| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Atherosclerosis and Lipoproteins |
From the Donald W. Reynolds Centers of the Brigham and Womens Hospital (A.Z., N.G., L.M., P.L.), Harvard Medical School, Boston, Mass; and the University of Texas Southwestern Medical Center (S.A., A.K., D.K.M., G.L.V., S.G., J.d.L.), Dallas. Current affiliations: Department of Cardiology (A.Z.), University of Freiburg, Germany; Karolinska Institute (N.G.), Stockholm, Sweden; and Cardiovascular Disease (U.S.), Boehringer Ingelheim Pharmaceuticals, Ridgefield, Conn.
Correspondence to Andreas Zirlik, MD, University of Freiburg, Department of Cardiology and Angiology, Breisacherstrasse 33, 79106 Freiburg, Germany. E-mail andreas.zirlik{at}uniklinik-freiburg.de
| Abstract |
|---|
|
|
|---|
Methods and Results— IL-18 plasma levels were determined by ELISA in 2231 subjects from the Dallas Heart Study. In univariable analysis, IL-18 levels associated with traditional cardiovascular risk factors and particularly with components of the metabolic syndrome (MS, P<0.01 for trend across the number of MS components); IL-18 also associated with coronary artery calcium (CAC) scores measured by electron beam computed tomography and aortic plaque measured by MRI (P<0.01 for each). In multivariable analyses, IL-18 remained associated with multiple components of the MS but not with CAC or aortic plaque.
Conclusions— In a large population-based sample, elevated IL-18 plasma levels associated with risk factors for atherosclerosis and with the metabolic syndrome. The association between IL-18 and atherosclerosis diminished after accounting for traditional cardiovascular risk factors. These data suggest that IL-18 does not add independently to detection of atherosclerotic burden in asymptomatic individuals.
In the Dallas Heart Study, IL-18 levels associated with traditional cardiovascular risk factors, components of the metabolic syndrome, and surrogate markers of subclinical atherosclerosis. The association of IL-18 and atherosclerosis diminished in multivariate analysis suggesting that IL-18 does not predict atherosclerotic burden in this collective.
Key Words: atherosclerosis imaging interleukins risk factors metabolic syndrome
| Introduction |
|---|
|
|
|---|
The proinflammatory cytokine IL-18 participates in adaptive immunity,3 and several lines of evidence support a proatherogenic role for IL-18. Expression of IL-18 and the IL-18 receptor subunits is increased in atherosclerotic arteries compared with normal arterial segments.4,5 Stimulation of cell types found in atheromata with IL-18 induces proinflammatory cytokines such as interferon (IFN)-
and IL-6, adhesion molecules such as intercellular adhesion molecule-1 (ICAM-1), and enzymes capable of degrading extracellular matrix (MMPs).4,6,7 In addition, studies in mice demonstrate that lack of IL-18 or inhibition of IL-18 signaling decreases atherosclerotic plaque formation, whereas administration of exogenous IL-18 promotes atherogenesis and yields more fatty than fibrous lesions.8–11
Recent evidence also suggests that IL-18 may serve as a biomarker of cardiovascular risk. IL-18 levels increase in patients with stable and unstable angina and in those with myocardial infarction (MI),12–14 and higher IL-18 levels correlate with poorer prognosis in patients with established coronary artery disease (CAD).15–17 Several recent studies demonstrated a strong association between IL-18 and several CV risk factors, especially those linked with the metabolic syndrome.18 Although IL-18 levels carry prognostic information in patients with and without established CAD,15 conflicting data have been reported regarding the association of IL-18 and the burden of atherosclerosis.19,20 This study investigates whether IL-18 adds to traditional risk factors in determining atherosclerotic burden in the large population-based cohort of the Dallas Heart Study.
| Methods |
|---|
|
|
|---|
50% of the final cohort. Details of the DHS are described elsewhere.21 Briefly, 6101 subjects participated in an initial detailed in-home health survey (visit 1). Of these, 3557 subjects returned for blood and urine collection (visit 2), and 3072 of these subjects returned for various imaging procedures, including MRI for the presence of aortic plaque, electron-beam computer tomography (EBCT) for assessment of coronary artery calcium (CAC), and dual energy X-ray absorbitometry (DEXA) to determine whole body fat and lean mass (visit 3). Subjects who completed each of the 3 visits had similar medical history, demographic data, body mass index (BMI), and blood pressure, and laboratory parameters were similar among those completing visit 2 and visit 3.21 Patients with sufficient plasma volume for the quantification of IL-18 were included in the present study (n=2231). All participants provided informed consent, and the study protocol was approved by the Institutional Review Board of UT-Southwestern Medical Center.
Variable Definition
Detailed definitions of variables have been reported previously.22 Metabolic syndrome (MS), as well as its individual components, were defined as established by the National Heart, Lung, and Blood Institute/American Heart Association consensus statement.23 BMI was defined as weight (in kilograms)/height (meters). Angina pectoris was defined as self-reported chest pain with exertion plus at least 1 affirmative answer to an additional criteria from the Rose Angina Questionaire: location, effect of exertion effort on chest pain, improvement of pain with rest, or duration of chest pain.24 History of myocardial infarction (MI) was determined by self-report. Tobacco use was defined as tobacco use within 30 days before visit 1, and alcohol use was defined as any alcohol consumption in the last year. Glomerular filtration rate (GFR) was estimated using the MDRD calculation.25 The homeostasis model assessment of insulin resistance index (HOMA-IR) was calculated as described previously [fasting insulin (µIU/mL)xfasting glucose (mmol/L)/22.5].26
Imaging Procedures
EBCT scans were performed using an Imatron C-150XP EBCT scanner (Imatron Inc), using a previously described protocol.27 CAC was quantified and expressed as Agatston score, and the mean of 2 consecutive scans was used unless only 1 scan was performed.28 To maximize signal-to-noise ratio and reproducibility, a cutoff of 10 Agatston units was selected as a data-derived threshold to define the presence of calcium. CAC scores were also classified using an ordinal score according to a previously defined scheme: no calcium (0 to
10), mild (>10 to
100), moderate (>100 to
400), and severe (>400).29 MRI of the abdominal aorta used a Philips Medical Systems 1.5 Tesla Intera magnet (Philips Medical Systems), as described previously.30,31 Images were analyzed by trained observers using the Magnetic Resonance Analytical Software Systems (MASS) cardiac analysis software package (Version 4.2 beta, Medis Medical Imaging Systems Inc) according to previously defined criteria.32 DEXA was used to divide total body mass into 3 compartments: total fat mass, lean mass, and bone mass, as described previously.33
Quantification of IL-18 and Other Analytes
Venous blood was collected into ethylendiaminetetraacetic acid (EDTA) tubes at visit 2, centrifuged at 1430g for 15 minutes at 4°C, and plasma was stored in aliquots at –80°C until analysis. Samples were thawed and CRP (Roche Diagnostics), MCP-1 (Biosite Inc), and sCD40L (Bender Med Systems) were quantified as previously described.22,30 IL-18 levels were assayed in duplicates by an "in-house" ELISA using an unlabeled primary and biotinylated secondary monoclonal mouse anti-human IL-18 antibody (MBL). The intra- and interassay variation coefficients were below 10% and 18%, respectively. Leptin was measured with a commercial radioimmunoassay (Linco Research Inc). The lowest level of leptin detectable by this assay is 0.5 µg/L and all values <0.5 µg/L (n=133) were designated 0.5 µg/L (intraassay and interassay coefficient of variation of 8.3% and 6.2%, respectively, in men and 3.4% and 4.6% in women).
Statistical Analysis
Statistical analyses were performed using SAS (Version 9.1, SAS Corporation) statistical software package. For univariable analyses, subjects were divided into quartiles based on IL-18 levels. The Cochrane-Armitage test for trend was used to compare categorical variables, and the
2 trend test across ordered groups was used to compare continuous variables across increasing IL-18 quartiles. Spearman correlation coefficients were calculated to evaluate the correlation between IL-18 levels and different biomarkers. In multivariable linear regression models for log (IL-18), a measure of atherosclerotic burden (either aortic plaque, CAC score >10, or ordinal CAC score), as well as other confounders, were entered as independent variables. To determine which risk factors independently associated with IL-18, clinical factors were added in stepwise fashion using a multivariable linear regression model for log (IL-18); criteria to enter the model was P<0.2 and to remain in the model was P<0.1. Prevalent CAC (CAC >10) or aortic plaque was then entered into the resulting model to determine whether these markers of atherosclerostic burden were associated with IL-18 independent of CV risk factors. To determine whether IL-18 predicted atherosclerotic burden independently of established cardiovascular risk factors, logistic regression models were created with a measure of atherosclerotic burden (CAC 10, ordinal CAC scores, and aortic plaque prevalence) as the dependent variable and IL-18 and other potential confounders as independent variables. Based on our sample size, there was 88% power to detect a 50% increase in the presence of CAC (>10) between the 1st and 4th quartiles of IL-18, and >99% power to detect a 50% increase in the prevalence of aortic plaque between the 1st and 4th quartiles of IL-18. In all analyses, 2-sided probability values of <0.05 were considered significant.
| Results |
|---|
|
|
|---|
|
Factors Associated With IL-18 Values
Subjects with higher levels of IL-18 were older, more frequently male, Caucasian, and obese. Furthermore, they were more likely to have history of diabetes, hypertension, tobacco use, low HDL, hypertriglyceridemia, previous myocardial infarction, and a lower GFR (Table 2). In contrast, IL-18 did not associate with alcohol use, hypercholesterolemia, family history of MI, or stable angina. Subjects with higher IL-18 levels more frequently took aspirin. No significant associations were observed between IL-18 quartiles and other drugs such as statins and antihypertensives. The associations observed in the overall cohort were similar in Caucasians (n=608) and African American (n=1213). Among the smaller Hispanic subgroup (n=369), the associations between IL-18 and several of the CV risk factors such as obesity, body mass index, tobacco use, hypercholesterolemia, and hypertriglyceridemia were not statistically significant (supplemental Tables I through III, available online at http://atvb.ahajournals.org). When looking at females and males separately the associations between IL-18 quartiles and cardiovascular risk factors were largely the same (supplemental Tables IV and V).
|
In univariable analyses, IL-18 levels associated with other inflammatory markers including MCP-1 (Rho 0.15, P<0.0001), sCD40L (Rho 0.11, P<0.0001), and with CRP (Rho 0.17, P<0.0001). Plasma levels of IL-18 were not associated with urine microalbumin levels (Rho 0.04, P=0.06).
IL-18 and the Metabolic Syndrome
IL-18 levels correlated significantly with components of the metabolic syndrome. Geometric mean concentrations of IL-18 rose progressively with increasing numbers of metabolic risk factors (Figure). Measures of obesity, such as BMI, total fat mass, total lean mass, WTHR, and waist circumference correlated significantly with IL-18 levels (Table 3). IL-18 also showed strong correlations with measures of insulin resistance, including HOMA and glucose levels. In stratified analyses, correlations between measures of obesity and IL-18 were present in those without but not with diabetes, whereas the association between IL-18 and HOMA-IR was present in both nondiabetic and diabetic subjects. IL-18 associated most strongly with WTHR (Rho 0.20, P=0.0001) and insulin resistance (Rho=0.17, P<0.0001), a relationship that remained significant on stratification for gender (Table 4). Median fasting glucose levels increased modestly but significantly across IL-18 quartiles (89, 91, 91, and 92 mg/dL; P=0.003). However, after adjusting for HOMA-IR, these differences were attenuated and no longer significant (P=0.81). Stratification for the different ethnic groups did not reveal an ethnic bias of this association.
|
|
|
IL-18 as a Biomarker of Subclinical Atherosclerosis
IL-18 levels were significantly higher in subjects with increased CAC scores and prevalent aortic plaque as assessed by EBCT (561 versus 503 pg/mL, P=0.0053) and MRI (536 versus 478 pg/mL), respectively. After adjusting for tobacco use, male sex, low HDL, hypercholesterolemia, black race, family history of MI, glomerular filtration rate (GFR), hypertension (HTN), and diabetes mellitus (DM), increasing IL-18 quartiles did not associate with either CAC >10, ordinal CAC scores, or aortic plaque (data not shown). Similar findings were observed when IL-18 was added as continuous variable for all 3 models. Interestingly, when stratifying for males and females, we only found a statistically significant association between IL-18 levels and parameters of subclinical atherosclerosis in females but not in males (supplemental Tables IV and V). However, as in the overall cohort these associations did not persist in a multivariable model (data nor shown).
Influence of Atherosclerosis and Other Variables on Plasma Levels of IL-18
In a multivariable linear regression model with log IL-18 as the dependent variable and clinical factors associated with IL-18 in univariable analysis added in stepwise fashion, only non–African American race, WTHR, HOMA-IR, smoking, low HDL cholesterol, and GFR associated independently with log IL-18 (Table 4). When added to this model, CAC >10 (P=0.1526) and aortic plaque (P=0.6873) were not independently associated with log IL-18.
| Discussion |
|---|
|
|
|---|
We observed associations between IL-18 levels and a multitude of traditional CV risk factors including many components of the MS, concurring with a recent report by Hung et al.18 Several other reports observed robust correlations between IL-18 and diabetes, a major contributor to CV disease.35–37 We additionally identified associations between IL-18 levels and measures of insulin resistance, such as HOMA-IR and hyperglycemia. IL-18 levels correlate with fasting glucose levels, however on adjustment for HOMA-IR this association disappears. Furthermore, in diabetic subjects, the association of IL-18 with measures of obesity were not significant, suggesting that insulin resistance is a strong determinant of IL-18. In our study, IL-18 levels also associated with several markers of obesity such as BMI, WTHR, and total fat mass. Adipose tissue provides an important source of the IL-18 blood pool, and fat mass correlates with IL-18 plasma levels.38–41 Our results suggest that fat in the central portions of the body contributes more to IL-18 levels and that elevated IL-18 in patients with the MS derives predominantly from these fat accumulations, reflecting increased CV risk of such patient populations.
Our observation that associations between IL-18 and atherosclerosis depend closely on the presence of CV risk factors such as obesity and diabetes, supports the hypothesis that IL-18 mediates some of the detrimental proatherogenic effects of these factors. Several lines of laboratory evidence support this hypothesis. Gerdes et al demonstrated that IL-18 expressed by macrophages evokes a proinflammatory response in vascular endothelial cells and smooth muscle cells.4 Several other proinflammatory effects of IL-18 have since been described, including stimulation of leukocyte adhesion molecules, induction of chemokines, and expression of MMPs, factors implicated in atherogenesis.1,4,6,7,42 Several in vivo studies suggested that disruption of the IL-18 gene results in reduced atherosclerosis in mice.8,9 Interestingly, exogenous IL-18 not only accelerates atherosclerosis in mice but also promotes features of diabetes.10,11,43,44 Conversely, Esposito et al found that hyperglycemia directly stimulates IL-18 expression in humans via an oxidative mechanism.45 These data suggest that IL-18 not only links diabetes and the metabolic syndrome with atherogenesis but may also play a critical role in the pathogenesis upstream of both diseases. This concept has however recently been challenged by a report by Netea et al demonstrating the surprising finding that IL-18 deficiency results in obesity and insulin resistance in mice, a phenotype reversed by exogenous administration of IL-18 or overexpression of IL-18 binding protein.46 These apparently conflicting data suggest that the biological effect of IL-18 may be threshold-dependent and vary for different genetic backgrounds.
Finally, our data show significantly higher IL-18 plasma levels in Caucasians compared with African-Americans, who accounted for approximately 50% of our study. Although some studies have reported similar risk factors in both ethnic groups, only scarce data exist about the ethnic differences in levels of CV biomarkers.47,48 Biomarker levels and their clinical significance may vary between different ethnicities. Our data indicate that associations of IL-18 with CV risk factors are similar in Caucasians and African Americans. Among the smaller subgroup of Hispanic subjects, the associations between IL-18 and some major CV risk factors such as obesity, body mass index, tobacco use, hypercholesterolemia, and hypertriglyceridemia were not statistically significant, likely because of lower power compared with the other ethnic groups. The associations between IL-18, the metabolic syndrome, and subclinical atherosclerosis were similar in the respective ethnic subgroups.
The present study has several limitations that require consideration. Because some samples lacked sufficient volume for quantification of IL-18, we cannot eliminate the influence of selection bias. However, our data indicate no difference in the distribution of baseline characteristics between those who were initially included in the study and those who had enough volume for IL-18 quantification. In addition, we estimated atherosclerotic burden by measuring CAC determined by EBCT and the determined presence of aortic plaque by MRI, techniques that may lack sufficient sensitivity to detect low-grade atherosclerosis. Also, the ELISA system used measures total IL-18 levels. Total IL-18 levels may not reflect the biologically active free IL-18 because there exist several natural inhibiting binding proteins. Our study cannot eliminate the possibility that measures of free or active IL-18 would perform better than IL-18 for estimating the burden of atherosclerosis. However, in previous studies IL-18 and its binding proteins most frequently both increased in inflammatory diseases.49,50 Finally, the cross-sectional design of the study does not allow investigation of temporal relationships between IL-18 levels, atherosclerosis, and CV risk factors. In contrast, the strengths of this study lie in its population-based character, extensive cardiovascular phenotype characterization, and large sample size.
In sum, this study demonstrates that IL-18 associates with several CV risk factors, including many components of the MS, in a large population based cohort of apparently healthy subjects. Although IL-18 associates with subclinical atherosclerosis in univariable analysis, this association does not persist after adjustment for traditional CV risk factors. Thus IL-18 does not appear to add to currently accepted risk markers as a diagnostic tool for assessment of atherosclerotic burden in a community-based population.
| Acknowledgments |
|---|
Sources of Funding
This work was supported by grants from the Donald W. Reynolds Foundation to P.L. and J.d.L., the NIH (HL-66086) to U.S., the Ernst Schering Research Foundation to N.G., and the Deutsche Forschungsgemeinschaft to A.Z. (ZI743/1-1 and ZI743/3-1).
Disclosures
None.
| Footnotes |
|---|
Original received June 7, 2007; final version accepted June 28, 2007.
| References |
|---|
|
|
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
ATVB Home | Subscriptions | Archives | Feedback | Authors | Help | AHA Journals Home | Search Copyright © 2007 American Heart Association, Inc. All rights reserved. Unauthorized use prohibited. |