(Circulation. 1996;93:1809-1817.)
© 1996 American Heart Association, Inc.
Articles |
Insulin Sensitivity and Atherosclerosis
From the Department of Public Health Sciences, Bowman Gray School of Medicine of Wake Forest University (G.H., D.Z.), Winston-Salem, NC; Department of Radiology, Tufts-New England Medical Center (D.H.O.), Boston, Mass; Division of Clinical Epidemiology, University of Texas at San Antonio (S.H.); Department of Preventive Medicine and Biometrics, University of Colorado Health Sciences Center (M.R., R.H.), Denver; Division of Research, Kaiser Permanente (J.V.S.), Oakland, Calif; Division of Diabetes, Hypertension, and Nutrition, Department of Medicine, University of Southern California Medical Center (M.F.S.), Los Angeles; Division of Epidemiology and Clinical Applications, National Heart, Lung, and Blood Institute (P.S.), Bethesda, Md; and Department of Physiology and Biophysics, University of Southern California School of Medicine (R.B.), Los Angeles.
Correspondence to George Howard, DrPH, Department of Public Health Sciences, Bowman Gray School of Medicine of Wake Forest University, Medical Center Blvd, Winston Salem, NC 27157-1063. E-mail howard@phs.bgsm.wfu.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionAppendix 1References |
|---|
Background Reduced insulin sensitivity has been proposed as an important risk factor in the development of atherosclerosis. However, insulin sensitivity is related to many other cardiovascular risk factors, including plasma insulin levels, and it is unclear whether an independent role of insulin sensitivity exists. Large epidemiological studies that measure insulin sensitivity directly have not been conducted.
Methods and Results The Insulin Resistance Atherosclerosis Study (IRAS) evaluated insulin sensitivity (SI) by the frequently sampled intravenous glucose tolerance test with analysis by the minimal model of Bergman. IRAS measured intimal-medial thickness (IMT) of the carotid artery as an index of atherosclerosis by use of noninvasive B-mode ultrasonography. These measures, as well as factors that may potentially confound or mediate the relationship between insulin sensitivity and atherosclerosis, were available in relation to 398 black, 457 Hispanic, and 542 non-Hispanic white IRAS participants. There was a significant negative association between SI and the IMT of the carotid artery both in Hispanics and in non-Hispanic whites. This effect was reduced but not totally explained by adjustment for traditional cardiovascular disease risk factors, glucose tolerance, measures of adiposity, and fasting insulin levels. There was no association between SI and the IMT of the carotid artery in blacks. The association between SI and the IMT was stronger for the internal carotid artery than for the common carotid artery in all ethnic groups.
Conclusions Higher levels of insulin sensitivity are associated with less atherosclerosis in Hispanics and non-Hispanic whites but not in blacks. This effect is partially mediated by traditional cardiovascular risk factors.
Key Words: atherosclerosis • diabetes mellitus • insulin • carotid arteries
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
One large, cross-sectional study has established an association of fasting insulin levels and atherosclerosis,1 and at least two relatively small studies have shown a relationship of low SI to atherosclerosis.2 3 Low SI in this context can be defined as a relative inability of insulin to enhance glucose disposal. Insulin resistance is thought to promote atherosclerosis, in part through associated metabolic abnormalities, eg, hypertension, hyperglycemia, hyperinsulinemia, and dyslipidemia. It is important to measure insulin resistance directly and to focus future interventions on this underlying defect rather than to treat and monitor each associated abnormality separately. In fact, some drugs that effectively lower blood pressure (thiazides, ß-blockers), hyperglycemia (insulin), or lipid levels (nicotinic acid) may actually increase insulin resistance and worsen long-term outcomes. The IRAS directly measured insulin resistance in generally healthy adults and related it to subclinical atherosclerosis, with control for the levels of associated metabolic abnormalities.
Direct assessment of SI is expensive and time consuming, which makes studies of associations of SI with atherosclerosis in larger cohorts a difficult task. SI is correlated with both fasting insulin and 2-hour insulin levels after a glucose load4 5 and other "traditional" cardiovascular risk factors previously shown to cluster with SI in the insulin-resistance syndrome (Syndrome X).6 This correlation between SI and other factors related to the development of atherosclerosis makes assessment of the independent role of SI a challenge that requires a substantial sample size.
The relationship between SI and atherosclerosis may differ by ethnic group. Non-Hispanic whites are more insulin sensitive than Hispanics7 8 and blacks,8 and yet carotid wall thickness is similar or thinner in Hispanics than in non-Hispanic whites.9 SI has been shown to be related to lower blood pressure in whites but not in blacks in some10 but not all studies.11 Thus, it is possible that the relationship between SI and atherosclerosis may differ in various ethnic groups.
The IRAS is the first large epidemiological study to assess both SI and atherosclerosis directly. The study evaluated a large (1600+) triethnic (black, Hispanic, and non-Hispanic white) cohort in four US locations. The present report describes the relationship between SI and atherosclerosis as defined by the IMT of the ICA and CCA assessed by B-mode ultrasound. We hypothesized that a negative relationship exists between SI and atherosclerosis and that this relationship may be mediated by the other cardiovascular risk factors that make up the insulin-resistance syndrome, such as elevated blood pressure, larger body adiposity, and lower HDL levels. In addition, we evaluated whether the association of SI and atherosclerosis is mediated through circulating insulin.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
The IRAS is the first large epidemiological study conducted to examine the relationship between insulin levels and SI with measures of atherosclerosis and prevalent cardiovascular disease. The study recruited a triethnic population of 1625 subjects and was conducted at four clinical centers. The San Antonio, Tex, and San Luis Valley, Colo, centers recruited Hispanics and non-Hispanic whites from two ongoing population-based studies (the San Antonio Heart Study12 and the San Luis Valley Diabetes Study,13 respectively). The Los Angeles and Oakland, Calif, centers recruited non-Hispanic whites and blacks from Kaiser Permanente, a nonprofit HMO. Recruitment was tailored to yield approximately equal numbers of participants by ethnicity, sex, and glucose tolerance categories (NIDDM, impaired glucose tolerance, and normal). Persons with impaired glucose tolerance were the most difficult to find; hence, sampling strategies were focused on methods that would maximize the number of these participants. The San Antonio and San Luis centers drew random samples from participant lists from their respective ongoing studies, which resulted in selection of a disproportionate number of participants with previously documented impaired glucose tolerance. The Oakland and Los Angeles centers enriched their sample of persons with impaired glucose tolerance by oversampling from lists of nondiabetic persons with elevated levels of fasting plasma glucose (ie, 6.1 to 7.2 mmol/L). A total of 3416 potential participants were contacted to obtain the final sample of 1625, for an overall recruitment rate of 48%. The most common reason for exclusion at the screening interview was recent use of corticosteroids (n=101). Other reasons for exclusion included insulin treatment in the previous 5 years, severe limitation of caloric intake (<3360 kJ), decompensated congestive heart failure, decompensated emphysema or chronic lung disease, unstable angina, current treatment for cancer (other than skin cancer), seizure disorder or epilepsy, pregnancy, cognitive or psychological dysfunction (interviewer assessed), or other serious illnesses. The IRAS examination required two visits conducted approximately 1 week apart, each of which lasted approximately 4 hours. The oral glucose tolerance test and ultrasound examination were conducted during the first visit, and the FSIGT was performed during the second visit. Patients were asked to fast for 12 hours before each visit. Further details of the IRAS study and the recruitment process are available elsewhere.14
This report includes only 1397 subjects because (1) the ultrasound machines were transferred from the IRAS to another NHLBI-funded project for approximately the final 2 months of patient evaluation, which resulted in deletion of 111 participants from analysis, and (2) some subjects failed to complete the FSIGT evaluation, which resulted in deletion of an additional 117 participants from analysis.
B-mode real-time ultrasound was used to assess the IMT of the carotid artery wall by use of a protocol identical to that used in the Cardiovascular Health Study.15 Briefly, a bilateral assessment of the wall thickness was made in the ICA and CCA. For the ICA, the sonographer sought the site of maximal IMT thickness in the region between the dilatation of the carotid bulb and the ICA 1 cm distal to the tip of the flow divider. For the ICA, three images were obtained (bilaterally) at the site of maximal thickness at different interrogation angles (proximal, lateral, and anterior). For the CCA, bilateral images were obtained 1 cm proximal to the dilatation of the carotid bulb at a single (lateral) angle.
Ultrasound images were recorded on videotape and transferred to a central reading facility (D.H.O., principal investigator) for measurement of the IMT. For each of the eight available images, the maximal IMT was taken over a 1-cm segment of the arterial wall distant from the skin surface ("far wall"). Two summary measures were calculated: (1) the mean of the six ICA sites and (2) the mean of the two CCA sites. To allow equal weighting of the left and right arteries in the presence of missing data, the mean value of the available measures on the left ICA and the mean value of the available measures on the right ICA were calculated, and then the mean of these two means was used in analysis. This approach is identical or similar to that used to provide an index of atherosclerosis in other epidemiological studies15 16 17 18 19 and clinical trials.20 21 22
SI was assessed by an insulin-enhanced, FSIGT (12 samples)23 24 with minimal model analysis.25 From the minimal model, insulin sensitivity is expressed as parameter SI. In the IRAS, FSIGTs were performed with insulin injection (0.03 U/kg). The patterns of insulin and glucose levels during this test were assayed by our central laboratory. These insulin and glucose values were then used to estimate the parameters of the minimal model. This model is composed of two differential equations that are implemented on the digital computer. The overall model has four parameters that are estimated from the FSIGT; it has been documented previously that the ratio of two of these parameters (p3/p2) is an accurate SI measure, which is correlated with measures obtained with the more difficult glucose clamp procedure.25 Specifically, glucose (0.3 g/kg) and insulin (0.03 U/kg) were injected intravenously at 0 and 20 minutes, respectively. Blood samples were collected at 0, 2, 4, 8, 19, 22, 30, 40, 50, 70, 100, and 180 minutes for determination of glucose and insulin levels. This modified protocol has been proved to be a valid and reliable index of SI compared with the gold standard euglycemic clamp method, both overall (r=.62) and in subsets of persons with normal (r=.53) or impaired glucose tolerance (r=.48); however, its reliability is marginally lower among diabetic subjects (r=.41).26
Linear regression was used to estimate the relation between
SI and IMT. Early in the analysis, it became
apparent that there were clear ethnic differences in the
SI/IMT relationship verified by examining
interaction terms and that the mean IMT for participants with an
SI of zero was not adequately described by the simple
linear relation between SI and IMT (verified by examination
of regression residuals and by polynomial regression). To address the
ethnic differences in the SI/IMT relation, separate
linear models were fit for each of the three ethnic groups. The
SI for approximately 15% of IRAS participants was
estimated to be zero. Most of these individuals were known diabetics or
had diabetic glucose tolerance tests at the IRAS exam. To address the
nonlinearity at SI=0, we used the model:
 |
is the regression error term,
and ßk is the estimated regression parameter
to describe the relation between SI and IMT. Given this
parameterization, ß1 should be interpreted as
the regression slope that describes the relation of SI and
IMT among those participants with a nonzero SI value. The
ß2 parameter should be interpreted as the
difference between the mean IMT for participants with a 0 value for
SI and the IMT expected (for SI=0) from the
linear extrapolation of this relation among participants with nonzero
values of SI. Whereas the primary analysis was
performed for the entire IRAS cohort, confirmatory models for
analysis of the diabetic and nondiabetic subgroups were also
performed (see data in "Appendix").
The goal of the analysis was to assess the relation of
SI and IMT and to evaluate the impact of potential
confounding or mediating variables on this relation. As such, a
statistical modeling approach was used in which the
SI/IMT relation was first estimated after adjustment
for demographic factors only. Additional potential confounders,
suggested as a potential mechanism for SI, were
added to this model in a stepwise manner (the order of which was
determined a priori) to evaluate the potential mediating effect of
these variables on the SI/IMT relation. Six
models were considered. The basic model included demographic
variables of age, clinical center, and sex. The second model (model
2) evaluated the SI/IMT relation after adjustment
for the demographic variables plus the well-established
cardiovascular risk factors of HDL
cholesterol level, LDL cholesterol level,
smoking status (current smoker, past smoker, or never smoker), and
hypertension (systolic blood pressure 
140,
diastolic blood pressure 90, or current use of
hypertensive medications). The third model (model 3) adjusted for
demographic factors, established risk factors, and glucose tolerance
status categorized by use of the World Heath Organization
criteria27 (normal, impaired, or diabetic). The fourth
model (model 4) considered the relation of SI and IMT after
control for demographic factors, cardiovascular disease
risk factors, glucose tolerance status, and adiposity as indexed by
body mass index and waist-hip ratio. The fifth model (model 5)
adjusted for all factors in model 4 and added the fasting insulin
level. Finally, model 6 adjusted for all factors in model 5 and added
the 2-hour insulin level; hence, this final model adjusted for
demographic factors, cardiovascular risk factors,
glucose tolerance status, adiposity measures, and fasting and 2-hour
insulin levels.
There were clearly significant (P
.05) interactions between
SI and ethnicity noted in many models, which indicated that
the SI/IMT relation differed substantially among the
three ethnic groups. This interaction was not present for other
demographic factors such as sex (P>.1). Because of these
SI-by-ethnicity interactions, separate models fit for
each of the ethnic groups are presented herein. To provide
comparable models across ethnic groups, the model with an indicator
variable for SI=0 (described above) was used for all
analyses regardless of the statistical significance of the
indicator variable.
Because the distribution of SI is right skewed (skewness=2.8), there was some concern that individuals with high SI values could exert undue influence (leverage) on the estimated SI/IMT relation. A number of analyses were conducted to ensure this was not the case, including analysis of the relation of IMT with the log of SI, deletion of the 10 most highly leveraged points (approximately 1% of the sample), and deletion of the 10 highest SI values (again, approximately 1% of the sample). Each of these alternative analyses provided results that did not differ substantially from the primary analysis (data not shown).
To compare the independent predictive value of fasting insulin levels and 2-hour insulin levels after a glucose challenge with the predictive value of SI, models that related both fasting and 2-hour insulin levels with carotid artery wall thickness were fit after demographic adjustment for each ethnic group. Because neither fasting nor 2-hour insulin levels proved to be significantly related to IMT in the demographic model, other models were not considered.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
A description of the IRAS cohort available for the present analysis is provided in Table 1
. Non-Hispanic
white participants were more insulin sensitive and had lower body
adiposity than their black or Hispanic counterparts. Black participants
had higher blood pressure and lipid levels and were more likely to have
been categorized as hypertensive or diabetic than either the Hispanic
or non-Hispanic white participants. The highest rate of smoking was
among the Hispanic participants. There were substantial ethnic
differences in the mean IMT, with blacks having generally thicker
carotid artery walls. Many of these differences reflect established
population differences.28 We caution against making ethnic
comparisons, because these differences are not adjusted for clinical
center (and as such may be confounded by geography) and are of marginal
importance in that analyses were performed separately for each
ethnic group.
|
Because the multiple metabolic abnormalities associated with diabetes could alter the relation between SI and IMT, the SI/IMT relation was examined separately within the nondiabetic (normal plus impaired glucose tolerance) and diabetic IRAS subgroups (data shown in "Appendix"; none of the coefficients for the SI=0 terms were significant, and these were omitted from the "Appendix"). Differences in the SI/IMT relation between diabetics and nondiabetics were tested via an SI-by-diabetes interaction, which proved to be significant (0.01<P<.05) only for the CCA in blacks in the demographic model (model 1). In this case, there was a statistically nonsignificant negative association between SI and CCA IMT in nondiabetic participants and a statistically nonsignificant positive association between SI and CCA IMT in diabetic participants. This single significant difference between diabetics and nondiabetics in the association between SI and IMT does not substantially affect the interpretation (because both trends were statistically nonsignificant) and could have occurred by chance alone (the interaction was of borderline significance, and a substantial number of interactions were tested, which increased the chance of spurious findings). Because there was little evidence that the SI/IMT relation differs between diabetics and nondiabetics, the emphasis of the present report is on the combined analysis.
As shown in Table 2 and Fig 1, after
adjustment for demographic factors (model 1), the average ICA IMT in
both Hispanic and non-Hispanic white participants was 
30 µm
thinner in association with a 1-U increase in SI
(P.003) among those participants with nonzero
SI values. The magnitude of this effect was reduced by
20%, to 23 µm, by statistical adjustment for major
cardiovascular risk factors (model 2). The magnitude of
the association between ICA IMT and SI was similar after
this adjustment for non-Hispanic whites and Hispanics, and the relation
remained significant for non-Hispanic whites (P=.012) and
Hispanics (P=.050). Statistical adjustment for glucose
tolerance status reduced the magnitude of the association by an
additional 20%, to 17 µm per unit difference in
SI, which reduced the significance of the
association for both non-Hispanic whites (P=.072) and
Hispanics (P=.162). Additional statistical adjustment for
adiposity (model 4), fasting insulin level (model 5), and 2-hour
insulin level (model 6) had a very small effect on the estimated
magnitude of association and statistical significance of the
SI/ICA-IMT relation. As seen in Fig 1, for black
participants, the relation was nonsignificant (P>.5) in the
unanticipated direction that a thicker ICA IMT was associated with
higher SI values in all models.
|
|
For non-Hispanic whites, there was significant evidence (in all models
except model 1, in which there was marginally significant evidence)
that the average ICA IMT was 110 µm thinner among those
participants with a zero SI value than would be expected
from trend on the basis of those participants with a nonzero
SI value. This implied that after control for demographic
factors (model 1), the average ICA IMT for non-Hispanic white
participants with a zero SI was similar to that for
participants with an SI of 3.2 (SI-zero
coefficient/ßSI=-96.7/-30.1).
For both blacks and Hispanics, there was no evidence
(P>.05) that the average ICA IMT for participants with a
zero SI value differed from the trend established among the
participants with a nonzero SI value
(P>.4).
In the CCA, the average IMT was approximately 10 µm thinner for each
unit increase in SI in all three ethnic groups after
control for demographic factors (see model 1 in Table 2
and Fig 2), a difference that proved significant (in
non-Hispanic whites) or marginally significant (in blacks or
Hispanics). Adjustment for major cardiovascular risk
factors (model 2) had a relatively minor impact on the magnitude of the
association for the Hispanic and non-Hispanic white participants but
accounted for nearly all of the association in blacks. The association
between SI and CCA IMT remained significant for
non-Hispanic whites after adjustment for these
cardiovascular risk factors (P=.022);
however, the association for Hispanics became nonsignificant
(P=.256). Adjustment for glucose tolerance status,
adiposity, and fasting and 2-hour insulin levels (models 3, 4, 5, and
6) accounted for most of the remaining observed association for
Hispanics (P>.6). For non-Hispanic whites, adjustment for
glucose tolerance status marginally reduced the magnitude of the
association (P=.064). Additional adjustment for adiposity
(model 4), fasting insulin level (model 5), and 2-hour insulin level
(model 6) reduced the magnitude of the association to 7 µm per
unit difference in SI, a difference that was not
statistically significant (P>.18). For the CCA IMT,
participants with a zero SI differed little from the trend
established among those with a nonzero SI.
|
In Table 3, the relation between fasting and 2-hour
insulin levels with IMT is shown. None of the models indicated a
relation between insulin levels and IMT in either the CCA or ICA, and
other models were not considered. Because insulin levels do not have a
linear relation with glucose over the entire range of glucose tolerance
from normal to diabetic, linear regression techniques may not describe
the relation between insulin levels and IMT optimally. To address this
concern, the relation between fasting and 2-hour insulin levels with
IMT was estimated for diabetics and nondiabetics separately. Among the
24 models considered (separately for fasting and 2-hour insulin levels
among two diabetes status strata, three ethnic group strata, and the
CCA and ICA walls), there was a significant association
(P=.03) between the 2-hour insulin level and IMT only for
the CCA bed of diabetic Hispanic participants. These additional
analyses failed to support the hypothesis of a link between
insulin levels and IMT because they lacked a consistent pattern
of significance, there was a lack of any significant finding in
nondiabetics (in whom an association was more anticipated), and it was
likely that this single finding was a spurious result of the relatively
large number of statistical tests performed.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
These results indicate that for non-Hispanic whites and Hispanics, higher SI is associated with thinner IMT of the carotid artery wall. The association between decreased SI and carotid wall thickness (an indicator of atherosclerosis) is dependent in part on traditional cardiovascular risk factors, such as dyslipidemia, hypertension, and smoking, as might be predicted by studies of the insulin-resistance syndrome.6 However, the association between decreased SI and carotid wall thickness remains significant in non-Hispanic whites (and is of similar magnitude in Hispanics) even after adjustment for traditional cardiovascular risk factors, which suggests that insulin resistance may have an independent effect on atherogenesis.
The magnitude of this association in Hispanic and non-Hispanic whites
is relatively large. A difference in SI equivalent to the
interquartile difference (75th percentile minus 25th percentile,
2.21-0.41=1.8
min-1·µU-1 ·
mL-1) was associated with an estimated
difference in mean ICA IMT of 55 µm after adjustment for
demographic factors and 32 µm after adjustment for other major
cardiovascular risk factors (adiposity and fasting and
2-hour insulin levels). This contrasts with the reported difference of
60 µm to 70 µm between persons with and without prevalent heart
disease, hypertension, or smoking.29 30 In the IRAS data,
a 60-µm difference was also seen between hypertensives and
normotensives, and a difference of 90 µm was seen between current
smokers and nonsmokers, after adjustment for age, ethnicity, sex, and
clinic (data not shown). Therefore, moderate differences in
SI were associated with differences in IMT nearly as large
as those associated with many of the traditional
cardiovascular risk factors.
For Hispanic and non-Hispanic white participants, statistical
adjustment for a number of covariates did move the statistical
relations from significance (P.05) to nonsignificance
(P>.05) in each group. However, the consistency
of the magnitude of the association for Hispanics and non-Hispanic
whites across the spectrum of models considered is striking (Fig 1).
This similarity offers an internal replication of the results.
Combining the Hispanic and non-Hispanic white groups may be justified,
because it has been reported that the predominant ethnic admixture of
Hispanics is white.31 32 The combining of these groups is
also supported by the lack of statistical evidence of an
SI-by-ethnicity interaction between Hispanics and
non-Hispanic whites (P>.5). When these two groups are
combined, the association between SI and ICA IMT is highly
significant in the most basic model with adjustment for demographic
factors (regression coefficient=-28.4, P.0001) and
remains significant up to and including the model with adjustment for
demographic factors, traditional risk factors, diabetes status, and
measures of adiposity (model 4; regression coefficient=-18.3,
P=.028). With additional adjustment for fasting and 2-hour
insulin levels (model 6), the magnitude of the
SI/IMT relation is reduced marginally (regression
coefficient=-17.4) and the relation becomes marginally
significant (P=.058). Hence, even with adjustment for a wide
range of factors that may potentially mediate the effect of
SI, the magnitude of the SI/IMT
relation is reduced by only 39% and marginal significance is
maintained. Thus, these data suggest the observed association of
increased SI with decreased IMT in the model with only
demographic adjustment is not entirely mediated by associations of
SI with traditional cardiovascular disease
risk factors and insulin levels in Hispanics and non-Hispanic
whites.
In another group of whites, Laakso and coworkers2 also
reported an association between greater SI and less
atherosclerosis in a case-control study of 43
normoglycemic Finns.2 In that study, the SI of
30 participants with thick CCA IMT as measured by B-mode ultrasound was
contrasted with 13 control participants with thin IMT. The study by
Laakso and coworkers did not adjust for cardiovascular
risk factors, adiposity, or insulin levels; hence, it is most
comparable to the demographic model (model 1) reported in the
present study. In model 1, we found a significant
(P.0001) association between CCA IMT and SI
among non-Hispanic whites, which is consistent with the
findings of Laakso et al.2 CCA IMT was also significantly
correlated with SI measured by the euglycemic
clamp in 25 Swedish men considered to be at high risk of
atherosclerosis and 23 Swedish men considered to be at
low risk of atherosclerosis.3 By
stratifying for risk of atherosclerosis, this
report3 "adjusted" for major
cardiovascular risk factors and is most comparable to
our model after adjustment for cardiovascular risk
factors (model 2). We found that the association between CCA IMT and
SI remained significant (P=.022) in non-Hispanic
whites after adjustment for demographic and
cardiovascular risk factors. The current report
confirms and extends earlier findings by examination of the association
between SI and IMT in larger samples from three ethnic
groups before and after adjustment for additional potential
confounders.
It has been suggested that higher insulin levels (either by themselves or as a reflection of higher levels of insulin resistance) are related to the development of atherosclerosis both through a direct effect on the arterial wall and indirectly through their effects on lipids and blood pressure.33 A direct mechanism for the atherogenicity of insulin could involve its ability to stimulate both lipid synthesis in arterial tissue and proliferation of arterial smooth muscle cells. Alternatively, insulin and insulin resistance have been shown to be related to a clustering of major cardiovascular risk factors that are part of the insulin-resistance syndrome that includes hypertension and dyslipidemia.6 Although our analysis showed a substantial relation for SI with IMT among non-Hispanic whites and Hispanics, we failed to document a notable relation of either fasting or 2-hour insulin levels with IMT. These data suggest that the relation of SI with atherosclerosis is stronger than that between insulin measures and atherosclerosis.
Control for lipids, hypertension, and obesity, each of which is a factor in the insulin-resistance syndrome, substantially reduced the estimated impact of SI, which suggests that these factors may be intermediate factors in the relation between SI and IMT. However, even with further adjustment for fasting and 2-hour insulin levels, approximately half of the SI/IMT relation remained unexplained. It is possible that control for hypertension, HDL cholesterol levels, LDL cholesterol levels, smoking, and glucose tolerance status provided an inadequate control for the factors in the insulin-resistance syndrome. These factors, as well as fasting and 2-hour insulin levels, are measured with error. More precise measurement and modeling may account for an additional portion of the estimated SI/IMT relation. Other risk factors, including plasma fibrinogen and plasminogen activator inhibitor-1, have been shown to be associated with insulin resistance and may contribute to the residual SI/IMT association. Alternatively, a more direct effect of SI on the development of atherosclerosis may be operative. Hence, we suggest that insulin resistance is related to the development of atherosclerosis in Hispanics and non-Hispanic whites not only through its association with factors in the insulin resistance syndrome, but also through some other mechanism.
The strength of the SI/ICA-IMT association in both
Hispanics and non-Hispanic whites contrasts with that observed for
blacks. For the black IRAS participants, the estimated association
exists nonsignificantly in the unanticipated direction (Fig 1; Table 2). Whereas the association of SI and IMT for blacks did
not differ significantly from zero (P>.5), it did differ
significantly from that observed for Hispanics and non-Hispanic whites
(P<.05). This difference between Hispanics/non-Hispanic
whites and blacks in the association of SI and IMT is
consistent with a similar difference observed in blacks and
non-Hispanic whites in the association of SI and blood
pressure.10 34 It has been suggested that SI
and glucose tolerance status are less associated in blacks, as up to
50% of blacks with NIDDM remain insulin sensitive.34 35
In the IRAS, control for hypertension did not substantially modify the
SI/IMT relation in blacks. Therefore, the mechanism
for the ethnic difference in the SI/IMT relation
remains unexplained.
A sizable proportion of the IRAS participants had an SI
value of zero: 16% of blacks, 16% of Hispanics, and 15% of
non-Hispanic whites. Most of the zero SI values were
observed in diabetic participants who did not require insulin
(diabetics who required insulin were excluded from the IRAS because
they would respond differently to the insulin and glucose challenges
used in the FSIGT). There was a tendency for the average IMT of these
participants to be thinner than the IMT that would be expected on the
basis of the trend estimated among the participants with a nonzero
SI, although it reached statistical significance in
less than half of the models considered. The underlying reasons for
this finding could include at least two possibilities. First, the IRAS
recruitment process could have resulted in the selection of
resistant subjects who, for other, unmeasured reasons, were
relatively protected from atherosclerosis. For example,
the IRAS excluded all NIDDM subjects who had ever used insulin.
Patients with established diabetes who have not required insulin
therapy may have less severe disease and be at lower biological risk
for atherosclerotic complications than their insulin-using
counterparts. Similarly, persons with both low SI and NIDDM
may have particularly severe atherosclerosis that would
lead to early mortality or severe morbidity. The IRAS may have excluded
this group by the obvious selective mortality and by exclusion of
patients with unstable angina and congestive heart disease. Second, it
is possible that there exists a subset of subjects who are truly very
insulin resistant (as estimated by a zero SI) and
who have an unknown protective mechanism to deter the development of
atherosclerosis. The relatively frequent occurrence of
a zero SI, particularly among diabetic participants,
also has several possible explanations. These estimates may reflect an
inability of the FSIGT protocol to fully stimulate
insulin-sensitive tissues in some subjects. It is possible that
some subjects would show no discernible response to the dose of insulin
used in the FSIGT (0.03 U/kg or 2 U over 3 hours) or would have an
SI that is actually very close to zero at the level of
insulinemia achieved during the test. Also, the frequent occurrence of
a zero SI could have resulted from a failure of the
analysis program (MINMOD) to achieve numerical convergence at
an appropriate point on the estimation surface. A substudy is currently
under way within the IRAS to further understand the factors that
underlie an estimate of a zero SI. In the present
report, the effect of these zero-SI participants has been
removed statistically from the estimation of the relation between the
measured SI and IMT among those participants with nonzero
SI values by the introduction of the indicator variable
for SI=0.
There were also notable differences in the SI/IMT relation between the ICA and CCA beds. These differences may reflect differences in the likelihood of atherosclerotic development at the two sites or differences in the IRAS scanning protocol between the two beds. It has been noted that development of atherosclerosis tends to begin in the region of the carotid bifurcation and then to extend in either direction (distal or proximal) to other sites.36 Therefore, in the relatively young IRAS cohort, increased IMT associated with the development of atherosclerosis is more likely to be observed in the internal carotid images, including the bifurcation region, than in the common carotid images, proximal to the dilatation of the carotid bulb. The idea that a thickened IMT associated with atherosclerosis is more likely to be observed in the ICA than the CCA is further strengthened by the IRAS scanning protocol. The ICA images are taken at the point of maximal intimal-medial thickening, whereas the CCA images were taken at a fixed point below the dilatation of the carotid bulb. Because there are established correlations between IMT at different sites within the carotid artery bed,37 the similar pattern observed in these analyses between the ICA and CCA was expected. However, the ultrasound protocol used in the ICA compared with the CCA should have higher accuracy, because a larger number of images were averaged, and it may be more likely to reflect atherosclerotic thickening, because it more actively sought "maximal thickening."
The present study has several limitations. First, the SI/IMT relation was estimated in a cross-sectional manner, and therefore causation of the protective effect of increased sensitivity cannot be addressed directly. Clearly, it is important to confirm these cross-sectional relations in a longitudinal study. Second, attempts to collect a sufficient number of IRAS participants across the spectrum of glucose tolerance did not permit the IRAS to be a strictly population-based study. We believe this is a minor concern, because the IRAS population was drawn from existing population-based studies (San Antonio Heart Study and San Luis Valley Diabetes Study) or from HMO populations (Oakland and Los Angeles, Calif), so that the population was basically representative of the general population, and the focus of the present report is on the relationship between factors measured in the study participants, rather than a description of the distribution of the factors in the general population. We also assume that cross-sectional analyses will tend to underestimate the magnitude of associations between risk factors and atherosclerosis, as did those analyses presented here, because those individuals most at risk will not be available for recruitment ("survivor" bias). Finally, we cannot rule out that there may have been misclassification of conventional cardiovascular risk factors, which would lead to an underestimation of their effect, or that other relevant confounders (eg, plasminogen activator inhibitor-1, sex hormone–binding globulin, and others) were not considered.
In conclusion, data from the initial IRAS examination indicate an inverse association between SI and atherosclerosis as assessed by measures of carotid wall thickness in Hispanics and non-Hispanic whites. This association is reduced but not completely eliminated by confounding factors, including HDL, LDL, smoking, hypertension, glucose tolerance status, body mass index, waist-hip ratio, and fasting insulin; this suggests that the association is mediated in part by the relationship of diminished SI with several established cardiovascular disease risk factors. For unknown reasons, this relationship of SI with atherosclerosis was not seen in black subjects, a finding reminiscent of the lack of association of SI with blood pressure in blacks in some studies. Longitudinal studies will be necessary to further clarify these relationships.
|
Selected Abbreviations and Acronyms |
|---|
|
|
Acknowledgments |
|---|
This work was supported by NHLBI grants HL47887, HL47889, HL47890, HL47892, and HL47902 and by the General Clinical Research Centers Program (National Center for Research Resources GCRC, M01 RR431, M01 RR01346).
|
Appendix 1 |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
|
Received October 24, 1995; revision received January 18, 1996; accepted January 22, 1996.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
-
Folsom AR, Eckfeldt JH, Weitzman S, Ma J,
Chambless LE, Barnes RW, Cram KB, Hutchinson RG, for the
Atherosclerosis Risk in Communities (ARIC) Study
Investigators. Relation of carotid artery wall thickness to diabetes
mellitus, fasting glucose and insulin, body size, and physical
activity. Stroke. 1994;25:66-73. [Abstract]
-
Laakso M, Sarlund H, Salonen R, Suhonen M,
Pyörälä K, Salonen JT, Karhapää P.
Asymptomatic atherosclerosis and insulin
resistance. Arterioscler Thromb. 1991;11:1068-1076.
[Abstract/Free Full Text]
-
Agewall S, Fagerbern B, Attvall S, Wendelhag I,
Urbanavicius V, Wikstrand J. Carotid artery intima-media
thickness is associated with insulin-mediated glucose disposal in
men at high and low coronary risk. Stroke. 1995;26:956-960.
[Abstract/Free Full Text]
-
Laakso M. How good a marker is insulin level
for insulin resistance? Am J Epidemiol. 1993;137:959-965.
[Abstract/Free Full Text]
-
Hollenbeck CB, Chen N, Chen Y-DI, Reaven GM.
Relationship between the plasma insulin response to oral glucose and
insulin-stimulated glucose utilization in normal subjects.
Diabetes. 1984;33:460-463. [Abstract]
-
Reaven GM. 1988 Banting lecture: role of
insulin resistance in human disease. Diabetes. 1988;37:1595-1607. [Abstract]
-
Haffner SM, Stern MP, Dunn JR, Mobley M, Blackwell J,
Bergman RN. Diminished insulin sensitivity and increased insulin
response in non-obese non-diabetic Mexican Americans.
Metabolism. 1990;39:842-847. [Medline]
[Order article via Infotrieve]
-
Haffner SM, D'Agostino R Jr, Saad MF.
Increased insulin resistance and insulin secretion in non-diabetic
African Americans and Hispanics compared to non-Hispanic whites: the
Insulin Resistance Atherosclerosis Study.
Diabetes. In press.
-
D'Agostino R, Burke G, O'Leary D, Haffner S, Rewers
M, Howard G, for the IRAS Investigators. Ethnic differences in carotid
wall thickness: the Insulin Resistance Atherosclerosis
Study (IRAS). Circulation. 1996;93:630. Abstract.
-
Saad MF, Lillioja S, Nyomba BL, Castillo C, Ferraro R,
de Gregorio M, Ravussin E, Knowler WC, Bennett PH, Howard BV, Borardus
C. Racial differences in the relation between blood pressure and
insulin resistance. N Engl J Med. 1991;324:733-739. [Abstract]
-
Falkner B, Hulman D, Tannenbaum J, Kushner K.
Insulin resistance and blood pressure in young black men.
Hypertension. 1990;16:706-711.
[Abstract/Free Full Text]
-
Stern MP, Rosenthal M, Haffner SM, Hazuda HP, Franco
LJ. Sex difference in the effects of sociocultural status on
diabetes and cardiovascular risk factors in Mexican
Americans: the San Antonio Heart Study. Am J
Epidemiol. 1984;120:834-851.
[Abstract/Free Full Text]
-
Hamman RF, Marshall JA, Baxter J, Kahn LB, Mayer EJ,
Orleans M, Murphy JR, Lezotte DC. Methods and prevalence of
non-insulin dependent diabetes mellitus in a biethnic Colorado
population: the San Luis Valley Diabetes Study. Am J
Epidemiol.. 1989;129:296-311.
-
Wagenknecht LE, Mayer EJ, Rewers M, Haffner S, Selby J,
Borok GM, Henkin L, Howard G, Savage PJ, Saad MF, Bergman RN, Hamman
R. The Insulin Resistance Atherosclerosis Study
(IRAS): objectives, design, and recruitment results. Ann
Epidemiol.. 1995;5:464-472. [Medline]
[Order article via Infotrieve]
-
O'Leary DH, Polak JF, Wolfson SK Jr, Bond MG, Bommer
W, Sheth S, Psaty BM, Sharrett AR, Manolio TA, on behalf of the CHS
Collaborative Research Group. Use of sonography to evaluate carotid
atherosclerosis in the elderly: the
Cardiovascular Health Study. Stroke. 1991;22:1155-1163.
[Abstract/Free Full Text]
-
Bond MG, Barnes RW, Riley WA, Wilmoth SK, Chambless LE,
Howard G, Owens B, for the ARIC Study Group. High-resolution B-mode
ultrasound scanning methods in the Atherosclerosis Risk
in Communities (ARIC) study. J Neuroimaging. 1991;1:68-73. [Medline]
[Order article via Infotrieve]
-
Riley WA, Barnes RW, Bond MG, Evans GW, Chambless LE,
Heiss G. High-resolution B-mode ultrasound reading methods
in the Atherosclerosis Risk in Communities (ARIC)
cohort. J Neuroimaging. 1991;1:168-172. [Medline]
[Order article via Infotrieve]
-
Salonen R, Salonen JT. Determinants of carotid
intima-media thickness: a population-based ultrasonography
study in eastern Finnish men. J Intern
Med. 1991;229:225-231. [Medline]
[Order article via Infotrieve]
-
Bots ML, Hofman A, de Bruyn AM, de Jong PTVM, Grobbee
DE. Isolated systolic hypertension and vessel wall
thickness of the carotid artery: the Rotterdam Elderly Study.
Arterioscler Thromb. 1993;13:64-69.
[Abstract/Free Full Text]
-
The ACAPS Group. Rationale and design for the
Asymptomatic Carotid Artery Plaque Study (ACAPS).
Control Clin Trials. 1992;13:293-314. [Medline]
[Order article via Infotrieve]
-
Crouse JF, Byington RP, Bond MG, Espeland MA, Sprinkle
JW, McGovern M, Furberg CD. Pravastatin, lipids, and
atherosclerosis in the carotid arteries: design
features of a clinical trial with carotid
atherosclerosis outcome. Control Clin
Trials. 1992;13:495-506. [Medline]
[Order article via Infotrieve]
-
Mack WJ, Selzer RH, Hodis HN, Erickson JK, Liu CR, Liu
CH, Crawford DW, Blankenhorn DH. One-year reduction and
longitudinal analysis of carotid intima-media thickness
associated with colestipol/niacin therapy. Stroke. 1993;24:1779-1783.
[Abstract/Free Full Text]
-
Bergman RN, Finegood DT, Ader M. Assessment of
insulin sensitivity in vivo. Endocr Rev. 1985;6:45-86. [Medline]
[Order article via Infotrieve]
-
Yang Y, Youn JH, Bergman RN. Modified protocols
improve insulin sensitivity estimation using the minimal model.
Am J Physiol. 1987;253:E595-E602.
[Abstract/Free Full Text]
-
Pacini G, Bergman RN. MINMOD: a computer program
to calculate insulin sensitivity and pancreatic responsivity from the
frequently sampled intravenous glucose tolerance
test. Comput Methods Programs Biomed. 1986;23:113-122. [Medline]
[Order article via Infotrieve]
-
Saad MF, Anderson RL, Laws A, Watanabe RM, Kades WW,
Chen YDI, Sands RE, Pei D, Savage PH, Bergman RN. A comparison
between the minimal model and the glucose clamp in the assessment of
insulin sensitivity across the spectrum of glucose tolerance.
Diabetes. 1994;43:1114-1121. [Abstract]
-
World Health Organization. Diabetes
Mellitus. Geneva, Switzerland; World Health Organization; 1985.
Technical Report Series No. 727.
-
Keil JE, Sutherland SE, Knapp RG, Lackland DT, Gazes
PC, Tyroler HA. Mortality rates and risk factors for
coronary disease in black as compared with white men and
women. N Engl J Med. 1993;329:73-78.
[Abstract/Free Full Text]
-
Burke GL, Evans GW, Riley WA, Sharrett AR, Howard G,
Barnes RW, Rosemond W, Crow RS, Rautaharju PM, Heiss G.
Arterial wall thickness is associated with prevalent
cardiovascular disease in middle-aged adults: the
Atherosclerosis Risk in Communities (ARIC)
study. Stroke. 1995;26:386-391.
[Abstract/Free Full Text]
-
Howard G, Burke GL, Szklo M, Tell GS, Eckfeldt J, Evans
GW, Heiss G. Active and passive smoking are associated with
increased carotid wall thickness: the Atherosclerosis
Risk in Communities (ARIC) study. Arch Intern Med. 1994;154:1277-1282. [Medline]
[Order article via Infotrieve]
-
Chakraborty R, Ferrell RE, Stern MP, Haffner SM, Hazuda
HP, Rosenthal M. Relationship of prevalence of non-insulin
dependent diabetes mellitus with Amerindian admixture in the Mexican
Americans of San Antonio, Texas. Genet Epidemiol. 1986;3:435-454. [Medline]
[Order article via Infotrieve]
-
Gardner LL, Stern MP, Haffner SM, Gaskill SP, Hazuda
HP, Relethford JH, Eiffler CW. Prevalence of diabetes in Mexican
Americans: relationship to percent of gene pool derived from Native
American sources. Diabetes. 1984;33:86-92. [Abstract]
-
Stout RW. Insulin and atheroma:
20-yr perspective. Diabetes Care. 1990;13:631-654. [Abstract]
-
Chaiken RL, Banerji MA, Huey H, Lebovitz H. Do
blacks with NIDDM have an insulin-resistance syndrome?
Diabetes. 1993;42:444-449. [Abstract]
-
Banerji MA, Lebovitz HE. Insulin action in black
Americans with NIDDM. Diabetes Care. 1992;15:1295-1302. [Abstract]
-
Solberg LA, Eggen DA. Localization and sequence
of development of atherosclerotic lesions in the carotid and vertebral
arteries. Circulation. 1971;43:711-724.
[Abstract/Free Full Text]
-
Howard G, Burke GL, Evans GW, Crouse JR, Riley W,
Arnett D, de Lacy R, Heiss G. Relations of intimal-medial
thickness among sites within the carotid artery as evaluated by B-mode
ultrasound. Stroke. 1994;25:1581-1587.[Abstract]
This article has been cited by other articles:
 |
|
 |
|
  P. Y. Hsue, K. Squires, A. F. Bolger, B. Capili, G. A. Mensah, Z. Temesgen, C. A. Wanke, D. A. Wohl, and for Working Group 4 Screening and Assessment of Coronary Heart Disease in HIV-Infected Patients Circulation, July 8, 2008; 118(2): e41 - e47. [Full Text] [PDF]
|
|
|
|
|
|
R. L. Pande, T. S. Perlstein, J. A. Beckman, and M. A. Creager Association of Insulin Resistance and Inflammation With Peripheral Arterial Disease: The National Health and Nutrition Examination Survey, 1999 to 2004 Circulation, July 1, 2008; 118(1): 33 - 41. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Sanchez, T. Priego, M. Palou, A. Tobaruela, A. Palou, and C. Pico Oral Supplementation with Physiological Doses of Leptin During Lactation in Rats Improves Insulin Sensitivity and Affects Food Preferences Later in Life Endocrinology, February 1, 2008; 149(2): 733 - 740. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Muniyappa, S. Lee, H. Chen, and M. J. Quon Current approaches for assessing insulin sensitivity and resistance in vivo: advantages, limitations, and appropriate usage Am J Physiol Endocrinol Metab, January 1, 2008; 294(1): E15 - E26. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. G. Bertoni, N. D. Wong, S. Shea, S. Ma, K. Liu, S. Preethi, D. R. Jacobs Jr, C. Wu, M. F. Saad, and M. Szklo Insulin Resistance, Metabolic Syndrome, and Subclinical Atherosclerosis: The Multi-Ethnic Study of Atherosclerosis (MESA) Diabetes Care, November 1, 2007; 30(11): 2951 - 2956. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. A. Marini, E. Succurro, S. Frontoni, M. L. Hribal, F. Andreozzi, R. Lauro, F. Perticone, and G. Sesti Metabolically Healthy but Obese Women Have an Intermediate Cardiovascular Risk Profile Between Healthy Nonobese Women and Obese Insulin-Resistant Women Diabetes Care, August 1, 2007; 30(8): 2145 - 2147. [Full Text] [PDF]
|
|
|
|
|
|
M. F. Crutchlow, B. Robinson, B. Pappachen, N. Wimmer, A. J. Cucchiara, D. Cohen, and R. Townsend Validation of Steady-State Insulin Sensitivity Indices in Chronic Kidney Disease Diabetes Care, July 1, 2007; 30(7): 1813 - 1818. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Beysen, E. J. Murphy, T. McLaughlin, T. Riiff, C. Lamendola, H. C. Turner, M. Awada, S. M. Turner, G. Reaven, and M. K. Hellerstein Whole-Body Glycolysis Measured by the Deuterated-Glucose Disposal Test Correlates Highly With Insulin Resistance In Vivo Diabetes Care, May 1, 2007; 30(5): 1143 - 1149. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Bruemmer C-Peptide in Insulin Resistance and Vascular Complications: Teaching an Old Dog New Tricks Circ. Res., November 24, 2006; 99(11): 1149 - 1151. [Full Text] [PDF]
|
|
|
|
|
|
G. S. Johansson and H. J. Arnqvist Insulin and IGF-I action on insulin receptors, IGF-I receptors, and hybrid insulin/IGF-I receptors in vascular smooth muscle cells Am J Physiol Endocrinol Metab, November 1, 2006; 291(5): E1124 - E1130. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. R. Crouse III Thematic review series: Patient-Oriented Research. Imaging atherosclerosis: state of the art J. Lipid Res., August 1, 2006; 47(8): 1677 - 1699. |