0013-7227/03/$15.00/0
Printed in U.S.A.
The Journal of Clinical Endocrinology & Metabolism 88(4):1785–1791
Copyright © 2003 by The Endocrine Society
doi: 10.1210/jc.2002-021674
Intramyocellular Lipids: Anthropometric Determinants
and Relationships with Maximal Aerobic Capacity and
Insulin Sensitivity
CLAUS THAMER, JÜRGEN MACHANN, OLIVER BACHMANN, MICHAEL HAAP, DOMINIK DAHL,
BEATE WIETEK, OTTO TSCHRITTER, ANDREAS NIESS, KLAUS BRECHTEL, ANDREAS FRITSCHE,
CLAUS CLAUSSEN, STEPHAN JACOB, FRITZ SCHICK, HANS-ULRICH HÄRING, AND
MICHAEL STUMVOLL
Department of Endocrinology and Metabolism (C.T., O.B., M.H., D.D., O.T., A.F., S.J., H.-U.H., M.S.) and Section on
Experimental Radiology (J.M., B.W., K.B., C.C., F.S.), Department of Diagnostic Radiology, Eberhard-Karls-University,
D-72076 Tübingen, Germany; and Medical Clinic (A.N.), Department of Sports Medicine, University of Freiburg, D-79106
Freiburg, Germany
The existence of metabolically relevant intramyocellular
lipids (IMCL) as assessed by the noninvasive 1H-magnetic
resonance spectroscopy (MRS) has been established. In the
present studies, we analyzed the relationships between IMCL
in two muscle types [the predominantly nonoxidative tibialis
muscle (tib) and the predominantly oxidative soleus muscle
(sol)] and anthropometric data, aerobic capacity (VO2max,
bicycle ergometry, n ⴝ 77) and insulin sensitivity (hyperinsulinemic euglycemic clamp, n ⴝ 105) using regression
analysis.
In univariate regression, IMCL (tib) was weakly but significantly correlated with percentage of body fat (r ⴝ 0.28, P ⴝ
0.01), whereas IMCL (sol) was better correlated with waistto-hip ratio (r ⴝ 0.41, P < 0.0001). No significant univariate
correlation with age or maximal aerobic power was observed.
After adjusting for adiposity, IMCL (tib) was positively correlated with measures of aerobic fitness. A significant interaction term between VO2max and percentage of body fat on
T
HE EXISTENCE OF metabolically relevant lipid stores in
skeletal muscle has long been established (1). In addition to glycogen, lipids represent a storage form of energy in
muscle. Consequently, their regulation by physical exercise
and their relationship with insulin-stimulated glucose metabolism has been of interest. Endurance training, for example, was shown to increase the im lipid content. In insulinresistant Pima Indians, im lipids were also increased and
proposed to have some causal role in the pathogenesis of
type 2 diabetes (2).
More recently, a quantification method based on magnetic
resonance spectroscopy has greatly facilitated the study of im
lipids and their relationship with insulin sensitivity. Moreover, this noninvasive technique permitted distinction between lipids within muscle cells (intramyocellular lipids,
IMCL) from lipid interlaced between muscle fibers (extramyocellular, EMCL) and has been demonstrated to be as
accurate as biochemical or histological methods (3). Using
Abbreviations: EMCL, Extramyocellular lipids; IMCL, intramyocellular lipids; ISI, insulin sensitivity index; MRS, magnetic resonance
spectroscopy; sol, soleus muscle; tib, tibialis muscle; WHR, waist to hip
ratio; VO2max, maximal aerobic capacity.
IMCL (tib) (P ⴝ 0.04) existed (whole model r2 ⴝ 0.26, P ⴝ 0.001).
In contrast, aerobic fitness did not influence IMCL (sol). No
correlation between insulin sensitivity as such and IMCL (tib)
(r ⴝ ⴚ0.13, P ⴝ 0.2) or IMCL (sol) (r ⴝ 0.03, P ⴝ 0.72) was
observed. Nethertheless, a significant interaction term between VO2max and IMCL on insulin sensitivity existed [P ⴝ
0.04 (tib) and P ⴝ 0.02 (sol)]; [whole model (sol) r2 ⴝ 0.61, P <
0.0001, (tib) r2 ⴝ 0.60, P < 0.0001].
In conclusion, obesity and aerobic fitness are important
determinants of IMCL. IMCL and insulin sensitivity are negatively correlated in untrained subjects. The correlation between the two parameters is modified by the extent of aerobic
fitness and cannot be found in endurance trained subjects.
Thus, measurements of aerobic fitness and body fat are indispensable for the interpretation of IMCL and its relationship with insulin sensitivity. (J Clin Endocrinol Metab 88:
1785–1791, 2003)
this technique, several groups reported a negative association between the IMCL content and insulin sensitivity (4 –7).
In addition, IMCL appears to be acutely regulated. Experimental elevation of free fatty acids resulted in an increase in
IMCL (8, 9), whereas prolonged endurance exercise decreased IMCL (10).
It is still unclear, however, how an augmented aerobic
capacity increases insulin sensitivity while paradoxically being associated with increased IMCL. It is also not known
whether in obesity increased IMCL predicts insulin resistance independent of excess body fat as such and whether
genetic factors play a role in presetting the individual IMCL
content. We therefore analyzed statistical associations between anthropometric data [age, sex, percentage of body fat,
waist to hip ratio (WHR), maximal aerobic power] and IMCL
in a cohort of 105 healthy nondiabetic individuals covering
a broad range of those variables. IMCL was determined by
magnetic resonance spectroscopy both in tibialis anterior
muscle (tib), a fast-twitching, white muscle, and in soleus
muscle (sol), a slow-twitching, red muscle. Furthermore, we
studied the relationship between IMCL and insulin sensitivity (hyperinsulinemic-euglycemic), taking into account
the influence of anthropometric measures, especially the role
1785
1786
J Clin Endocrinol Metab, April 2003, 88(4):1785–1791
Thamer et al. • Determinants of IMCL
of maximal aerobic capacity (VO2max). The relatively large
size of our cohort permitted use of multivariate linear regression analysis to identify relevant and independent
relationships.
Subjects and Methods
Subjects
A total of 105 healthy nondiabetic subjects were recruited from the
Tübingen Family Study for Type 2 Diabetes (see Table 1 for details) and
underwent determination of IMCL as described below. In a subgroup
of 77 of these subjects, an exercise test to volitional exhaustion on a cycle
ergometer to determine maximal oxygen uptake (VO2max).
All subjects underwent the standard preparatory procedures and
investigations of the protocol of the Tübingen Family Study (medical
history, physical examination, routine blood test, and oral glucose tolerance test). The protocols were approved by the local ethics committee.
After the nature of the study was explained, all subjects gave informed
written consent.
Determination of IMCL by MRS
Neutral lipids within the muscle cell (IMCL) and those interlaced
between the muscle fibers (EMCL) can be differentiated due to their
geometrical arrangement using proton magnetic resonance spectroscopy (1H-MRS; Refs. 11 and 12). Localized image guided proton spectra
of the tib anterior muscle representing a muscle of mixed type I and II
fibers and of the sol representing a muscle of predominantly type I fibers
with high oxidative capacity were acquired on a 1.5-Tesla whole body
imager (Magnetom Vision, Siemens, Erlangen, Germany). For volume
selection, a single voxel STEAM technique was applied. Measurement
parameters: echo time ⫽ 10 msec, repetition time ⫽ 2 sec, volume of
interest 11 ⫻ 11 ⫻ 20 mm3, 40 acquisitions. IMCL and EMCL were
quantified as previously described (11, 12).
Euglycemic hyperinsulinemic clamp
After a 12-h overnight fast around 0700 h, an antecubital vein was
cannulated for infusion of insulin and glucose. A dorsal hand vein of the
contra lateral arm was cannulated and placed under a heating device to
permit sampling of arterialized blood. After basal blood was drawn,
subjects received a primed insulin infusion at a rate of 1.0
mU䡠kg⫺1䡠min⫺1 for 2 h. Blood was drawn every 5 min for determination
of blood glucose, and a glucose infusion was adjusted appropriately to
maintain the fasting glucose level. An insulin sensitivity index (ISI; in
␮mol䡠kg⫺1䡠min⫺1䡠pm⫺1) for systemic glucose uptake was calculated as
the mean infusion rate of glucose (in ␮mol䡠kg⫺1䡠min –1) necessary to
maintain euglycemia during the last 60 min of the euglycemic hyperinsulinemic clamp divided by the steady state serum insulin concentration.
Measurement of VO2max
Subjects underwent continuous, incremental exercise test to volitional
exhaustion on a cycle ergometer. The cycle ergometer test was per-
formed on an electromagnetically braked cycle ergometer (ergometrics
800 S, Ergoline, Bitz, Germany). Oxygen consumption was measured
using a spiroergometer (MedGraphics System Breese Ex 3.02 A, St. Paul,
MN). Subjects were asked to choose a pedaling rate of 60 rpm and to
maintain that rate throughout the test. After a 2-min warm-up period at
0 W, the test was initiated at an initial power output of 20 W. Stepwise
increments of 40 W were made every minute until exhaustion. VO2max
is expressed as percentage of predicted peak VO2 (ml/min) based on sex,
age, and body mass index as described previously (13). For the purpose
of the present analyses, we used this parameter as measure of physical
fitness.
Statistical analyses
All data are given as mean ⫾ sem unless otherwise stated. The
statistical software package JMP (SAS Institute Inc., Cary, NC) was used
for statistical analyses. Age, percentage body fat, VO2max, and ISI were
not normally distributed and log transformed before analyses.
Simple linear regression was applied to examine the relationships
between age, percentage body fat, WHR, VO2max, and IMCL, and
Pearson’s regression coefficient was calculated. Multiple linear regression models with sex, age, percentage body fat, WHR and VO2max as
independent variables were used to calculate the predictive effect on the
dependent variable IMCL. The effect of the interaction term between
IMCL and VO2max on insulin sensitivity was calculated using multiple
linear regression models after adjusting for sex, age, percentage body fat,
and WHR. Stepwise linear regression was used to determine the variation in ISI explained by the interaction term between IMCL and
VO2max.
Three-dimensional mesh graphs were used to illustrate interaction
effects in a qualitative fashion. The x and y variable were divided into
a 4 ⫻ 4 grid, resulting in 16 categories. The mean z variable of the points
contained in each category was used to construct the mesh. z values for
empty categories were interpolated. The inverse distance method as
provided by the software package SigmaPlot (SPSS, Inc., Richmond, CA)
was used with values for weight of three and for intervals of four. Several
iterations were performed until a smooth surface was obtained.
Results
Determinants of IMCL
IMCL was found to be higher in sol compared with tib
[12.0 ⫾ 0.5 arbitrary units (AU); range 3.3–34.0 AU] vs. 4.0 ⫾
0.2 AU (range 0.7 – 10.1 AU; P ⬍ 0.0001). In the whole cohort,
IMCL (tib) was highly correlated with IMCL (sol, r ⫽ 0.43;
P ⬍ 0.0001; Fig. 1). IMCL (tib) was slightly higher in females
than males (4.3 ⫾ 0.3 vs. 3.8 ⫾ 0.2 AU), but this did not reach
statistical significance (P ⫽ 0.1). IMCL (sol) was significantly
higher in males than in females (12.8 ⫾ 0.7 vs. 9.8 ⫾ 0.6 AU,
P ⫽ 0.01). IMCL (tib) was significantly correlated with percentage body fat (r ⫽ 0.28, P ⫽ 0.01) but not with age (P ⫽
0.77), WHR (P ⫽ 0.15) or waist circumference (P ⫽ 0.11).
IMCL (sol) was correlated with WHR (r ⫽ 0.41, P ⬍ 0.0001),
TABLE 1. Subject characteristics
All subjects (mean ⫾
n (female/male)
Age (yr)
BMI (kg/m2)
WHR
% Body fat
Fasting glucose [OGTT (mmol䡠l⫺1)]
2-h Glucose [OGTT (mmol䡠l⫺1)]
IMCL (tib) (AU)
IMCL (sol) (AU)
ISI (␮mol䡠(kg䡠min䡠pM)⫺1
Maximal aerobic power (% predicted)
BMI, Body mass index; OGTT, Oral glucose tolerance test.
105 (28/77)
29 ⫾ 6
24.6 ⫾ 5
0.85 ⫾ 0.08
21 ⫾ 8
4.8 ⫾ 0.5
5.2 ⫾ 1.4
4.0 ⫾ 2
12.0 ⫾ 5
0.107 ⫾ 0.06
SD)
Subjects with data on maximal
aerobic power (mean ⫾ SD)
77 (25/52)
29 ⫾ 6
24.2 ⫾ 4.7
0.84 ⫾ 0.08
21 ⫾ 9
4.7 ⫾ 0.5
5.1 ⫾ 1.3
3.8 ⫾ 2
11.3 ⫾ 4
0.113 ⫾ 0.06
109 ⫾ 29
Thamer et al. • Determinants of IMCL
J Clin Endocrinol Metab, April 2003, 88(4):1785–1791 1787
waist circumference (r ⫽ 0.31, P ⫽ 0.001) and age (r ⫽ 0.17,
P ⫽ 0.06) but not with percentage body fat (P ⫽ 0.64).
VO2max was not correlated with IMCL neither in the tib (P ⫽
0.46) nor in the sol (P ⫽ 0.54).
In multivariate regression analyses, IMCL (tib) was found
to be significantly higher in females. Percentage of body fat
remained an independent determinant of IMCL in both muscle types. Because WHR and percentage of body fat are
highly correlated, the models were similar when replacing
percentage of body fat by WHR. Interestingly, WHR had a
stronger effect than percentage of body fat when both variables were included (last model in Tables 2 and 3). In sol,
WHR explained most of the univariate sex difference. Values
predicted by a model with sex, age, percentage of body fat,
and WHR as independent variables explained 14% and 18%
of the variation in IMCL (tib) and IMCL (sol), respectively.
The r2 for the models predicting IMCL (tib) improved upon
inclusion of VO2max as additional independent variable
(r2 ⫽ 0.19, P ⫽ 0.01) with aerobic power being positively
correlated. The relationships with percentage of body fat and
WHR were unaffected for IMCL (tib). VO2max was not an
independent predictor of IMCL (sol; Tables 2 and 3).
Adding an interaction term for percentage body fat and
VO2max as independent variable further improved the quality of the model with IMCL (tib) as dependent variable (r2 ⫽
0.26, P ⫽ 0.001) but not those with IMCL (sol) as dependent
variable (Table 3). The relationship between adiposity, aerobic power and IMCL (tib) is illustrated in Fig. 2. In lean
subjects IMCL (tib) increases with increasing VO2max. With
increasing obesity this relationship disappears.
IMCL and insulin sensitivity
In the entire group (n ⫽ 105), insulin sensitivity was not
correlated with IMCL (sol) (r ⫽ ⫺0.03, P ⫽ 0.72) or IMCL (tib)
(r ⫽ ⫺0.13, P ⫽ 0.2). This was similar in the subgroup with
aerobic power measurements available (n ⫽ 77, r ⫽ 0.1, P ⫽
0.35, sol; and r ⫽ ⫺0.07 P ⫽ 0.53, tib). In a multivariate
regression analysis with ISI as dependent variable neither
IMCL (tib; P ⫽ 0.76) nor IMCL (sol; P ⫽ 0.35) were independent determinants of ISI after adjusting for age, percentage of body fat and VO2max (Table 4, model 1). Interestingly,
for both muscle types an interaction term between IMCL and
VO2max was an independent predictor of insulin sensitivity.
This relationship remained significant after adjusting for sex,
age, percentage of body fat, and VO2max (see Table 4). The
interaction effect between IMCL (tib) and VO2max on ISI is
illustrated in Fig. 3. While in subjects with low aerobic power,
ISI was negatively correlated with IMCL (tib), it was positively correlated in subjects with high aerobic fitness.
In stepwise regression analysis, the interaction term for
IMCL (tib) and VO2max entered first explaining 52% of the
variation in ISI. The analogous interaction term for sol was
weaker and entered third after VO2max and percentage of
body fat explaining an extra 3% of the variation in ISI.
Discussion
Aim of the present study was to identify anthropometric
determinants of the IMCL content as measured by 1H-MRS
and to analyze the correlation with insulin action in a large
cohort. IMCL was measured in the predominantly nonoxidative tib anterior muscle and in the predominantly oxidative sol. Although the IMCL content in the two muscles was
significantly correlated, a high individual variability remained (Fig. 1). Moreover, the anthropometric parameters as
statistical determinants of IMCL were clearly different between the two muscle types. This probably reflects differences in fiber type composition. For example, VO2max was
identified as an important determinant of IMCL (tib) but not
FIG. 1. Correlation between IMCL (tib) and IMCL (sol).
TABLE 2. Results of multivariate regression with IMCL as dependent variable (n ⫽ 105)
Independent variable
M. tibialis anterior
Estimate
SE
Sex
Age (log)
0.09
⫺0.08
0.06
0.24
0.10
0.74
Sex
Age (log)
% Body fat (log)
0.03
⫺0.09
0.32
0.06
0.24
0.13
0.58
0.69
0.02
Sex
Age (log)
WHR
0.22
⫺0.47
2.66
0.07
0.26
0.80
0.001
0.07
0.001
Sex
Age (log)
% Body fat (log)
WHR
0.17
⫺0.41
0.15
2.21
0.08
0.27
0.15
0.92
0.04
0.13
0.32
0.02
P
M. soleus
Model r2
Estimate
SE
P
0.25
0.03
⫺0.13
0.37
0.04
0.18
0.002
0.04
0.04
0.08
⫺0.17
0.36
0.19
0.05
0.18
0.10
⬍0.01
0.05
0.06
⬍0.01
0.12
⫺0.05
0.12
1.68
0.05
0.20
0.61
0.28
0.53
0.01
⬍0.01
0.18
0.01
0.14
⫺0.08
0.15
0.07
1.46
0.06
0.20
0.11
0.70
0.21
0.45
0.52
0.04
⬍0.01
0.18
Model P
Model P
Model r2
0.002
0.12
0.001
0.15
1788
J Clin Endocrinol Metab, April 2003, 88(4):1785–1791
Thamer et al. • Determinants of IMCL
TABLE 3. Results of multivariate regression with IMCL as dependent variable in subjects with VO2max measurement available (n ⫽ 77)
Independent variable
M. tibialis anterior
M. soleus
Model r2
Estimate
SE
P
Model P
Model r2
0.04
0.13
⫺0.14
0.30
0.11
0.08
0.05
0.21
0.13
0.20
0.01
0.17
0.39
0.70
0.06
0.12
0.002
0.04
0.01
0.03
0.02
0.15
⫺0.06
0.12
1.29
0.13
0.06
0.24
0.77
0.19
0.30
0.63
0.10
0.50
0.02
0.15
0.09
0.32
1.05
0.18
0.27
0.08
0.04
0.03
0.08
0.01
0.01
0.19
⫺0.08
0.12
1.20
0.06
0.16
0.07
0.24
0.80
0.13
0.20
0.28
0.62
0.14
0.68
0.44
0.05
0.15
0.05
⫺0.34
0.37
0.54
⫺1.63
0.07
0.27
0.17
0.25
0.55
0.48
0.22
0.03
0.04
⬍0.01
0.002
0.22
⫺0.14
0.30
0.11
0.08
0.06
0.05
0.21
0.13
0.20
0.44
0.01
0.17
0.39
0.7
0.9
0.11
0.12
0.15
⫺0.62
1.87
0.29
0.67
⫺1.43
0.09
0.31
1.04
0.17
0.26
0.56
0.09
0.05
0.08
0.10
0.01
0.01
0.001
0.26
⫺0.07
0.11
1.27
0.06
0.16
0.19
0.07
0.25
0.82
0.14
0.21
0.44
0.29
0.65
0.13
0.66
0.43
0.66
0.08
0.15
Estimate
SE
P
Sex
Age (log)
% Body fat (log)
VO2max (log)
0.03
⫺0.30
0.43
0.56
0.07
0.29
0.17
0.27
0.67
0.30
0.02
0.04
Sex
Age (log)
WHR
VO2max (log)
0.25
⫺0.69
2.94
0.56
0.08
0.33
1.03
0.26
Sex
Age (log)
WHR
% Body fat (log)
VO2max (log)
0.16
⫺0.67
2.41
0.32
0.72
Sex
Age (log)
% Body fat (log)
VO2max (log)
VO2max (log) ⫻ % Body fat (log)
Sex
Age (log)
WHR
% Body fat (log)
VO2max (log)
VO2max (log) ⫻ % Body fat (log)
Model P
FIG. 2. Three dimensional relationship between
VO2max, percentage of body fat and IMCL in tib
anterior muscle. The 4 ⫻ 4 grid was generated by
dividing both VO2max and percentage of body fat
into four equal intervals. The mean IMCL value of
the subjects contained in one square was used to
construct the mesh graph. In lean subjects,
VO2max was positively correlated with IMCL. In
untrained individuals, increased percentage of
body fat was strongly associated with increased
IMCL (please note that the category with highest
VO2max and highest percentage of body fat contained no data points).
of IMCL (sol; Table 3). Soleus muscle is rich in oxidative,
slow-twitch type I fibers, whereas tibialis anterior muscle is
rich in fast-twitch type 2 fibers (14). Therefore, the lack of an
association with the predominantly oxidative muscle may
reflect the higher and possibly compensating oxidative capacity of sol in response to endurance training that could
prevent accumulation of lipids. In addition, the MR spectroscopic measurement of IMCL in sol is generally more
difficult and characterized by a greater experimental variation (15).
Determinants of the IMCL content differed between the
two muscle types. In this study, a relatively weak correlation
between muscle lipids and measures of obesity was observed. IMCL (tib) was correlated with percentage of body
fat, whereas the best single predictor of IMCL (sol) was
WHR. This suggests that adiposity as such reflecting caloric
oversupply has a weak but significant effect on the IMCL
content. This appears to be at variance with a previous report
finding no significant association between percentage of
body fat and skeletal muscle triglyceride contents in 37 Pima
Thamer et al. • Determinants of IMCL
J Clin Endocrinol Metab, April 2003, 88(4):1785–1791 1789
TABLE 4. Results of multivariate regression with ISI as dependent variable and IMCL (tib) (left part of the table) and IMCL (sol) (right
part of the table) as independent variable
Independent variable
IMCL (tib) as independent
IMCL (sol) as independent
Estimate
SE
P
Model P
Model r2
Estimate
SE
P
Model P
Model r2
Sex
Age (log)
% Body fat (log)
VO2max (log)
IMCL (log)
0.07
0.15
⫺0.58
1.21
⫺0.03
0.06
0.23
0.14
0.22
0.09
0.25
0.53
0.0001
⬍0.0001
0.76
⬍0.0001
0.58
0.08
0.12
⫺0.61
1.18
0.12
0.06
0.23
0.14
0.21
0.13
0.17
0.61
⬍0.0001
⬍0.0001
0.35
⬍0.0001
0.58
VO2max (log)
IMCL (log)
IMCL (log) ⫻ VO2max (log)
1.54
⫺0.12
0.96
0.20
0.09
0.35
⬍0.0001
0.19
0.01
⬍0.0001
0.52
1.55
0.20
0.95
0.21
0.14
0.53
⬍0.0001
0.16
0.08
⬍0.0001
0.48
% Body fat (log)
VO2max (log)
IMCL (log)
IMCL (log) ⫻ VO2max (log)
⫺0.44
1.23
⫺0.02
0.72
0.13
0.21
0.09
0.33
⬍0.001
⬍0.0001
0.79
0.03
⬍0.0001
0.59
⫺0.51
1.18
0.19
1.01
0.12
0.20
0.13
0.47
⬍0.0001
⬍0.0001
0.14
0.04
⬍0.0001
0.59
Sex
Age (log)
% Body fat (log)
VO2max (log)
IMCL (log)
IMCL (log) ⫻ VO2max (log)
0.06
0.10
⫺0.52
1.18
⫺0.02
0.70
0.06
0.23
0.14
0.21
0.09
0.33
0.26
0.66
0.001
⬍0.0001
0.79
0.04
⬍0.0001
0.60
0.10
0.07
⫺0.63
1.11
0.26
1.10
0.06
0.23
0.14
0.21
0.14
0.47
0.09
0.76
⬍0.0001
⬍0.0001
0.06
0.02
⬍0.0001
0.61
FIG. 3. Three-dimensional
relationship between VO2max, IMCL in tib anterior muscle and insulin sensitivity.
The 4 ⫻ 4 grid was generated by dividing both VO2max and IMCL (tib) into
four equal intervals. The mean IMCL
value of the subjects contained in one
square was used to construct the mesh
graph. In untrained individuals high
IMCL predicts low insulin sensitivity.
In highly trained individuals, this relationship is reversed and high IMCL predicts high insulin sensitivity.
Indians (2). In this study, however, only total muscle triglycerides were measured by a lipid extraction method and no
distinction between intramyocellular vs. extramyocellular localizations could be made. Moreover, the relatively weak
association with adiposity may become detectable only with
greater samples sizes.
The weak association of IMCL (tib) with sex disappeared
in multivariate analyses. The association of IMCL (sol) with
sex became weaker and depended on the parameters included in the model. In some instances, sex effects were
explained by WHR, a crude measure of body fat distribution.
In multivariate regression analyses WHR was the best predictor of both IMCL (tib) and IMCL (sol) (model 2, Table 3).
It is possible that, just like for other fat depots, hormonal
factors may be involved in the regulation of IMCL.
The performance of the models predicting IMCL (tib) improved markedly upon inclusion of VO2max as independent
variable. After adjusting for percentage of body fat VO2max
was positively correlated with IMCL. Moreover, the model
performed even better after including an interaction term
1790 J Clin Endocrinol Metab, April 2003, 88(4):1785–1791
between VO2max and percentage of body fat as indicated by
the increase in r2 from 0.19 – 0.26 (Table 3). In plain words, the
correlation between IMCL (tib) and adiposity depends on
VO2max (Fig. 2).
In our entire study, population insulin sensitivity and
IMCL in either muscle were not correlated. This contrasts
with previous reports including one from our laboratory on
small well-matched groups in which IMCL were shown to be
increased in insulin-resistant individuals (4, 6, 7). The underlying mechanisms were proposed to include elevation of
intramyocellular fatty acid metabolites such as diacyl-glycerol, fatty acyl coenzyme A and ceramides and, consequently, reduced insulin-induced tyrosine phosphorylation
of the insulin receptor and phosphatidylinositol 3-kinase
activity (7, 16).
More recent studies indicated that the relationship between IMCL and insulin resistance is more complex and
influenced by the oxidative capacity of skeletal muscle (17).
The oxidative capacity of skeletal muscle, which increases
with training, is in part assessed by our measurement of
VO2max. We identified a striking and statistically significant
interaction effect between the VO2max and the IMCL content
on insulin sensitivity in both muscle types (Fig. 3). The interaction effect indicates that the relationship between IMCL
and ISI is modified by aerobic fitness. Only in untrained
individuals, high IMCL predicted low insulin sensitivity. In
trained athletes, this relationship was reversed and high
IMCL predicted high insulin sensitivity. Our findings are
generally compatible with results of Goodpaster and colleagues (17, 18), demonstrating increased intramyocellular
lipid using histochemical lipid staining in endurance trained
athletes compared with lean controls, whereas insulin sensitivity was comparable. The authors subsequently hypothesized that the association between IMCL content and insulin
sensitivity may be influenced by the oxidative capacity of
skeletal muscle.
A possible explanation for this apparently paradoxical
finding may have to do with the increased glucose transport
capacity in endurance trained muscle (19, 20). Increased glucose transport would facilitate im triglyceride synthesis and
at the same time improve insulin sensitivity. This mechanism
is not operative or not visible in untrained subjects because
here IMCL accumulation driven by chronic caloric (particularly fat) oversupply dominates the effect of training. Consistent with this hypothesis, in healthy individuals an increase in IMCL could be rapidly induced by intralipid plus
heparin infusion and high fat diet (8, 21). In addition, IMCL
may exist in more than one compartment—spatial or functional—that are indistinguishable by the MRS technique.
And IMCL of a trained athlete may be localized in a different
compartment than IMCL of an obese, insulin-resistant
individual.
In conclusion, in this large heterogeneous cohort of
healthy, nondiabetic subjects, percentage of body fat, and
VO2max were independent determinants of IMCL (tib) but
not of IMCL (sol). It is possible that the ratio of IMCL (sol)
over IMCL (tib) harbors additional information yet to be
uncovered. A significant interaction effect between VO2max
and percentage of body fat on IMCL (tib) was detected suggesting that the association between VO2max and IMCL (tib)
Thamer et al. • Determinants of IMCL
is influenced by presence or absence of obesity. Insulin sensitivity is a negative function of IMCL only in untrained
subjects, whereas in endurance-trained subjects high IMCL
actually predicted high insulin sensitivity. Thus, measurements of aerobic fitness and body fat are indispensable for
the interpretation of IMCL measurements and its relationship with insulin sensitivity.
Acknowledgments
We thank our volunteers for their participation.
Received October 28, 2002. Accepted January 15, 2003.
Address all correspondence and requests for reprints to: Dr. Michael
Stumvoll, Medizinische Universitätsklinik, Otfried-Müller-Strasse 10,
D-72076 Tübingen, Germany. E-mail: [email protected].
This work was supported by a grant from the Deutsche Forschungsgemeinschaft (DFG No. JA-1005/1-1, DFG Stu-192/9-1), the Federal
Ministry of Education and Research (Fö. 01KS9602), a grant from the
European Community (QLRT-1999-00674) and the Interdisciplinary
Center of Clinical Research Tübingen (IZKF).
References
1. Dagenais GR, Tancredi RG, Zierler KL 1976 Free fatty acid oxidation by
forearm muscle at rest, and evidence for an intramuscular lipid pool in the
human forearm. J Clin Invest 58:421– 431
2. Pan DA, Lillioja S, Kriketos AD, Milner MR, Baur LA, Bogardus C, Jenkins
AB, Storlien LH 1997 Skeletal muscle triglyceride levels are inversely related
to insulin action. Diabetes 46:983–988
3. Szczepaniak LS, Babcock EE, Schick F, Dobbins RL, Garg A, Burns DK,
McGarry JD, Stein DT 1999 Measurement of intracellular triglyceride stores
by H spectroscopy: validation in vivo. Am J Physiol 276(5 Part 1):E977–E989
4. Jacob S, Machann J, Rett K, Brechtel K, Volk A, Renn W, Maerker E, Matthaei
S, Schick F, Claussen CD, Haring HU 1999 Association of increased intramyocellular lipid content with insulin resistance in lean nondiabetic offspring of
type 2 diabetic subjects. Diabetes 48:1113–1119
5. Krssak M, Falk Petersen K, Dresner A, DiPietro L, Vogel SM, Rothman DL,
Roden M, Shulman GI 1999 Intramyocellular lipid concentrations are correlated with insulin sensitivity in humans: a 1H NMR spectroscopy study.
Diabetologia 42:113–116
6. Perseghin G, Scifo P, De Cobelli F, Pagliato E, Battezzati A, Arcelloni C,
Vanzulli A, Testolin G, Pozza G, Del Maschio A, Luzi L 1999 Intramyocellular triglyceride content is a determinant of in vivo insulin resistance in
humans: a 1H–13C nuclear magnetic resonance spectroscopy assessment in
offspring of type 2 diabetic parents. Diabetes 48:1600 –1606
7. Virkamaki A, Korsheninnikova E, Seppala-Lindroos A, Vehkavaara S, Goto
T, Halavaara J, Hakkinen AM, Yki-Jarvinen H 2001 Intramyocellular lipid is
associated with resistance to in vivo insulin actions on glucose uptake, antilipolysis, and early insulin signaling pathways in human skeletal muscle.
Diabetes 50:2337–2343
8. Bachmann OP, Dahl DB, Brechtel K, Machann J, Haap M, Maier T, Loviscach
M, Stumvoll M, Claussen CD, Schick F, Haring HU, Jacob S 2001 Effects of
intravenous and dietary lipid challenge on intramyocellular lipid content and
the relation with insulin sensitivity in humans. Diabetes 50:2579 –2584
9. Boden G 1997 Role of fatty acids in the pathogenesis of insulin resistance and
NIDDM. Diabetes 46:3–10
10. Brechtel K, Niess AM, Machann J, Rett K, Schick F, Claussen CD, Dickhuth
HH, Haering HU, Jacob S 2001 Utilisation of intramyocellular lipids (IMCLs)
during exercise as assessed by proton magnetic resonance spectroscopy (1HMRS). Horm Metab Res 33:63– 66
11. Schick F, Eismann B, Jung WI, Bongers H, Bunse M, Lutz O 1993 Comparison
of localized proton NMR signals of skeletal muscle and fat tissue in vivo: two
lipid compartments in muscle tissue. Magn Reson Med 29:158 –167
12. Boesch C, Slotboom J, Hoppeler H, Kreis R 1997 In vivo determination of
intra-myocellular lipids in human muscle by means of localized 1H-MRspectroscopy. Magn Reson Med 37:484 – 493
13. Hansen JE, Sue DY, Wasserman K 1984 Predicted values for clinical exercise
testing. Am Rev Respir Dis 129(2 Part 2):S49 –S55
14. Polgar J, Johnson MA, Weightman D, Appleton D 1973 Data on fibre size in
thirty-six human muscles. An autopsy study. J Neurol Sci 19:307–318
15. Brechtel K, Machann J, Jacob S, Strempfer A, Schick F, Haring HU, Claussen
CD 1999 In-vivo 1H-MR spectroscopy: the determination of the intra- and
extramyocellular lipid content depending on the insulin effect in the direct
Thamer et al. • Determinants of IMCL
J Clin Endocrinol Metab, April 2003, 88(4):1785–1791 1791
offspring of type-2 diabetics. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb
Verfahr 171:113–120
16. Shulman GI 2000 Cellular mechanisms of insulin resistance. J Clin Invest
106:171–176
17. Goodpaster BH, He J, Watkins S, Kelley DE 2001 Skeletal muscle lipid content
and insulin resistance: evidence for a paradox in endurance-trained athletes.
J Clin Endocrinol Metab 86:5755–5761
18. Kelley DE, Goodpaster B, Wing RR, Simoneau JA 1999 Skeletal muscle fatty
acid metabolism in association with insulin resistance, obesity, and weight loss.
Am J Physiol 277(6 Part 1):E1130 –E1141
19. Etgen Jr GJ, Jensen J, Wilson CM, Hunt DG, Cushman SW, Ivy JL 1997
Exercise training reverses insulin resistance in muscle by enhanced recruitment of GLUT-4 to the cell surface. Am J Physiol 272(5 Part 1):E864 –E869
20. Ploug T, Stallknecht BM, Pedersen O, Kahn BB, Ohkuwa T, Vinten J, Galbo
H 1990 Effect of endurance training on glucose transport capacity and glucose
transporter expression in rat skeletal muscle. Am J Physiol 259(6 Part 1):E778 –
E786
21. Boden G, Lebed B, Schatz M, Homko C, Lemieux S 2001 Effects of acute
changes of plasma free fatty acids on intramyocellular fat content and insulin
resistance in healthy subjects. Diabetes 50:1612–1617
IOF World Congress on Osteoporosis
May 14 –18, 2004
Rio de Janeiro, Brazil
Abstract deadline: November 14, 2003
For further information, registration and electronic abstract submission, visit www.osteofound.org or contact
the Congress Secretariat at [email protected]. For information about fellowship awards for research
available through the IOF, please contact www.osteofound.org.