Diabetes Care In Press, published online January 26, 2007
Impact of Insulin Resistance on Risk of Type 2 Diabetes and Cardiovascular Disease
in People with Metabolic Syndrome
Received for publication 7 December 2006 and accepted in revised form 18 January 2007.
Short title: Insulin Resistance and Metabolic Syndrome
James B. Meigs, MD, MPH
Martin K. Rutter, MD
Lisa M. Sullivan, PhD
Caroline S Fox, MD, MPH
Ralph B. D’Agostino Sr, PhD
Peter W. F. Wilson, MD
From Harvard Medical School and the General Medicine Division, Department of Medicine,
Massachusetts General Hospital, Boston, MA (JBM); The Manchester Diabetes Centre, Manchester
Royal Infirmary and The Division of Cardiovascular and Endocrine Sciences, University of Manchester,
UK (MKR); Department of Biostatistics, Boston University School of Public Health, Boston, MA (LMS);
Department of Mathematics, Statistics, and Consulting Unit, Boston University, Boston, MA (RDA); the
National Heart, Lung, and Blood Institute’s Framingham (Mass) Heart Study (CSF), and Emory
University School of Medicine, Atlanta, GA (PWFW).
Address for Correspondence: James B Meigs MD MPH, General Medicine Division, Massachusetts
General Hospital, 50 Staniford Street, 9th Floor, Boston, MA 02114, USA. Email: jmeigs@partners.org.
Copyright American Diabetes Association, Inc., 2007
Abstract
OBJECTIVE - Metabolic syndrome (MetS) increases risk for type 2 DM and CVD and may be
associated with insulin resistance (IR).
RESEARCH DESIGN AND METHODS - We tested the hypothesis that MetS confers risk with or
without concomitant IR among 2803 Framingham Offspring Study subjects followed up to 11 years for
new DM (135 cases) or CVD (240). We classified subjects by presence of MetS (using ATP3, IDF, or
EGIR criteria) and IR (HOMA-IR ≥75%ile) and used separate risk factor-adjusted proportional hazards
models to estimate relative risks (RR) for DM or CVD using as referents those without IR, MetS, or
without both.
RESULTS – Fifty-six percent with ATP3, 52% with IDF, and 100% with EGIR MetS had IR. IR
increased risk for DM (RR 2.6, 95% CI 1.7-4.0) and CVD (1.8, 1.4-2.3), as did MetS (DM: ATP3, 3.5,
2.2-5.6; IDF, 4.6, 2.7-7.7; EGIR, 3.3, 2.1-5.1; CVD: ATP3, 1.8, 1.4-2.3; IDF, 1.7, 1.3-2.3; EGIR, 2.1, 1.62.7). Relative to those without either Mets or IR, MetS and IR increased risk for DM (ATP3, 6.0, 3.310.8; IDF 6.9, 3.7-13.0) and CVD (ATP3, 2.3, 1.7-3.1; IDF 2.2, 1.6-3.0). Any MetS without IR increased
risk for DM ~3-fold (p<0.001); IDF MetS without IR (RR 1.6, p=0.01) but not ATP3 MetS without IR
(RR 1.3, p=0.2) increased risk for CVD.
CONCLUSIONS - MetS increased risk for DM regardless of IR. ATP3 MetS may need IR to increase
risk for CVD. The simultaneous presence of MetS and IR identify an especially high risk individual.
Background
People with the cluster of risk factors
including obesity, impaired fasting glucose,
hypertension, low HDL cholesterol and elevated
triglycerides are thought to have a ‘metabolic
syndrome’ (MetS) reflecting underlying insulin
resistance (IR). Both MetS and IR are factors in
the development of type 2 diabetes and
cardiovascular disease (CVD)1. Several competing
definitions of MetS are in use, and each is linked
differently to the presence of IR. These definitions
include that of the National Cholesterol Education
Program’s Third Adult Treatment Panel (ATP3),2;
the International Diabetes Federation (IDF) 3, and
the European Group for the Study of Insulin
Resistance (EGIR)4. The EGIR definition requires
the presence of IR plus any two other metabolic
traits; ATP3 and IDF require at least three
metabolic traits but do not require the presence of
IR. In studies of ATP3 MetS as many as half of
subjects do not have IR 5-7
There are few population-based data
comparing how well the ATP3 or IDF MetS
definitions identify subjects with IR 8 or
comparing how well ATP3, IDF, or EGIR MetS
predict subsequent risk for incident diabetes 9, 10 or
CVD 11-13. In addition, while it has been implied
that the presence of MetS is a surrogate for the
presence of IR, there are few data on DM or CVD
risk associated with MetS in the absence of IR, or
in the presence of both MetS and IR. With this
background in mind we performed an analysis in
the Framingham Offspring Study (FOS) using
three MetS definitions in a test of the hypothesis
that MetS confers risk for subsequent development
of diabetes or CVD with or without concomitant
IR.
Research Design and Methods
Study Subjects
The FOS is a community-based
prospective observational study of CVD and its
risk factors 14. During the fifth exam cycle (the
baseline exam, 1991-1995), 3799 participants
fasted overnight and had a standardized medical
examination, including a 2-hour oral glucose
tolerance test (OGTT). From 3,799 participants,
we excluded those with prevalent diabetes
(n=429), prevalent CVD (n=269), or missing
information on covariates (n=298), which left
2,803 subjects for analysis. Subjects were
followed from baseline over a mean of 6.8 years
for new cases of diabetes and a mean of 11.6 years
for first CVD events. The Institutional Review
Board of Boston University approved the study
protocol, and all subjects gave written informed
consent at each examination.
Clinical Definitions and Laboratory Methods
We defined MetS according to updated
2005 NCEP ATP3 criteria 2, IDF criteria 3, and
EGIR criteria 4. Key features and differences
among these three MetS definitions are as follows.
ATP3 MetS requires the presence of any three of
the following five traits: a large waist
circumference, impaired fasting glucose (IFG),
low HDL cholesterol, high triglycerides, or
hypertension; IDF requires a large waist
circumference plus any two of the preceding
metabolic traits, and EGIR requires IR plus any
two of the preceding traits, with the exception that
lipid traits are not counted separately (low HDL-C
or high triglycerides count as one trait). The other
major differences are that a larger waist defines
‘large waist circumference’ for ATP3 (≥102 cm in
men or ≥88 cm in women) than for IDF or EGIR
(≥94 cm in white men or ≥80 cm in white
women), and by current criteria IFG is defined by
a fasting plasma glucose (FPG) of 5.6-6.9 mmol/l
in the ATP3 and IDF definitions and a FPG of 6.16.9 mmol/l in the EGIR definition. We measured
IR with the homeostasis model using the
following validated formula: HOMA-IR = (fasting
glucose (mmol/l) x fasting insulin (µU/mL)) / 22.5
15, 16
. Values of HOMA-IR in each quarter were:
Q1, 2.21-5.07; Q2, 5.08-6.18; Q3, 6.19-7.84; and
Q4, 7.85-30.80 units. We defined IR as a HOMAIR value greater than the 75th percentile (7.84)
among non-diabetic subjects 4. We measured
height, weight, and waist circumference with the
subject standing in light clothes. Waist
circumference was measured at the level of the
umbilicus. Pair-wise inter-technician (three
technicians) intra-class correlations performed
periodically for waist circumference quality
control ranged from 0.96 to 0.99. Blood pressure
values were taken as the mean of two
measurements after the subject had been seated for
at least five minutes. Those who reported smoking
cigarettes regularly during the year prior to the
exam were considered current smokers. We based
a positive parental history of diabetes on selfreport of diabetes in one or both parents 17. We
defined impaired glucose tolerance (IGT) as a 2hour OGTT glucose level of 7.8–11.0 mmol/l.
Laboratory methods for glucose, insulin, and lipid
assays have been published previously 18
Diabetes and CVD Assessment
We defined diabetes at the baseline exam
as a FPG ≥ 7.0 mmol/l, a 2-hour OGTT glucose of
≥ 11.1 mmol/l, or use of hypoglycemic drug
therapy. We defined diabetes at follow-up as
development of a FPG ≥ 7.0 mmol/l or new use of
hypoglycemic drug therapy during the study
interval. Over 99% of diabetes among
Framingham Offspring is type 2 diabetes 18. We
defined baseline and follow-up CVD by standard
Framingham Heart Study criteria as any of the
following: new onset angina, fatal and non-fatal
myocardial infarction or stroke, transient ischemic
attack, heart failure, or intermittent claudication 19.
Statistical Analysis
We used chi-square tests or ANOVA to
test differences in baseline characteristics by MetS
and IR categories. We used the kappa (ê ) statistic
to assess the level of agreement between MetS
definitions, where poor agreement is considered a
ê < 0.20, fair, ê = 0.21 to 0.40; moderate, ê = 0.41
to 0.60; substantial, ê = 0.61 to 0.80; and very
good agreement, ê > 0.80 20. Subjects were
followed from baseline through the seventh (19982001) exam for diabetes and through December,
2004 for CVD events. Risk for diabetes or CVD
was examined in separate analyses. We calculated
incidence rates for diabetes or CVD as the number
of diabetes or CVD events divided by personyears of follow-up in each category. For diabetes
incidence, we used the exam visit date that a new
case of diabetes was identified as the date of
diagnosis. For CVD events, we used the actual
date of the event as the date of diagnosis, and for
subjects without events, the date of their last
follow-up exam as the censoring date. We used
hazard ratios from proportional hazards regression
models (accounting for interval censoring for
diabetes events) to estimate relative risks (RR) and
95% confidence intervals (CIs) for incident
diabetes or CVD conditioned on baseline MetS or
IR categories. Models were adjusted for age and
sex, and used those with without MetS or IR as the
referent groups. Multivariable models were
adjusted for major disease risk factors beyond
those comprising metabolic syndrome. Models
predicting incident diabetes included covariates
for age, sex, parental history of diabetes, body
mass index (BMI, as kg weight per meter height
squared) and IGT. Multivariable models
predicting incident CVD included covariates for
age, sex, LDL cholesterol level, and smoking.
First-order sex-by-MetS interaction terms were
not significant (in part because there were too few
events in some subgroups to calculate stable sexspecific risk estimates) so we did not conduct sexspecific analyses. For instance, even in the sexcombined analysis, in the smallest groups (no
MetS but with IR) we had only 30–40% power to
detect a difference in the observed proportions at
alpha=0.05. We used areas under the receiver
operating characteristic curve (aROC) to compare
the ability of MetS and/or IR to discriminate
future diabetes or CVD risk. The aROC is
interpreted as the probability that the modeled
phenotype(s) correctly discriminate subjects
developing endpoints from those without
endpoints, where 0.5 is chance discrimination and
1.0 is perfect discrimination. We estimated the
population attributable risk percent for diabetes or
CVD associated with exposure categories (for
instance, the ATP3 MetS and IR) as PAR% =
(proportion of cases in the exposure category x
((relative risk exposure category – 1) / relative risk
21
exposure category)) x 100 . We performed all analyses
using SAS (SAS Institute, Cary, NC) and
considered a two-sided value of p< 0.05 to be
statistically significant.
Results
The mean age of the study population
overall was 54 (range 26-82) years, and 55% were
women. Baseline characteristics of study subjects
are shown in Table 1. Diabetes and CVD risk
factor levels were generally more adverse among
people with IR compared to those without IR, and
most adverse among those with MetS and IR.
Among 2803 people the prevalence of ATP3 MetS
was 27.8%, of IDF MetS 34.2%, and of EGIR
MetS 19.1%. By definition, the prevalence of IR
was 25%. Among those with ATP3 MetS the
prevalence of IR was 56.4%, among those with
IDF MetS, 52%, and, by definition, 100% among
those with EGIR MetS. The prevalence of IR in
those without ATP3 MetS was 12.8% and among
those without IDF MetS, 11.0%. There was
substantial agreement in MetS classification by
ATP3 vs. IDF criteria (ê statistic 0.77 overall;
women, 0.81; men, 0.71) and moderate agreement
between IDF and EGIR or ATP3 criteria (IDF vs.
EGIR: ê statistic 0.50; EGIR vs. ATP3: ê statistic
0.53).
Incidence rates for diabetes stratified by
the presence or absence of MetS and IR were
generally similar for all three MetS definitions
(Figure). The diabetes incidence rates were
dramatically higher for those with both MetS and
IR compared with the other categories. Similar
relationships were apparent for the incidence of
CVD,.
Regression models confirmed that all three
MetS definitions conferred generally similar risk
for incident diabetes (Table 2). Of the three, IDF
MetS was associated with a perhaps slightly
higher age-sex-risk factor adjusted relative risk for
diabetes (4.6) than was ATP3 MetS (3.5) or EGIR
MetS (3.3), and all had a somewhat higher relative
risk for diabetes than did IR (2.6). In fully
adjusted models IR without MetS (by any
definition) was not associated with a significantly
increased risk for diabetes, but this was the most
uncommon subgroup contributing the fewest
events (Figure). MetS without IR was associated
with a significant ~3-fold increased risk. ATP3 or
IDF MetS and IR were associated with a 6-7-fold
increased risk for diabetes, consistent with an
additive (on the log scale) effect of MetS and IR
on diabetes risk. In all models, first-order
interaction terms for MetS-by-IR were not
significant (all p > 0.6), confirming that MetS did
not confer greater diabetes risk in the presence of
IR than in the absence of IR. As we have
previously shown that the number of MetS-related
traits is positively associated with risk for diabetes
1
it was perhaps not surprising that EGIR MetS
(which although requires IR is a sum of as few as
3 to as many as 5 traits) conferred a lower relative
risk (3.3) than ATP3 or IDF MetS and IR (which
at minimum represent a sum of as few as 4 and as
many as 6 traits). To demonstrate this point we
conducted a subsidiary analysis of ATP3 or IDF
MetS and IR, but with MetS defined as any two or
more component traits. In this analysis relative
risks were very similar as for EGIR MetS. For
instance, the risk for diabetes relative to all
subjects without ATP3 MetS (2 or more traits) and
without IR was 3.03 (95% CI 1.93- 4.74). In fully
adjusted models all three MetS definitions, IR, and
their joint combinations were associated with
similar discriminatory capacity for diabetes
(adjusted aROCs 0.83-0.85), and MetS with IR
accounted for 42-66% of diabetes risk in the
population (Table 2).
Also shown in Table 2 are relative risks for
CVD. As reported previously 1 adjusted relative
risks for CVD associated with MetS were
substantially lower than for diabetes (Table 2),
ranging from 1.7–2.1. IR conferred similar (1.8)
risk for CVD as did MetS by any definition. In
fully adjusted models IR without MetS did not
confer significantly increased risk for CVD, and
only IDF MetS without IR significantly increased
risk for CVD. As for diabetes, relative risk sizes
were consistent with an additive effect of MetS
and IR on risk of CVD, and first-order interaction
terms for MetS-by-IR were not significant (all p
>0.3). In fully adjusted models, all MetS
definitions were associated with similar
discriminatory capacity for CVD (adjusted aROCs
0.73 – 0.74) and MetS and IR accounted for 14% 23% of CVD risk in the population.
Additional Subsidiary Analyses
FPG is a component of the MetS
definitions, and it is also a component of HOMAIR. This could potentially lead to overfitting FPG
in statistical models that included both MetS and
HOMA-IR.
We
therefore
we
used
hyperinsulinemia (≥upper quartile of fasting
serum insulin in subjects without diabetes) as an
alternative measure of IR. The results of this
analysis were virtually identical to those using to
HOMA-IR (not shown).
The increased CVD risk associated with
MetS might be explained, at least in part, by
incident diabetes and therefore we performed an
analysis that excluded those who developed
diabetes from the analysis of CVD events. This
analysis yielded a modest attenuation (~10-15%
lower) of the relative risks associated with MetS
and IR for incident diabetes and CVD but the
results of the analysis, including significance
levels, were essentially unchanged (not shown).
The increased diabetes risk associated with
MetS could be largely explained by the IFG
component of the definition. To explore this
further we excluded subjects with IFG from the
analysis and we used hyperinsulinemia as a
measure of IR (rather than HOMA-IR) to remove
fasting glucose as an exposure variable. In this
analysis ATP III and IDF MetS remained
significant predictors of incident diabetes in ageand sex-adjusted analyses (ATP III MetS (no
IFG): RR=2.2, p=0.02; IDF MetS (no IFG):
RR=3.2, p=0.0003) but neither remained
significant predictors in multivariable-adjusted
analyses (ATP III MetS (no IFG): RR=1.1, p=0.9;
IDF MetS (no IFG): RR=1.9, p=0.06). Compared
to the referent group, hyperinsulinemic subjects
with ATP III (no IFG) or IDF MetS (no IFG) were
at increased risk for incident diabetes in age- and
sex-adjusted analyses (ATP III MetS (no IFG) +
hyperinsulinemia: RR=4.2, p=0.0003; IDF MetS
(no IFG) + hyperinsulinemia: RR=6.0, p<0.0001)
but only IDF MetS (no IFG)/hyperinsulinemic
subjects remained at increased risk in
multivariable-adjusted models (ATP III MetS (no
IFG) + hyperinsulinemia: RR=1.7, p=0.3; IDF
MetS (no IFG) + hyperinsulinemia: RR=3.1,
p<0.01). This analysis suggests that the covariates
in the multivariate models, parental history of
diabetes and IGT, largely accounted for the
association of non-glucose MetS traits with risk of
diabetes.
Conclusions
This study provides several insights: a) the
level of agreement among ATP3, IDF and EGIR
MetS definitions is moderate or better, but only
half of individuals with ATP3 or IDF MetS are
insulin resistant defined by the conventional top
quartile of the HOMA-IR distribution; b)
individuals with ATP3 or IDF MetS but without
IR are at increased risk for diabetes, and those
with IDF MetS but without IR are at increased risk
for CVD, however, the joint presence of IR and
MetS indicates substantially increased risk for
diabetes or CVD; and c) all three MetS
definitions, IR, and their joint combinations are
associated with similar discriminatory capacity
and PAR% for incident diabetes or CVD. The data
support the general concept of risk factor
clustering as a diabetes and CVD risk factor, and
suggest that adding measurement of IR helps to
identify increased risk in individuals with MetS,
but on a population level, diagnosis of risk factor
clustering with or without IR leads to equivalent
ability to sort groups into higher and lower-risk
categories.
Our data confirm those of other population
studies showing a higher prevalence of IDF MetS
compared with ATP3 MetS 22-26 , largely due to
lower thresholds for elevated waist circumference
in IDF Mets vs. ATP3 MetS. Few studies have
examined the prevalence of EGIR MetS 27. Its
relatively low prevalence in Framingham is
accounted for by the requirement for IR and
higher thresholds defining elevated glucose,
triglycerides and blood pressure. A relatively high
level of agreement between ATP3 and IDF MetS
definitions is not surprising since both definitions
share common risk factors defined by generally
similar thresholds. Our data also confirm other
studies showing that only about half of people
with MetS have evidence of IR. 8.5, 6 In a prior
analysis from Framingham we showed that even
among obese subjects with ATP3 MetS the
prevalence of IR was only 68% 7. This present
study extends these data, and we show that the
widely-promoted ATP3 and IDF MetS definitions
are not, as commonly stated, 2, 3 synonymous with
IR. Only the EGIR definition, which requires the
presence of IR, represents a true ‘insulin resistance
syndrome’.
Several groups have promoted different definitions of
American, white, black, and Chinese samples
suggest that ATP3 and IDF MetS confer roughly
equivalent risk for diabetes, 9, 10 an observation
that we confirm in a large, unselected community
clustering might have value on the individual
based white sample. We extend this observation to
clinical level in people with MetS but is probably
show that MetS alone, but perhaps not the
unnecessary to further discriminate high risk
uncommon IR alone, increases risk for diabetes,
groups on the population level.
and that MetS and IR have additive effects
of this
study include
a derived
large, from fact
at independent
physiologic
domains
increasing diabetes risk. Data from Pima Indians also shows thStrengths
prospectively evaluated, community-based sample
be a significant diabetes risk factor. However the
assessed for standardized exposures and outcomes.
IR effect was in the direction of increased risk,
Limitations include that we did not have adequate
and the additive pattern of risk in the group with
sample size to subdivide the sample by sex. We
MetS and IR suggests that low power in the group
did not directly measure IR; use of proxy
with MetS but without IR accounted in part for the
measures will misclassify some people and
statistically non-significant association with
diminish the true magnitude of associations of IR
diabetes. In addition, a non-significant effect in
with outcomes. Finally, the Framingham
this group might be explained by imprecision in
population is white, so findings may have limited
the true estimation of IR inherent in use of proxy
generalizability.
measures, or the possibility that metabolic
In
summary,
prospective
analysis
correlates of IR confer much of its diabetogenic
demonstrates that ATP3 and IDF MetS are not
effect. 9 However, that MetS is a risk factor for
synonymous with an insulin resistant phenotype.
diabetes even in the absence of IR, and the clearly
The presence of IR substantially increases
additive effects of MetS and IR on diabetes risk
individual diabetes or CVD risk in people with
suggest that both are independent determinants of
MetS, but population risk prediction using MetS is
diabetes, and may operate by at least partially
similar with or without concurrent IR. A clinical
distinct pathways.
trial may be needed to test whether clinical use of
The foregoing discussion concerning risk
IR adds value to care aimed at reducing individual
for diabetes also applies to risk for CVD. In
metabolic risk. No one MetS definition offers a
addition, the present study underscores that, by
clear advantage for diabetes or CVD risk
any definition, and regardless of the presence or
detection. As such consensus on a definition for
absence of IR, MetS is a far more powerful risk
risk factor clustering would be helpful, as
factor for diabetes than for CVD. 1 Several studies
regardless how defined, risk factor clustering
have examined the risk for CVD associated with
identifies individuals and groups at marked
MetS by various definitions, and they have found
increased risk for future diabetes and modest risk
that all definitions of MetS are generally
for future CVD.
associated with a two-fold increased relative risk
for new CVD.10-13, 26 One can conclude that none
of the competing MetS definitions provides a
Acknowledgements and Disclosures
distinct advantage over the others as a CVD risk
Supported by the NHLBI’s Framingham Heart
prediction tool. We previously published an
Study (Contract No. N01-HC-25195), a grant from
analysis of Framingham data demonstrating that
GlaxoSmithKline, and by an American Diabetes
ATP3 MetS and IR were independently associated
Association Career Development Award to Dr.
with incident CVD over seven years of follow-up
29
Meigs. Dr. Meigs currently has research grants
. The present study advances that analysis by
from GlaxoSmithKline, Wyeth and sanofi-aventis,
extending surveillance up to 11 years,
and serves on safety or advisory boards for
dichotomizing MetS and IR into clinically
GlaxoSmithKline, Merck, and Lilly. Dr. Wilson is
identifiable groups, and demonstrating a clear
supported by grants from GlaxoSmithKline and
additive effect of MetS and IR on both diabetes
Wyeth. The funding agencies had no influence
and CVD risk. Elsewhere we have advocated the
over the decision to publish the findings. The
measurement of IR as part of MetS, to render it a
authors thank Peter Shrader MS for assistance
true insulin resistance syndrome, as in the EGIR
30
with the statistical analyses.
definition . The present study suggests that
measurement of IR as a component of risk factor
References
1.
Wilson PW, D'Agostino RB, Parise H, Sullivan L, Meigs JB. Metabolic syndrome as a precursor
of cardiovascular disease and type 2 diabetes mellitus. Circulation. 2005;112:3066-3072.
2.
Grundy SM, Cleeman JI, Daniels SR, Donato KA, Eckel RH, Franklin BA, Gordon DJ, Krauss
RM, Savage PJ, Smith SC, Jr., Spertus JA, Costa F. Diagnosis and Management of the Metabolic
Syndrome. An American Heart Association/National Heart, Lung, and Blood Institute Scientific
Statement. Circulation. 2005;112: 2735-52.
3.
Alberti KG, Zimmet P, Shaw J. Metabolic syndrome--a new world-wide definition. A Consensus
Statement from the International Diabetes Federation. Diabet Med. 2006;23:469-480.
4.
Balkau B, Charles MA. Comment on the provisional report from the WHO consultation. European
Group for the Study of Insulin Resistance (EGIR). Diabet Med. 1999;16:442-443.
5.
Liao Y, Kwon S, Shaughnessy S, Wallace P, Hutto A, Jenkins AJ, Klein RL, Garvey WT. Critical
evaluation of adult treatment panel III criteria in identifying insulin resistance with dyslipidemia.
Diabetes Care. 2004;27:978-983.
6.
Cheal KL, Abbasi F, Lamendola C, McLaughlin T, Reaven GM, Ford ES. Relationship to Insulin
Resistance of the Adult Treatment Panel III Diagnostic Criteria for Identification of the Metabolic
Syndrome. Diabetes. 2004;53:1195-1200.
7.
Meigs JB, Wilson PW, Fox CS, Vasan RS, Nathan DM, Sullivan L, D'Agostino RB. Body Mass
Index, Metabolic Syndrome and Risk of Type 2 Diabetes or Cardiovascular Disease. J Clin
Endocrinol Metab. 2006;91: 2906-12.
8.
Rathmann W, Haastert B, Icks A, Giani G, Holle R, Koenig W, Lowel H, Meisinger C. Prevalence
of the metabolic syndrome in the elderly population according to IDF, WHO, and NCEP
definitions and associations with C-reactive protein: the KORA Survey 2000. Diabetes Care.
2006;29:461-6
9.
Hanley AJ, Karter AJ, Williams K, Festa A, D'Agostino RB, Jr., Wagenknecht LE, Haffner SM.
Prediction of type 2 diabetes mellitus with alternative definitions of the metabolic syndrome: the
Insulin Resistance Atherosclerosis Study. Circulation. 2005;112:3713-3721.
10.
Wang JJ, Li HB, Kinnunen L, Hu G, Jarvinen TM, Miettinen ME, Yuan S, Tuomilehto J. How
well does the metabolic syndrome defined by five definitions predict incident diabetes and
incident coronary heart disease in a Chinese population? Atherosclerosis. 2006;[Epub ahead of
print].
11.
Dekker JM, Girman C, Rhodes T, Nijpels G, Stehouwer CD, Bouter LM, Heine RJ. Metabolic
syndrome and 10-year cardiovascular disease risk in the Hoorn Study. Circulation.
2005;112(5):666-673.
12.
Katzmarzyk PT, Janssen I, Ross R, Church TS, Blair SN. The importance of waist circumference
in the definition of metabolic syndrome: prospective analyses of mortality in men. Diabetes Care.
2006;29:404-409.
13.
Lawlor DA, Smith GD, Ebrahim S. Does the new International Diabetes Federation definition of
the metabolic syndrome predict CHD any more strongly than older definitions? Findings from the
British Women's Heart and Health Study. Diabetologia. 2006;49:41-48.
14.
Kannel WB, Feinleib M, McNamara PM, Garrison RJ, Castelli WP. An investigation of coronary
heart disease in families. The Framingham offspring study. Am J Epidemiol. 1979;110:281-290.
15.
Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis
model assessment: insulin resistance and beta-cell function from fasting plasma glucose and
insulin concentrations in man. Diabetologia. 1985;28:412-419.
16.
Bonora E, Targher G, Alberiche M, Bonadonna RC, Saggiani F, Zenere MB, Monauni T, Muggeo
M. Homeostasis model assessment closely mirrors the glucose clamp technique in the assessment
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
of insulin sensitivity: studies in subjects with various degrees of glucose tolerance and insulin
sensitivity. Diabetes Care. 2000;23:57-63.
Murabito JM, Nam BH, D'Agostino RB, Sr., Lloyd-Jones DM, O'Donnell CJ, Wilson PW.
Accuracy of offspring reports of parental cardiovascular disease history: the Framingham
Offspring Study. Ann Intern Med. 2004;140:434-440.
Meigs JB, Mittleman MA, Nathan DM, Tofler GH, Singer DE, Murphy-Sheehy PM, Lipinska I,
D'Agostino RB, Wilson PW. Hyperinsulinemia, hyperglycemia, and impaired hemostasis: the
Framingham Offspring Study. JAMA. 2000;283:221-228.
Cupples LA, D'Agostino RB. Section 34: Some risk factors related to the annual incidence of
cardiovascular disease and death using pooled repeated biennial measurements: Framingham
Heart Study, 30-year followup. Washington D.C: U.S. Department of Commerce; 1988.
Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics.
1977;33:159-174.
Rockhill B, Newman B, Weinberg C. Use and misuse of population attributable fractions. Am J
Public Health. 1998;88:15-19.
Adams RJ, Appleton S, Wilson DH, Taylor AW, Dal Grande E, Chittleborough C, Gill T, Ruffin
R. Population comparison of two clinical approaches to the metabolic syndrome: implications of
the new International Diabetes Federation consensus definition. Diabetes Care. 2005;28:27772779.
Athyros VG, Ganotakis ES, Elisaf M, Mikhailidis DP. The prevalence of the metabolic syndrome
using the National Cholesterol Educational Program and International Diabetes Federation
definitions. Curr Med Res Opin. 2005;21:1157-1159.
Ford ES. Prevalence of the metabolic syndrome defined by the international diabetes Federation
among adults in the u.s. Diabetes Care. 2005;28:2745-2749.
Lorenzo C, Serrano-Rios M, Martinez-Larrad MT, Gonzalez-Sanchez JL, Seclen S, Villena A,
Gonzalez-Villalpando C, Williams K, Haffner SM. Geographic variations of the International
Diabetes Federation and the National Cholesterol Education Program-Adult Treatment Panel III
definitions of the metabolic syndrome in nondiabetic subjects. Diabetes Care. 2006;29:685-691.
Saely CH, Koch L, Schmid F, Marte T, Aczel S, Langer P, Hoefle G, Drexel H. Adult Treatment
Panel III 2001 but not International Diabetes Federation 2005 criteria of the metabolic syndrome
predict clinical cardiovascular events in subjects who underwent coronary angiography. Diabetes
Care. 2006;29:901-907.
Balkau B, Charles MA, Drivsholm T, Borch-Johnsen K, Wareham N, Yudkin JS, Morris R,
Zavaroni I, van Dam R, Feskins E, Gabriel R, Diet M, Nilsson P, Hedblad B. Frequency of the
WHO metabolic syndrome in European cohorts, and an alternative definition of an insulin
resistance syndrome. Diabetes Metab. 2002;28:364-376.
Hanson RL, Imperatore G, Bennett PH, Knowler WC. Components of the "Metabolic Syndrome"
and incidence of type 2 diabetes. Diabetes. 2002;51:3120-3127.
Rutter MK, Meigs JB, Sullivan LM, D'Agostino RB, Sr., Wilson PW. Insulin resistance, the
metabolic syndrome, and incident cardiovascular events in the Framingham Offspring Study.
Diabetes. 2005;54:3252-3257.
Meigs JB. Definitions and mechanisms of the metabolic syndrome. Curr Opin Endocrinol
Diabetes. 2006;13:103-110.
Table 1: Baseline Characteristics According to ATP3 Metabolic Syndrome and Insulin Resistance Category
No ATP3 MetS
ATP3 MetS
No IR
IR
No IR
IR
p-value
N
1764
259Ź
340
440
Age (years)
52.3
53.1
56.6
56.2
<.0001
Sex (% female)
60.7
40.5
54.1
40.9
<.0001
Waist circumference (cm)
86
96
98.3
105.5
<.0001
Systolic Blood Pressure (mm Hg)
119
124
135
137
<.0001
Diastolic Blood Pressure (mm Hg)
72
76
79
80
<.0001
HDL cholesterol (mmol/l)
1.45
1.24
1.12
1.04
<.0001
Triglycerides (mmol/l)
1.08
1.39
2.02
2.13
<.0001
Fasting Glucose (mmol/l)
5.1
5.4
5.4
5.7
<.0001
2-hr OGTT glucose (mmol/l)
5.4
6.1
6.3
6.9
<.0001
BMI (kg/m2)
25.3
28.5
29
31.8
<.0001
Parental History of DM (%)
15.2
22.8
16.8
21.8
0.0006
LDL cholesterol (mmol/l)
3.2
3.4
3.6
3.4
<.0001
Smoking (%)
18.9
14.7
22.1
16.6
0.08
Data are mean values for continuous variables, proportions (%) for categorical variables and p-values from 3 d.f.
ANOVA for overall comparisons
Table 2. Relative Risks for Incident Type 2 Diabetes or CVD by ATP3, IDF or EGIR Metabolic Syndrome and/or Insulin Resistance Category
Multivariable-Adjusted†
Age-Sex-Adjusted
ATP3
IDF
EGIR
ATP3
IDF
EGIR
10.5 (6.5-16.9) d
0.78
76
8.1 (5.6-11.7) d
0.78
57
3.5 (2.2-5.6) d
0.84
55
2.6 (1.7-4.0) d
0.83
42
4.6 (2.7-7.7) d
0.85
66
3.3 (2.1-5.1) d
0.84
45
1.0
2.0 (0.9-4.3)
3.0 (1.6-5.7) c
6.0 (3.3-10.8) d
0.85
50
1.0
1.5 (0.5-4.1)
3.6 (1.9-6.8) d
6.9 (3.7-13.0) d
0.85
55
1.0
0.9 (0.4-2.4)
§
3.2 (2.0-5.1) d
0.84
44
1.8 (1.4-2.3) d
0.73
21
1.8 (1.4-2.3) d
0.73
18
1.7 (1.3-2.3) d
0.74
23
2.1 (1.6-2.7) d
0.73
19
1.0
1.2 (0.7-1.9)
1.3 (0.9-1.9)
2.3 (1.7-3.1) d
0.73
18
1.0
1.6 (1.0-2.7)
1.6 (1.1-2.2) a
2.2 (1.6-3.0) d
0.73
17
1.0
0.7 (0.4-1.5)
§
2.0 (1.6-2.7) d
0.74
19
aROC
PAR%
Insulin Resistance*
aROC
PAR%
Type 2 Diabetes
8.6 (5.7-12.9) d
0.78
67
6.5 (4.4-9.4) d
0.76
58
No MetS/No IR
No MetS/IR*
MetS/No IR*
MetS/IR*
aROC (MetS/IR)
PAR% (MetS/IR)
1.0
3.4 (1.6-7.1) b
5.4 (2.9-9.9) d
16.7 (10.2-27.4) d
0.82
56
1.0
2.4 (0.9-6.7)
5.9 (3.2-11.0) d
19.0 (11.0-32.8) d
0.82
61
1.0
1.4 (0.6-3.6)
§
8.4 (5.7-12.3) d
0.78
57
MetS*
aROC
PAR%
Insulin Resistance*
aROC
PAR%
1.8 (1.4-2.4) d
0.71
21
1.8 (1.4-2.3) d
0.70
18
1.8 (1.4-2.3) d
0.71
24
2.0 (1.6-2.7) d
0.71
19
No MetS/No IR
No MetS/IR*
MetS/No IR*
MetS/IR*
aROC (MetS/IR)
PAR% (MetS/IR)
1.0
1.2 (0.8-1.9)
1.4 (1.0-2.1)
2.3 (1.7-3.1) d
0.71
18
1.0
1.7 (1.1-2.8) a
1.7 (1.2-2.4) b
2.2 (1.6-3.0) d
0.71
17
1.0
0.8 (0.4-1.6)
§
2.0 (1.5-2.7) d
0.71
19
MetS*
CVD
P values are a, p<0.05; b, p<0.01; c, p<0.001, d, p<0.0001
* Referent is no MetS, no IR, or no MetS/No IR; aROC = area under the receiver operating characteristic curve; PAR% = population attributable risk
† Multivariable-adjusted = for DM: parental history of diabetes, BMI, and 2hr OGTT glucose; for CVD: LDL cholesterol and smoking
§ All with EGIR MetS are insulin resistant; there is no EGIR MetS/No IR category
3.5
DM
ATP3
IDF
EGIR
CVD
Percent per Year
3.0
2.5
2.0
1.5
1.0
*
*
0.5
0.0
No MetSNo IR
ATP3
IDF
EGIR
21
16
43
No MetSIR
MetSNo IR
11
5
5
22
27
*
MetSIR
81
87
87
No MetSNo IR
No MetSIR
MetSNo IR
107
91
143
22
21
9
36
52
*
MetSIR
75
76
88
Figure. Annualized incidence rates of type 2 diabetes (DM, left-hand panel) or cardiovascular disease (CVD,
right-hand panel) by ATP3, IDF, or EGIR metabolic syndrome (MetS) with or without insulin resistance (IR).
Number of DM or CVD events are given below each category. * By defintion all subjects with EGIR metabolic
syndrome had insulin resistance.