Disclosure: The authors declared no conflict of interest.
Article first published online: 12 MAR 2013
Copyright © 2012 The Obesity Society
Volume 21, Issue 1, pages E118–E123, January 2013
How to Cite
Bose, S., Krishnamoorthy, P., Varanasi, A., Nair, J., Schutta, M., Braunstein, S., Iqbal, N., Schwartz, S., Clair, C. St., Master, S. R., Rader, D. J., Reilly, M. P. and Mehta, N. N. (2013), Measurement of waist circumference predicts coronary atherosclerosis beyond plasma adipokines. Obesity, 21: E118–E123. doi: 10.1002/oby.20086
Funding agencies: Dr Mehta is supported by K23HL097151-01 and is a recipient of the American College of Cardiology Young Investigator Award for the Metabolic Syndrome and the National Psoriasis Foundation Award for Young Investigators.
- Issue published online: 12 MAR 2013
- Article first published online: 12 MAR 2013
- Accepted manuscript online: 18 OCT 2012 12:59PM EST
- Manuscript Accepted: 16 JUL 2012
- Manuscript Received: 13 OCT 2011
The association of plasma adipokines beyond waist circumference (WC) with coronary artery calcification (CAC), a measure of subclinical atherosclerosis, is unknown.
Design and Methods:
Asymptomatic Caucasian individuals from two community-based cross-sectional studies (n = 1,285) were examined and multivariate analysis of traditional risk factors was performed, then WC and adipokines (adiponectin and leptin) were added. Incremental value of each was tested with likelihood ratio testing.
Beyond traditional risk factors, WC (Tobit regression ratio 1.69, P < 0.001) and plasma leptin (1.57, P < 0.001) but not plasma adiponectin (P = 0.75) were independently associated with CAC. In nested models, neither adiponectin (χ2 = 0.76, P = 0.38) nor leptin (χ2 = 1.32, P = 0.25) added value to WC beyond traditional risk factors, whereas WC added incremental value to adiponectin (χ2 = 28.02, P < 0.0001) and leptin (χ2 = 13.58, P = 0.0002).
In the face of important biomarkers such as plasma adiponectin and leptin, WC remained a significant predictor of CAC beyond traditional risk factors underscoring the importance of WC measurement during cardiovascular risk assessment.
The prevalence of obesity is increasing rapidly in most industrialized countries. Obesity is associated with increased risk of atherosclerosis [1, 2] and is associated with type 2 diabetes (DM) , a potent risk factor for cardiovascular disease (CVD) . Coronary artery calcification (CAC) score has been widely accepted as a strong non-invasive surrogate marker of future cardiovascular event [5, 6]. Although addressing the treatable traditional cardiac risk factors (hypertension, hypercholesterolemia, smoking, and diabetes) improves cardiovascular outcomes in patients, residual risk remains in many patients. Various biomarkers have been evaluated to address the residual risk associated with CVD after treating traditional risk factors. Among these, plasma leptin, adiponectin, apoB , and insulin resistance estimation  have been studied in relationship with markers of asymptomatic atherosclerosis such as CAC and intimal medial thickness (IMT) , and indeed demonstrate strong association. However, these are not extensively available in most clinical settings.
Obesity may account for a portion of the increased risk of CVD, and is captured by easy office-based measurements such as BMI and waist circumference (WC). WC is thought to be better than BMI in discriminating elevated risk of CVD [10, 11]. Indeed, WC is independently associated with atherosclerosis and cardiovascular events beyond traditional risk factors (smoking, hypertension, diabetes, hyperlipidemia, and family history of CVD)  and therefore may be a key metric when evaluating cardiovascular risk. Furthermore, high WC often is a result of large central deposition of adipose tissue that secretes adipokines, bioactive peptides with endocrine, and paracrine activity, which have been associated with atherosclerosis . Although adiponectin and leptin are associated with cardiovascular risk factors [14, 15], CAC , and CVD [16-18], how these adipokines are related to CVD in the presence of WC remains poorly understood. In prior studies by our research group, we have demonstrated that plasma leptin levels were associated with CAC in DM  and non-DM , after adjusting for BMI and CRP. However, distinct from these two studies, in this article, we present data from a larger sample of both diabetic and nondiabetic participants and analyze waist as a potential independent predictor of CAC with relation to adipokines because how these plasma markers of adiposity interplay with simple office-based anthropometric markers of adiposity (i.e., WC or BMI) has not been assessed. Therefore, the objective of this study was to examine the relationship of WC and adipokines with subclinical atherosclerosis measured by CAC, and to examine whether adipokines add information beyond WC in predicting CAC. We hypothesized that adipokines would not add value beyond WC in predicting CAC.
Our study participants are composed of unrelated individuals from Penn Diabetes Heart Study (PDHS) and the Study of Inherited Risk of Coronary Atherosclerosis (SIRCA). The details of PDHS [14, 18] and SIRCA [13, 15] have been published. In brief, both are contemporary, single center, cross-sectional, community-based studies of subjects without clinical evidence of coronary heart disease, recruited at the University of Pennsylvania. The studies also utilized the same clinical research center, research staff, and electron beam CT scanner. Exclusion criteria in both studies were the presence of clinical coronary heart disease (defined as myocardial infarction, coronary revascularization, angiographic disease, or positive stress test), elevated creatinine (>1.2 mg/dl), and in SIRCA, the presence of diabetes. Our study focused on unrelated Caucasian subjects (total N = 1,285: DM = 536; non-DM = 749).
Participants were evaluated at the General Clinical Research Center at the University of Pennsylvania Medical Center after a 12-hour overnight fast . A questionnaire regarding medical, family, social history, medication use, and exercise was completed and verified by a study physician. Hypertension was defined as taking antihypertensive medications or blood pressure higher than 130/80 mmHg. Height, weight, WC at the level of the umbilicus and resting bilateral systolic and diastolic blood pressure were performed. BMI was calculated using the standard formula [BMI = (body weight in kilogram)/(height in meters)2]. Exercise was defined as none, intermittent (<2 times/week), or regular (>4 times/week), and alcohol use was defined as yes or no. Plasma levels of lipids were measured enzymatically (Cobas Fara II; Roche Diagnostic Systems, Somervile, NJ) in lipoprotein fractions after ultracentrifugation (β-quantification technique) in PDHS, and in whole plasma in SIRCA. Plasma levels of adiponectin and leptin were measured ((Linco, St. Charles, MO) using enzyme-linked immunosorbent assays. The C-reactive protein (CRP; high sensitivity) levels were assayed using immunoturbidimetry. The intra-assay and interassay coefficients of variance for pooled human plasma were 5.7% and 9.9% for adiponectin; 5.5% and 12.4% for leptin; 4.1% and 11.6% for insulin; and 8.0% and 8.3% for CRP, respectively. Framingham risk scores (FRS) were calculated as described by Wilson et al. . Participants were classified as having metabolic syndrome using the National Cholesterol Education Program (NCEP) definition . The homeostasis model assessment of insulin resistance (HOMA-IR) index = fasting glucose (mmol/l) × fasting insulin (μU/ml)/22.5  was used as a measure of insulin resistance. Global Agatston CAC scores , measured using electron beam tomography, were determined from 40 continuous 3-mm-thick computed tomograms collected on an EBT scanner (Imatron, San Francisco, CA). Scoring was performed by a single experienced radiological technologist, blinded to clinical and laboratory characteristics, using customized software (Imatron) .
Summaries of continuous variables are reported as mean ± SD for normally distributed data and median with interquartile range for skewed data. Categorical variables are reported as proportions and percentages. The crude association of WC, BMI, plasma adiponectin, and leptin with serum lipids, metabolic, inflammatory, and other CAC risk factors and parameters were examined using Spearman correlation. Multivariable analysis of CAC scores was performed using Tobit conditional regression of natural log (CAC+1) because of the distribution of CAC data (many zero scores with a marked right skew) . Tobit regression models the dichotomous outcome of zero versus nonzero and then assumes normality conditional on the presence of nonzero score data. The Tobit model is designed to assess the relationship between explanatory variables and a censored dependent variable at one end, at which many observations are clustered. We chose this modeling because CAC scores are censored at zero and the use of ordinary least-squares regression on such a non-normal distribution would produce biased estimates and invalid inference. Tobit modeling has otherwise similar assumptions about error distributions as the linear regression model . The Tobit regression model was designed using the Tobit command in STATA with the ll(0) option to indicate left censoring at a CAC score of 0.
Coefficients from Tobit regression analysis for WC predicting CAC were multiplied with SD of WC (6.33), before deriving the Tobit ratios and 95% confidence intervals. The association was tested for a 1 SD change in WC to reflect implication of significant change of WC on prediction of CAC. In all regressions, plasma hsCRP, adiponectin, and leptin were logarithmically transformed when included as continuous variables given non-Gaussian distributions. The association of WC (1 SD change), log-transformed plasma adiponectin, and leptin with CAC was assessed in Tobit models with confounding factors: age, gender, medications, hypertension (defined as SBP>140 or DBP > 90 mmHg or use of antihypertensive therapy), hypercholesterolemia (defined as serum total cholesterol > 5.18 mmol/L or use of dyslipidemia therapy), family history of CVD, presence of DM, cigarette smoking, alcohol use, exercise, and plasma hsCRP. We stratified the Tobit models by gender and by DM status. We have also performed interaction analyses for all three main variables of interest (i.e., waist, leptin, and adiponectin). Because we observed similar patterns in stratified analyses and did not observe any interactions between gender and independent variables defined a priori as P < 0.1 (Supporting Information Table 2B), we present results for the entire sample. Finally, we applied likelihood ratio tests (LRT) in nested models to assess the value of each parameter (WC, plasma adiponectin, and leptin) relative to each other in predicting CAC. Statistical analyses were performed using STATA 10.0 software (STATA Corp, College Station, TX).
Characteristics of study sample
Table 1 summarizes study sample characteristics in all subjects and stratified by DM status. Participants with DM were older, more likely to be male, and had greater prevalence of metabolic syndrome. High-density lipoprotein (HDL)-C levels were lower in DM versus non-DM subjects. Total and low-density lipoprotein (LDL)-C were lower in those with DM most likely because of greater statin use in the diabetic sample. BMI, WC, and serum hs-CRP level were all higher in those with DM. As expected, HOMA-IR values were higher in DM (all subjects on insulin were excluded from our study). Plasma adiponectin level was lower and leptin level higher in DM subjects as expected. Consistent with greater CVD risk, FRS and CAC scores were greater in DM compared with that in non-DM subjects.
|Variable||Total sample (N = 1,285)||DM group (N = 536)||Non-DM group (N = 749)||DM versus non-DM group (P-value)|
|Age, y||53 (46-61)||60 (54-68)||48 (42-54)||<0.0001|
|Male, %||793 (61.7)||396 (73.9)||397 (53.0)||<0.0001|
|CVD family history, %||603 (46.9)||126 (23.5)||477 (63.7%)||<0.0001|
|Total cholesterol, mg/dL||192 (166-220)||176 (153-201)||205 (177-229)||<0.0001|
|HDL cholesterol, mg/dL||46 (38-56)||44 (37-52)||48 (39-59)||<0.0001|
|Triglycerides, mg/dL||126 (90-179)||97 (138-204)||116 (87-158)||<0.0001|
|LDL cholesterol, mg/dL||115 (92-138.8)||99 (80-120)||126.8 (104.2-148.4)||<0.0001|
|Glucose, mg/dL||98 (89-116)||120 (100-145)||93 (86-100)||<0.0001|
|Insulin, IU/mL||8.8 (5.3-14.4)||14.4 (10.1-19.9)||6.3 (4.1-9.1)||<0.0001|
|HOMA-IR||2.2 (1.2-4.0)||4.2 (2.8-6.4)||1.4 (0.9-2.2)||<0.0001|
|Statin, %||402 (31.3)||296 (55.2)||106 (14.2)||<0.0001|
|ASA, %||361 (28.1)||245 (45.7)||116 (15.5)||<0.0001|
|ACE inhibitor, %||353 (27.5)||307 (57.3)||46 (6.1)||<0.0001|
|Ten-year Framingham Risk||8 (4-13)||13 (10-20)||5 (3-8)||<0.0001|
|Current smoking, %||141 (11.0)||51 (9.5)||90 (12.0)||0.158|
|Alcohol use, %||822 (64.0)||315 (58.8)||507 (67.7)||<0.0001|
|Blood pressure, mm Hg|
|Systolic||129 (119-138)||130 (121-140)||128 (117-137)||<0.0001|
|Diastolic||76 (71-84)||76 (70-83)||78 (71-84)||0.044|
|BMI, kg/m2||28.8 (25.3-32.8)||31.5 (28.3-35.5)||26.8 (24.1-30.2)||<0.0001|
|Waist circumference, in||38 (34-42.5)||42 (39-46)||35.3 (31.8-39)||<0.0001|
|Metabolic syndrome, %||611 (47.6)||419 (78.2)||192 (25.6)||<0.0001|
|hs C-reactive protein, mg/dL||1.4 (0.6-2.9)||1.6 (0.8-3.4)||1.2 (0.5-2.6)||<0.0001|
|Adiponectin, μg/mL||13.1 (8.3-21.0)||9.2 (6.2-14.6)||16.4 (11.5-24.6)||<0.0001|
|Leptin, ng/mL||9.4 (5.4-18.3)||11.2 (6.3-20.3)||8.4 (4.5-16.3)||<0.001|
|Coronary artery calcification|
|Mean score (±SD)||225.2 ± 559.4||414.0 ± 766.0||90.0 ± 271.5||<0.0001|
|Median (IQR)||13 (0-167)||100 (4-427)||3 (0-47)||<0.0001|
Interrelationship of WC, plasma adiponectin, plasma leptin, and their relationship with other CHD risk factors
Spearman correlations of WC, plasma adiponectin plasma leptin with other CVD risk parameters across DM status, and gender were assessed (Supporting Information Table 1A-1C). The association of WC, plasma adiponectin, and leptin with other CVD risk factors (including BMI) revealed fairly consistent inverse correlations for adiponectin and HDL-C with all other variables, as expected. Interestingly, WC, adiponectin, and leptin had strong consistent correlations across DM status with HDL-C, triglyceride, and hs-CRP but not with total cholesterol, LDL-C, or FRS. Adiponectin and leptin were strongly associated with other cardiovascular risk factors consistently in non-DM subjects but not in DM group.
WC and adipokines are associated with CAC in fully adjusted model
We stratified analyses by DM status (Supporting Information Table 2A) and also tested for interaction between gender and independent variables (Supporting Information Table 2B). Because we observed no significant interaction pattern (defined a priori as P < 0.1), we present results for the entire sample (Table 2). In a CAC model adjusted for age, sex, family history of CVD, DM status, exercise, and major cardiometabolic medications, WC and leptin but not plasma adiponectin were associated with CAC (Table 2). After further adjusting for traditional risk factors (smoking, hypertension, hypercholesterolemia), plasma hsCRP, and DM status, WC was still independently associated with CAC (Tobit regression ratio, TRR 1.69; P < 0.001). In this fully adjusted model, plasma leptin (TRR 1.57, P < 0.001) was also associated with CAC. The association of WC and leptin with CAC was relatively stronger in non-DM subjects (Supporting Information Table 2A) and was not significant in women beyond traditional risk factors (Supporting Information Table 2B).
|N = 1,285|
|Variables adjusted for||P||TR (95% CI)|
|Basic model (i.e., age, gender)||<0.001||1.46 (1.24-1.72)|
|Basic model, FHx, Exr, Medsa||<0.001||1.47 (1.23-1.74)|
|Basic model, FHx, Exr, Meds, Major RFsb, hsCRP||<0.001||1.47 (1.22-1.77)|
|Basic model, BMI, FHx, Exr, Meds, Major RFsb, hsCRP||0.42||1.19 (0.78-1.84)|
|Basic model, FHx, Exr, Meds, Major RFsb, hsCRP, DM||<0.001||1.69 (1.39-2.05)|
|Basic model (i.e., Age, Gender)||0.542||1.08 (0.85-1.36)|
|Basic model, FHx, Exr, Medsa||0.325||1.13 (0.89-1.43)|
|Basic model, FHx, Exr, Meds, Major RFsb, hsCRP||0.329||1.13 (0.89-1.43)|
|Basic model, BMI, FHx, Exr, Meds, Major RFsb, hsCRP||0.719||0.93 (0.64-1.36)|
|Basic model, FHx, Exr, Meds, Major RFsb, hsCRP, DM||0.75||1.04 (0.81-1.34)|
|Basic model (i.e., Age, Gender)||<0.001||1.59 (1.30-1.96)|
|Basic model, FHx, Exr, Medsa||<0.001||1.55 (1.26-1.92)|
|Basic model, FHx, Exr, Meds, Major RFsb, hsCRP||<0.001||1.52 (1.21-1.90)|
|Basic model, BMI, FHx, Exr, Meds, Major RFsb, hsCRP||0.456||1.19 (0.75-1.89)|
|Basic model, FHx, Exr, Meds, Major RFsb, hsCRP, DM||<0.001||1.57 (1.25-1.98)|
As a positive control to test the findings of the association of waist with CAC, we added BMI as a covariate in the model. As expected, because of the collinear nature of BMI and WC (spearman ρ = 0.82) and similar phenotypic characteristics captured by both, the significance was attenuated in Tobit regression models (Table 2).
Adipokines do not add value to WC
In nested models, the trend of association was similar across both DM status and gender despite subtle differences in the Tobit regression models. In nested model analysis (Table 3), WC added incremental information beyond traditional risk factors (χ2 27.36; P < 0.001). Plasma leptin (χ2 15.10; P < 0.001) but not adiponectin (χ2 0.10; P = 0.75) added to traditional risk factors in the same model. Neither plasma adiponectin (χ2 0.76; P = 0.38) nor leptin (χ2 1.32; P = 0.25) added any significant incremental value as a predictor of CAC beyond WC in nested model analysis, whereas WC remained significant after being added to adiponectin (χ2 = 28.02; P < 0.0001) as well as leptin (χ2 = 13.58; P = 0.0002).
|Variable||N = 1,285|
|WC added to modela||27.36 (<0.001)|
|Adiponectin added to modela||0.10 (0.75)|
|Leptin added to modela||15.10 (<0.001)|
|Adiponectin added to WC in modela||0.76 (0.38)|
|WC added to adiponectin in modela||28.02 (<0.0001)|
|Leptin added to WC in modela||1.32 (0.25)|
|WC added to leptin in modela||13.58 (0.0002)|
In this study, we demonstrate that WC is associated with CAC beyond traditional risk factors. Most importantly, we have shown that the measurement of plasma adipokines did not add value in CAC prediction beyond WC, but instead WC added incremental value to these adipokines. These findings underscore the importance of anthropometric measurement in CVD risk assessment.
WC has been associated with increased cardiovascular events in a range of populations beyond that predicted by BMI alone  and perhaps better than BMI . Indeed, central obesity is important in metabolic derangement as observed in the metabolic syndrome . Central obesity is partly contributed by visceral adipose tissue (VAT), a detrimental depot of fat  that secretes a large variety of inflammatory cytokines, chemokines, and adipokines such as adiponectin and leptin, providing a compelling mechanistic link between metabolic syndrome, obesity, and vascular complications . Indeed, in our current study, WC was found to have fairly uniform relation with other CVD risk factors across genders and DM status, was an independent strong predictor of CAC beyond traditional CVD risk factors, and added incremental value to adiponectin and leptin.
In addition to WC, other markers may capture a similar metabolically deranged phenotype that WC captures, such as BMI , adipokines , insulin resistance estimation , and imaging various fat depots to quantify visceral and subcutaneous adipose tissue . In fact, each of these markers has been shown to be independently associated with CVDs, and in prior studies from our group [13, 18] we have further confirmed these findings that adipokines relate to CAC in both a DM and non-DM sample. For example, adiponectin and leptin were shown to be associated with subclinical atherosclerosis measured by CAC , and in smaller studies, low plasma adiponectin levels have been reported to be associated with progression of CAC in type 1 diabetic and non-DM subjects independent of other cardiovascular risk factors . Plasma leptin levels were shown to be associated with CAC in DM after controlling adiposity and hs-CRP in an ethnically diverse patient group . Another study with smaller number of urban black South African CVD patients revealed association of leptin with BMI, WC, and hs-CRP as well as the closest link with metabolic syndrome status . Although adiponectin and leptin are involved in adipose pathways associated with CVD risk, they do not fully capture the global effect of visceral adiposity through multiple other intermediate effectors including cytokines, chemokines, insulin resistance itself, and lipoprotein-mediated pathways. In addition to adipokines, insulin resistance estimation also was associated with subclinical atherosclerosis measured by CAC beyond traditional risk factors . Additionally, Fox et al and Lear et al found in their studies that VAT was strongly associated with adverse cardiometabolic risk profile [27, 28].
Our study provides a systematic assessment of how established markers of cardiometabolic risk add to WC and demonstrate that WC may capture information beyond adipokines and traditional CVD risk factors when assessing atherosclerotic burden measured by CAC. Although WC is a crude estimate of central adiposity, it has been demonstrated to be a good surrogate for VAT , a reservoir of bioactive, deleterious, adipose tissue. Several modalities of imaging demonstrated strong correlation of WC and abdominal adiposity [30, 31]. Findings of our study are logical as WC may encompass a more broadly representative process of adipose inflammation including insulin resistance and lipoprotein dysfunction not captured by adipokines alone. Furthermore, we did not find significant differences in the associations of CAC with biomarkers and WC based on DM status, which supports the notion that pathways involved with adiposity-mediated inflammation and atherosclerosis may not differ tremendously, and may be well represented by measurement of WC.
We emphasize that the clinical implications of these findings are substantial. Integrating WC measurement into clinical practice provides additional challenge. Should the presence of metabolic syndrome be assessed in addition to WC given the strong association of metabolic syndrome with CVD risk [15, 32] despite the need for additional laboratory-based measured (e.g., TGs, HDL, glucose)? Furthermore, estimation of insulin resistance also may provide complimentary information to WC regarding cardiometabolic risk ; however, it is neither widely utilized nor available in most centers. Finally, if WC alone is used to assess CVD risk within established risk scores such as the Framingham score , should it be modeled as a continuous variable or dichotomized at a cutpoint? Future studies should systematically address these analytical issues carefully to best utilize the incremental information gained from WC.
Our study demonstrated the largest systematic assessment of WC in the presence of plasma adipokines with CAC, an established marker of burden of atherosclerosis. Our study has several limitations. Analyses were cross sectional, thus causal and longitudinal relationships were not addressed. We also did not examine clinical outcomes. The total adiponectin assay we used may fail to capture the bioactive molecular weight forms. Lower plasma level of high molecular weight (HMW) adiponectin has been reported to be associated with the presence, extent, and vulnerable characteristics of coronary plaques assessed using CT angiogram . Mangge et al  demonstrated HMW subfraction showed a better correlation to IMT compared with total adiponectin. Given CAC variability by race , our findings cannot be generalized beyond Caucasians. In addition, CAC is an estimate, and not a direct measure of coronary atherosclerosis , thus it may fail to detect lipid-rich, softer, noncalcified coronary plaques. Despite these limitations, however, CAC scores are strong, independent predictors of events , including in DM . Waist-to-hip ratio (WHR) has been validated in a number of studies as a strong predictor of cardiovascular event, perhaps a better predictor than WC as indicated in certain studies . Unfortunately, we did not have WHR data for all our patients making it impossible to use that as our anthropometric marker in this study.
In the face of important biomarkers of adiposity such as plasma adiponectin and leptin, WC remained significantly associated with CAC. WC added incremental value beyond the adipokines in predicting CAC in a fully adjusted model (including but not limited to adjustment for traditional CVD risk factors such as hypertension, hypercholesterolemia, diabetes, smoking, and family history of CVD). However, adipokines failed to add any value as a predictor of CAC beyond WC. This underscores the importance of incorporating WC measurement in the assessment of CVD risk, which may capture multiple risk factors for CVD simultaneously. Further research studies should focus on the measurement of WC in the face of multiple CVD biomarkers in a large outcome-based study to understand the prognostic characteristics, which WC adds beyond well-established CVD risk biomarkers.
- 29Body mass index and waist circumference independently contribute to the prediction of nonabdominal, abdominal subcutaneous, and visceral fat. Am J Clin Nutr 2001; 75: 683-688., , , , .
Additional Supporting Information may be found in the online version of this article.
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.