International Journal of Cardiology 86 (2002) 61–69

www.elsevier.com / locate / ijcard

Fasting insulin levels independently associated with coronary heart disease

in non-diabetic Turkish men and women q

Altan Onat a , *, Koksal

¨ Ceyhan d , Vedat Sansoy b , Omer
¨ Basar a , Burak Erer a , Omer
¨ Uysal c ,
e
¨
Gulay Hergenç
a
Departments of Cardiology and Internal Medicine, Cerrahpasa Medical Faculty, Istanbul University, Istanbul, Turkey
b
Cardiology Institute, Istanbul University, Istanbul, Turkey
c
Division of Biostatistics, Cerrahpasa Medical Faculty, Istanbul University, Istanbul, Turkey
d
Ersek Center for Cardiovascular Surgery, Istanbul, Turkey
e
` ` Technical University, Istanbul, Turkey
Department of Biochemistry, Yıldız

Received in revised form 6 March 2002; accepted 6 April 2002

Abstract

Background: Levels of plasma insulin have been recognized as a weak risk indicator for coronary or cardiovascular risk in the general
population with ethnic background and gender modifying this relationship. We assessed whether insulin concentrations are associated with
or would serve as a marker of prevalent coronary heart disease risk in a cross-sectional study of a population having low cholesterol levels
(just under 5 mmol / l) but higher prevalence of components of the metabolic syndrome. Methods: In 688 participants of the Turkish Adult
Risk Factor Survey in 2001, plasma insulin values as well as other risk variables were evaluated, and coronary heart disease was
diagnosed based on clinical findings and Minnesota coding of resting electrocardiograms. Nearly equal numbers of men and women (.30
years of age) constituted the population sample from the two largest regions of Turkey. Concentrations of insulin were determined by the
chemiluminescent immunometric method. Results: Geometric mean value was 50 pmol / l (interquartile range 37–68 pmol / l), without
revealing a significant difference in genders. Fasting insulin was correlated in both genders with many variables, notably those involving
central obesity, triglycerides, blood pressure, physical inactivity and, inversely, with high-density lipoprotein (HDL)-cholesterol. In a
regression model, waist circumference and body mass index were strongly associated with log insulin, after controlling for age and
presence of coronary heart disease. The age- and obesity-adjusted odds ratio for coronary heart disease in the highest as opposed to the
lowest quartile was 2-fold in both genders (P,0.05). Even after adjustment for dyslipidemia, blood pressure, glucose intolerance, physical
activity and smoking status, an over 2-fold increased coronary heart disease risk still persisted with regard to hyperinsulinemia ($10
mU / l, 69.5 pmol / l). When C-reactive protein which was correlated with fasting insulin only in women, was added to the model, the
impact of hyperinsulinemia on coronary heart disease risk remained unchanged. Conclusion: Hyperinsulinemia (i) may provide
information on the coronary heart disease likelihood over and above that provided by the other risk factors, including HDL-cholesterol,
and (ii) may contribute, within the frame of insulin resistance, to the coronary heart disease risk independently of the classical risk factors.
 2002 Elsevier Science Ireland Ltd. All rights reserved.

Keywords: Hyperinsulinemia; Coronary heart disease; Inflammation; Risk assessment

1. Introduction
q
From the Turkish Society of Cardiology, Istanbul, Turkey.
*Corresponding author. Present address: Nisbetiye cad. 37 / 24, Etiler
Postmeal or fasting hyperinsulinemia has been
80630, Istanbul, Turkey. Tel.: 190-212-288-4455; fax: 190-212-288-
4433. shown to be associated with increased coronary or
E-mail address: tkd@ixir.com (A. Onat). cardiovascular risk [1–3]. As concluded in a review

0167-5273 / 02 / $ – see front matter  2002 Elsevier Science Ireland Ltd. All rights reserved.
PII: S0167-5273( 02 )00190-0
62 A. Onat et al. / International Journal of Cardiology 86 (2002) 61–69
[4], while older prospective studies of endogenous
(postmeal) insulin levels in non-diabetic adults found
significantly increased coronary heart disease risk,
few subsequent studies have done so [5,6]. Overall, it
has been considered in a meta-analysis as a weak risk
indicator for the occurrence of cardiovascular disease
[7]. More recently, several studies on fasting insulin
levels have reported a positive association with
coronary heart disease risk either in men [8–10], or
among women but not in men [11]. Uncertainty
prevails on its value in risk prediction among women
and in the elderly [12,13]. It is generally agreed that
the relationship between hyperinsulinemia and car-
diovascular risk is modified by ethnic background
[7,14]. Paucity of information exists with regard to
populations of developing countries and, in particular,
to populations having normal cholesterol levels but a
high prevalence of the metabolic syndrome. We
report our findings and analysis in a cross-sectional
study with respect to fasting hyperinsulinemia as a
risk factor for prevalent coronary heart disease in an
unselected population sample of non-diabetic Turkish
adults. Distribution of insulin levels and their interre-
lation with other risk factors in a general population
having distinguishing features in risk profile [15,16]
will also be described.
2. Materials and methods
Participants of this study form part of the cohort of
the Turkish Adult Risk Factor Study, a prospective
survey on the prevalence of cardiac disease and risk
factors in adults in Turkey carried out periodically
since 1990 in seven geographical regions of the
country [15,17]. The last follow-up in May 2001 was
confined to the cohort residing in the two largest
regions of Turkey, namely the Marmara and the
Central Anatolian regions which make up 43% of the
entire sample population surveyed. These regions are
slightly more developed in terms of urbanization and
industrialization than the remaining regions of the
country. Participants were 31 years of age or older,
and 95% of the people invited were tracked, while
78% of those invited were examined. All subjects
gave informed consent. The survey is representatively
stratified for sex and age, with rural distribution
slightly more preponderant (45%) than currently
existing. Out of a total of 541 men and 581 women,
plasma insulin levels were assayed in the non-fasting
state in 15 subjects and under fasting conditions in
745 individuals. When 57 subjects with diabetes were
also excluded, 320 non-diabetic men and 368 women
remained who formed the study population for analy-
sis of fasting insulin levels.
The field study protocol included a questionnaire
on the past history of cardiovascular disease, smoking
habits, levels of physical activity, and family income,
while physical examination of the cardiovascular
system including measurement of blood pressure, a
set of anthropometric measurements, recording of a
resting electrocardiogram and appropriate collection
and shipment of blood samples.
Diagnosis of coronary heart disease was based on
the presence of angina pectoris, of a history of
myocardial infarction with or without accompanying
Minnesota codes of electrocardiograms [18], or on a
history of myocardial revascularization. Among
women typical angina before age 45 and atypical
angina at any age precluded the diagnosis. Isolated
typical angina in women and atypical angina in men
were considered as suspect diagnoses. These criteria
resulted from the fact that electrocardiographic
changes of ‘ischemic type’ (codes 1.1–3, 4.1–3, 5.1–
3, 7.1) were absent in only one-third of all patients
(in seven men and 17 women).
2.1. Measurement of risk factors and validation
Blood pressure was measured with an aneroid
sphygmomanometer in the sitting position on the
right arm, and the mean of two recordings 3 min
apart was recorded. Waist circumference was mea-
sured with the subject standing and wearing only
underwear, at the level midway between the lower rib
margin and the iliac crest, while that of the hip was
measured at the level of the great trochanters. Body
mass index was calculated by the computer as weight
divided by height squared (kg / m 2 ). With regard to
cigarette smoking, non-smokers, past smokers and
three increments of current smoking formed the five
categories. Physical activity was graded by the par-
ticipant himself into four categories of increasing
order with the aid of the following scheme: grade 1:
white-collar worker, sewing-knitting, walking #1 km
daily; grade 2: repair worker, house work, walking
 A. Onat et al. / International Journal of Cardiology 86 (2002) 61–69
1–2 km daily; grade 3: mason, carpenter, truck
driver, cleaning floors and windows, walking 4 km
daily; grade 4: heavy labor, farming, regular sports
activity [15].
Blood samples were centrifuged at 2000 rpm for
15 min within 30 min after collection. Serum par-
titioned to aliquots was kept at 4 8C and sent to
Istanbul the same day at 4–8 8C where it was kept in
a freezer at 280 8C for less than 1 month until
measurements. Serum concentrations of total choles-
terol, triglycerides and glucose were measured using
enzymatic Roche Diagnostics kits (Mannheim, Ger-
many) with Hitachi 902 autoanalyzer. Direct auto-
mated HDL-cholesterol measurements (Roche Diag-
nostics homogeneous ‘HDL-C plus’ kit which uses
PEG-modified enzymes, sulfated a-cyclodextrin and
dextran sulfate) were done with homogeneous en-
zymatic colorimetric test. Plasma concentrations of
insulin were determined by the chemiluminescent
immunometric method using Diagnostic Products kits
and the autoimmunanalyzer Immulite (Los Angeles,
CA). All insulin measurements were performed on
the same day. Calibrators and controls were supplied
by the manufacturer. Calibration range was up to 400
mIU / l and analytical sensitivity 2 mIU / l. Values of
high-sensitivity C-reactive protein were measured
with the Dade-Behring nephelometer system using N
Latex CRP mono reagent by particle-enhanced im-
munonephelometric method (Behring Werke, Mar-
burg, Germany). Both the reference laboratory and
the laboratory in which the measurements were done
participated in external quality control programs of
the College of American Pathologists.
Within run coefficients of variation for insulin
regarding normal and pathological control samples
were 4.2 and 5.3%, respectively; for the remaining
variables, they ranged between 1.3 and 2.4%. Day to
day coefficients of variation for glucose, total choles-
terol, HDL-cholesterol, triglycerides, insulin and C-
reactive protein were 1.5–3%.
Plasma apolipoprotein B values were measured by
the turbidimetric method (Turbitimer, Behring) in a
previous survey 10 months before. Mean values for
concentrations of insulin and C-reactive protein are
geometric means6S.D. calculated from the log-trans-
formed distribution. Insulin values were also pro-
vided in pmol / l whereby 1 mU / l was converted to
6.945 pmol / l. External quality control was performed
63
in a reference laboratory in a random selection of 3%
of participants. Plasma measurements and coronary
heart disease diagnosis were executed in an entirely
blinded fashion. Collected survey data were checked
by the primary investigator before and after being
included in the database to ensure data quality and
completeness. Individuals with diabetes mellitus or
glucose intolerance were identified using the criteria
of World Heart Organization [19].
2.2. Statistical analysis
The significance of the correlation between log
insulin values and other variables was assessed using
the Spearman test. Concentrations of fasting insulin
were statistically analyzed by quartiles. Tests for
trend were used to assess any relation of increasing
insulin values and risk of coronary heart disease after
dividing the study sample into quartiles. Linear
regression models were fit for log insulin as a
dependent variable, and variables of interest as
independent variables to demonstrate their contribu-
tion. Likelihood estimates of coronary heart disease
and confidence intervals were obtained by use of
logistic regression analyses. Adjusted odds ratios
were obtained for these models that accounted for
confounding variables. A value of P,0.05 (two-
tailed) was considered statistically significant. Multi-
ple regression analyses of the data were carried out
using STATA 5-0 for Windows package.
3. Results
The clinical characteristics of the sample popula-
tion are shown in Table 1 separately for men and
women. The mean ages of 320 men and 368 women
were almost identical (51612 years). Tendency to
obesity (mean body mass index 28.3 kg / m 2 ), in
particular to central obesity (mean waist circumfer-
ence 91 cm) is clearly apparent among women and is
suggested in men. This is paralleled by high systolic
and diastolic pressures in women and by high plasma
triglyceride levels in both genders. The most striking
features of the study sample are the low mean levels
of total cholesterol (|192 mg / dl, or 4.97 mmol / l),
LDL-cholesterol (around 121 mg / dl or 3.15 mmol / l),
and especially HDL-cholesterol (1.06 mmol / l), dis-
64 A. Onat et al. / International Journal of Cardiology 86 (2002) 61–69

Table 1
Characteristics of the non-diabetic study population (n5688)
Men (n5320) Women (n5368)
n Mean S.D. n Mean S.D.
Age, years 320 51.6 12.6 368 50.5 12
Waist circumference, cm 318 89.2 9.8 361 92.5 10.8
Body mass index, kg / m 2 306 27.5 4.2 360 29.1 5.3
Systolic BP, mmHg 320 130.9 23.1 367 139.6 29
Diastolic BP, mmHg 320 82 13.2 367 85 14.3
Total cholesterol, mg / dl 318 187.9 36.7 367 196.2 37.6
HDL-cholesterol, mg / dl 318 37.4 8.8 367 45 9.8
Triglycerides, mg / dl 317 158.6 87.3 367 130.7 65.6
Apolipoprotein B a , mg / dl 116 110.6 33.7 222 118.1 49
Fasting glucose, mg / dl 317 78.3 12.9 367 77.6 12.1
Fasting insulin, pmol / l 320 60.2 b 7.66 368 60.8 b 5.6
C-reactive protein, mg / l 289 4.11 b 14.1 335 3.99 b 5.3
Current smokers, % 320 43.8 367 16.6
Physical activity grade 307 2.44 0.7 350 2.08 0.47
BP, blood pressure.
a
Values from 2000.
b
Arithmetic means.

tinctly lower than in Western populations. In addition,
smoking is prevalent among men (44%) and seden-
tary lifestyle among women. Altogether in 74 sub-
jects (35 men, 39 women) in this study, coronary
heart disease was considered to be prevalent at the
time of the survey, including 16 subjects (5 men, 11
women) with suspect coronary heart disease diag-
nosis.
3.1. Insulin values by sex
The distribution of insulin values ranged from ,2
to 95 mU / l. The median (and interquartile) values of
fasting insulin in men were 6.8 mU / l (47.2 pmol / l)
(5.02 and 9.22 mU / l, or 35–64 pmol / l). Respective
values among women were higher by 12%, namely,
7.6 mU / l (53 pmol / l) (5.63 and 10.26 mU / l, or
39–71.3 pmol / l), P.0.05. Some 24% of men and
27% of women had concentrations $10 mU / l.
3.2. Relationship of insulin values with other risk
parameters
Pearson correlation coefficients and significance
between associations of logarithm of fasting insulin
on the one hand and 12 parameters measured or
estimated in the 688 non-diabetic men and women
under study are shown in Table 2. These correlations
were uniformly highly significant in both genders
with the exception of total cholesterol and apo B.
Associations of insulin levels were inverse, as ex-
pected, with regard to physical activity and HDL-
cholesterol. Correlation was strongest with waist
circumference, men exhibiting a powerful correlation
(r50.45 vs. 0.33). Smoking status was significantly
associated inversely with log insulin levels in men
but not in women. This unadjusted finding would lose
its significance or reverse the relation when adjust-
ment was made for the difference of 5.3 cm in waist
girth and nearly 7 years in age between male smokers
and non-smokers, a difference which in our analysis
would account for a variation of 26% in insulin
concentrations and thus be in agreement with the
experience that chronic smoking is associated with
high age- and body mass index-adjusted plasma
insulin levels [20]. Apo B was measured in one-half
of all participants, in three-fifths of those with
hyperinsulinemia. Among the latter group of 97
subjects (insulin levels$10 mU / l), 39 had apo B$
120 mg / dl (and 30 subjects presumably had com-
bined hyperlipidemia as indicated by the presence of
levels of apo B$120 mg / dl combined with
triglycerides$1.5 mmol / l). Prevalence of apo B$
120 mg / dl in the 242 remaining subjects was lower
(32.6%), interaction between insulin and apo B levels
existing only in men, and not in women.
 A. Onat et al. / International Journal of Cardiology 86 (2002) 61–69 65

Table 2
Partial correlation coefficient (r) and significance (P) between log fasting insulin and 12 risk parameters in 688 Turkish men and women, aged .30 years
Versus a Both genders Men, n5320 Women, n5368
r P, r P, r P,
Waist circumference (cm), n5318 / 361 0.390 0.000 0.449 0.000 0.333 0.000
Body mass index (kg / m 2 ), n5306 / 360 0.366 0.000 0.442 0.000 0.311 0.000
Triglycerides (mg / dl), n5317 / 367 0.276 0.000 0.262 0.000 0.337 0.000
Diastolic blood pressure (mmHg) 0.191 0.000 0.178 0.001 0.194 0.000
Systolic blood pressure (mmHg) 0.172 0.000 0.186 0.001 0.151 0.004
Smoking status 20.124 0.000 20.165 0.005 NS
HDL-cholesterol (mg / dl), n5318 / 367 20.165 0.000 20.211 0.000 20.202 0.000
Age 0.127 0.001 0.133 0.001 0.128 0.014
C-reactive protein, n5309 / 362 0.095 0.014 0.094 0.1 0.126 0.017
Apolipoprotein B (mg / dl), n5116 / 222 0.08 0.141 0.151 0.105 NS
Total cholesterol (mg / dl), n5318 / 367 NS NS NS
Physical activity grade 20.160 0.000 20.099 0.1 20.117 0.039
a
n indicates number of men and women with present data.

Log C-reactive protein concentrations were sig- (diastolic) blood pressure, central obesity, smoking
nificantly associated (F53.61, P50.014, ANOVA) habit and physical activity grade. When the effect of
with insulin quartiles, geometric means ranging from log C-reactive protein was further included into this
1.73, 1.53, 2.01 and 2.65 mg / l from the lowest to the model (Table 5), logistic regression revealed that the
highest quartile, respectively. When separately ana- association between hyperinsulinemia and coronary
lyzed for genders, the association was not significant heart disease was unchanged among women and
in men, but was of borderline significance in women. marginally weakened in men. Overall, the indepen-
In a multiple linear regression model (F535, P, dent association with coronary heart disease risk was
0.001) body mass index in men and, particularly, somewhat stronger in hyperinsulinemic men than
waist circumference (P,0.02) in both genders were women.
significantly and independently associated with log Included in addition in the logistic regression
insulin concentrations, after adjustment for age which model were: sex, waist circumference, diastolic pres-
did not appear to affect these levels independently of
central obesity (Table 3).
Table 3
Obesity correlates of log fasting insulin adjusted for age and presence of
3.3. Odds ratios for fasting insulin associated with coronary heart disease (n5659)
coronary heart disease Coeff. b S.E. Std b P,
Adults F535.2; P,0.001
When concentrations of fasting insulin of the entire Waist circumference 0.0054 0.001 0.251 0.000
sample comprising 659 subjects were divided into Body mass index 0.0073 0.003 0.158 0.008
quartiles, the number of those with prevalent cor- Presence of CHD 0.0668 0.028 0.090 0.017
Age 0.00117 0.001 0.064 0.094
onary heart disease rose with each quartile, and after
adjustment for age, obesity and glucose intolerance, Men F523; P,0.001
the odds ratio for coronary heart disease in the upper Waist circumference 0.00617 0.002 0.255 0.013
Body mass index 0.0122 0.006 0.214 0.036
quartile (hyperinsulinemia) rose abruptly to 1.93 as
Presence of CHD 0.0685 0.041 0.090 0.093
compared to the lowest quartile. Adjustment for the Age 0.00132 0.001 0.069 NS
ten covariates of fasting insulin concentrations in the
model led to the retainment of the odds ratio just over Women F513.4; P,0.001
Waist circumference 0.0041 0.002 0.205 0.009
2 and suppressed the emergence of any salient Body mass index 0.0059 0.003 0.148 0.056
modifiable variable, other than systolic blood pres- Presence of CHD 0.0609 0.038 0.085 0.111
sure in men (Table 4 and Fig. 1). The covariates in Age 0.00112 0.001 0.063 NS
the model were such strong variables as dyslipidemia, Std, standardized.
66 A. Onat et al. / International Journal of Cardiology 86 (2002) 61–69

Table 4 which comprised women as well as men and middle-

Odds ratios a and 95% confidence intervals for prevalent coronary heart
disease by quartiles of fasting insulin (n5644) aged and elderly participants, hyperinsulinemia
proved to be significantly and independently associ-
Quartiles of insulin Parameter estimates Confidence
intervals ated with prevalent coronary heart disease. The study
b S.E. OR
is only of intermediate size though one of few studies
Adults P trend 0.034 on women examining the relationship between insulin
Lowest, ,5.5 mU / l 1
Second, 5.6–7.3 mU / l 0.85 NS levels and coronary heart disease risk, along with the
Third, 7.4–9.9 mU / l 1.04 NS ARIC [11], the San Luis Valley [21] and the Kuopio
Highest, .10 mU / l 0.707 0.245 2.03 1,25; 3,28 [13] studies. In adults aged 31 or over, we found
Age 0.056 0.016 1.058 1,02; 1,09
Systolic blood pressure 0.016 0.009 1.016 0,999; 1,034
geometric mean fasting insulin values of 51 pmol / l
(with interquartile ranges 38–68 pmol / l) which did
Men, n5303 P trend 0.127 not differ significantly with regard to gender. These
Lowest, ,5.5 mU / l 1 values are slightly lower than those (mean 68 pmol / l)
Second, 5.6–7.3 mU / l 0.85 NS
Third, 7.4–9.9 mU / l 1.04 NS reported in several recent surveys [13,22], though
Highest, .10 mU / l 0.802 0.376 2.23 1,664; 4,66 higher than that obtained in the Helsinki Policemen
Age 0.054 0.026 1.056 1,002; 1,112 study [2], and similar to the population samples
Systolic blood pressure 0.037 0.015 1.038 1,008; 1,069
reported by Hergenç et al. [23]. Previous studies
Women, n5341 P trend 0.157 utilizing non-specific immunoassays may have some-
Lowest, ,5.5 mU / l 1 what overestimated insulin values; furthermore,
Second, 5.6–7.3 mU / l 0.85 NS
Third, 7.4–9.9 mU / l 1.04 NS
rather than the absolute levels in a community, the
Highest, .10 mU / l 0.700 0.348 2.01 1,018; 3,98 relative concentrations are critical in assessing the
Age 0.054 0.022 1.056 1,01; 1,10 impact on the risk of coronary heart disease.
a
Included in the logistic regression model were in addition: sex, waist Similar to previous studies [11,21,24,25], insulin
circumference, diastolic pressure, total cholesterol, triglycerides, HDL- levels were correlated with numerous risk variables
cholesterol, glucose intolerance, smoking status and physical activity
grade. Model comprised 64 subjects with CHD (male 31, female 33).
of the insulin resistance syndrome in our cohort in
both genders, namely with measures of (central)
obesity, blood pressure, triglycerides and, inversely
sure, total cholesterol, triglycerides, HDL-cholesterol, with HDL-cholesterol and physical activity. Waist
glucose intolerance, smoking status and physical circumference was the best correlate of fasting insulin
activity grade. The model included 63 subjects with levels as noted by others [25]. Though the relation to
CHD (male 30, female 33). apo B did not attain statistical significance which may
A cut-off point of 10.0 mU / l for hyperinsulinemia be ascribed to the determination of these levels only
included 30 patients with coronary heart disease out in part of our cohort, interaction between hyperin-
of 388 subjects (18.2%), whereas there were only 38 sulinemia and hyper-apo B did exist among men. A
coronary heart disease patients among 494 individ- significant but weak correlation (r50.15) was ob-
uals (7.7%) below the cut point, resulting in an tained between levels of insulin and apo B in a
unadjusted 2.4-fold discrimination in coronary heart cross-sectional study of elderly men [26].
disease likelihood. C-reactive protein, also related to the insulin
resistance syndrome, was significantly though weakly
correlated with insulin levels in women, as it was, in
4. Discussion a previous analysis of our cohort [27], the variable
having the strongest correlation with prevalent cor-
The present cross-sectional population-based study onary heart disease. Our finding that the addition of
evaluated the distribution of and the association with the variable C-reactive protein to the logistic regres-
coronary heart disease risk of fasting plasma insulin sion model failed to attenuate insulin’s association
concentrations in a general population which differed with coronary heart disease among women, suggests
in the risk profile from those studied to date in highly that hyperinsulinemia is closely associated with the
industrialized communities. In a study population subclinical inflammatory marker. The hypothesis that
 A. Onat et al. / International Journal of Cardiology 86 (2002) 61–69 67

Fig. 1. Odds ratio in each quartile (Qtl) of fasting insulin level adjusted for nine risk factors: 2-fold distinction of risk in the quartile$10 mU / l is obtained.
The aggregate number of men and women in each quartile is indicated in the abcissa.

chronic subclinical inflammation is part of the insulin events and causal relations. Final confirmation that
resistance syndrome [28,29] seems supported in our hyperinsulinemia is a predictor of CHD in Turkish
population having a different ethnicity. adults can only be obtained from a prospective study.
It should be emphasized that the main limitation of Adjustment for antihypertensive medication was not
our study is its being cross-sectional in nature which made. The size of the population sample is smaller
does not allow definition of the time sequence of than many (but not all) related previous studies.
Nevertheless, possibly due to the risk profile of the
Table 5 cohort, this did not preclude the emergence of a clear
Independent associations on the likelihood of prevalent CHD of quartiles result. It should be pointed out that any potential bias
of fasting insulin, C-reactive protein and other variables (n5628)
in terms of inaccuracy in the insulin assays or of
Quartiles of insulin Parameter estimates Confidence coronary heart disease diagnosis would tend to dilute
intervals
b S.E. OR rather than augment the association. The association
Adults P trend 0.056 between fasting hyperinsulinemia and coronary heart
Lowest, ,5.5 mU / l 1 disease tended to exhibit a threshold effect (at insulin
Second, 5.6–7.3 mU / l 0.84 NS
Third, 7.4–9.9 mU / l 1.02 NS level $10 mU / l) rather than exhibiting gradedness
Highest, .10 mU / l 0.707 0.245 1.95 1,20; 3,16 across the quartiles. Among the strengths of this
Age 0.053 0.017 1.054 1,02; 1,09 study are that adjustment for HDL-cholesterol was
Systolic blood pressure 0.018 0.009 1.018 1,001; 1,036
made and all diabetics diagnosed by the World Health
68 A. Onat et al. / International Journal of Cardiology 86 (2002) 61–69
Organization criteria were excluded. Moreover, fast-
ing insulin has been used which according to insulin
clamp studies correlates better with insulin resistance
than postchallenge insulin [4].
The odds ratio of hyperinsulinemia on the likeli-
hood of coronary heart disease was more conspicuous
than in most previous studies, namely 2-fold between
the highest and lowest quartiles (representing a
gradient of 60 pmol / l), and this pertained to both
genders. In a meta-analysis of 12 prospective studies
published up to 1996 on the relationship between
insulin and cardiovascular disease [7], the summary
relative risk corresponding to the interquartile range
was 1.17 (95% CI, 1.09–1.26). The greater odds ratio
obtained herein may partly be due to a substantial
proportion of the cohort being insulin-resistant par-
ticipants which non-diabetic Turkish adults pre-
sumably often are. Yet, the Helsinki Policemen study
[2] in men also reported a high relative risk as did the
large biracial ARIC study [11] in the female hy-
perinsulinemic groups when not fully adjusted for
conventional factors. Finally, it is considered that
most recent investigations using sensitive and specific
insulin assays give higher risk ratios than older
studies [30] which applies to our study. Thus the
strength of the stated association might be ascribed to
the assumption that the contribution of hyperin-
sulinemia and / or insulin resistance to atherogenesis
in a population with low mean cholesterol levels but
with a higher prevalence of the insulin resistance
syndrome [31] might be greater than in communities
exhibiting high LDL-cholesterol levels.
The role of hyperinsulinemia as a coronary risk
factor remains controversial. Most investigators
shared the opinion that, since the modest association
between hyperinsulinemia and the raised coronary
heart disease risk often disappeared after adjustment
for dyslipidemia, obesity and hypertension, hyperin-
sulinemia, instead of being an independent coronary
risk factor, increases the coronary heart disease risk
through these factors [4,30,32,33]. Indeed, an editori-
al concluded that neither hyperinsulinemia, nor in-
sulin resistance is a major risk factor for the develop-
ment of atherosclerotic cardiovascular disease in the
absence of other risk factors [34]. However, workers
of the Quebec Cardiovascular study recently voiced
the view that the relation of hyperinsulinemia to
coronary heart disease may be largely independent of
alterations in body weight, blood pressure and plasma
lipoprotein concentrations [8], to which our results
lend support. Related findings of the Health Profes-
sionals Follow-up Study also suggested that the
relationship between obesity and cardiovascular risk
factors is partly mediated by plasma insulin [26].
To which mechanisms may a possible impact of
hyperinsulinemia on coronary heart disease risk be
ascribed? Aside from the recently described acylation
stimulating protein that determines the rate of fatty
acid uptake by adipocytes during lipolysis, insulin is
a prime determinant of the effectiveness of the
processes of fatty acid uptake and retention, which
has been designated by Sniderman and coworkers
[35] as fatty acid trapping. Ineffective fatty acid
trapping leads to an excess fatty acid flux to the
muscles and the liver. Increased fatty acid flux to the
liver enhances the rate of secretion of VLDL particles
leading to hypertriglyceridemic hyperapoB with ac-
companying raised vascular risk. Increased fatty acid
flux to skeletal muscle results in reduced utilization
of glucose by muscle and augments muscle tri-
glyceride content with concomitant reduced insulin
sensitivity [36] and diminished insulin removal by the
liver. This process further promotes insulin resistance
and hyperinsulinemia.
Our findings suggest that hyperinsulinemia (i) may
provide information on the coronary heart disease
likelihood over and above that provided by the other
risk factors, including HDL-cholesterol, and (ii) may
contribute to the coronary heart disease risk indepen-
dently of the classical risk factors. This knowledge
may have implications for prevention inasmuch as
increased physical activity and reduction and control
of obesity are known to lead to a decline in elevated
plasma insulin levels and underlying insulin resist-
ance.
Acknowledgements
This study was supported in part by the Pfizer and
Astra companies (both Istanbul, Turkey).
References
¨ ¨ ¨ K. Relationship of glucose tolerance and plasma insulin to
[1] Pyorala
the incidence of coronary heart disease: results from two population
studies. Diabetes Care 1979;2:131–41.
 A. Onat et al. / International Journal of Cardiology 86 (2002) 61–69
¨ ¨ ¨ M, Miettinen H, Laakso M, Pyorala
[2] Pyorala ¨ ¨ ¨ K. Hyperinsulinemia
predicts coronary heart disease risk in healthy middle-aged men: the
22-year follow-up results of the Helsinki Policemen Study. Circula-
tion 1998;98:398–404.
[3] Fontbonne A, Charles MA, Thibult N et al. Hyperinsulinemia as a
predictor of coronary heart disease mortality in a healthy population:
the Paris Prospective Study, 15-year follow-up. Diabetologia
1991;34:356–61.
[4] Wingard DL, Ferrara A, Barrett-Connor E. Is insulin really a heart
disease risk factor? Diabetes Care 1995;18:1299–304.
[5] Yarnell JWG, Sweetnam PM, Marks V, Teale JD, Bolton CH. Insulin
in ischaemic heart disease: are associations explained by triglyceride
concentrations? The Caerphilly Prospective Study. Br Heart J
1994;71:293–6.
[6] Ferrara A, Barrett-Connor EL, Edelstein SL. Hyperinsulinemia does
not increase the risk of fatal cardiovascular disease in elderly men or
women without diabetes: the Rancho Bernardo Study, 1984–1991.
Am J Epidemiol 1994;140:857–69.
[7] Ruige JB, Assendelft WJJ, Dekker JM, Kostense PJ, Heine RJ,
Bouter LM. Insulin and risk of cardiovascular disease: a meta-
analysis. Circulation 1998;97:996–1001.
[8] Despres´ J-P, Lamarche B, Mauriege P et al. Hyperinsulinemia as an
independent risk factor for ischemic heart disease. N Engl J Med
1996;334:952–7.
[9] Møller LF, Jespersen J. Fasting serum insulin levels and coronary
heart disease in a Danish cohort: 17-year follow-up. J Cardiovasc
Risk 1995;2:235–40.
[10] Perry IJ, Wannamethee SG, Whincup PH, Shaper AG, Walker MK,
Alberti KGMM. Serum insulin and coronary heart disease in
middle-aged British men. Am J Epidemiol 1996;144:224–34.
[11] Folsom AR, Liao F, Szklo M, Smith R, Stevens J, Eckfeldt JH. A
prospective study of coronary heart disease in relation to fasting
insulin, glucose and diabetes. The Atherosclerosis Risk in Com-
munities (ARIC) Study. Diabetes Care 1997;20:935–42.
[12] Welin L, Eriksson H, Larsson B, Ohlson LO, Svardsudd K, Tibblin
G. Hyperinsulinemia is not a major risk factor in elderly men: the
study of men born in 1912. Diabetologia 1992;35:766–70.
¨
[13] Lempiainen ¨
P, Mykkanen ¨ ¨ ¨ K, Laakso M, Kuusisto J.
L, Pyorala
Insulin resistance syndrome predicts coronary heart disease events in
elderly non-diabetic men. Circulation 1999;100:123–8.
[14] Howard G, O’Leary DH, Zaccaro D et al. Insulin sensitivity and
atherosclerosis. Circulation 1996;93:1809–17.
[15] Onat A, Avcy´ G3, 3enocak M, Ornek ¨ ¨ ¨
E, Gozukara Y. Plasma lipids
and their interrelation in Turkish adults. J Epidemiol Commun
Health 1992;46:470–6.
[16] Onat A. Risk factors and cardiovascular disease in Turkey. Athero-
sclerosis 2001;156:1–10.
[17] Onat A, 3enocak M. Relative risk of factors for coronary heart
disease in population with low cholesterol levels. Int J Cardiol
1994;43:51–60.
[18] Rose G, Blackburn H, Gillum RF, Prineas RJ. In: Cardiovascular
survey methods, 2nd ed., WHO: Geneva, 1982, pp. 124–7.
[19] WHO Study Group on Prevention of Diabetes Mellitus. In: Preven-
tion of diabetes mellitus: report of a WHO study group, Geneva:
WHO, 1994, p. 17, WHO Technical Report Series No. 844.
¨
[20] Ronnemaa ¨ ¨ ¨ K, Ronnemaa
T, Pyorala ¨ EM, Laakso M, Puukka P.
Smoking is independently associated with high plasma insulin levels
in non-diabetic men. Diabetes Care 1996;19:1229–32.
69
[21] Rewers M, Shetterly SM, Baxter J, Hamman RF. Insulin and
cardiovascular disease in Hispanics and non-Hispanic whites: the
San Luis Valley Diabetes Study. Circulation 1992;85:865, abstract.
[22] Haffner SM, Mykkanen ¨ L, Festa A, Burke JP, Stern MP. Insulin-
resistant prediabetic subjects have more atherogenic risk factors than
insulin-sensitive prediabetic subjects: implications for preventing
coronary heart disease during the prediabetic state. Circulation
2000;101:975–80.
[23] Hergenç G, Schulte H, Assmann G, von Eckardstein A. Associations
of obesity markers, insulin, and sex hormones with HDL-cholesterol
levels in Turkish and German individuals. Atherosclerosis
1999;145:147–56.
[24] Zavaroni I, Bonora E, Pagliara M et al. Risk factors for coronary
artery disease in healthy persons with hyperinsulinemia and normal
glucose tolerance. N Engl J Med 1989;320:702–6.
[25] Lemieux I, Pascot A, Couillard C et al. Hypertriglyceridemic waist:
a marker of atherogenic metabolic triad (hyperinsulinemia; hy-
perapolipoprotein B; small, dense LDL) in men? Circulation
2000;102:179–84.
[26] Chu N-F, Spiegelman D, Hotamisligil G, Rifai N, Stampfer M,
Rimm EB. Plasma insulin, leptin, and soluble TNF receptors levels
in relation to obesity-related atherogenic and thrombogenic car-
diovascular disease risk factors among men. Atherosclerosis
2001;157:495–503.
´ ´ ´ B, Kele3 Y,
[27] Onat A, Sansoy V, Yyldyrym ´ Uysal O,
¨ Hergenç G.
C-reactive protein and coronary heart disease in Western Turkey.
Am J Cardiol 2001;88:601–7.
[28] Festa A, D’Agostino R, Howard G, Mykkanen ¨ L, Tracy RP, Haffner
SM. Chronic subclinical inflammation as part of the insulin resist-
ance syndrome. The Insulin Resistance Atherosclerosis Study
(IRAS). Circulation 2000;102:42–7.
[29] Yudkin JS, Kumari M, Humphries SE, Mohamed-Ali V. Inflamma-
tion, obesity, stress and coronary heart disease: is interleukin-6 the
link? Atherosclerosis 2000;148:209–14.
¨ ¨ ¨ K. Hyperinsulinemia and insulin resistance as predictors of
[30] Pyorala
cardiovascular risk. CVD Prevention 2000;3:28–36.
[31] Onat A, Sansoy V. Systolic and diastolic blood pressure related to
six other parameters in Turkish adults: strong correlation with
relative weight. Int J Cardiol 1998;63:295–303.
[32] Lakka TA, Lakka H-M, Salonen JT. Hyperinsulinemia and the risk
of coronary heart disease? N Engl J Med 1996;335:976–7, letter.
[33] Yarnell JW, Sweetnam PM, Marks V, Teale JD, Bolton CH. Insulin
in ischaemic heart disease: are associations explained by triglyceride
concentrations? The Caerphilly prospective study. Br Heart J
1994;71:293–6.
[34] Savage PJ, Saad MF. Insulin and atherosclerosis: villain, accom-
plice, or innocent bystander? Br Heart J 1993;69:473–5, editorial.
[35] Sniderman AD, Cianflone K, Arner P, Summers L, Frayn K. The
adipocyte, fatty acid trapping, and atherogenesis. Arterioscler
Thromb Vasc Biol 1998;18:147–51.
[36] Pan DA, Lillioja S, Kriketos AD et al. Skeletal muscle triglyceride
levels are inversely related to insulin action. Diabetes 1997;46:983–
8.