Metabolic syndrome and inflammatory biomarkers: a community-based cross-sectional study at the Framingham Heart Study

The covariates were defined at examination cycle 7 through assessment of questionnaires, physicals and laboratory tests. Current smoking status was classified by self-report of cigarette smoking during the year prior to examination. Resting blood pressure was measured in a seated position by the physician using a mercury column sphygmomanometer; blood pressure represented the average of two readings. Hypertension was defined as systolic blood pressure ≥140 mmHg or diastolic blood pressure ≥90 mmHg or use of antihypertensive medications. Waist circumference in centimeters was measured by trained technicians at the umbilicus level according to a standard protocol. BMI was defined as the individual’s body weight divided by the height squared, expressed in kg/m². Obesity was defined as BMI ≥30 kg/m². Lipid profile, plasma glucose and insulin levels were measured from morning fasting blood samples using standardized assays. We calculated homeostasis model assessment insulin resistance index (HOMA-IR) by applying the formula [(fasting insulin)(fasting glucose)]/22.5. Insulin resistance was considered present if the insulin resistance index (HOMA-IR) was ≥75^th percentile. Aspirin use was defined as 3 or more doses per week.

Fasting samples were frozen at −80° Celsius until testing. Serum concentrations were measured for CRP, interleukin-6, intercellular adhesion molecule-1, monocyte chemoattractant protein-1. Plasma concentrations were estimated for CD40 ligand, osteoprotegerin, P-selectin, tumor necrosis factor-alpha, and tumor necrosis factor receptor 2. CRP was measured through high sensitivity Dade Behring BN100 nephelometer. Other biomarkers were assessed by enzyme linked immunoassay; all intra-assay coefficients of variation were ≤9.1%. Details regarding marker selection and measurements have been reported [24].

Metabolic syndrome was defined according to the National Cholesterol Education Program Adult Treatment Panel III guidelines [5] as elucidated in Table1. Metabolically healthy but obese individuals were defined as participants with BMI ≥30 kg/m², but without the metabolic syndrome. Metabolically obese but normal weight individuals were identified as those with the metabolic syndrome and BMI <25 kg/m².

The inflammatory markers concentrations showed skewed distributions and were natural log-transformed before further analysis. We performed multivariable linear regression with biomarkers as dependent variables adjusting for age, sex, smoking, aspirin use, and hormone replacement therapy. In model 1, metabolic syndrome was the key exposure. In model 2, we added adjustment for metabolic syndrome components as continuous traits while adjusting for age, sex, smoking, aspirin use, hormone replacement, lipid lowering treatment and hypertension therapy. In model 3, we examined inflammatory biomarkers with interaction between metabolic syndrome and BMI: normal weight was defined as BMI <25 kg/m², overweight as BMI 25–29.99 kg/m² and obese as BMI ≥30 kg/m². With model 4, we analyzed inflammatory biomarkers with the interaction between metabolic syndrome and insulin resistance. We estimated fold increment (and 95% CI) comparing adjusted biomarker levels in groups with and without metabolic syndrome. We considered P <0.05 as statistically significant. The statistical analysis was performed with SAS version 8.1 (SAS Institute, Cary, NC, USA).

Multiple linear regression models demonstrated highly statistically significant associations (P <0.0001) between prevalent metabolic syndrome and CRP, interleukin-6, intercellular adhesion molecule-1, P-selectin, tumor necrosis factor-alpha, and tumor necrosis factor receptor-2,after adjusting for age, sex, smoke, aspirin and hormone replacement therapy. CD40 ligand and monocyte chemoattractant-1 also were nominally associated (P ≤ 0.02), but osteoprotegerin was not associated with the metabolic syndrome. Except for osteoprotegerin, inflammatory biomarkers showed higher mean concentrations in participants with versus without metabolic syndrome (Table2, model 1).

Table2 (model 2) shows the relation of the metabolic syndrome to inflammatory biomarkers after adjusting for the components of the metabolic syndrome. In that setting, metabolic syndrome remained significantly associated only with P-selectin (P = 0.005). Subjects with metabolic syndrome had a 1.06 fold (i.e., 6%; (95% CI 1.02, 1.10, P = 0.005)) increase of P-selectin compared to those without metabolic syndrome.

Only 10% (n = 101) of individuals with metabolic syndrome were normal weight, 43% (n = 421) were overweight and 47% (n = 462) were obese (Additional file1 Table S1). In participants without metabolic syndrome, the distribution was shifted to lower mean BMI (45% normal weight, 41% overweight and 14% obese). Among normal weight participants, 12% had the metabolic syndrome. Obesity was observed in 680 participants; 32% (n = 218) were metabolically healthy but obese. Among normal weight individuals, the metabolic syndrome was associated with higher mean concentrations of the following biomarkers: CRP, intercellular adhesion molecule-1, interleukin-6, P-selectin, tumor necrosis factor-alpha and tumor necrosis factor receptor 2 when compared to healthy normal weight individuals (Additional file1 Table S1).

The interaction between metabolic syndrome and BMI category was statistically significant for CRP (P = 0.02). The proportional increase in CRP, comparing those with to those without metabolic syndrome, decreased across BMI categories (Figure1). The presence of metabolic syndrome was associated with a 1.60 fold (i.e. 60%; (95% CI 1.31, 1.95 fold)) increase of the mean CRP concentration than would be anticipated in normal weight individuals, with a 1.27 fold (i.e. 27%; (95% CI 1.13, 1.43 fold)) increase among overweight subjects, but with a non-significant 1.13 fold (i.e. 13%; (95% CI 0.96, 1.31 fold)) increment in obese individuals (Additional file1 Table S1).

In individuals with metabolic syndrome 49 % had insulin resistance, compared to 10 % in those without metabolic syndrome The interaction between metabolic syndrome and insulin resistance was statistically significant only for CRP (P = 0.008). Metabolic syndrome without insulin resistance was associated with a 1.67 fold (i.e. 67%; (95% CI 1.50, 1.86 fold)) increase of mean CRP levels compared to those individuals without metabolic syndrome and insulin resistance. Among those with insulin resistance, metabolic syndrome was associated with a 25% increment (1.25 fold (95% CI 1.04, 1.51 fold)) of mean CRP levels compared to those without the metabolic syndrome (Additional file2 Table S2). When evaluating participants without metabolic syndrome, those with insulin resistance had statistically significant higher mean concentrations of CRP compared to those without insulin resistance, and their levels were similar to those individuals with metabolic syndrome but without insulin resistance. Mean CRP levels were highest in individuals with both conditions (Additional file3 Figure S1).

The interaction of metabolic syndrome and sex was statistically significant for CRP (P < 0.0001) and tumor necrosis factor receptor 2 (P = 0.002). Among women, we observed in the presence of metabolic syndrome a statistically significant 2.17 fold increment of mean CRP levels (95% CI 1.95, 2.43), whereas in men the observed increment was 1.47 fold (95% CI 1.30, 1.65). Regarding tumor necrosis factor receptor 2, women with metabolic syndrome had higher mean concentrations (fold increment 1.12, 95% CI 1.09, 1.15) compared to those without; among men the corresponding increment was 1.05 fold (95% CI 1.01, 1.08) (Additional file4 Table S3).

We observed a significant association between the metabolic syndrome and all inflammatory biomarkers except osteoprotegerin, which is consistent with the hypothesis that the metabolic syndrome is accompanied by an inflammatory state. We report an interaction between BMI and metabolic syndrome for CRP; in individuals with obesity the presence of the metabolic syndrome did not appear to be associated with additional elevation in mean CRP concentrations. We also detected a significant interaction for CRP in relation to the metabolic syndrome and insulin resistance. Among those without the metabolic syndrome, the presence of insulin resistance was associated with higher mean concentrations of CRP. When evaluating metabolically obese but normal weight individuals, we observed higher mean concentrations of CRP, intercellular adhesion molecule-1, interleukin 6, P-selectin, tumor necrosis factor-alpha and tumor necrosis receptor 2 compared to healthy normal weight individuals. Our results reinforce the concept that the metabolic syndrome even in the absence of obesity is associated with an inflammatory state. Finally, we demonstrated that adjusting for all the components of the metabolic syndrome attenuated the association between the metabolic syndrome with all biomarkers, except P-selectin.

The association between metabolic syndrome and some of the inflammatory biomarkers has been examined in the past [2, 9, 10, 12, 16]. The current literature provides evidence of elevated levels of CRP, tumor necrosis factor alpha, interleukin 6 in individuals with central fat when compared to those with normal fat distribution [25, 26]. In the same cohort at the Framingham Heart Study, we demonstrated that tumor necrosis factor alpha and tumor necrosis factor alpha receptor 2 remained associated with insulin resistance after adjusting for central obesity, adiponectin and resistin [27].

Consistent with our results increased levels of P-selectin have been described among individuals with as compared to without the metabolic syndrome [28, 29]. An increased expression of cell adhesion molecules such as intercellular adhesion molecule-1 and P-selectin have also been associated in a smaller cohort with increased waist circumference, low HDL cholesterol and elevated fasting glucose [16]. P-selectin is known to be involved in the attachment of circulating leukocytes to the vascular endothelium, contributing to the early development of atherosclerotic lesions, even before a metabolic disorder would be detected. It is expressed on activated platelets as well as by endothelial cells. The secretion of P-selectin can be induced through atherogenic factors such as oxidized LDL. Nevertheless the association between P-selectin and the metabolic syndrome after adjusting for its components although statistically significant, warrants cautious interpretation. We may have increased the chance of introducing false positive results by multiple testing. The clinical significance of the reported association merits further study.

We recognize the controversy surrounding the use of the metabolic syndrome as a diagnostic or management tool, understanding its role as a pre-morbid condition rather than a clinical diagnosis [13]. In this regard it has been estimated that about one fifth of the US population fulfills the criteria of the metabolic syndrome [15]. Further studies evaluating the role of the inflammatory biomarkers among metabolically healthy but obese and metabolically obese but normal weight individuals compared to their counterparts are needed in order to enhance our understanding regarding the pathophysiology behind the observed clustering of abnormal metabolic traits. We acknowledge that the clinical significance of our findings is uncertain. Further work should investigate whether inflammatory markers will prove useful in the early identification of individuals at risk for the development of the metabolic traits, and whether such risk stratification will be associated with the ability to reduce or delay the incidence of associated morbidity and mortality.

Given our cross-sectional observational design, our study cannot prove causality. It is possible that metabolic features lead to inflammation, or that inflammation predisposes to the development of metabolic perturbations, or that a complex feedback loop exists wherein each fuels the development and progression of the other. Alternatively both inflammation and metabolic traits may be both related to additional untested features.

Of the various available definitions for the metabolic syndrome, we used the National Cholesterol Education Program Adult Treatment Panel III criteria. However, one should consider the possibility that any other available scheme to define the metabolic risk could be equally valid and produce different results. We did not account for the multiple testing inherent in examining 9 biomarkers, increasing the chance to introduce false positive findings. Because our sample represents mostly white individuals, the generalization of our findings to other ethnic/racial groups is uncertain. Although we selected a robust panel of inflammatory biomarkers, we recognized the limitation caused by missing information on biomarkers such as E-selectin, VCAM-1 or adiponectin. The strengths of the present study includes a large, community-based sample, a routine ascertainment of potential confounders and the availability of a robust set of inflammatory markers, using precise techniques to quantify their concentrations.