Data are limited regarding prevalence and prognostic significance of subclinical cardiovascular disease (CVD) in individuals with metabolic syndrome (MetS). We investigated prevalence of subclinical CVD in 1,945 Framingham Offspring Study participants (mean age 58 years; 59% women) using electrocardiography, echocardiography, carotid ultrasound, ankle-brachial blood pressure, and urinary albumin excretion. We prospectively evaluated the incidence of CVD associated with MetS and diabetes according to presence versus absence of subclinical disease. Cross-sectionally, 51% of 581 participants with MetS had subclinical disease in at least one test, a frequency higher than individuals without MetS (multivariable-adjusted odds ratio 2.06 [95% CI 1.67–2.55]; P < 0.0001). On follow-up (mean 7.2 years), 139 individuals developed overt CVD, including 59 with MetS (10.2%). Overall, MetS was associated with increased CVD risk (multivariable-adjusted hazards ratio [HR] 1.61 [95% CI 1.12–2.33]). Participants with MetS and subclinical disease experienced increased risk of overt CVD (2.67 [1.62–4.41] compared with those without MetS, diabetes, or subclinical disease), whereas the association of MetS with CVD risk was attenuated in absence of subclinical disease (HR 1.59 [95% CI 0.87–2.90]). A similar attenuation of CVD risk in absence of subclinical disease was observed also for diabetes. Subclinical disease was a significant predictor of overt CVD in participants without MetS or diabetes (1.93 [1.15–3.24]). In our community-based sample, individuals with MetS have a high prevalence of subclinical atherosclerosis that likely contributes to the increased risk of overt CVD associated with the condition.
Presence of subclinical disease in multiple vascular beds has been suggested as an indicator of overall atherosclerotic burden (1). Consistent with this concept, investigators have reported an increased risk of overt cardiovascular disease (CVD) events in individuals with subclinical disease (2–4) or target organ damage (as evidenced by left ventricular hypertrophy [LVH] [5,6] or microalbuminuria [7–9]). It is also widely acknowledged that established risk factors for overt CVD promote the development of subclinical CVD (10).
In this context, the metabolic syndrome (MetS) is a condition that is associated with the clustering of risk factors including high blood pressure, abdominal obesity, glucose intolerance, and dyslipidemia. Whereas the definition and clinical utility of MetS has been the subject of recent debate (11,12), it is accepted that MetS carries an increased risk of CVD (13,14). Given the clustering of risk factors that characterizes MetS, it is likely that individuals with MetS have a high burden of subclinical disease (a term used herein to refer to both subclinical atherosclerosis and target organ damage). It is also likely that subclinical disease would contribute to the increased risk of CVD associated with MetS. Yet, data examining this premise comprehensively are lacking in the published literature. Of note, whereas several studies (15–25) have documented the increased prevalence of subclinical disease in MetS, they have not investigated the potential role of such disease in mediating the vascular risk associated with the condition. Additionally, previous studies (15–25) assessing subclinical disease in MetS have typically focused on single measures of subclinical disease.
Accordingly, we characterized comprehensively the cross-sectional prevalence of subclinical disease in individuals with prevalent MetS in the community. Additionally, we tested the hypothesis that the presence of subclinical disease contributes to the increased risk of overt CVD associated with MetS prospectively.
RESEARCH DESIGN AND METHODS
The design and selection criteria of the Framingham Offspring Study have been described previously (26). Participants who attended the sixth examination cycle (1995–1998) were eligible for the present study (n = 3,532). The participants underwent routine medical history, physical examination including blood pressure measurement, anthropometry, laboratory assessment of CVD risk factors, and testing for the presence of subclinical CVD (see section below). The institutional review board at Boston Medical Center approved the study, and all participants gave written informed consent.
Participants were excluded from the present investigation for the following reasons: prevalent CVD at baseline (n = 415), unavailable electrocardiography data (n = 6), unavailable measurement of urinary albumin (n = 460), unavailable ankle-brachial blood pressure data (n = 49), unavailable or inadequate carotid ultrasonography data (n = 70), and unavailable or inadequate echocardiographic left ventricular mass data (n = 587). After these exclusions, 1,945 individuals (mean age 58 years; 59% women) were eligible and constituted the study sample.
Definition of risk factors and MetS.
Cigarette smoking was defined by self-report of cigarette use within the year preceding the heart study baseline examination. Diabetes was defined as a fasting plasma glucose ≥126 mg/dl or use of insulin or oral hypoglycemic agents (27). The MetS was defined according to the modified National Cholesterol Education Program Adult Treatment Panel III criteria (28) by the presence of three or more of the following: increased waist circumference (≥102 cm for men, ≥88 cm for women), elevated blood pressure (≥130 mmHg systolic or ≥85 mmHg diastolic or treatment for hypertension), hyperglycemia (fasting blood glucose ≥100 mg/dl or treatment for elevated glucose), hypertriglyceridemia (≥150 mg/dl or treatment with nicotinic acid or fibrates), or low HDL cholesterol (<40 mg/dl in men, <50 mg/dl in women).
Subclinical disease measures and score.
Measures of subclinical vascular disease and target organ damage were chosen based on a review of the published literature. The five tests used to characterize the prevalence of subclinical disease are detailed in Table 1 and described briefly below. A standard 12-lead computerized resting electrocardiogram was obtained with the participants in a supine position. The sex-specific Cornell voltage criteria were used to assess the presence of electrocardiographic LVH (29). All participants underwent routine transthoracic echocardiographic examination. M-mode measurements of left ventricular dimensions were obtained by the leading-edge-to-leading-edge technique (30). Left ventricular ejection fraction was estimated by experienced observers based on the visual assessment of left ventricular contractile performance and wall motion in multiple two-dimensional views. Carotid ultrasound readings were acquired and images analyzed according to a standard protocol (31). Imaging was conducted using a high-resolution 7.5-MHz transducer for the common carotid artery and a 5.0-MHz transducer for the carotid bulb and internal carotid artery (Toshiba Medical Systems), as described previously (32). Carotid intima-media thickness (IMT) measurements were made from gated diastolic images of the left and right carotid artery at the level of the distal common carotid artery, the carotid artery bulb, and the proximal 2 cm of the internal carotid artery. The maximal IMT at each site was defined as the mean of the maximal IMT measured at the near and far walls of the vessel. The internal carotid artery IMT was defined as the mean of the maximal IMT measurements for the carotid artery bulb and the internal carotid artery on both the right and left side. Ankle-brachial systolic blood pressure measurements were obtained by trained technicians according to a standard protocol, using an 8-MHz Doppler pen probe and an ultrasonic Doppler flow detector (Parks Medical Electronics, Aloha, OR), as previously described (33). Microalbuminuria was assessed by estimating the urine albumin-to-creatinine ratio in a single urine sample. The urinary albumin concentration was determined by immunoturbidometry (Tina-Quant Albumin Assay; Roche Diagnostics), and the urinary creatinine concentration was measured with a modified Jaffe method. The urine albumin-to-creatinine ratio measured in a spot urine sample is highly correlated with 24-h urine albumin excretion (7).
Although several subclinical disease indicators have a continuous distribution of values, we dichotomized values using thresholds described in the literature that have been validated against outcome events in other studies. Such an approach permits characterization of the presence versus absence of subclinical disease (10). Thus, participants could have abnormalities signifying subclinical disease on any of the components of the five tests described in Table 1. For LVH and carotid ultrasound abnormalities, presence of any specific abnormality (of two to three eligible ones; see Table 1) was considered indicative of subclinical disease. Additionally, for each participant we constructed a subclinical disease score, ranging from 0 to 5, calculated from the number of tests (LVH by electrocardiogram or echocardiography, left ventricular systolic dysfunction by echocardiography, carotid ultrasound abnormality, peripheral arterial disease by ankle-brachial blood pressure index, and glomerular endothelial dysfunction by urinary albumin excretion rate) with evidence of subclinical disease. We weighted each of the five tests equally for the sake of simplicity, an approach paralleling that of the subclinical disease index formulated by Kuller et al. (10). This approach assumes that the hazards posed by abnormality of any of the five tests (or of subclinical disease measures defined by a test) are similar. We chose this analytical strategy to facilitate interpretation of results in a simple, yet meaningful, way.
Follow-up and outcome events.
All participants were under longitudinal surveillance for the occurrence of CVD events, through periodic examinations at the Framingham Heart Study and via biennial health history updates between examinations. Three experienced investigators obtained and reviewed hospitalization and physician office visit records for all suspected CVD events. A separate review committee that included a neurologist adjudicated cerebrovascular events, and a heart study neurologist independently examined most participants with suspected stroke.
The outcome in this study was the incidence of a first overt CVD event, defined as a composite of coronary heart disease (recognized or unrecognized myocardial infarction, angina pectoris, coronary insufficiency, or coronary heart disease death), cerebrovascular disease (stroke or transient ischemic attack), heart failure (by Framingham criteria), and intermittent claudication. Diagnosis criteria for these events have been described previously (34). Follow-up was from the sixth examination (baseline) up to 31 December 2005.
Statistical methods.
Participants were categorized into three mutually exclusive groups for analyses: those without either MetS or diabetes (the referent group), those with MetS but no diabetes, and those with diabetes. We evaluated individuals with diabetes as a comparison group because MetS is a risk factor for diabetes. Furthermore, prior research suggests that people with diabetes have a high prevalence of subclinical disease, which is a primary determinant of CVD risk in this group (35).
First, we characterized the prevalence of subclinical disease in each of the three groups. Because age is a major determinant of subclinical disease, we also assessed the sex-specific prevalence of subclinical disease in the three groups stratified by age (<60 vs. ≥60 years). Second, we performed multivariable logistic regression to assess the associations of MetS and diabetes with the prevalence of subclinical disease, adjusting for age and sex. Odds ratios (and their 95% CIs) for the presence of any subclinical disease (a score of one or more) and for the prevalence of each individual subclinical disease measure was estimated for participants with MetS and diabetes, with those who had neither condition serving as referent. Third, we evaluated the prognostic significance of subclinical disease in the three groups prospectively. Age- and sex-adjusted incidence rates of CVD were calculated for the three groups overall and stratifying each group by presence of subclinical disease.
We used Cox proportional hazards regression to assess the risk of CVD incidence in participants with MetS and in those with diabetes, with the group with neither condition serving as referent. We constructed two sets of models (age- and sex-adjusted and multivariable-adjusted [incorporating only age, sex, smoking, and LDL cholesterol to keep the models parsimonious]) and performed the following analyses hierarchically to investigate the role of subclinical disease in individuals with MetS or diabetes: A) without adjustment for subclinical vascular disease, B) with adjustment for subclinical disease score modeled as a dichotomous variable (subclinical disease present [score ≥1] versus absent [score = 0]), C) with adjustment for subclinical disease score as an ordinal variable, and D) stratified by presence versus absence of any subclinical vascular disease.
We also examined effect modification by testing the statistical significance of the following two-way interaction terms in the multivariable models: subclinical disease by age, subclinical disease by sex, subclinical disease by MetS, and subclinical disease by diabetes. To assess the incremental utility of subclinical disease for predicting CVD risk, we calculated the c-statistic for models A–C. Further, we created a subclinical disease score using “office-based” tests that included electrocardiography, ankle-brachial index, and microalbuminuria (rendering a score from 0 to 3) and defined presence versus absence of subclinical disease based on these tests. The association between this office-based measure of subclinical disease and overt CVD was examined in secondary analyses. Two-sided P values of <0.05 were considered statistically significant. All analyses were performed using SAS 9.1 (SAS Institute, Cary, NC).
RESULTS
The baseline clinical characteristics for the three groups of participants are shown in Table 2. The groups with diabetes or MetS at baseline were older, had a higher proportion of men, and had higher levels of blood pressure, fasting glucose, triglycerides, and BMI but lower HDL levels compared with the referent group without diabetes or MetS (Table 2).
Subclinical vascular disease in participants with prevalent MetS or diabetes.
The prevalence of subclinical disease in the three groups is shown in Table 2 (lower half). Individuals with MetS or diabetes had a higher prevalence of subclinical disease (Table 2). The prevalence of subclinical disease increased with age in both sexes in all three groups (Fig. 1). In age- and sex-adjusted logistic regression models, MetS and diabetes were both strongly and significantly associated with the presence of electrocardiographic and echocardiographic LVH, increased carotid IMT and stenosis, and microalbuminuria (Table 3). Diabetes was associated positively with left ventricular systolic dysfunction and higher prevalence of a low ankle-brachial index, whereas MetS was not significantly associated. Overall, MetS was associated with a twofold odds and diabetes with an over fourfold odds of having at least one subclinical disease abnormality compared with individuals without MetS or diabetes.
Prognostic significance of subclinical disease in participants with MetS or diabetes.
On follow-up (mean 7.2 years [range 0.1–9.0]), 139 of 1,945 individuals (7.1%; 46% women) developed a first overt CVD event, including 59 of 581 participants (10.2%) with MetS. The age- and sex-adjusted incidence of CVD rose across the three groups, with the highest rates in participants with diabetes (Table 4). Further, the presence of subclinical disease was consistently associated with an increased CVD event rate in all three groups (Table 4). Participants with MetS and subclinical disease had CVD incidence rates comparable with that for the group of individuals with diabetes (including both individuals with and without subclinical disease).
In both age- and sex-adjusted and in multivariable-adjusted analyses, MetS and diabetes were associated with an increased CVD risk in models without adjustment for subclinical disease (Table 5, model A). In multivariable models adjusting for presence of subclinical vascular disease, the association of MetS and diabetes with CVD risk was attenuated, becoming borderline statistically significant for MetS (Table 5, model B). The presence of subclinical disease was significantly associated with a twofold increased risk of CVD (compared with absence of subclinical disease) in these models (Table 5, model B). Each point increase in the subclinical disease score was associated with a 49% increased risk of CVD in multivariable-adjusted analyses (Table 5, model C). In these models, the associations of MetS and diabetes with CVD risk were further attenuated. Similarly, in models stratifying groups by presence versus absence of subclinical vascular disease (Table 5, model D), individuals with MetS and diabetes who had subclinical disease experienced a 2.7- and 4.0-fold increased CVD risk, respectively, compared with participants without subclinical disease, MetS, or diabetes (who served as referent). Further, presence of subclinical disease was a significant predictor of CVD in the group of participants without MetS or diabetes at baseline. In contrast, individuals with the presence of MetS or diabetes but without any subclinical vascular disease were not at a statistically significant increased risk of overt CVD compared with the referent group.
Additional analyses.
None of the interaction terms evaluated reached statistical significance (P > 0.60 for all), suggesting that the association of subclinical disease with incident overt CVD was not modified by age, sex, or the presence of MetS or diabetes.
In the whole study sample, the c-statistics for models that included MetS and diabetes as covariates, but not subclinical disease (model A), were 0.71 for the age- and sex-adjusted model and 0.73 for the multivariable model (adjusting also for LDL cholesterol and smoking). When subclinical disease was added as a dichotomous variable (model B), the c-statistic increased to 0.73 for the age- and sex-adjusted model and 0.74 for the multivariable model (P = 0.03 and 0.12, respectively, for comparisons with models without subclinical disease). When subclinical disease score was incorporated as an ordinal variable (model C), the c-statistic was 0.74 for both the age- and sex-adjusted and the multivariable models (P = 0.01 and 0.07, respectively, for comparisons with models without subclinical disease).
The c-statistics for models including subclinical disease score, but not MetS or diabetes as covariates, were 0.73 for the age- and sex-adjusted model and 0.74 for the multivariable model. When MetS and diabetes were added as covariates to the models, the c-statistic increased to 0.74 for the age- and sex-adjusted model and 0.75 for the multivariable model (P = 0.14 and 0.21, respectively, for comparisons with models without MetS and diabetes).
In secondary analyses, we assessed the associations between a subclinical disease score using office-based tests (electrocardiography, ankle-brachial index, and microalbuminuria) and overt CVD. The age- and sex-adjusted CVD rates by presence or absence of subclinical disease in individuals with or without MetS and diabetes (online appendix Table 1 [http://dx.doi.org/10.2337/db07-0078]), and the associations between subclinical disease and overt CVD (online appendix Table 2), were similar to those estimated using the more comprehensive subclinical disease definition. However, the associations between subclinical disease defined by office-based measures only and overt CVD were somewhat weaker than the associations using the original subclinical disease definition.
To examine a potential selection bias resulting from exclusion of participants with nonavailable data on subclinical disease measures, we compared the clinical features and risk of developing overt CVD in participants excluded from our investigation and those included in our sample. Participants who were excluded because of missing data on subclinical disease measures were older, more likely to be men, and had higher BMI and triglyceride levels (online appendix Table 3). However, the CVD incidence rates in those excluded due to missing data on subclinical disease (age- and sex-adjusted rate 8.0 [95% CI 6.4–9.6]) and in the “included” study sample (7.0 [5.8–8.2]) did not differ significantly (P = 0.28).
DISCUSSION
Principal findings.
Our principal findings are fourfold. First, in our community-based sample, over half of the individuals with MetS had subclinical disease upon comprehensive assessment using a panel of clinical tests that are routinely available. Prevalence increased markedly with age, with nearly two-thirds of people with MetS aged >60 years having evidence of subclinical disease. Second, individuals with MetS with evidence of subclinical disease experienced overt CVD incidence rates comparable with those with diabetes (considered a coronary heart disease risk equivalent) (36) and a risk nearly threefold that of participants without subclinical disease, MetS, or diabetes. Third, the presence of subclinical disease conferred approximately a twofold risk of overt CVD even in those without either MetS or diabetes (compared with their counterparts without subclinical disease). Fourth, adjustment for subclinical disease presence attenuated the association of MetS and diabetes with CVD risk. This observation suggests an important role of subclinical disease in mediating the vascular risks associated with MetS, extending prior work that established this premise for diabetes (35). The incorporation of subclinical disease measure in multivariable models resulted in a modest increase in the c-statistic.
Comparison with previous studies of subclinical disease in MetS.
As noted previously, several investigations have documented the greater burden of subclinical atherosclerosis in individuals with MetS (15–25,35). These prior studies were limited in their focus on specific measures or target organs, such as presence of increased left ventricular mass (24,25), left ventricular dysfunction (25), increased carotid IMT (15–20), or increased coronary artery calcium score (21–23). None used a panel of tests to comprehensively characterize the burden of subclinical disease associated with the condition. Data on the prognostic significance of subclinical disease in MetS are even more limited. One previous study (37) reported that presence of MetS predicts myocardial ischemia in subjects with a high coronary artery calcium score.
Investigators from the Cardiovascular Health Study have underscored the prognostic significance of subclinical disease in the elderly in a series of reports, although those reports did not focus on MetS (1–4,35). Indeed, Kuller et al. (10) formulated a subclinical disease index that combined measures of low ankle-brachial blood pressure index, carotid artery stenosis, increased carotid IMT, major electrocardiogram abnormalities, abnormal wall motion or ejection fraction on echocardiography, and positive responses to the Rose angina or claudication questionnaire without clinical evidence of angina or claudication. These investigators have established that the subclinical index is a powerful predictor of CVD in individuals without CVD (2,3) and in people with diabetes or impaired glucose tolerance (35). In the present study, we modified the subclinical disease index proposed by Kuller et al. (10) by adding two additional measures linked to CVD risk in numerous reports (i.e., echocardiographic LVH [5,6], which is more prevalent than its electrocardiographic counterpart, and microalbuminuria [7–9], an indicator of endothelial dysfunction and/or target organ damage). The cut points and definitions for the components of the subclinical disease score were developed a priori after careful review of the existing literature with the objective of formulating a score with high predictive utility. We also extend the observations reported by Kuller et al. (10) on the adverse prognostic implications of subclinical disease to the context of the MetS.
Strengths and limitations.
The strengths of this study are the moderate-to-large community-based sample, the continuous surveillance for CVD events blinded to subclinical disease status, and the routine assessment of a comprehensive set of tests of subclinical disease reflecting atherosclerotic burden and target organ damage across the cardiovascular system. Several limitations of our investigation should be noted. Our sample was middle-aged and predominantly white, limiting the generalizability of our findings to other ethnicities and age-groups. Moreover, we could not evaluate the role of coronary artery calcification as a marker of subclinical disease in MetS because such imaging was not performed at the sixth examination cycle. We have reported elsewhere the greater burden of coronary calcification in individuals with diabetes and pre-diabetes (38). Additionally, we evaluated a composite of CVD outcomes to maximize statistical power and did not analyze individual vascular outcomes. On a parallel note, we did not explore the relative prognostic significance of individual components of the subclinical disease score. We have noted earlier an inherent limitation of our statistical modeling that weighted the hazard associated with different subclinical disease measures equally for the sake of simplicity. Also, in the present study, we included microalbuminuria among the subclinical disease measures, even though it is part of the World Health Organization definition of the MetS (39). This was possible because we used the National Cholesterol Education Program definition of the MetS, the reason being that we consider microalbuminuria as a valuable marker of target organ damage (7–9). Further, to avoid overfitting of the multivariable models, we adjusted only for a smaller set of standard CVD risk factors, not including factors constituting the MetS, physical activity, medication like aspirin or β-blockers, or alcohol consumption. Last, exclusion of a substantial proportion of individuals due to nonavailable subclinical disease measures is an unavoidable limitation of large epidemiological studies, associated with the requirement for availability of data for all five tests of subclinical disease measures. However, this means that our results must be interpreted with caution and must be replicated in other samples.
Implications.
There has been considerable debate recently regarding the definition of the MetS and its clinical utility (11,12). Our data suggest that MetS (as defined by a cluster of risk factors) is associated with a high prevalence of subclinical disease. Our observations provide insights that the condition marks a group at high risk of overt CVD, possibly on the basis of a high burden of subclinical disease. In fact, individuals with MetS and subclinical disease experienced overt CVD incidence rates comparable with those with diabetes, which is considered a coronary heart disease risk equivalent (36). The finding of a high prevalence of subclinical disease even in younger individuals (aged <60 years) with MetS underscores the need for aggressive treatment of risk factors in young adulthood and the importance of primary prevention. This is important given that subclinical atherosclerosis is the primary underlying etiology of overt CVD. The higher CVD risk associated with the presence of subclinical disease even in individuals without MetS or diabetes suggests that lesser degrees of elevation of risk factors than that defined by MetS may be pathogenetically related to the development of subclinical atherosclerosis and its evolution to overt disease.
The presence of subclinical disease did not add substantially in terms of model discrimination. This shows that our findings are primarily interesting from a pathophysiological perspective. Our results do not imply that individuals with MetS or diabetes should undergo extensive screening for subclinical disease. The clinical utility of assessment of subclinical disease burden in these individuals still needs to be examined comprehensively in other settings before any changes to the current guidelines can be suggested (28). Also, it is important to assess other measures of test performance, such as model calibration or reclassification of CVD risk in additional studies.
Conclusions.
In our community-based cohort, individuals with MetS have a high prevalence of subclinical cardiovascular disease, which likely contributes to the increased risk of overt CVD associated with the condition.
Prevalence of any subclinical atherosclerosis or target organ damage in individuals without diabetes or MetS, those with MetS (but no diabetes), and participants with diabetes, by age.
Prevalence of any subclinical atherosclerosis or target organ damage in individuals without diabetes or MetS, those with MetS (but no diabetes), and participants with diabetes, by age.
Definitions of subclinical vascular disease
| Characteristic | Definition of subclinical disease component | Cut points for subclinical disease presence used in the present study |
|---|---|---|
| LV hypertrophy by electrocardiography or echocardiography | ||
| LV hypertrophy by Cornell criteria using electrocardiography | Sum of the R wave in a VL and the S wave in lead V3 exceeding 2.8 mV in men and 2.0 mV in women (29) | Presence of LV hypertrophy by Cornell criteria using electrocardiography or a height-adjusted LV mass using echocardiography that met or exceeded the sex-specific 80th percentile (40) |
| LV hypertrophy by echocardiography | LV mass was calculated as 0.8 (1.04 [IVS + LVEDD + PW]3 − [LVEDD]3) + 0.6 g (42). LV mass values were then adjusted for height using the ratio of LV mass to height. | |
| LV systolic dysfunction by echocardiography | ||
| LV systolic dysfunction | LV fractional shortening was calculated as (LVEDD − LVESD)/LVEDD (41) | A fractional shortening of <0.29 by M mode or by evidence on two- dimensional echocardiography of mild or greater systolic dysfunction on visual assessment in multiple views (corresponding to ejection fraction <50%) or by both criteria (30) |
| Carotid ultrasound abnormality | ||
| Increased carotid artery IMT | A composite measure that combined the maximal common carotid artery IMT and maximal internal carotid artery IMT was obtained by averaging these two measurements after standardization (subtraction of the mean and division by the SD for the measurement) (31) | A standardized carotid IMT that met or exceeded the sex-specific 80th percentiles in the sample (2), an extreme increase of common carotid IMT, or presence of carotid artery stenosis ≥25% |
| Extreme increase of common carotid artery IMT | An extreme increase of common carotid IMT ≥1 mm (43) | |
| Carotid artery stenosis ≥25% | Presence of a stenosis of ≥25% in the internal or common carotid artery (2) | |
| Peripheral arterial disease | ||
| Ankle-brachial index ≤0.9 | Defined as the ratio of the average systolic blood pressure at the ankle of each leg divided by the average systolic blood pressure in the arm with the highest blood pressure. | An ankle-brachial index at or below 0.9 in either leg (2,33) |
| Glomerular endothelial dysfunction | ||
| Microalbuminuria | An urine albumin-to-creatinine ratio ≥25 μg/mg in men and ≥35 μg/mg in women (7) | Presence of microalbuminuria according to the definition |
| Characteristic | Definition of subclinical disease component | Cut points for subclinical disease presence used in the present study |
|---|---|---|
| LV hypertrophy by electrocardiography or echocardiography | ||
| LV hypertrophy by Cornell criteria using electrocardiography | Sum of the R wave in a VL and the S wave in lead V3 exceeding 2.8 mV in men and 2.0 mV in women (29) | Presence of LV hypertrophy by Cornell criteria using electrocardiography or a height-adjusted LV mass using echocardiography that met or exceeded the sex-specific 80th percentile (40) |
| LV hypertrophy by echocardiography | LV mass was calculated as 0.8 (1.04 [IVS + LVEDD + PW]3 − [LVEDD]3) + 0.6 g (42). LV mass values were then adjusted for height using the ratio of LV mass to height. | |
| LV systolic dysfunction by echocardiography | ||
| LV systolic dysfunction | LV fractional shortening was calculated as (LVEDD − LVESD)/LVEDD (41) | A fractional shortening of <0.29 by M mode or by evidence on two- dimensional echocardiography of mild or greater systolic dysfunction on visual assessment in multiple views (corresponding to ejection fraction <50%) or by both criteria (30) |
| Carotid ultrasound abnormality | ||
| Increased carotid artery IMT | A composite measure that combined the maximal common carotid artery IMT and maximal internal carotid artery IMT was obtained by averaging these two measurements after standardization (subtraction of the mean and division by the SD for the measurement) (31) | A standardized carotid IMT that met or exceeded the sex-specific 80th percentiles in the sample (2), an extreme increase of common carotid IMT, or presence of carotid artery stenosis ≥25% |
| Extreme increase of common carotid artery IMT | An extreme increase of common carotid IMT ≥1 mm (43) | |
| Carotid artery stenosis ≥25% | Presence of a stenosis of ≥25% in the internal or common carotid artery (2) | |
| Peripheral arterial disease | ||
| Ankle-brachial index ≤0.9 | Defined as the ratio of the average systolic blood pressure at the ankle of each leg divided by the average systolic blood pressure in the arm with the highest blood pressure. | An ankle-brachial index at or below 0.9 in either leg (2,33) |
| Glomerular endothelial dysfunction | ||
| Microalbuminuria | An urine albumin-to-creatinine ratio ≥25 μg/mg in men and ≥35 μg/mg in women (7) | Presence of microalbuminuria according to the definition |
IVT, interventricular septum thickness; LV, left ventricular; LVEDD, left ventricular end-diastolic diameter; LVESD, left ventricular end-systolic diameter; PW, posterior left ventricular wall thickness.
Baseline characteristics of the study sample
| Characteristic | Referent* | MetS† | Diabetes |
|---|---|---|---|
| n | 1,249 | 581 | 115 |
| Clinical features‡ | |||
| Age (years) | 56 ± 9 | 60 ± 9 | 62 ± 8 |
| Women (%) | 61.7 | 53.7 | 48.7 |
| Systolic blood pressure (mmHg) | 122 ± 17 | 135 ± 18 | 137 ± 19 |
| Diastolic blood pressure (mmHg) | 74 ± 9 | 79 ± 9 | 75 ± 10 |
| High blood pressure (%)§ | 33.2 | 79.5 | 78.3 |
| Total cholesterol (mg/dl) | 204 ± 37 | 214 ± 39 | 201 ± 44 |
| LDL cholesterol (mg/dl) | 125 ± 34 | 133 ± 33 | 120 ± 36 |
| HDL cholesterol (mg/dl) | 58 ± 15 | 44 ± 13 | 45 ± 15 |
| Low HDL (%)‖ | 15.3 | 65.2 | 55.7 |
| Triglycerides (mg/dl) | 103 ± 50 | 187 ± 101 | 190 ± 173 |
| High triglycerides (%)¶ | 11.5 | 68.0 | 55.3 |
| Fasting blood glucose (mg/dl) | 93 ± 8 | 102 ± 10 | 165 ± 53 |
| Impaired fasting glucose (%)# | 17.8 | 60.4 | NA |
| BMI (kg/m2) | 25.8 ± 3.8 | 29.7 ± 4.6 | 29.4 ± 5.2 |
| BMI ≥30 kg/m2 (%) | 12.4 | 39.9 | 41.7 |
| Increased waist circumference (%)** | 33.7 | 80.3 | 74.6 |
| Current smokers (%) | 14.4 | 14.1 | 13.0 |
| Subclinical disease†† | |||
| LV hypertrophy by electrocardiography or echocardiography | |||
| LV hypertrophy by Cornell criteria (%) | 0.64 | 2.8 | 4.4 |
| LV mass-to-height ratio (g/m men/women) | 105 ± 21/83 ± 16 | 115 ± 23/95 ± 21 | 121 ± 32/98 ± 20 |
| LV hypertrophy by echocardiography (%) | 10.7 | 25.1 | 34.8 |
| LV systolic dysfunction by echocardiography | |||
| Fractional shortening | 0.38 ± 0.05 | 0.37 ± 0.05 | 0.37 ± 0.06 |
| LV systolic dysfunction (%) | 2.7 | 3.1 | 7.0 |
| Carotid ultrasound abnormality | |||
| Carotid artery IMT (standardized; men/women) | −0.05 ± 0.26/−0.14 ± 0.22 | 0.05 ± 0.29/−0.04 ± 0.25 | 0.21 ± 0.35/0.03 ± 0.29 |
| Increased carotid artery IMT (%) | 12.0 | 24.1 | 40..9 |
| Extreme increase of common carotid artery IMT (%) | 1.8 | 6.7 | 14.8 |
| Carotid artery stenosis ≥25% (%) | 9.2 | 19.8 | 32.2 |
| Peripheral arterial disease | |||
| Ankle-brachial index | 1.13 ± 0.09 | 1.13 ± 0.10 | 1.13 ± 0.13 |
| Ankle-brachial index ≤0.9 (%) | 1.0 | 1.9 | 4.4 |
| Glomerular endothelial dysfunction | |||
| Microalbuminuria (%) | 6.2 | 10.8 | 24.4 |
| Composite of subclinical disease measures | |||
| At least one (%) | 29.8 | 50.8 | 70.4 |
| At least two (%) | 5.4 | 17.6 | 36.5 |
| Three or more (%) | 0.9 | 3.4 | 11.3 |
| Mean score | 0.36 ± 0.62 | 0.72 ± 0.85 | 1.20 ± 1.03 |
| Characteristic | Referent* | MetS† | Diabetes |
|---|---|---|---|
| n | 1,249 | 581 | 115 |
| Clinical features‡ | |||
| Age (years) | 56 ± 9 | 60 ± 9 | 62 ± 8 |
| Women (%) | 61.7 | 53.7 | 48.7 |
| Systolic blood pressure (mmHg) | 122 ± 17 | 135 ± 18 | 137 ± 19 |
| Diastolic blood pressure (mmHg) | 74 ± 9 | 79 ± 9 | 75 ± 10 |
| High blood pressure (%)§ | 33.2 | 79.5 | 78.3 |
| Total cholesterol (mg/dl) | 204 ± 37 | 214 ± 39 | 201 ± 44 |
| LDL cholesterol (mg/dl) | 125 ± 34 | 133 ± 33 | 120 ± 36 |
| HDL cholesterol (mg/dl) | 58 ± 15 | 44 ± 13 | 45 ± 15 |
| Low HDL (%)‖ | 15.3 | 65.2 | 55.7 |
| Triglycerides (mg/dl) | 103 ± 50 | 187 ± 101 | 190 ± 173 |
| High triglycerides (%)¶ | 11.5 | 68.0 | 55.3 |
| Fasting blood glucose (mg/dl) | 93 ± 8 | 102 ± 10 | 165 ± 53 |
| Impaired fasting glucose (%)# | 17.8 | 60.4 | NA |
| BMI (kg/m2) | 25.8 ± 3.8 | 29.7 ± 4.6 | 29.4 ± 5.2 |
| BMI ≥30 kg/m2 (%) | 12.4 | 39.9 | 41.7 |
| Increased waist circumference (%)** | 33.7 | 80.3 | 74.6 |
| Current smokers (%) | 14.4 | 14.1 | 13.0 |
| Subclinical disease†† | |||
| LV hypertrophy by electrocardiography or echocardiography | |||
| LV hypertrophy by Cornell criteria (%) | 0.64 | 2.8 | 4.4 |
| LV mass-to-height ratio (g/m men/women) | 105 ± 21/83 ± 16 | 115 ± 23/95 ± 21 | 121 ± 32/98 ± 20 |
| LV hypertrophy by echocardiography (%) | 10.7 | 25.1 | 34.8 |
| LV systolic dysfunction by echocardiography | |||
| Fractional shortening | 0.38 ± 0.05 | 0.37 ± 0.05 | 0.37 ± 0.06 |
| LV systolic dysfunction (%) | 2.7 | 3.1 | 7.0 |
| Carotid ultrasound abnormality | |||
| Carotid artery IMT (standardized; men/women) | −0.05 ± 0.26/−0.14 ± 0.22 | 0.05 ± 0.29/−0.04 ± 0.25 | 0.21 ± 0.35/0.03 ± 0.29 |
| Increased carotid artery IMT (%) | 12.0 | 24.1 | 40..9 |
| Extreme increase of common carotid artery IMT (%) | 1.8 | 6.7 | 14.8 |
| Carotid artery stenosis ≥25% (%) | 9.2 | 19.8 | 32.2 |
| Peripheral arterial disease | |||
| Ankle-brachial index | 1.13 ± 0.09 | 1.13 ± 0.10 | 1.13 ± 0.13 |
| Ankle-brachial index ≤0.9 (%) | 1.0 | 1.9 | 4.4 |
| Glomerular endothelial dysfunction | |||
| Microalbuminuria (%) | 6.2 | 10.8 | 24.4 |
| Composite of subclinical disease measures | |||
| At least one (%) | 29.8 | 50.8 | 70.4 |
| At least two (%) | 5.4 | 17.6 | 36.5 |
| Three or more (%) | 0.9 | 3.4 | 11.3 |
| Mean score | 0.36 ± 0.62 | 0.72 ± 0.85 | 1.20 ± 1.03 |
Data are means ± SD or percentages.
No MetS or diabetes.
No diabetes.
P <0.0001 for all comparisons of clinical features between the group with MetS and the reference group and between the group with diabetes and reference group in age- and sex-adjusted linear (for means) or logistic (for proportions) regression models.
Systolic blood pressure ≥130 mmHg, diastolic blood pressure ≥85 mmHg, or antihypertensive medication use.
<40 mg/dl in men, <50 mg/dl in women.
>150 mg/dl or lipid-lowering medication use.
≥100 mg/dl.
≥102 cm (40 inches) in men, 88 cm (35 inches) in women.
Definitions of subclinical disease (see Table 1). LV, left ventricular; NA, not applicable.
Odds of subclinical vascular disease in participants with prevalent MetS and diabetes
| Characteristic* | Referent (n = 1,249)† | MetS (n = 581)‡ | Diabetes (n = 115) | ||||
|---|---|---|---|---|---|---|---|
| Odds ratio (95% CI) | P | Odds ratio (95% CI) | P | ||||
| LV hypertrophy by electrocardiography or echocardiography | |||||||
| LV hypertrophy by Cornell criteria | Referent | 3.79 (1.59–9.02) | 0.003 | 5.91 (1.86–18.84) | 0.003 | ||
| LV hypertrophy by echocardiography | Referent | 2.54 (1.95–3.32) | <0.0001 | 3.80 (2.46–5.86) | <0.0001 | ||
| LV systolic dysfunction by echocardiography | |||||||
| LV systolic dysfunction | Referent | 1.08 (0.60–1.96) | 0.79 | 2.47 (1.09–5.60) | 0.03 | ||
| Carotid ultrasound abnormality | |||||||
| Increased carotid artery IMT | Referent | 1.86 (1.42–2.45) | <0.0001 | 3.75 (2.41–5.82) | <0.0001 | ||
| Extreme increase of common carotid artery IMT | Referent | 2.78 (1.62–4.78) | 0.0002 | 5.60 (2.79–11.22) | <0.0001 | ||
| Carotid artery stenosis ≥25% | Referent | 1.89 (1.41–2.53) | <0.0001 | 3.13 (1.97–4.97) | <0.0001 | ||
| Peripheral arterial disease | |||||||
| Ankle-brachial index ≤0.9 | Referent | 1.36 (0.60–3.10) | 0.47 | 2.93 (0.99–8.64) | 0.05 | ||
| Glomerular endothelial dysfunction | |||||||
| Microalbuminuria | Referent | 1.64 (1.15–2.34) | 0.007 | 4.16 (2.52–6.84) | <0.0001 | ||
| Composite of subclinical disease measures | |||||||
| At least one | Referent | 2.06 (1.67–2.55) | <0.0001 | 4.33 (2.81–6.68) | <0.0001 | ||
| Characteristic* | Referent (n = 1,249)† | MetS (n = 581)‡ | Diabetes (n = 115) | ||||
|---|---|---|---|---|---|---|---|
| Odds ratio (95% CI) | P | Odds ratio (95% CI) | P | ||||
| LV hypertrophy by electrocardiography or echocardiography | |||||||
| LV hypertrophy by Cornell criteria | Referent | 3.79 (1.59–9.02) | 0.003 | 5.91 (1.86–18.84) | 0.003 | ||
| LV hypertrophy by echocardiography | Referent | 2.54 (1.95–3.32) | <0.0001 | 3.80 (2.46–5.86) | <0.0001 | ||
| LV systolic dysfunction by echocardiography | |||||||
| LV systolic dysfunction | Referent | 1.08 (0.60–1.96) | 0.79 | 2.47 (1.09–5.60) | 0.03 | ||
| Carotid ultrasound abnormality | |||||||
| Increased carotid artery IMT | Referent | 1.86 (1.42–2.45) | <0.0001 | 3.75 (2.41–5.82) | <0.0001 | ||
| Extreme increase of common carotid artery IMT | Referent | 2.78 (1.62–4.78) | 0.0002 | 5.60 (2.79–11.22) | <0.0001 | ||
| Carotid artery stenosis ≥25% | Referent | 1.89 (1.41–2.53) | <0.0001 | 3.13 (1.97–4.97) | <0.0001 | ||
| Peripheral arterial disease | |||||||
| Ankle-brachial index ≤0.9 | Referent | 1.36 (0.60–3.10) | 0.47 | 2.93 (0.99–8.64) | 0.05 | ||
| Glomerular endothelial dysfunction | |||||||
| Microalbuminuria | Referent | 1.64 (1.15–2.34) | 0.007 | 4.16 (2.52–6.84) | <0.0001 | ||
| Composite of subclinical disease measures | |||||||
| At least one | Referent | 2.06 (1.67–2.55) | <0.0001 | 4.33 (2.81–6.68) | <0.0001 | ||
Data are age- and sex-adjusted odds ratio (95% CI) of subclinical disease.
For definitions of subclinical disease characteristics, see Table 1.
No MetS or diabetes.
No diabetes. LV, left ventricular.
Incidence of CVD during 8 years of follow-up
| Characteristic | Number of events/number at risk | Person-years at risk | Age- and sex-adjusted rate |
|---|---|---|---|
| Referent* | |||
| All | 61/1,249 | 9,138 | 5.90 (3.69–8.03) |
| No subclinical disease | 29/877 | 6,506 | 4.38 (2.32–6.38) |
| Any subclinical disease present | 32/372 | 2,632 | 8.30 (4.62–11.78) |
| MetS† | |||
| All | 59/581 | 4,113 | 9.82 (6.17–13.26) |
| No subclinical disease | 17/286 | 2,085 | 6.91 (3.02–10.60) |
| Any subclinical disease present | 42/295 | 2,028 | 12.33 (7.27–17.05) |
| Diabetes | |||
| All | 19/115 | 767 | 13.74 (6.65–20.10) |
| No subclinical disease | 2/34 | 241 | 7.52 (0–16.95) |
| Any subclinical disease present | 17/81 | 525 | 16.32 (7.45–24.16) |
| Characteristic | Number of events/number at risk | Person-years at risk | Age- and sex-adjusted rate |
|---|---|---|---|
| Referent* | |||
| All | 61/1,249 | 9,138 | 5.90 (3.69–8.03) |
| No subclinical disease | 29/877 | 6,506 | 4.38 (2.32–6.38) |
| Any subclinical disease present | 32/372 | 2,632 | 8.30 (4.62–11.78) |
| MetS† | |||
| All | 59/581 | 4,113 | 9.82 (6.17–13.26) |
| No subclinical disease | 17/286 | 2,085 | 6.91 (3.02–10.60) |
| Any subclinical disease present | 42/295 | 2,028 | 12.33 (7.27–17.05) |
| Diabetes | |||
| All | 19/115 | 767 | 13.74 (6.65–20.10) |
| No subclinical disease | 2/34 | 241 | 7.52 (0–16.95) |
| Any subclinical disease present | 17/81 | 525 | 16.32 (7.45–24.16) |
Data are percent (95% CI).
No MetS or diabetes.
No diabetes.
Presence of the MetS, diabetes, and subclinical vascular disease and risk of CVD
| Age- and sex-adjusted models | Multivariable-adjusted models* | |||||
|---|---|---|---|---|---|---|
| Hazard ratio (95% CI) | P | Hazard ratio (95% CI) | P | |||
| Model A: CVD risks associated with MetS and diabetes, not adjusting for subclinical disease | ||||||
| Referent group† | Referent | Referent | ||||
| MetS‡ | 1.70 (1.19–2.45) | 0.004 | 1.61 (1.12–2.33) | 0.01 | ||
| Diabetes | 2.52 (1.49–4.26) | 0.0005 | 2.37 (1.55–3.60) | 0.003 | ||
| Model B: CVD risks associated with MetS and diabetes, adjusting for subclinical disease as a dichotomous variable (subclinical disease in at least 1 territory) | ||||||
| Referent group† | Referent | Referent | ||||
| Metabolic syndrome‡ | 1.51 (1.05–2.19) | 0.03 | 1.44 (1.00–2.10) | 0.05 | ||
| Diabetes | 2.01 (1.17–3.43) | 0.01 | 1.93 (1.09–3.42) | 0.02 | ||
| Subclinical disease present | 2.09 (1.43–3.06) | <0.0001 | 1.90 (1.29–2.79) | 0.0012 | ||
| Model C: CVD risks associated with MetS and diabetes, adjusting for subclinical disease as a ordinal variable (sum of affected territories) | ||||||
| Referent group† | Referent | Referent | ||||
| Metabolic syndrome‡ | 1.45 (1.00–2.10) | 0.05 | 1.40 (0.96–2.04) | 0.08 | ||
| Diabetes | 1.78 (1.03–3.06) | 0.04 | 1.74 (0.98–3.12) | 0.06 | ||
| Subclinical disease sum | 1.60 (1.34–1.92) | <0.0001 | 1.49 (1.23–1.80) | <0.0001 | ||
| Model D: CVD risks associated with MetS and diabetes, by presence versus absence of subclinical disease | ||||||
| Referent group (subclinical disease absent)† | Referent | Referent | ||||
| Referent group (subclinical disease present)† | 2.10 (1.25–3.51) | 0.002 | 1.93 (1.15–3.24) | 0.01 | ||
| MetS (subclinical disease absent)‡ | 1.57 (0.86–2.86) | 0.14 | 1.59 (0.87–2.90) | 0.13 | ||
| MetS (subclinical disease present)‡ | 3.13 (1.91–5.14) | <0.0001 | 2.67 (1.62–4.41) | <0.0001 | ||
| Diabetes (subclinical disease absent) | 1.63 (0.39–6.82) | 0.51 | 0.91 (0.12–6.71) | 0.93 | ||
| Diabetes (subclinical disease present) | 4.33 (2.32–8.07) | <0.0001 | 4.01 (2.09–7.66) | <0.0001 | ||
| Age- and sex-adjusted models | Multivariable-adjusted models* | |||||
|---|---|---|---|---|---|---|
| Hazard ratio (95% CI) | P | Hazard ratio (95% CI) | P | |||
| Model A: CVD risks associated with MetS and diabetes, not adjusting for subclinical disease | ||||||
| Referent group† | Referent | Referent | ||||
| MetS‡ | 1.70 (1.19–2.45) | 0.004 | 1.61 (1.12–2.33) | 0.01 | ||
| Diabetes | 2.52 (1.49–4.26) | 0.0005 | 2.37 (1.55–3.60) | 0.003 | ||
| Model B: CVD risks associated with MetS and diabetes, adjusting for subclinical disease as a dichotomous variable (subclinical disease in at least 1 territory) | ||||||
| Referent group† | Referent | Referent | ||||
| Metabolic syndrome‡ | 1.51 (1.05–2.19) | 0.03 | 1.44 (1.00–2.10) | 0.05 | ||
| Diabetes | 2.01 (1.17–3.43) | 0.01 | 1.93 (1.09–3.42) | 0.02 | ||
| Subclinical disease present | 2.09 (1.43–3.06) | <0.0001 | 1.90 (1.29–2.79) | 0.0012 | ||
| Model C: CVD risks associated with MetS and diabetes, adjusting for subclinical disease as a ordinal variable (sum of affected territories) | ||||||
| Referent group† | Referent | Referent | ||||
| Metabolic syndrome‡ | 1.45 (1.00–2.10) | 0.05 | 1.40 (0.96–2.04) | 0.08 | ||
| Diabetes | 1.78 (1.03–3.06) | 0.04 | 1.74 (0.98–3.12) | 0.06 | ||
| Subclinical disease sum | 1.60 (1.34–1.92) | <0.0001 | 1.49 (1.23–1.80) | <0.0001 | ||
| Model D: CVD risks associated with MetS and diabetes, by presence versus absence of subclinical disease | ||||||
| Referent group (subclinical disease absent)† | Referent | Referent | ||||
| Referent group (subclinical disease present)† | 2.10 (1.25–3.51) | 0.002 | 1.93 (1.15–3.24) | 0.01 | ||
| MetS (subclinical disease absent)‡ | 1.57 (0.86–2.86) | 0.14 | 1.59 (0.87–2.90) | 0.13 | ||
| MetS (subclinical disease present)‡ | 3.13 (1.91–5.14) | <0.0001 | 2.67 (1.62–4.41) | <0.0001 | ||
| Diabetes (subclinical disease absent) | 1.63 (0.39–6.82) | 0.51 | 0.91 (0.12–6.71) | 0.93 | ||
| Diabetes (subclinical disease present) | 4.33 (2.32–8.07) | <0.0001 | 4.01 (2.09–7.66) | <0.0001 | ||
Values are Cox proportional hazard ratio (95% CI).
Adjusted for age, sex, LDL, and smoking, in addition to MetS, diabetes, and subclinical vascular disease.
No MetS or diabetes.
No diabetes.
Published ahead of print at http://diabetes.diabetesjournals.org on 30 March 2007. DOI: 10.2337/db07-0078.
Additional information for this article can be found in an online appendix at http://dx.doi.org/10.2337/db07-0078.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Article Information
This work was supported through the Swedish Heart-Lung Foundation and the Swedish Society of Medicine (to E.I.); the National Institutes of Health/National Heart, Lung, and Blood Institute (NHLBI) (contract no. N01-HC-25195, 1R01HL080124, and 2K24HL04334 to R.S.V.); and a career development award from the American Diabetes Association (to J.B.M.).
The NHLBI had no role in the study design, analyses, or drafting of the manuscript. The NHLBI reviews all manuscripts submitted for publication, but it was not involved in the decision to publish.
The study was also supported by donation of microalbuminuria assay reagents from Roche Diagnostics.
Parts of this article was presented as an oral presentation at the American Heart Association's 47th Annual Conference on Cardiovascular Disease Epidemiology and Prevention in association with the Council on Nutrition, Physical Activity, and Metabolism in Orlando, Florida, 28 February to 3 March 2007. The abstract from the conference has been published in a supplement of Circulation.
