Quartiles of clamp glucose disposal rate reflect insulin sensitivity.
Error bars indicate 95% confidence intervals.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Ingelsson E, Sundström J, Ärnlöv J, Zethelius B, Lind L. Insulin Resistance and Risk of Congestive Heart Failure. JAMA. 2005;294(3):334–341. doi:10.1001/jama.294.3.334
Context Diabetes and obesity are established risk factors for congestive heart
failure (CHF) and are both associated with insulin resistance.
Objective To explore if insulin resistance may predict CHF and may provide the
link between obesity and CHF.
Design, Setting, and Participants The Uppsala Longitudinal Study of Adult Men, a prospective, community-based,
observational cohort in Uppsala, Sweden. We investigated 1187 elderly (≥70
years) men free from CHF and valvular disease at baseline between 1990 and
1995, with follow-up until the end of 2002. Variables reflecting insulin sensitivity
(including euglycemic insulin clamp glucose disposal rate) and obesity were
analyzed together with established risk factors (prior myocardial infarction,
hypertension, diabetes, electrocardiographic left ventricular hypertrophy,
smoking, and serum cholesterol level) as predictors of subsequent incidence
of CHF, using Cox proportional hazards analyses.
Main Outcome Measure First hospitalization for heart failure.
Results One hundred four men developed CHF during a median follow-up of 8.9
(range, 0.01-11.4) years. In multivariable Cox proportional hazards models
adjusted for established risk factors for CHF, increased risk of CHF was associated
with a 1-SD increase in the 2-hour glucose value of an oral glucose tolerance
test (hazard ratio [HR], 1.44; 95% confidence interval [CI], 1.08-1.93), fasting
serum proinsulin level (HR, 1.29; 95% CI, 1.02-1.64), body mass index (HR,
1.35; 95% CI, 1.11-1.65), and waist circumference (HR, 1.36; 95% CI, 1.10-1.69),
whereas a 1-SD increase in clamp glucose disposal rate decreased the risk
(HR, 0.66; 95% CI, 0.51-0.86). When adding clamp glucose disposal rate to
these models as a covariate, the obesity variables were no longer significant
predictors of subsequent CHF.
Conclusions Insulin resistance predicted CHF incidence independently of established
risk factors including diabetes in our large community-based sample of elderly
men. The previously described association between obesity and subsequent CHF
may be mediated largely by insulin resistance.
Congestive heart failure (CHF) is a major cause of morbidity and mortality.
The age-adjusted mortality for patients with CHF is 4 to 8 times that of the
general population.1 The predominant causes
of heart failure are hypertension and coronary heart disease. Other established
risk factors for CHF include left ventricular hypertrophy (LVH), valvular
heart disease, diabetes, cigarette smoking, obesity, and dyslipidemia.1-4
Diabetes as a predictor of subsequent CHF was first described in the
Framingham Heart Study 3 decades ago,5 and
the disease is frequently cited as a risk factor for CHF.1,2,6,7 Yet,
more detailed characterizations of the association between diabetes and subsequent
CHF are still lacking. In recent years, associations between diabetes or impaired
glucose regulation and altered left ventricular geometry and function have
been reported.8-10 Furthermore,
in patients with manifest CHF, insulin resistance is associated with more
severe disease and a worse prognosis,11-13 but
insulin resistance has not been investigated as a predictor of CHF. Obesity
is a more recently described risk factor for CHF3,6,7 and
is also associated with changes in left ventricular geometry and function.14 Abdominal obesity is closely associated with insulin
resistance and manifest diabetes.15
We hence hypothesized that insulin resistance may predict CHF and may
provide the link between obesity and CHF. Our primary aim was to analyze measures
of insulin sensitivity (including euglycemic insulin clamp glucose disposal
rate) and secretion as predictors of CHF incidence in a community-based sample
of elderly men, adjusting for diabetes and other traditional risk factors
for CHF. Our secondary aim was to analyze if the previously described association
between obesity and CHF may be mediated by insulin resistance.
The study is based on the Uppsala Longitudinal Study of Adult Men cohort
(http://www.pubcare.uu.se/ULSAM/), a health investigation focusing
on identifying metabolic risk factors for cardiovascular disease, to which
all 50-year-old men living in Uppsala, Sweden, in 1970-1974 were invited.
Of these, 2322 (82%) participated in the investigation.16 The
cohort was reinvestigated 20 years later (1990-1995, ie, the baseline of the
present study). Of the 1681 available 70-year-old men invited to the follow-up
investigation, 1221 (73%) attended. For the present study, 20 participants
were excluded due to a previous diagnosis of CHF and 14 due to a diagnosis
of valvular disease in the hospital discharge register at baseline. Thus,
1187 men were eligible for the present investigation. We examined a subsample
of 1061 nondiabetic men after exclusion of all participants with diabetes
at baseline (n = 126). Furthermore, we examined a subsample of 1034
nonobese men after exclusion of all men with body mass index (BMI) (calculated
as weight in kilograms divided by the square of height in meters) greater
than 30 at baseline (n = 153) and another subsample of 433 normal-weight
men after exclusion of all men with BMI greater than 25 at baseline (n = 754).
All participants gave written informed consent, and the ethics committee of
Uppsala University approved the study.
Examinations performed when the participants were 70 years of age included
a medical examination, a questionnaire, blood sampling (after an overnight
fast), supine blood pressure measurement, anthropometric measurements, a euglycemic
insulin clamp, an oral glucose tolerance test (OGTT), and measurement of insulin,
proinsulin, and lipid levels as previously described.10,17 Insulin
sensitivity was determined using the euglycemic insulin clamp technique, according
to DeFronzo et al,18 with a slight modification:
insulin was infused at a constant rate of 56 instead of 40 mU/(min × m2) to achieve nearly complete suppression of hepatic glucose output.19 Glucose disposal rate, representing insulin sensitivity,
was calculated as the amount of glucose taken up during the last 60 minutes
of the clamp procedure and is presented in mg/kg of body weight per minute.
An OGTT was performed by measuring the concentrations of plasma glucose and
immunoreactive insulin immediately before and 120 minutes after ingestion
of 75 g of anhydrous dextrose. The OGTT and the clamp procedure were performed
at least 1 week apart. The concentrations of intact and 32-33 split proinsulin
were analyzed using a 2-site immunometric assay technique.20 Specific
insulin concentrations were determined using a chemiluminescent immunoenzymatic
assay. Homeostasis model assessment insulin resistance index was calculated
as fasting insulin concentration × fasting glucose concentration/22.5.21
Blood pressure was measured in the supine position after resting for
10 minutes. The values were recorded twice to the nearest even value, and
the means of the 2 values were given. The presence of hypertension at baseline
was defined as systolic blood pressure at least 140 mm Hg and/or diastolic
blood pressure at least 90 mm Hg, and/or use of antihypertensive medication.
At baseline, 46% of patients with hypertension were treated with antihypertensive
medication. The presence of diabetes at baseline was defined as fasting plasma
glucose level of 126.1 mg/dL (7.0 mmol/L) or more and/or the use of oral hypoglycemic
agents or insulin.22 Electrocardiographic LVH
was defined as high-amplitude R waves according to the revised Minnesota code23 together with left ventricular strain pattern.4 Coding of smoking was based on interview reports,
and coding of demographic data was based on the questionnaire. The Swedish
hospital discharge register was used to assess the presence of valvular disease
(International Classification of Diseases, Ninth Revision [ICD-9] codes 394-397 and 424 or ICD-10 codes I05-I08 and I34-I37) and prior myocardial infarction (MI)
(ICD-9 code 410 or ICD-10 code
I21). The precision of the diagnosis of MI in the discharge register is high.24,25
The participants had a median follow-up time of 8.9 years (range, 0.01-11.4
years), contributing to 9899 person-years at risk. One hundred thirty-two
men had a hospital discharge register diagnosis of heart failure between the
age 70 baseline and the censor date at the end of 2002. As a possible diagnosis
of heart failure, we considered ICD heart failure
codes 428 (ICD-9) and I50 (ICD-10) and hypertensive heart disease with heart failure, I11.0 (ICD-10). The medical records from the hospitalization were reviewed
by 2 physicians (E.I. and L.L.) blinded to the baseline data, who classified
the cases as definite, questionable, or miscoded. The classification relied
on the definition proposed by the European Society of Cardiology,26 and the review process has been described extensively.27 After this validation, 104 definitive cases of heart
failure were included in the total cohort, 87 cases in the subsample without
diabetes, and 80 cases in the subsample without obesity. None of the participants
was lost to follow-up.
All analyses were defined a priori. Data are presented as mean (SD)
or percentage. Logarithmic transformation was performed to achieve normal
distribution if necessary. The residuals of all regression analyses were examined
and found to be normally distributed. Proportionality of hazards was confirmed
by visually examining Nelson-Aalen curves. We examined incidence rates in
quartiles of all continuous independent variables, and no obvious deviations
from linearity were observed. All variables were treated as continuous, except
for prior acute MI, hypertension, diabetes, electrocardiographic LVH, smoking,
and interim MI, which were treated as dichotomous. The prognostic values for
CHF incidence of a 1-SD increase in the continuous variables, or a transfer
from one level to another of the dichotomous variables, were investigated
with Cox proportional hazards analyses.
We investigated the independent variables in 5 sets of models in a hierarchical
fashion: unadjusted; adjusted for diabetes at baseline; adjusted for diabetes
plus other established risk factors for CHF (prior acute MI, hypertension,
electrocardiographic LVH, smoking, and serum cholesterol level) determined
at baseline; adjusted for diabetes and established risk factors, plus interim
MI during the follow-up period; and adjusted for diabetes and established
risk factors, plus clamp glucose disposal rate (to examine whether the obesity
measures remained predictors of CHF independent of the criterion standard
measure of insulin sensitivity). The models were repeated in a subsample excluding
all participants with diabetes at baseline and in 2 subsamples excluding all
men with BMI greater than 30 and BMI greater than 25 at baseline, respectively.
Pearson correlation coefficients were examined to evaluate the correlations
between variables reflecting glucose metabolism and those reflecting obesity.
In accordance with our a priori analysis plan, missing data were handled such
that only participants missing data on a covariate needed for a particular
model were excluded from the analyses, to maximize the statistical power.
To rule out an effect modification by established risk factors on the relation
of insulin sensitivity to CHF, we investigated interaction terms of each of
the established risk factors and clamp glucose disposal rate. Two-tailed 95%
confidence intervals (CIs) and P values were given,
with P<.05 regarded as significant (P values not shown). Analyses were performed using STATA version 8.2
(Stata Corp, College Station, Tex).
One hundred four participants developed CHF during follow-up, and the
incidence rate was 10.5 per 1000 person-years at risk. Table 1 shows the clinical characteristics at baseline.
In unadjusted Cox proportional hazards analyses, all examined variables
reflecting impaired glucose regulation and obesity were significant predictors
of heart failure incidence (Table 2,
second column). Incidence rates by quartiles of clamp glucose disposal rate
are shown in the Figure. When adjusting
for the presence of diabetes at baseline, the following variables remained
significant: clamp glucose disposal rate; OGTT 2-hour glucose level; fasting
levels of insulin, proinsulin, and 32-33 split proinsulin; BMI; and waist
circumference (Table 2, third column).
When adjusting also for other established baseline risk factors for CHF (prior
acute MI, hypertension, diabetes, electrocardiographic LVH, smoking, and serum
cholesterol level), the significant independent predictors of subsequent CHF
in separate models were clamp glucose disposal rate, OGTT 2-hour glucose level,
fasting proinsulin level, BMI, and waist circumference (Table 2, last column). These 5 variables each remained significant
predictors of subsequent CHF, with essentially the same point estimates and
CIs, when adding interim MI during follow-up to the covariates (Table 3, second column).
In unadjusted Cox proportional hazards analyses in the subsample excluding
participants with diabetes, the significant predictors of CHF incidence were
clamp glucose disposal rate, OGTT 2-hour glucose level, fasting levels of
proinsulin and 32-33 split proinsulin, BMI, and waist circumference (Table 4, second column). When adjusting for established
risk factors for CHF (prior acute MI, hypertension, electrocardiographic LVH,
smoking, and serum cholesterol level), the following variables remained significant
predictors of CHF in separate models: clamp glucose disposal rate, fasting
levels of proinsulin and 32-33 split proinsulin, BMI, and waist circumference
(Table 4, last column). These variables,
except for fasting 32-33 split proinsulin level, remained significant predictors
of subsequent CHF, with essentially the same CIs and point estimates, when
adding interim MI to the covariates (Table 3,
When repeating the unadjusted Cox proportional hazards analyses in the
subsample of nonobese men, the significant predictors of CHF incidence were
clamp glucose disposal rate, fasting glucose level, OGTT 2-hour glucose level,
fasting proinsulin level, BMI, and waist circumference (Table 5, second column). When adjusting for the presence of diabetes,
the following variables remained significant: clamp glucose disposal rate,
OGTT 2-hour glucose level, BMI, and waist circumference (Table 5, third column). When adjusting also for other established
risk factors for CHF (prior acute MI, hypertension, diabetes, electrocardiographic
LVH, smoking, and serum cholesterol level), only clamp glucose disposal rate
remained a significant predictor of CHF (Table
5, last column). Clamp glucose disposal rate remained a significant
predictor of subsequent CHF in this subsample also when adjusting for interim
MI as well as diabetes plus established risk factors (hazard ratio [HR], 0.73;
95% CI, 0.55-0.97). We also examined a subsample of normal-weight men (excluding
all participants with BMI >25 [n = 754]), but this left us with
a sample too small (433 participants, 23 cases) to draw any firm conclusions.
Nevertheless, the point estimates for clamp glucose disposal rate remained
similar but with wider CIs due to low statistical power (HR, 0.78; 95% CI,
0.51-1.18, in the unadjusted model and HR, 0.75; 95% CI, 0.45-1.24, in the
models adjusted for diabetes plus established risk factors).
The variables describing impaired glucose regulation and obesity were
highly correlated (Pearson r = −0.60, P<.001 for clamp glucose disposal rate vs both BMI and
waist circumference). In the models including obesity variables, diabetes
plus established risk factors, and clamp glucose disposal rate, the obesity
variables were no longer significant, whereas clamp glucose disposal rate
remained significant (Table 6). When
performing the same analyses in the subsample excluding participants with
diabetes at baseline and in the subsample of nonobese men, the same patterns
were observed but with larger CIs, rendering some associations borderline
significant (Table 6). None of the interaction
terms were significant.
In this community-based sample of elderly men free of CHF and valvular
disease at baseline, insulin resistance predicted CHF incidence independently
of diabetes and other established risk factors for CHF. Furthermore, our observations
indicate that the previously described association between obesity and subsequent
CHF may be mediated partly by insulin resistance.
Several previous longitudinal studies have shown an association between
diabetes and CHF.1,2,5-7 In
the present study, clamp glucose disposal rate and fasting proinsulin level,
mainly reflecting insulin resistance, were the strongest glucometabolic predictors
of CHF, both when adjusting for diabetes and in a subsample without diabetes.
To our knowledge, this is the first study to demonstrate a relation between
milder states of impaired glucose regulation and CHF incidence. Because information
about diabetes incidence during follow-up was not collected in a systematic
manner, it is possible that impaired glucose regulation at baseline was a
sign of impending diabetes, which is a known risk factor for CHF. Still, we
show that impaired glucose regulation in healthy participants without diabetes
or obesity at baseline is a strong predictor of subsequent CHF, independent
of established risk factors. Our observations may indicate that the risk for
CHF is already increased in the long subclinical phase of impaired glucose
regulation that precedes clinically manifest diabetes.
In previous studies, signs of impaired glucose regulation have been
related to both left ventricular systolic28 and
diastolic29 dysfunction and left ventricular
are numerous possible explanations for the observed relation of insulin resistance
to CHF incidence: (1) The formation of advanced glycosylation end products
is greatly accelerated in patients with diabetes,30 which
in the myocardium leads to increased collagen cross-linking and myocardial
stiffness.31 Ventricular function can be improved
and myocardial stiffness reversed in diabetic dogs when they are treated with
a collagen cross-link breaker such as metformin.31 (2)
Insulin may act as a growth factor in the myocardium, which is supported by
the experimental observation that sustained hyperinsulinemia leads to increased
myocardial mass and decreased cardiac output in rats.32 (3)
Hyperinsulinemia leads to sodium retention,33 which
may lead to decompensation in persons with otherwise subclinical myocardial
dysfunction due to volume expansion. (4) Hyperinsulinemia also leads to sympathetic
nervous system activation,34 which is a presumed
causal factor for CHF.1,35 (5)
Insulin resistance is related to an increased pressor response to angiotensin
II36 and has recently been demonstrated to
increase the stimulating effects of angiotensin II on cellular hypertrophy
and collagen production37 in individuals with
hypertension, leading to myocardial hypertrophy and fibrosis35 and
likely subsequent CHF.
Obesity as a risk factor for CHF has been established within the last
decade.3,6,7 In the
present study, BMI and waist circumference were strong predictors of CHF independently
of established risk factors for CHF. This demonstrates that both truncal and
overall obesity increase the risk of CHF to about the same degree. However,
as obesity is also strongly associated with diabetes and insulin resistance,15 we investigated whether the relation between obesity
and CHF may be mediated by insulin resistance. When clamp glucose disposal
rate was included in the multivariable models with BMI and waist circumference,
the obesity variables were no longer significant predictors of CHF. This observation
would be expected if insulin resistance were in the causal pathway between
obesity and CHF, which was our hypothesis. Furthermore, in the subsample of
nonobese men, clamp glucose disposal rate was a significant predictor of subsequent
CHF independent of established risk factors, whereas the obesity variables
were no longer significant predictors of CHF. These findings demonstrate that
insulin resistance is a risk factor for CHF independent of both truncal and
overall obesity. It may imply either that insulin resistance forgoes obesity
in a causal pathway leading to CHF, or simply that the relation of obesity
to CHF is circumstantial and that obesity in this case may be regarded as
an indicator of the more important trait, insulin resistance. It should be
noted that it is not possible from our data to definitely disentangle the
causative relations between obesity, insulin resistance, and CHF, but our
data do add an important piece of knowledge and should stimulate further research
in the area.
The strengths of this study include the large, community-based population,
the long follow-up period, and the detailed metabolic characterization of
the cohort. To our knowledge, the Uppsala Longitudinal Study of Adult Men
cohort is the largest population that has been examined with the criterion
standard for measurement of insulin resistance, the euglycemic insulin clamp
method. Furthermore, all CHF cases were validated, limiting the inclusion
of false-positive cases.
There are some limitations to this study. As we only examined men of
the same age with a similar ethnic background, this study has an unknown generalizability
to women or other age and ethnic groups. On the other hand, we circumvent
the powerful effects of age on CHF incidence. Since the CHF diagnosis was
based on a review of medical records, it was not possible to distinguish between
systolic and diastolic heart failure because echocardiography was not available
at the time of diagnosis for many of the cases. Thus, in our material it is
not possible to examine whether the impact of insulin resistance is different
on systolic vs diastolic heart failure. Finally, as noted above, we do not
have information on diagnosis of diabetes during follow-up.
Insulin resistance predicted CHF incidence independently of established
risk factors in our large community-based sample of elderly men. The previously
described association between obesity and subsequent CHF may be mediated largely
by insulin resistance. Further studies are needed to confirm our findings.
Corresponding Author: Erik Ingelsson, MD,
Department of Public Health and Caring Sciences, Section of Geriatrics, Uppsala
University, Box 609, SE-75125 Uppsala, Sweden (email@example.com).
Author Contributions: Dr Ingelsson had full
access to all of the data in the study and takes responsibility for the integrity
of the data and the accuracy of the data analysis.
Study concept and design: Ingelsson, Sundström,
Acquisition of data: Ingelsson.
Analysis and interpretation of data: Ingelsson,
Sundström, Ärnlöv, Zethelius, Lind.
Drafting of the manuscript: Ingelsson.
Critical analysis of the manuscript for important
intellectual content: Ingelsson, Sundström, Ärnlöv,
Statistical analysis: Ingelsson.
Administrative, technical, or material support:
Study supervision: Sundström, Ärnlöv,
Financial Disclosures: None reported.
Funding/Support: This study was supported by
Primary Health Care in Uppsala County, Swedish Heart Lung Foundation (Hjärt-Lungfonden),
the Ernfors Family Foundation, and Thuréus Foundation.
Role of the Sponsors: The funding sources had
no role in the design and conduct of the study; the collection, analysis,
and interpretation of the data; or the preparation, review, or approval of
Create a personal account or sign in to: