Clinical Science (2002) 102, 345–352 (Printed in Great Britain)
Alcohol consumption, the metabolic syndrome
and insulin resistance in 58-year-old clinically
healthy men (AIR study)
Daniel GOUDE*, Bjo$ rn FAGERBERG† and Johannes HULTHE*, on behalf of the AIR
study group
*Wallenberg Laboratory for Cardiovascular Research, Sahlgrenska University Hospital, Go$ teborg University, 413 45 Gothenburg,
Sweden, and †Department of Medicine, Sahlgrenska University Hospital, Go$ teborg University, 413 45 Gothenburg, Sweden
A
B
S
T
R
A
C
T
It has been shown that a light-to-moderate intake of alcohol may enhance insulin sensitivity ; a
decrease in insulin sensitivity is a component of the clustering of risk factors known as the
metabolic syndrome. However, previous studies have been limited to relatively small or
heterogeneous study groups, or have used suboptimal methods of measuring insulin action.
Hence the aim of the present study was to examine whether the metabolic syndrome (as recently
defined), components of this syndrome and smoking are associated with alcohol consumption.
The study was performed in a population-based sample of clinically healthy men (n l 391), all 58
years old and not undergoing any treatment with cardiovascular drugs. Insulin-mediated glucose
uptake (euglycaemic hyperinsulinaemic clamp) was measured in a subgroup of these subjects
(n l 104). Trend analysis showed no difference in alcohol intake across the groups of men with
none of the criteria in the definition of the metabolic syndrome (n l 77), men with one or more
of the criteria (n l 252) and men fulfilling all criteria (n l 62). However, in the whole group
(n l 391), alcohol consumption was significantly positively associated with serum triacylglycerols
(triglycerides), high-density lipoprotein (HDL) cholesterol and cigarette-years. Furthermore,
alcohol consumption was positively associated with insulin-mediated glucose uptake (r l 0.20,
P 0.05). In multiple regression analyses, body mass index, alcohol consumption and serum
triacylglycerols were independent co-variates to insulin-mediated glucose uptake. Thus, in 58year-old healthy men recruited from the general population, there was a significant association
between alcohol consumption, serum triacylglycerols, HDL cholesterol and cigarette-years. In a
subgroup of 104 subjects, alcohol consumption was independently and positively associated with
insulin-mediated glucose uptake. To our knowledge, this is the first study to show an independent
relationship between insulin sensitivity, as measured by the clamp technique, and alcohol intake.
INTRODUCTION
A growing number of studies support the hypothesis that
regular light-to-moderate alcohol consumption is beneficial in lowering the risk of coronary heart disease [1,2].
It has also been shown previously that chronic light-to-
moderate alcohol intake may enhance insulin sensitivity ;
a decrease in insulin sensitivity is a component of the
clustering of risk factors known as the metabolic syndrome. However, these studies have been limited to
relatively small or heterogeneous study groups, or have
involved indirect methods of measuring insulin action.
Key words: alcohol consumption, insulin resistance, metabolic syndrome, smoking.
Abbreviations: BMI, body mass index ; FFM, fat-free mass ; GIR, glucose infusion rate ; GIR
, GIR adjusted for FFM ; HDL,
FFM
high-density lipoprotein ; LDL, low-density lipoprotein ; WHR, waist\hip ratio.
Correspondence : Dr Johannes Hulthe (e-mail johannes.hulthe!wlab.wall.gu.se).
# 2002 The Biochemical Society and the Medical Research Society
345
346
D. Goude, B. Fagerberg and J. Hulthe
The methods used to disclose insulin sensitivity have
varied from quantification of surrogate variables such as
fasting plasma insulin [3] and insulin levels before and
after oral glucose tolerance tests [4,5], to the insulin
suppression test [5]. Yet authorities concur that the
‘ reference technique ’ for measuring insulin sensitivity is
by utilizing the euglycaemic hyperinsulinaemic clamp
technique [6,7].
It has also been postulated that alcohol conveys its
positive effects on coronary heart disease by producing
an altered lipid and lipoprotein profile. Indeed, it has
been shown that regular moderate alcohol intake may be
linked to raised levels of high-density liporotein (HDL)
[2,3,8,9]. Alcohol intake has, however, also been shown
to be associated with raised triacylglycerol (triglyceride)
levels [10]. Interestingly, both of these variables are
included in the definition of the metabolic syndrome.
These observations raise the question : is there an association between the aforementioned variables and
alcohol intake ?
Hence the aim of the present study was to assess the
relationship between alcohol intake and the presence of
factors that comprise the metabolic syndrome. Furthermore we aimed to investigate the relationship between alcohol consumption and insulin resistance, as
measured by the euglycaemic clamp technique. The
results were obtained in a population-based sample of
healthy men of Swedish ancestry, all 58 years old, selected
in order to minimize the effects of confounding factors
such as race, sex, age and treatment with different drugs
for cardiovascular disease.
METHODS
Study subjects
This study has been described in detail elsewhere [11].
The inclusion criteria were age 58 years, male sex, and
Swedish ancestry. Exclusion criteria were cardiovascular
or other clinically overt disease, treatment with cardiovascular drugs that might disturb the measurements
performed in the study, or unwillingness to participate.
The subjects were selected randomly from among men in
the County Council register and were invited to a
screening examination.
A power calculation indicated that it was necessary to
recruit at least 390 men into the study, with the main
objective being to examine the relationship between
insulin sensitivity and ultrasound-assessed atherosclerosis. The present report is a substudy within that project.
The screening protocol aimed at enriching the population
sample with men with low or high insulin sensitivity.
Thus, in connection with the screening examination, the
subjects were divided into quintiles of a body mass index
(BMI)\blood glucose score, which allowed immediate
# 2002 The Biochemical Society and the Medical Research Society
stratification and selection for further studies (n l 818).
The following equation was used :
BMI\blood glucose score l
46.22k21.27iBMIk0.84i(whole-body glucose)
This algorithm was based on results from a previous
study of clinically healthy men of similar age who had
undergone a euglycaemic hyperinsulinaemic clamp examination. The correlation coefficient between the BMI\
blood glucose score and the observed insulin sensitivity
was 0.81. In the present population sample this score was
correlated significantly with insulin sensitivity measured
with the euglycaemic hyperinsulinaemic clamp method,
when expressed as insulin-mediated glucose uptake
adjusted either for body weight (r l 0.69, P 0.001) or
for fat-free mass (FFM) (r l 0.59, P 0.001) (n l 104).
Following the screening examination (n l 818), every
subject in quintile 1 (indicating low insulin sensitivity ;
n l 153 ; 39 % of the final study population of 391) and
quintile 5 (indicating high sensitivity ; n l 144 ; 37 %),
and every fifth man in quintiles 2–4 (indicating intermediate sensitivity ; n l 94 ; 24 %), was invited to
further examinations (n l 391). Among these 391 subjects, 22 were found to have a fasting blood glucose
concentration of 6.1 mmol\l. However, none of the
subjects had overt diabetes mellitus. From the main study
group (n l 391), every fourth subject was randomly
selected for a euglycaemic hyperinsulinaemic clamp
examination (n l 104). Subjects in the clamp subgroup
(n l 104) were distributed as follows : quintile 1, n l 36
(35 % of the subpopulation of 104) ; quintile 2–4, n l 28
(27 %) ; quintile 5, n l 40 (38 %). Insulin-mediated glucose uptake was 4.9p2.7, 9.7p2.4 and 9.9p2.8 mg :
min−" : kg−" FFM for those in score-based quintiles 1, 2–4
and 5 respectively (P 0.001 for trend). This classification was only used in order to recruit subjects with a
wide range of insulin sensitivities, and was not used when
analysing study results.
The subjects received both written and oral information before they gave their consent to participate. The
study was approved by the Ethics Committee at Sahlgrenska University Hospital.
Definition of the metabolic syndrome
The definition suggested by a working group consulted
by WHO in 1998 [12] was used. The metabolic syndrome
is defined as glucose intolerance and\or insulin resistance
(see below), together with two or more of the following
risk factors. (1) Raised arterial pressure 160\90 mmHg
(either value). (2) Raised triacylglycerols ( 1.7 mmol\l)
and\or low HDL cholesterol ( 0.9 mmol\l). (3) Abdominal obesity [waist\hip ratio (WHR) 0.90] and\or
BMI 30. (4) Microalbuminuria (urinary albumin excretion rate 20 µg\min or albumin\creatinine ratio 20 mg\g). In the present study subjects with a dipstick
Alcohol consumption and the metabolic syndrome
score of 1j were also included in this category, whereas
those with a dipstick score of more than 1j or a urinary
excretion rate 208 µg\min were excluded from this
category.
Glucose intolerance was defined as fasting blood
glucose 5.6 mmol\l. Insulin resistance was defined as
fasting plasma insulin 14.86 m-units\l. This definition
was obtained by using the euglycaemic hyperinsulinaemic clamp method to define glucose uptake below the
lowest quartile for the background population under
investigation in a representative sample of 50 subjects
from the present study. The plasma insulin level corresponding to the lowest quartile for glucose uptake was
calculated and used as a cut-off point when defining
insulin resistance. The positive and negative predictive
values of this plasma insulin cut-off point were 0.88 and
0.86 respectively in 52 subjects from the same background
population that had undergone a clamp examination and
who were not included in the first calculation above.
The study group (n l 391) was divided into three
subgroups : (1) subjects fulfilling the criteria for the
metabolic syndrome (n l 62) ; (2) subjects not fulfilling
the criteria, but with one or several criteria used in the
definition of the metabolic syndrome (n l 252) ; and
(3) subjects with none of the risk factors constituting
the metabolic syndrome (n l 77). Furthermore, a subgroup was randomly selected in which a euglycaemic
hyperinsulinaemic clamp examination was carried out
(n l 104).
Measurements
Established questionnaires were used to evaluate history
of previous and current disease, smoking habits and
alcohol consumption [13]. The subjects were asked
whether they or any parent or sibling had suffered from
cardiovascular diseases, diabetes mellitus, hypertension
or obesity.
Body weight was measured on a balance scale with the
subject dressed in underwear. Measurements of waist and
hip circumferences were used to calculate WHR.
At the ultrasound examination, resting blood pressure
was measured phonographically in the right arm after
supine rest. Blood pressure was calculated to the nearest
1 mmHg, and the mean of two recordings was used in the
statistical analyses. A 12-lead standard ECG was recorded. Heart rate was recorded from the ECG. Blood
samples for serum cholesterol, serum triacylglycerols and
lipoprotein fractions were drawn after a fasting period of
at least 6 h and were frozen thereafter in aliquots at
k70 mC within 4 h.
The total number of smoking years was multiplied by
the number of cigarettes smoked daily, and the product
was termed ‘ cigarette-years ’.
The subjects were asked to estimate their consumption
of different types of alcoholic beverages. This infor-
mation was then used in conjunction with data on
percentual alcohol content in order to calculate the total
daily intake of ethanol.
Biochemical analysis
Serum concentrations of total cholesterol and triacylglycerols were determined by fully enzymic techniques.
HDL was determined after precipitation of apolipoprotein B-containing lipoproteins with MnCl and dextran
#
sulphate. Low-density lipoprotein (LDL) was calculated
as described by Friedewald et al. [14]. Whole-blood
glucose was measured with the glucose oxidase technique
[15]. Total plasma insulin was determined in all subjects
by RIA (Pharmacia Insulin RIA ; Pharmacia Diagnostics,
Uppsala, Sweden).
Euglycaemic hyperinsulinaemic clamp
examination
Several weeks before each examination, the individual
subject was asked not to change any habits and, during
the 2 days preceding the day of the examination, to avoid
unusual physical exercise, alcohol consumption or any
major change in caloric intake. Instructions were given to
avoid food intake, as well as any medication, smoking or
snuff-taking, from midnight on the day preceding the
clamp. The subject was allowed to drink water in the
morning. Before the examination started, a questionnaire
was completed in order to verify that the subject had
followed the instructions regarding physical activity,
food intake, alcohol intake and tobacco use. The subject
was also asked if he had any disease symptoms, such as
those indicating respiratory tract infection or fever. In
unclear cases, the doctor responsible decided whether to
cancel the examination. A euglycaemic hyperinsulinaemic clamp examination was then performed as described
by DeFronzo et al. [16].
After the clamp examination, FFM was measured
using the dual-energy X-ray absorptiometry body composition model [17] (Lunar DPX-L). Insulin sensitivity
was calculated as the glucose infusion rate per min during
the final 60 min of infusion, adjusted for FFM (GIRFFM)
[18].
Statistical analysis
All statistical computations were carried out using SPSS
10 for Windows (SPSS Inc., Chicago, IL, U.S.A.). Nonparametric Spearman’s rank correlation test was used in
the correlation analysis. The Mann–Whitney test was
used when comparing means in subjects with varying
degrees of alcohol consumption, subjects with the metabolic syndrome and subjects with no risk factors.
Furthermore, a t-distributed variable was used to calculate 95 % confidence intervals for differences. Mantel’s
# 2002 The Biochemical Society and the Medical Research Society
347
348
D. Goude, B. Fagerberg and J. Hulthe
test for linear association was used to test the relationship
between alcohol intake and the presence of factors that
constitute the metabolic syndrome.
A stepwise multiple regression model was used to
study the determinants of the euglycaemic GIR. A twotailed P value of
0.05 was considered as statistically
significant. All results were calculated for the study
group of 391 men unless otherwise noted.
RESULTS
Characteristics of the subjects by tertiles of
alcohol consumption
The 391 subjects were divided into tertiles according to
alcohol consumption (total median intake 9.77 g\day ;
range 0–129 g\day). The subjects in tertile III, with the
highest intake (median 23.2 g\day ; range 14.2–129 g\
day), differed from those in tertile I (median 0 g\day ;
range 0–5.6 g\day) by having significantly higher mean
values for HDL, serum triacylglycerols, serum cholesterol, pulse pressure, WHR and cigarette-years (Table
1). When tested for trend, HDL, serum triacylglycerols,
serum cholesterol and pulse pressure turned out to be
associated with alcohol intake. There were no significant
differences between the two groups in mean BMI, systolic
or diastolic blood pressure, heart rate, serum LDL
cholesterol or whole-blood glucose.
Alcohol consumption in subjects with the
metabolic syndrome compared with
subjects with no risk factors
When the subjects were divided into groups according to
the presence of risk factors for the metabolic syndrome,
Table 1
Table 2 Alcohol consumption in relation to presence of risk
factors that constitute the metabolic syndrome
Group 1 comprises subjects with the metabolic syndrome ; group 2 comprises
subjects with one or more risk factors for the syndrome ; and group 3 comprises
subjects with no risk factors. Values are meanspS.D. P values for differences are
as follows : group 1 compared with group 3, P l 0.424 ; group 1 compared with
mean of group 2 and group 3, P l 0.472.
Group
Alcohol consumption (g/day)
1 (n l 62)
2 (n l 252)
3 (n l 77)
15.5p20.6
12.9p14.1
11.4p10.5
Mean difference (group 3 minus group 1)
4.09
no significant differences in mean alcohol consumption
were found between those subjects with the metabolic
syndrome (as defined previously) and those with no risk
factors (Table 2). A test for linear association on the same
data did not give further evidence to indicate a statistical
relationship (P l 0.11 for trend).
Alcohol consumption in relation to
metabolic variables, serum lipids, smoking
and insulin sensitivity
In the univariate correlation analyses, alcohol consumption was found to be significantly and positively correlated with HDL cholesterol, serum triacylglycerols and
cigarette-years (Table 3). There was a borderline significant inverse relationship between fasting plasma insulin
and alcohol consumption (r l k0.09, P l 0.065). There
were no significant correlations between alcohol consumption and systolic or diastolic blood pressure, pulse
pressure, heart rate, serum cholesterol, LDL cholesterol,
Comparison of subjects divided into tertiles on the basis of alcohol intake
The 391 subjects were divided into tertiles (I–III) according to alcohol consumption. Alcohol intake is expressed as median (range) ; all other values are meanspS.D.
P values were determined using the Mann–Whitney test. Abbreviations : SBP/DBP, systolic/diastolic blood pressure ; HR, heart rate.
Parameter
I (n l 130)
II (n l 130)
III (n l 131)
Mean difference (I minus III)
P
Alcohol intake (g/day)
BMI (kg/m2)
WHR
SBP (mmHg)
DBP (mmHg)
HR (beats/min)
Pulse pressure (mmHg)
Serum cholesterol (mmol/l)
Serum triacylglycerols (mmol/l)
HDL cholesterol (mmol/l)
LDL cholesterol (mmol/l)
Blood glucose (mmol/l)
Plasma insulin (m-units/l)
Cigarette-years
0 (0–5.6)
26.3p4.86
0.94p0.074
122p17
71.9p10.8
59.9p8.53
49.8p12.6
5.84p1.14
1.39p0.66
1.21p0.33
4.00p1.01
4.84p0.87
9.86p5.20
291p413
9.81 (5.7–14.1)
26.5p4.27
0.94p0.062
123p17
71.8p10.6
60.4p9.11
51.2p11.5
6.01p1.01
1.51p0.77
1.25p0.34
4.09p0.88
5.04p1.71
10.1p5.41
264p376
23.2 (14.2–129)
26.5p4.03
0.95p0.062
126p18
72.4p10.9
61.0p7.99
53.2p13.0
6.19p1.15
1.80p1.48
1.36p0.42
4.06p1.00
4.85p0.75
9.97p7.93
450p426
–
k0.20
k0.014
k3.85
k0.50
k1.11
k3.43
k0.35
k0.41
k0.15
k0.058
k0.013
k0.10
k158
–
0.336
0.020
0.101
0.791
0.441
0.027
0.018
0.016
0.004
0.448
0.520
0.184
0.000
# 2002 The Biochemical Society and the Medical Research Society
Alcohol consumption and the metabolic syndrome
Table 3 Correlation analysis of alcohol consumption in
relation to metabolic variables and insulin sensitivity
Table 4 Stepwise multiple regression analysis showing contributions to the variance in GIRFFM
Spearman’s correlation coefficients are given for the relationships between the
listed variables and alcohol consumption (g/day) ; n l 391 except for GIRFFM
(n l 104). Significance : *P 0.05, **P 0.01. Abbreviations : SBP/DBP,
systolic/diastolic blood pressure ; HR, heart rate.
The model also included serum HDL cholesterol, WHR and cigarette-years.
Significance : *P 0.001.
Variable
BMI (kg/m2)
WHR
SBP (mmHg)
DBP (mmHg)
HR (beats/min)
Pulse pressure (mmHg)
Serum cholesterol (mmol/l)
Serum triacylglycerols (mmol/l)
HDL cholesterol (mmol/l)
LDL cholesterol (mmol/l)
Blood glucose (mmol/l)
Plasma insulin (m-units/l)
Cigarette-years
GIRFFM
r
0.021
0.084
0.068
0.010
0.037
0.086
0.090
0.102*
0.145**
0.012
0.004
k0.094
0.158**
0.203*
Variable
β-coefficient (standard error)
P
GIRFFM
BMI
Alcohol intake
log (Triacylglycerols)
k0.469 (0.059)
0.326 (0.019)
k0.220 (1.302)
0.000
0.000
0.008
R2
0.458*
Multiple regression
In a multiple regression analysis of the subpopulation
that underwent a euglycaemic hyperinsulinaemic clamp
examination (n l 104), GIRFFM was the dependent variable and the independent variables were continuous
variables that showed a statistically significant univariate
association with insulin-mediated glucose uptake. As
presented in Table 4, BMI, alcohol intake and serum
triacylglycerol levels (but not serum HDL cholesterol,
WHR or cigarette-years) independently of each other
explained 46 % of the variation in glucose uptake.
DISCUSSION
Figure 1 Scatterplot of GIRFFM against daily alcohol consumption
BMI, WHR or whole-blood glucose. However, in the
group of subjects in which a euglycaemic hyperinsulinaemic clamp examination was carried out (n l 104), alcohol
consumption was correlated significantly with fasting
plasma insulin (r l k0.24, P l 0.017) and with GIRFFM
(r l 0.203, P l 0.038) (Figure 1).
The results of the present study showed that, in a group
of clinically healthy 58-year-old men, the subjects with
the highest alcohol intake (tertile III) had higher mean
values for HDL cholesterol, serum triacylglycerols, total
cholesterol, pulse pressure, WHR and cigarette-years
compared with subjects with a low alcohol intake (tertile
I). When comparing subjects with the clustering of risk
factors typical for the metabolic syndrome with subjects
with no risk factors for this syndrome, no significant
difference in alcohol consumption was found. In a subgroup of subjects that underwent a clamp examination,
BMI, alcohol intake and serum triacylglycerols, but not
HDL cholesterol, WHR or cigarette-years, independently of each other explained 46 % of the variation in
GIR FFM.
In the present study of healthy 58-year-old men, a
suggested operative definition of the metabolic syndrome
was used, based on the combination of impaired fasting
glucose or insulin resistance and the presence of at least
two of six factors that are typical characteristics of the
metabolic syndrome [11,12]. When using this definition,
no significant differences between subject groups could
be found when comparing mean alcohol intake.
As expected, we found that subjects with the highest
alcohol intake had higher serum HDL cholesterol concentrations compared with low-level consumers, but also
a mainly negative metabolic profile, with higher mean
values for serum triacylglycerols, total cholesterol, pulse
pressure and WHR. An improvement in insulin sen# 2002 The Biochemical Society and the Medical Research Society
349
350
D. Goude, B. Fagerberg and J. Hulthe
sitivity has been found to be associated with higher
HDL-cholesterol levels [19]. It has been suggested that
this is one of the routes by which alcohol acts upon HDL
metabolism, although the exact mechanism remains to be
elucidated [20]. In accordance with this, the serum HDLcholesterol concentration has been shown to increase
with alcohol consumption [21]. The present study corroborates these findings.
Moderate regular alcohol consumption has been found
to be associated with increased insulin sensitivity in
several studies [22]. Women consuming one or two
drinks (10–20 g of alcohol) daily had a 15 % lower BMI
than non-drinkers, despite higher energy consumption,
which may have reflected enhanced insulin sensitivity
[23,24]. Several measures of insulin sensitivity have been
used when investigating the relationship between alcohol
intake and insulin sensitivity. For example, in the
Bruneck study, comprising 820 healthy, non-diabetic
men and women aged 40–79 years, low-to-moderate
amounts of alcohol (defined as 0–100 g\day) taken on a
regular basis were associated with significant decreases in
fasting insulin concentrations, post-glucose insulin concentrations and estimates for insulin resistance obtained
using the homoeostasis model [4]. Two further studies
have also shown that an increased alcohol intake was
associated with lower post-glucose insulin concentrations [20,25]. Other studies have demonstrated that an
increased alcohol intake is associated with lower fasting
insulin concentrations [3,5,26]. However, no previous
study has examined the relationship between insulin
sensitivity, as measured by the clamp technique, and
chronic light-to-moderate alcohol intake. Although it
has a high variability, the euglycaemic hyperinsulinaemic
clamp technique is regarded as the ‘ reference technique ’
when measuring insulin resistance [6,7].
In the present study a subgroup of subjects (n l 104)
underwent a euglycaemic hyperinsulinaemic clamp examination. We chose to correct the GIR for FFM, as it
has been demonstrated that the severity of insulinmediated glucose uptake is overestimated when adjusted
for total body weight [6]. GIRFFM showed a statistically
significant positive correlation with alcohol intake. In a
multivariate analysis, BMI, alcohol intake and serum
triacylglycerols together and independently accounted
for 46 % of the variability in glucose uptake.
As mentioned above, we found an independent association between GIRFFM and alcohol intake even after
adjustment for metabolic variables such as BMI, triacylglycerols and HDL cholesterol.
Previous experimental studies dealing with acute and
chronic ethanol intake have shown divergent results.
After acute and relatively high alcohol intake, a decreased
insulin sensitivity has been observed in metabolic studies
[27,28]. An increased plasma insulin response to glucose
after acute alcohol administration has been shown [29],
but a biphasic effect has also been demonstrated, with
# 2002 The Biochemical Society and the Medical Research Society
hypoinsulinaemia following initial hyperinsulinaemia
[30]. In animal models, inhibition by alcohol of insulin
secretion by pancreatic β-cells has been seen [31–33] ; a
reduced hepatic uptake of insulin has also been reported
[34].
Evidence of enhanced insulin sensitivity after chronic
light-to-moderate alcohol intake is, on the other hand,
relatively consistent [3,5,20,25]. Similar findings have
been obtained in animal models [35]. Possible explanations for this phenomenon are as yet scarce. However,
the observed relationship between GIRFFM and alcohol
intake in the present study localizes the site of alcohol
action on insulin sensitivity to the cellular level, since the
clamp will tend to eliminate variation in insulin sensitivity
due to vascular effects. There are some limitations to the
present study. Women were not included, and the results
can only be applied for the population of clinically
healthy 58-year-old men. Our model does not take into
consideration different drinking patterns, or previous
alcohol abuse. Also, alcohol intake is a skewed variable,
and total abstainers are prevalent in the population. Even
though the distribution may be normalized by logarithmic transformation of the alcohol intake or, alternatively, by treating drinkers and non-drinkers as separate
populations, there is no agreement on the correct method
of analysing this type of data [36]. No consensus exists
for defining a moderate alcohol intake, although previous
studies have generally regarded this as 20–40 g\day [1],
which in our present study corresponds most closely to
the tertile with the highest consumption (median 23.2 g\
day).
Apart from cigarette smoking, which may have negative effects on insulin sensitivity [37], only components
of the metabolic syndrome were considered in the present
analyses. Of course, many other factors may affect insulin
resistance, including physical activity [38,39], diet [40]
and infection [41].
The use of alcohol is a habit that is prevalent in the
general population and also potentially modifiable.
Hence beneficial effects conferred by its consumption are
of interest. One must, however, be cautious in issuing
recommendations, given the drug’s well known adverse
effects, including social and health risks such as hypertension, liver damage, impairment of β-cell function,
diabetes due to pancreatitis, hypercortisolism, neurological damage and cardiomyopathy.
To summarize, in 58-year-old subjectively clinically
healthy men recruited from the general population, there
was a positive association between alcohol intake and
several metabolic variables, such as HDL cholesterol and
serum triacylglycerols. No significant difference in alcohol consumption was observed between subjects with
the metabolic syndrome and subjects with no risk factors
for this syndrome. To our knowledge this is the first
study to show an independent relationship between
insulin sensitivity, as measured by the clamp technique,
Alcohol consumption and the metabolic syndrome
and alcohol intake. Insulin sensitivity was also independently associated with BMI and serum triacylglycerols. As this was a cross-sectional study, no conclusions
can be drawn about causality.
15
ACKNOWLEDGMENTS
17
This work was supported by grants from the Swedish
Heart-Lung Foundation, the Swedish Medical Research
Council (12270, 10880), King Gustav V and Queen
Viktoria Foundation, and AstraZeneca (Mo$ ndal,
Sweden). We thank laboratory technicians Eva-Lena
Alenhag, Anna Fro$ de! n and Caroline Schmidt for excellent
research assistance. Other members of the AIR study
group are John Wikstrand and Lena Bokemark.
16
18
19
20
REFERENCES
21
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Corrao, G., Rubbiati, L., Bagnardi, V., Zambon, A. and
Poikolainen, K. (2000) Alcohol and coronary heart
disease : a meta-analysis. Addiction 95, 1505–1523
Fagrell, B., DeFaire, U., Bondy, S. et al. (1999) The
effects of light to moderate drinking on cardiovascular
diseases. J. Intern. Med. 246, 331–340
Razay, G., Heaton, K. W., Bolton, C. H. et al. (1992)
Alcohol consumption and its relation to cardiovascular
risk factors in British women. Br. Med. J. 304, 80–83
Kiechl, S., Willeit, J., Poewe, W. et al. (1996) Insulin
sensitivity and regular alcohol consumption : large,
prospective, cross sectional population study (Bruneck
study). Br. Med. J. 313, 1040–1044
Facchini, F., Chen, Y. D. and Reaven, G. M. (1994)
Light-to-moderate alcohol intake is associated with
enhanced insulin sensitivity Diabetes Care 17, 115–119
Ferrannini, E. and Mari, A. (1998) How to measure
insulin sensitivity. J. Hypertens. 16, 895–906
Groop, L. C., Widen, E. and Ferrannini, E. (1993) Insulin
resistance and insulin deficiency in the pathogenesis of
type 2 (non-insulin-dependent) diabetes mellitus : errors
of metabolism or of methods ? Diabetologia 36,
1326–1331
Godsland, I. F., Leyva, F., Walton, C. et al. (1998)
Associations of smoking, alcohol and physical activity
with risk factors for coronary heart disease and diabetes
in the first follow-up cohort of the Heart Disease and
Diabetes Risk Indicators in a Screened Cohort study
(HDDRISC-1). J. Intern. Med. 244, 33–41
Luoma, P. V., Sotaniemi, E. A. and Pelkonen, R. O.
(1984) Alcohol, HDL cholesterol, and liver microsomal
induction (letter). Lancet i, 47
Castelli, W. P., Doyle, J. T., Gordon, T. et al. (1977)
Alcohol and blood lipids. The cooperative lipoprotein
phenotyping study. Lancet ii, 153–155
Hulthe, J., Bokemark, L., Wikstrand, J. et al. (2000) The
metabolic syndrome, LDL particle size, and
atherosclerosis : the Atherosclerosis and Insulin
Resistance (AIR) study. Arterioscler. Thromb. Vasc. Biol.
20, 2140–2147
Alberti, K. G. and Zimmet, P. Z. (1998) Definition,
diagnosis and classification of diabetes mellitus and its
complications. Part 1 : diagnosis and classification of
diabetes mellitus : provisional report of a WHO
consultation. Diabet. Med. 15, 539–553
Fagerberg, B., Lindstedt, G. and Berglund, G. (1993)
Effects of alcohol intake and obesity on serum liver
enzyme activity in obese men with mild hypertension.
J. Intern. Med. 233, 477–484
Friedewald, W. T., Levy, R. I. and Fredrickson, D. S.
(1972) Estimation of the concentration of low-density
lipoprotein cholesterol in plasma, without use of the
preparative ultracentrifuge. Clin. Chem. 18, 499–502
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Kadish, A. H. and Hall, D. A. (1965) A new method for
the continous monitoring of blood glucose : measurement
of dissolved oxygen. Clin. Chem. 11, 869–875
DeFronzo, R. A., Tobin, J. D. and Andres, R. (1979)
Glucose clamp technique : a method for quantifying
insulin secretion and resistance. Am. J. Physiol. 237,
E214–E223
Pietrobelli, A., Formica, C., Wang, Z. et al. (1996) Dualenergy X-ray absorptiometry body composition model :
review of physical concepts. Am. J. Physiol. 271,
E941–E951
Bokemark, L., Froden, A., Attvall, S. et al. (2000) The
euglycemic hyperinsulinemic clamp examination :
variability and reproducibility. Scand. J. Clin. Lab. Invest.
60, 27–36
Langer, R. D., Criqui, M. H. and Reed, D. M. (1992)
Lipoproteins and blood pressure as biological pathways
for effect of moderate alcohol consumption on coronary
heart disease. Circulation 85, 910–915
Mayer, E. J., Newman, B., Quesenberry, Jr., C. P. et al.
(1993) Alcohol consumption and insulin concentrations.
Role of insulin in associations of alcohol intake with
high-density lipoprotein cholesterol and triglycerides.
Circulation 88, 2190–2197
Linn, S., Carroll, M., Johnson, C. et al. (1993) Highdensity lipoprotein cholesterol and alcohol consumption
in US white and black adults : data from NHANES II.
Am. J. Public Health 83, 811–816
McCarty, M. F. (1999) The alcohol paradox. Am. J. Clin.
Nutr. 70, 940–942
Colditz, G. A., Manson, J. E., Stampfer, M. J. et al. (1992)
Diet and risk of clinical diabetes in women. Am. J. Clin.
Nutr. 55, 1018–1023
Prentice, A. M. (1995) Alcohol and obesity. Int. J. Obes.
Relat. Metab. Disord. 19 (Suppl. 5), S44–S50
Sevak, L., McKeigue, P. M. and Marmot, M. G. (1994)
Relationship of hyperinsulinemia to dietary intake in
south Asian and European men. Am. J. Clin. Nutr. 59,
1069–1074
Lazarus, R., Sparrow, D. and Weiss, S. T. (1997) Alcohol
intake and insulin levels. The Normative Aging Study.
Am. J. Epidemiol. 145, 909–916
Yki-Ja$ rvinen, H. and Nikkila$ , J. (1985) Ethanol decreases
glucose utlization in healthy man. J. Clin. Endocrinol.
Metab. 61, 941–945
Avogaro, A., Fontana, P., Valerio, A. et al. (1987) Alcohol
impairs insulin sensitivity in normal subjects. Diabetes
Res. 5, 23–27
Metz, R., Berger, S. and Mako, M. (1969) Potentiation of
the plasma insulin response to glucose by prior
administration of alcohol. An apparent islet-priming
effect. Diabetes 18, 517–522
Puhakainen, I., Koivisto, V. A. and Yki-Jarvinen, H.
(1991) No reduction in total hepatic glucose output by
inhibition of gluconeogenesis with ethanol in NIDDM
patients. Diabetes 40, 1319–1327
Avogaro, A. and Tiengo, A. (1993) Alcohol, glucose
metabolism and diabetes. Diabetes Metab Rev. 9, 129–146
Patel, D. G. and Singh, S. P. (1979) Effect of ethanol and
its metabolites on glucose mediated insulin release from
isolated islets of rats. Metab. Clin. Exp. 28, 85–89
Tiengo, A., Valerio, A., Molinari, M. et al. (1981) Effect
of ethanol, acetaldehyde, and acetate on insulin and
glucagon secretion in the perfused rat pancreas. Diabetes
30, 705–709
Fawcett, J., Hammond, B. and Smith, G. D. (1993) Acute
effects of ethanol on hepatic endocytosis and processing
of insulin in perfused rat liver. Am. J. Physiol. 264,
E420–E427
Russell, J. C., Amy, R. M., Manickavel, V. et al. (1989)
Effects of chronic ethanol consumption in
atherosclerosis-prone JCR :LA-corpulent rat.
Arteriosclerosis 9, 122–128
Gordon, T. and Kannel, W. B. (1983) Drinking and its
relation to smoking, BP, blood lipids, and uric acid. The
Framingham study. Arch. Intern. Med. 143, 1366–1374
Facchini, F. S., Hollenbeck, C. B., Jeppesen, J. et al.
(1992) Insulin resistance and cigarette smoking Lancet
339, 1128–1130
# 2002 The Biochemical Society and the Medical Research Society
351
352
D. Goude, B. Fagerberg and J. Hulthe
38
39
Mayer-Davis, E. J., D ’Agostino, Jr., R., Karter, A. J. et al.
(1998) Intensity and amount of physical activity in
relation to insulin sensitivity : the Insulin Resistance
Atherosclerosis Study. JAMA J. Am. Med. Assoc. 279,
669–674
Rosenthal, M., Haskell, W. L., Solomon, R. et al. (1983)
Demonstration of a relationship between level of physical
training and insulin-stimulated glucose utilization in
normal humans. Diabetes 32, 408–411
40
41
Feskens, E. J., Virtanen, S. M., Rasanen, L. et al. (1995)
Dietary factors determining diabetes and impaired
glucose tolerance. A 20-year follow-up of the Finnish and
Dutch cohorts of the Seven Countries Study. Diabetes
Care 18, 1104–1112
Yki-Jarvinen, H., Sammalkorpi, K., Koivisto, V. A. et al.
(1989) Severity, duration, and mechanisms of insulin
resistance during acute infections. J. Clin. Endocrinol.
Metab. 69, 317–323
Received 8 June 2001/17 September 2001; accepted 13 November 2001
# 2002 The Biochemical Society and the Medical Research Society