucose
infusion measured during the last 30 min of the clamp
(steady state) and is expressed as milligrams per minute per
kilogram of lean body mass (LBM) (mgmin
–1
kg
–1
). Fasting
plasma glucose levels were determined using the mean of
three basal values of plasma with a Beckman glucose ana-
lyzer.
Blood samples
After an overni ght fast (12 h), venous blood samples were
collected for the measurement of plasma concentrations of
total cholesterol, HDL-C, TGs, CRP, and insulin. Low-
density lipoprotein cholesterol (LDL-C) was estimated us-
ing the Friedewald formula (Friedewald et al. 1972).
Plasma was analyzed on the day of collection except for
insulin and CRP, which were kept at –80 8C until analyses
were performed. Analyses were done on the COBAS
INTEGRA 400 (Roche Diagnostic, Montreal, Que.) ana-
lyzer for total cholesterol, HDL-C, and TGs. Total choles-
terol, HDL-C, and TGs were used in the Friedewald
formula (Friedewald et al. 1972) to calculate LDL-C con-
centration. Fasting insulin levels were determined using
the mean of three basal values of plasma and measured in
duplicate by automated radioimmunoassay (Medicorp,
Montreal, Que.). Homeostasis model assessment (HOMA)
was calculated according to the formula of Matthews et al.
(1985) using fasting basal values of plasma glucose and in-
sulin. Two-hour glucose levels were measured from an oral
glucose tolerance test (2 h glucose) with a Beckman glu-
cose analyzer. High-sensitivity CRP (hsCRP) was assessed
by immunonephelometry on an IMMAGE analyser
(Beckman-Coulter, Villepinte, France).
Body composition
Body mass was measured using an electronic scale (Bal-
ance Industrielles, Montreal, Que.) and standing height was
measured using a wall stadiometer (Perspective Enterprises,
Portage, Mich.). Thereafter, body mass index (BMI = body
mass/height
2
(with height measured in metres)) was calcu-
lated. Lean body mass and fat mass were evaluated by
dual-energy X-ray absorptiometry (General Electric Lunar
Corporation version 6.10.019, Madison, Wis.). The intra-
class coefficient correlation for test–retest for fat mass and
lean body mass was 0.99 (n = 18).
Computed tomography (CT)
A GE High Speed Advantage CT scanner (General Elec-
tric Medical Systems, Milwaukee, Wis.) was used to meas-
ure visceral fat content. The subjects were examined in the
supine position with both arms stretched above their head
and were asked to stop breathing momentarily during the
scan. The position of the scan was established at the L4–L5
Table 1. Subject characteristics in relation to the severity of TG:HDL-C in overweight and obese postmenopausal women.
Physical characteristics Age (y)
BMI
(kg/m
2
) LBM (kg) Fat mass (kg) %Body fat Visceral fat (cm
2
)
VO
2 peak
(mLmin
–1
kg LBM
–1
)
(n = 41, 42, 43)
Muscle strength
(kg/kg LBM)
(n = 30, 38, 36)
T1 (<0.86) (n = 43) 58.3±5.0 32.1±4.2 40.6±4.1* 38.3±9.0 46.6±4.9 163.7±52.5
{
35.8±5.3 3.4±0.8
T2 (0.86–1.35) (n = 44) 57.1±4.6 31.8±4.2 42.0±5.9 37.3±7.8 45.4±4.0 196.2±56.1 34.7±5.7 3.4±0.9
T3 (>1.35) (n = 44) 58.2±5.3 32.8±4.6 44.1±7.0 39.5±9.2 45.9±5.3 198.6±54.0 34.2±5.9 2.9±0.7
{
Note: Values are means ± SD. VO
2 peak
, peak oxygen uptake.
*Significantly different from T3 (p < 0.05).
{
Significantly different from T2 and T3 (p < 0.05).
{
Significantly different from T1 and T2 (p < 0.05).
Karelis et al. 1091
#
2007 NRC Canada
vertebral disc using a scout image of the body. Visceral
adipose tissue area was quantified by delineating the intra-
abdominal cavity at the internalmost aspect of the abdomi-
nal and oblique muscle walls surrounding the cavity and the
posterior aspect of the vertebral body. The software used to
calculate the area was an advantage workstation from GE
4.1; VOXta 3.0.64u volume viewer 3D soft. The techniques
factors are 120 kV and 400 UA. An 8 slice scan of
0.625 mm each at the level of L4–L5 was used. Values of
visceral fat were calculated with the consideration of all 8
slices. That is, a thickness of 5 mm. The test–retest meas-
ures of the different body fat distribution indices on 10 CT
scans yielded a mean absolute difference of ±1%.
Aerobic capacity (VO
2peak
)
Aerobic capacity was assessed on an ergocycle Ergoline
900 (Bitz, Germany), with an Ergocard (Medi Soft, Dinant,
Belguim) cardiopulmonary exercise test station. Aerobic ca-
pacity was tested by a progressive test starting at 25 W with
an augmentation of 25 W every 2 min. Subjects were asked
to maintain a constant speed during the procedure. O
2
and
CO
2
were measured by a direct system using a face mask.
VO
2peak
was achieved when the power output could no lon-
ger be maintained. VO
2peak
was defined as the highest 30 s
average of oxygen consumption. A test–retest reliability trial
(n = 19) for VO
2max
performed in our laboratory showed an
intra-class correlation coefficient of 0.96.
Muscle strength
Lower-body strength was assessed using leg press weight
training equipment from Atlantis Precision Series (Atlantis
Inc., Laval, Que.). Subjects were aligned with the ball of
their feet on the footplate of the machine at shoulder width
so that their knee angle approximated 908 of flexion. Keep-
ing their back flat against the chair, participants were as-
sisted to the starting position, which consisted of extending
their legs outward. A complete repetition consisted of flex-
ing the knees and slowly returning to a complete extension
of the legs stopping before full knee extension. Subjects
were asked to start their session with a light walk on a
treadmill for 10 min. Muscle strength was measured using a
one-repetition maximum (1RM) technique. The first set was
used as a warm up of 10 repetitions with a light initial load
set by the training supervisor. Thereafter, the load was in-
creased until maximal effort was achieved. The 1RM was
typically determined within 5 trials with a 4 min rest be-
tween each trial. Failure was defined as a lift falling short
of the full range of motion. If none of the trials yielded a
1RM, the Wathan equation (Wathan 1994) was used to ex-
trapolate the 1RM. A muscular strength index was calcu-
lated by dividing the weight lifted (1RM) in kilograms by
the LBM in kilograms.
Statistical analysis
The data are expressed as the mean ± SD. We first veri-
fied the normality of the distribution of variables with a
Kolmogorov–Smirnov test and found that only CRP was
not normally distributed. Therefore, we used the log-
transformed (base 10) CRP for this variable in the analysis.
Pearson correlations were then performed to examine the
relationship between the TG:HDL-C and the
hyperinsulinemic–euglycemic clamp as well as HOMA. Par-
tial correlations were also used to control for peak oxygen
uptake, visce ral fat, and muscle strength. Subjects were then
characterized by dividing the entire cohort into tertiles based
on TG:HDL-C (T1 < 0.86 vs. T2 = 0.86 to 1.35 vs. T3 >
1.35, respectively). A one-way ANOVA was performed to
analyze mean differences among the three groups. When
significant differences were found, Tukey’s post hoc test
was performed to identify group differences. A stepwise
multilinear regression model determined which variables ex-
plained unique var iance for both the hyperinsulinemic–
euglycemic clamp and HOMA values. Based on exploratory
analyses and using biologically plausible hypotheses, inde-
pendent variables considered in the final model for the
hyperinsulinemic–euglycemic clamp were visceral fat, tri-
glycerides, HDL-C, total cholesterol: HDL-C, TG:HDL-C,
CRP, muscle strength, 2 h glucose, and VO
2peak
. As for
HOMA, independent variables considered in the final
model were visceral fat, triglycerides, HDL-C, total
cholesterol:HDL -C, TG:HDL-C, CRP, lean body mass, and
VO
2 peak
. Statistical analysis was performed using SPSS for
Windows (Chicago, Ill.). Significance was accepted at p <
0.05.
Results
Table 1 shows the subjects’ physical characteristics
among overw eight and obese postmenopausal women in re-
lation to the severity of TG:HDL-C classified into 3 groups:
T1 < 0.86 vs. T2 = 0.86 to 1.35 vs. T3 > 1.35, respectively).
Lean body mass was significantly lower in the T1 group
when compared with the T3 group. In addition, visceral fat
was significantly lower in the T1 group when compared
with both the T2 and T3 group. Moreover, muscle strength
was significantly lower in the T3 group when compared
with both the T2 and T1 group. Finally, all three groups
1092 Appl. Physiol. Nutr. Metab. Vol. 32, 2007
#
2007 NRC Canada
Table 2. Metabolic risks factors in relation to the severity of TG:HDL-C in overweight and obese postmenopausal women.
Metabolic risk factors
Total cholesterol
(mmol/L)
LDL-C
(mmol/L)
HDL-C
(mmol/L)
TGs
(mmol/L)
Total
cholesterol:
HDL-C TG:HDL-C
Fasting glucose
(mmol/L)
(n = 41, 44, 42)
T1 (<0.86) (n = 43) 5.4±0.8 3.2±0.6 1.8±0.3* 1.0±0.2* 3.1±0.5* 0.6±0.2* 5.1±0.5
T2 (0.86–1.35) (n = 44) 5.5±0.8 3.3±0.7 1.4±0.2
{
1.6±0.2
{
3.9±0.5
{
1.1±0.1
{
5.2±0.5
T3 (>1.35) (n = 44) 5.6±0.9 3.2±0.8 1.2±0.2 2.5±0.7 4.5±0.8 2.0±0.7 5.1±0.5
Note: Values are means ± SD.
*Significantly different from T2 and T3 (p < 0.05).
{
Significantly different from T3 (p < 0.05).
{
Significantly different from T1 and T2 (p < 0.05)
were comparable for age, body mass index, fat mass, per-
cent body fat, and peak oxygen uptake.
Metabolic risk factors are shown in Table 2. No differen-
ces between groups were noted for total cholesterol, LDL-C,
fasting glucose, or CRP. Group T1 had lower values for
fasting insulin when compared with group T3. Moreover,
2 h glucose was significantly higher in the T3 group when
compared with both the T2 and T1 groups. By design, sub-
jects in the T1 group had lower TG:HDL-C, TGs, and total
cholesterol:HDL -C and higher HDL-C levels when com-
pared with the T2 and T3 groups. Furthermore, subjects in
the T2 group showed lower TG:HDL-C, TGs, and total cho-
lesterol:HDL-C and higher HDL-C values than group T3.
Insulin sensitivity measured with the hyperinsulinemic–
euglycemic clamp showed a graded decrease among sub-
jects. This difference attained statistical significance when
group T1 was compared with groups T2 and T3. HOMA
values were significantly lower in group T1 when com-
pared with group T3.
The correlations between TG:HDL-C, the hyperinsulinemic–
euglycemic clamp, or HOMA levels and various measured
variables in overweight and obese postmenopausal women
are reported in Table 3. Significant correlations were
observed between TG:HDL-C and the hyperinsulinemic–
euglycemic clamp (r = –0.45; p < 0.0001), as well as with
HOMA (r = 0.42; p < 0.0001). We also examined this rela-
tionship after statistical control for peak oxygen uptake, vis-
ceral fat, and muscle strength using partial correlation
analysis. The partial correlation between TG:HDL -C and
the hyperinsulinemic–euglycemic clamp (r = –0.38; p <
0.0001), as well as HOMA ( r = 0.31; p < 0.004), remained
significant.
We per formed stepwise regression analysis to examine the
independent predictors of glucose disposal. Table 4 illus-
trates the summary of the model. Our results show that the
variables of TG:HDL-C, CPR, 2 h glucose, and VO
2 peak
were independent predictors of glucose disposal, collectively
explaining 39.9% of the variance. As for HOMA, our results
show that the variables of visceral fat and TG:HDL-C were
independent predictors, collectively explaining 35.4% of the
variance. The variance inflation factors in the regression
models for all independent predictors were equal to or less
than 1.1.
It should be noted that, in a sub-analysis, we also exam-
ined the relationship between TG:HDL-C and insulin sensi-
tivity in overweight (n = 51) and obese (n = 80)
postmenopausal women separately from the entire cohort
(n = 131). Interestingly, results indicate that no significant
relationship was observed between TG:HDL-C and the
hyperinsulinemic–euglycemic clamp (r = –0.27; p = 0.061)
as well as HOMA (r = 0.10; p = 0.50) in only overweight
subjects. In contrast, a significant correlation was observed
between TG:HDL-C and the hyperinsulinemic–euglycemic
clamp (r = –0.51; p < 0.0001), as well as with HOMA (r =
0.50; p < 0.0001), in only obese individuals. Furthermore,
stepwise regression analysis showed that TG:HDL-C ex-
plained 26.6% of the variation in glucose disposal, which
accounted for the greatest source of unique variance. In ad-
dition, TG:HDL-C explained 23.5% of the variation in
HOMA, which also accounted for the greatest source of
unique variance. Finally, it should be noted that in the
present study, obese subjects had significantly lower insulin
sensitivity values (using the hyperinsulinemic–euglycemic
clamp) compared with overweight individuals (10.7 ± 2.8
vs. 12.2 ± 2.8 mgmin
–1
kg FFM
–1
, respectively).
Discussion
Results from the present study show a significant relation-
ship between TG:HDL-C and insulin sensitivity. These
findings were found irrespective of the method used
(hyperinsulinemic–euglycemic clamp and HOMA methods).
These results support and extend the results of Brehm et
al. (2004) who also reported a relationship between
TG:HDL-C and insulin sensitivity using an oral glucose in-
sulin sensitivity index and the quantitative insulin sensitiv-
ity check index (QUICKI) in severely obese non-diabetic
men and women. Furthermore, results from the present
study support our hypothesis. That is, greater levels of in-
sulin sensitivity were as sociated with lower levels of
TG:HDL-C. This suggests that lower TG:HDL-C levels,
despite similar high levels of body fat between groups,
could contribute to the favorable metabolic profile ob-
served in the lower tertile group. In support of this hypoth-
esis, TG:HDL-C levels explained 16.4% of the variation in
glucose disposal in our cohort, which accounted for the
greatest source of unique variance. Other independent pre-
dictors of glucose disposal were 2 h glucose, CRP, and
peak oxygen consumption. In addition, TG:HDL-C was
the second predictor of HOMA, accounting for 11.7% of
the variation. It should be noted that TGs or HDL-C levels
alone were not independent predictors of glucose disposal
and HOMA in our cohort, suggesting the particular clinical
relevance of using TG:HDL-C instead of single lipid
markers for identifyi ng insulin-resistant individuals. In ad-
dition, visceral fat was not a predictor of insulin sensitivity
as measured by the hyperinsulinemic–euglycemic clamp,
Fasting insulin
(pmol/L)
(n = 34, 38, 37)
2 h glucose
(mmol/L)
(n = 43, 44, 43)
CRP (mg/L)
(n = 26, 27, 31)
HOMA
(n = 33, 38, 37)
Insulin sensitivity
(mgmin
–1
kg FFM
–1
)
75.9±27.2
{
6.0±1.6 3.4±4.0 2.86±1.1
{
12.9±2.7*
87.0±27.6 6.0±1.3 3.5±2.8 3.35±1.2 11.1±2.4
105.1±48.2 7.1±2.1
{
4.5±3.2 3.96±1.9 9.9±2.8
Karelis et al. 1093
#
2007 NRC Canada
but was a predictor for HOMA. A possible explanation for
this may be that glucose disposal rates determined by the
hyperinsulinemic–euglycemic clamp mainly reflects periph-
eral insulin resistance, particularly in skeletal muscle, and
HOMA is essentially an indicator of hepatic insulin resist-
ance (Wallace et al. 2004).
In a sub-analysis, we also examined the relationship be-
tween TG:HDL-C and insulin sensitivity in overweight and
obese postmenopausal women separately from the entire
cohort. Interestingly, results indicate that no significant
relationship was observed between TG:HDL-C and the
hyperinsulinemic–euglycemic clamp, as well as HOMA in
only overweight subjects. In contrast, a significant
correlation was observed between TG:HDL-C and the
hyperinsulinemic–euglycemic clamp, as well as with
HOMA in only obese individuals. It should be noted that in
the present study, obese subjects had significantly lower
insulin sensitivity values (using the hyperinsulinemic–
euglycemic clamp) compared with overweight individuals.
Therefore, this could explain the different findings ob-
served between overweight and obese subjects. That is,
TG:HDL-C may be a more reliable tool in groups present-
ing a higher prevalence of insulin resistance.
Interestingly , the results from HOMA parallel those of the
hyperinsulinemic–euglycemic clamp technique. As already
suggested, HOMA could display a clinically relevant sensi-
tivity to unmask the relation between insulin sensitivity and
metabolic disturbances while offering a high degree of sim-
plicity (Hanley et al. 2002). The strength of this relationship
is reinforced by the fact that the relationship is present even
in a homogenous population of obese postmenopausal
women. That is, even within an overweight and obese popu-
lation of similar body fat content, the relationship between
higher levels of insulin sensitivity and lower levels of
TG:HDL-C is evident.
What are potential mediating factors for lower levels of
insulin sensitivity with higher levels of TG:HDL-C? High
levels of visceral fat have been associated with metabolic
disturbances such as insulin resistance, dyslipidemia, and
hypertension (Gastaldelli et al. 2002; Wagenknecht et al.
2003), which could lead to an increased risk of cardiovascu-
lar disease (Brochu et al. 2000). Furthermore, high visceral
fat content may be an indicator of defective free fatty acid
trapping by the adipose tissue, which may lead to metabolic
complications such as dyslipidemia and insulin resistance
(Lewis et al. 2002). In addition, low levels of muscle
strength are associated with metabolic complications (Jurca
et al. 2004; Sayer et al. 2005) and an increased risk of mor-
tality (Rantanen et al. 2000; Metter et al. 2002). Moreover,
several studies have reported that low cardiorespiratory
fitness levels are associated with cardiovascular disease
(Carnethon et al. 2005), metabolic abnormalities (Lee et al.
2005), and an increased risk of mortality (Katzmarzyk et al.
2005). Finally, high levels of peak oxygen uptake and
Table 3. Pearson correlations between TG:HDL-C and the hyperinsulinemic–
euglycemic (HE) clamp or HOMA levels and various measured parameters.
Physical and metabolic factors TG:HDL-C HE clamp HOMA
TG/HDL 1.00 –0.45* 0.42*
HE Clamp –0.45* 1.00 0.47*
HOMA 0.42* –0.47* 1.00
Lean body mass 0.25* –0.36* 0.44*
Waist circumference 0.17 –0.25* 0.38*
Visceral fat 0.26* –0.32* 0.47*
HDL-C –0.68* 0.37* –0.28*
TGs 0.96* –0.41* 0.35*
Total cholesterol:HDL-C 0.73* –0.32* 0.24
{
2 hour glucose 0.21
{
–0.38* 0.32*
CRP 0.25
{
–0.31* 0.38
{
VO
2peak
–0.10 0.24* –0.21
{
Muscle strength –0.24
{
0.27* –0.10
*p < 0.01.
{
p < 0.05.
Table 4. Stepwise regression analysis regarding independent predictors of glucose dis-
posal and HOMA in overweight and obese postmenopausal women.
Dependent variable Step
Independent
variable Partial r
2
Total r
2
cumulative p value
Glucose disposal
(mgmin
–1
kg FFM
–1
)
1 TG:HDL-C 0.164 0.164 <0.01
2 2 h glucose 0.101 0.265 <0.01
3 CRP 0.076 0.341 <0.05
4 VO
2 peak
0.058 0.399 <0.05
HOMA 1 Visceral fat 0.237 0.237 <0.01
2 TG:HDL-C 0.117 0.354 <0.01
Note: FFM, fat-free mass.
1094 Appl. Physiol. Nutr. Metab. Vol. 32, 2007
#
2007 NRC Canada
muscle strength could be indicators of physical fitness and
functional capacity, which may translate to higher levels of
energy expenditure and in turn improve metabolic complica-
tions such as insulin sensitivity and dyslipidemia (Young
1995). Furthermore, a physiological link may exist between
insulin sensitivity and muscle strength, since skeletal muscle
is one of the major sites of insulin-stimulated glucose dis-
posal. In the present study, we observed a significant rela-
tionship between TG:HDL-C and both muscle strength and
visceral fat. Furthermore, a significant association was re-
ported between the hyperinsulinemic–euglycemic clamp
with visceral fat, peak oxygen uptake, and muscle strength.
When the association between insulin sensitivity and
TG:HDL-C was controlled for visceral fat, peak oxygen
consumption, and muscle strength, the correlation between
insulin sensitivity and TG:HDL-C remained significant, sug-
gesting that these parameters are not potential mediators of
this relationship. The same findings were also observed for
HOMA. Potential mechanisms that may explain the relation-
ship between insulin sensitivity and TG:HDL-C could be
free fatty acid concentration, different adipose tissue hor-
mones, physical activity energy expend iture, gene expres-
sion, and diet. In addition, in the study of McLaughlin et al.
(2005), the authors discu ss that LDL-C particle diameter or
postprandial remnant concentrations are associated with in-
sulin resistance and that TG:HDL-C predicts the presence
of the small, dense, LDL-C phenotype. Therefore, LDL-C
phenotype could also explain the relationship between
TG:HDL-C and insulin sensitivity.
There are several limitations in the present study. First,
our cohort is only composed of non-diabetic sedentary over-
weight and obese postmenopausal women. Therefore, our
findings are limited to this population. Second, we used a
cross-sectional approach, which does not allow us to con-
clude to any causal associations between insu lin sensitivity
and TG:HDL-C in our cohort. Third, no habituation was
performed before the muscle strength test. Despite these
limitations, our results are strengthened by using gold stand-
ard techniques, as well as by studying a well-characterized
cohort in a relatively large samp le size.
In conclusion, results in the present study show a
significant relatio nship between TG:HDL-C and insulin
sensitivity using two measurements, the gold standard
hyperinsulinemic–euglycemic clamp and HOMA, in over-
weight and obese postmenopausal women. This relationship
was more evident in obese subjects than in overweight indi-
viduals. However, the mechanism(s) that could mediate this
association remain(s) unclear. From a clinical perspective,
the present findings suggest that TG:HDL-C alone, particu-
larly in an obese population, could be a simple and reliable
tool that health professional may use to identify insulin-
resistant patients.
Acknowledgements
This study was supported by grants from the Canadian In-
stitute of Health Research New Emerging Teams in Obesity
(University of Montreal and University of Ottawa; MONET
project). Dr. Rabasa-Lhoret is supported by the Fonds de la
Recherche en Sante
´
du Que
´
bec.
References
Abbasi, F., Brown, B.W., Jr., Lamendola, C., McLaughlin, T., and Re-
aven, G.M. 2002. Relationship between obesity, insulin resistance,
and coronary heart disease risk. J. Am. Coll. Cardiol. 40: 937–943.
doi:10.1016/S0735-1097(02)02051-X. PMID: 12225719.
Brehm, A., Pfeiler, G., Pacini, G., Vierhapper, H., and Roden, M.
2004. Relationship between serum lipoprotein ratios and insulin
resistance in obesity. Clin. Chem. 50: 2316–2322. doi:10.1373/
clinchem.2004.037556. PMID:15459091.
Brochu, M., Poehlman, E.T., and Ades, P.A. 2000. Obesity, body
fat distribution, and coronary artery disease. J. Cardiopulm. Re-
habil. 20: 96–108. doi:10.1097/00008483-200003000-00003.
PMID:10763157.
Carnethon, M.R., Gulati, M., and Greenland, P. 2005. Prevalence
and cardiovascular disease correlates of low cardiorespiratory
fitness in adolescents and adults. JAMA, 294: 2981–2988.
doi:10.1001/jama.294.23.2981. PMID:16414945.
DeFronzo, R.A., Tobin, J.D., and Andres, R. 1979. Glucose clamp
technique: a method for quantifying insulin secretion and resis-
tance. Am. J. Physiol. 237: E214–E223. PMID:382871.
Friedewald, W.T., Levy, R.I., and Fredrickson, D.S. 1972. Estima-
tion of the concentration of low-density lipoprotein cholesterol
in plasma, without use of the preparative ultracentrifuge. Clin.
Chem. 18: 499–502. PMID:4337382.
Gastaldelli, A., Miyazaki, Y., Pettiti, M., Matsuda, M., Mahankali,
S., Santini, E., et al. 2002. Metabolic effects of visceral fat accu-
mulation in type 2 diabetes. J. Clin. Endocrinol. Metab. 87:
5098–5103. doi:10.1210/jc.2002-020696. PMID:12414878.
Gaziano, J.M., Hennekens, C.H., O’Donnell, C.J., Breslow, J.L.,
and Buring, J.E. 1997. Fasting triglycerides, high-density lipo-
protein, and risk of myocardial infarction. Circulation, 96:
2520–2525. PMID:9355888.
Grundy, S.M., Cleeman, J.I., Daniels, S.R., Donato, K.A., Eckel,
R.H., Franklin, B.A., et al. 2005. Diagnosis and management of
the metabolic syndrome: an American Heart Association/Na-
tional Heart, Lung, and Blood Institute Scientific Statement. Cir-
culation, 112: 2735–2752. doi:10.1161/CIRCULATIONAHA.
105.169404. PMID:16157765.
Haffner, S.M., Stern, M.P., Hazuda, H.P., Mitchell, B.D., and Pat-
terson, J.K. 1990. Cardiovascular risk factors in confirmed pre-
diabetic individuals. Does the clock for coronary heart disease
start ticking before the onset of clinical diabetes? JAMA, 263:
2893–2898. doi:10.1001/jama.263.21.2893. PMID:2338751.
Hanley, A.J., Karter, A.J., Festa, A., D’Agostino, R., Jr., Wagen-
knecht, L.E., Savage, P., et al. 2002. Factor analysis of
metabolic syndrome using directly measured insulin sensitivity:
The Insulin Resistance Atherosclerosis Study. Diabetes, 51:
2642–2647. doi:10.2337/diabetes.51.8.2642. PMID:12145182.
Ingelsson, E., Sundstrom, J., Arnlov, J., Zethelius, B., and Lind, L.
2005. Insulin resistance and risk of congestive heart failure.
JAMA, 294: 334–341. doi:10.1001/jama.294.3.334. PMID:
16030278.
Jeppesen, J., Hein, H.O., Suadicani, P., and Gyntelberg, F. 1998.
Triglyceride concentration and ischemic heart disease: an eight-
year follow-up in the Copenhagen Male Study. Circulation, 97:
1029–1036. PMID:9531248.
Jurca, R., Lamonte, M.J., Church, T.S., Earnest, C.P., Fitzgerald,
S.J., Barlow, C.E., et al. 2004. Associations of muscle strength
and fitness with metabolic syndrome in men. Med. Sci. Sports
Exerc. 36: 1301–1307. doi:10.1249/01.MSS.0000135780.88930.
A9. PMID:15292736.
Katzmarzyk, P.T., Church, T.S., Janssen, I., Ross, R., and Blair,
S.N. 2005. Metabolic syndrome, obesity, and mortality: impact
Karelis et al. 1095
#
2007 NRC Canada
of cardiorespiratory fitness. Diabetes Care, 28: 391–397. doi:10.
2337/diacare.28.2.391. PMID:15677798.
Lee, S., Kuk, J.L., Katzmarzyk, P.T., Blair, S.N., Church, T.S., and
Ross, R. 2005. Cardiorespiratory fitness attenuates metabolic
risk independent of abdominal subcutaneous and visceral fat in
men. Diabetes Care, 28: 895–901. doi:10.2337/diacare.28.4.895.
PMID:15793192.
Lewis, G.F., Carpentier, A., Adeli, K., and Giacca, A. 2002. Disor-
dered fat storage and mobilization in the pathogenesis of insulin
resistance and type 2 diabetes. Endocr. Rev. 23: 201–229.
doi:10.1210/er.23.2.201. PMID:11943743.
Matthews, D.R., Hosker, J.P., Rudenski, A.S., Naylor, B.A., Trea-
cher, D.F., and Turner, R.C. 1985. Homeostasis model assess-
ment: insulin resistance and beta-cell function from fasting
plasma glucose and insulin concentrations in man. Diabetologia,
28: 412–419. doi:10.1007/BF00280883. PMID:3899825.
McLaughlin, T., Abbasi, F., Cheal, K., Chu, J., Lamendola, C., and
Reaven, G. 2003. Use of metabolic markers to identify over-
weight individuals who are insulin resistant. Ann. Intern. Med.
139: 802–809. PMID:14623617.
McLaughlin, T., Reaven, G., Abbasi, F., Lamendola, C., Saad, M.,
Waters, D., et al. 2005. Is there a simple way to identify insulin-
resistant individuals at increased risk of cardiovascular disease?
Am. J. Cardiol. 96: 399–404. doi:10.1016/j.amjcard.2005.03.
085. PMID:16054467.
Metter, E.J., Talbot, L.A., Schrager, M., and Conwit, R. 2002.
Skeletal muscle strength as a predictor of all-cause mortality in
healthy men. J. Gerontol. A Biol. Sci. Med. Sci. 57: B359–B365.
PMID:12242311.
Rantanen, T., Harris, T., Leveille, S.G., Visser, M., Foley, D.,
Masaki, K., et al. 2000. Muscle strength and body mass index
as long-term predictors of mortality in initially healthy men. J.
Gerontol. A Biol. Sci. Med. Sci. 55: M168–M173.
PMID:10795731.
Reaven, G.M. 1988. Banting lecture 1988. Role of insulin resis-
tance in human disease. Diabetes, 37: 1595–1607. doi:10.2337/
diabetes.37.12.1595. PMID:3056758.
Sayer, A.A., Dennison, E.M., Syddall, H.E., Gilbody, H.J., Phillips,
D.I., and Cooper, C. 2005. Type 2 diabetes, muscle strength, and
impaired physical function: the tip of the iceberg? Diabetes
Care, 28: 2541–2542. doi:10.2337/diacare.28.10.2541. PMID:
16186295.
Wagenknecht, L.E., Langefeld, C.D., Scherzinger, A.L., Norris,
J.M., Haffner, S.M., Saad, M.F., et al. 2003. Insulin sensitivity,
insulin secretion, and abdominal fat: the Insulin Resistance
Atherosclerosis Study (IRAS) Family Study. Diabetes, 52:
2490–2496. doi:10.2337/diabetes.52.10.2490. PMID:14514631.
Wallace, T.M., Levy, J.C., and Matthews, D.R. 2004. Use and
abuse of HOMA modeling. Diabetes Care, 27: 1487–1495.
doi:10.2337/diacare.27.6.1487. PMID:15161807.
Wathan, D. 1994. Essentials of strength training and conditioning.
Human Kinetics, Champaign, Ill.
Young, J.C. 1995. Exercise prescription for individuals with meta-
bolic disorders. Practical considerations. Sports Med. 19: 43–54.
PMID:7740246.
1096 Appl. Physiol. Nutr. Metab. Vol. 32, 2007
#
2007 NRC Canada