Inflammatory Mediators and Premature Coronary Atherosclerosis in Rheumatoid Arthritis

Abstract

Objectives

Rheumatoid arthritis (RA) is an inflammatory disease associated with premature atherosclerosis. We examined the hypothesis that mediators of inflammation associated with atherosclerosis in other populations IL-6, TNF-α, SAA, VEGF, neutrophil count, IL-1α, E-selectin, ICAM-1, MPO, MMP-9, and VCAM-1 were increased and associated with the severity of coronary atherosclerosis in patients with RA.

Methods

Clinical variables, concentrations of inflammatory mediators and coronary artery calcification were measured in 169 patients with RA and 92 control subjects. Differences in concentrations of inflammatory mediators were compared using median quantile regression. The relationship of inflammatory mediators with the severity of coronary calcification in RA and control subjects was examined using proportional odds logistic regression allowing for interaction with disease status. Models were adjusted for traditional cardiovascular risk factors.

Results

Median serum concentrations of IL-6, SAA, ICAM-1, E-selectin, TNF-α, and MPO and peripheral blood neutrophil count were higher in patients with RA than controls (all p<0.05) independent of Framingham risk score and diabetes. IL-6 (main effect OR 1.72, 95%CI 1.12–2.66) and TNF-α concentrations (main effect OR 1.49, 95%CI 1.16–1.90) were significantly associated with higher amounts of coronary calcium independent of Framingham risk score and diabetes, and such main effects significantly differed from controls (p-value for interaction=0.001 and 0.03, respectively).

Conclusion

TNF-α and IL-6 are significantly associated with the severity of subclinical atherosclerosis independent of Framingham risk score in RA.

Introduction

Coronary atherosclerosis(1) and cardiovascular mortality rates(2) are increased in patients with rheumatoid arthritis (RA). Clinical studies(3) indicate that traditional cardiovascular risk factors included in the Framingham risk score such as age, hypertension, smoking and dyslipidemia contribute to increased atherosclerosis and poorer cardiovascular outcomes in patients with RA, but explain only a component of this increased risk. Inflammation associated with RA is thought to contribute to the increased risk of ischemic heart disease.

Inflammation has been implicated in the pathogenesis of atherosclerosis and subsequent cardiovascular disease(4), and increased concentrations of mediators or markers of inflammation predict subsequent atherosclerotic cardiovascular disease in the general population(5). However, the mechanisms whereby inflammation promotes atherogenesis are poorly understood. One potentially informative approach is to study atherogenesis in patients with a chronic inflammatory disease such as RA.

In the general population several mediators of inflammation including interleukin-6 (IL-6)(6), tumor necrosis factor-alpha (TNF-α)(6), serum amyloid A (SAA)(7), vascular endothelial growth factor (VEGF)(8), peripheral blood neutrophil count(9), interleukin-1 alpha (IL-1α)(10), myeloperoxidase (MPO)(11), matrix metalloproteinase-9 (MMP-9)(12), and adhesion molecules such as vascular cell adhesion molecule (VCAM-1), E-selectin, and intercellular adhesion molecule (ICAM-1)(13,14) have been implicated in the pathogenesis of atherosclerosis. We have recently shown a relationship between inflammation and atherosclerosis in patients with systemic lupus erythematosus (SLE) specific to IL-6, TNF-α, ICAM-1, VCAM-1 and E-selectin concentrations(15,16). However, RA and SLE differ in pathogenesis and clinical manifestations, and little is known about the contribution of specific inflammatory mediators to atherogenesis in RA. Accordingly, we tested the hypothesis that mediators of inflammation are associated with atherosclerosis in patients with RA, independent of the effects of traditional cardiovascular risk factors.

Materials and Methods

Patients and Control Subjects

One hundred and sixty nine patients with RA and 92 control subjects were recruited through advertisements, referral from local rheumatologists, or from a volunteer database maintained by the General Clinical Research Center (GCRC) at Vanderbilt University. Subjects were older than 18 years and patients with RA fulfilled the ACR classification criteria for RA(17). The subjects are participants in ongoing studies of cardiovascular risk in a cohort of patients with RA. Further details concerning the cohort and methods have been described previously(1,3,18). The study was approved by the Vanderbilt University Institutional Review Board and subjects gave written informed consent.

Clinical Measurements and Scores

Clinical information, laboratory data, and Agatston coronary calcium scores were obtained as described in detail elsewhere(1,3,18). In brief, coronary calcium was measured by electron beam computed tomography (EBCT) imaging with an Imatron C-150 scanner (GE/Imatron, South San Francisco, CA, USA) and was quantified as described by Agatston et al (19) by a single reviewer (PR) blinded to the clinical status of the subjects. Hypertension was defined as systolic blood pressure ≥ 140 mmHg or diastolic blood pressure ≥ 90mmHg or both at enrollment or currently receiving antihypertensive treatment. Diabetes was defined as a fasting blood glucose concentration >126 mg/dl at enrollment or currently receiving anti-diabetic treatment. The Framingham risk score used in this study is a version defined in the ATP III report(20) and is a composite score of traditional cardiovascular risk factors that includes blood pressure, smoking status, lipid concentrations, age and sex, but not diabetes. C-reactive protein (CRP) concentrations were determined in the hospital clinical laboratory. Before 2003, the laboratory did not use a high-sensitivity CRP assay, and low concentrations were reported as < 3 mg/liter. In 40 patients with RA who had CRP concentrations < 3 mg/liter, concentrations were measured by ELISA (Millipore).

Blood was drawn and serum was stored at -70°C. E-selectin, VCAM-1, ICAM-1, TNF-α, IL-6, VEGF, MMP-9, MPO, SAA, and IL-1α were measured by multiplex ELISA (Lincoplex^® Multiplex Immunoassay Kit, Linco Research, St. Charles, MO, USA). The coefficient of variation (CV) ranged from 1.4 –12.3% (intra-assay) and 6.8% – 21% (inter-assay).

Statistical Analysis

Descriptive statistics were calculated as mean with standard deviation (SD) or median with the interquartile range (IQR) according to distributions of continuous variables. Differences in concentrations of each inflammatory marker were compared between RA and control groups using the Wilcoxon rank-sum test. The effect of disease status on the difference in median concentrations of inflammatory mediators, adjusted for Framingham risk score and presence of diabetes, were calculated using median quantile regression(21). Median regression is a form of quantile regression, which does not require any distributional assumption of residuals and is robust to extremes of the response variables. Lastly, multivariable proportional odds logistic regression models were applied to analyze associations between each inflammatory marker and the severity of coronary calcification, independent of Framingham risk score and the presence of diabetes.

Because more than 50% of coronary calcification scores are zero, a suitable mathematical transformation is not possible for achieving normality of the residuals required for ordinary linear regression. Thus, we used the proportional odds model, also known as the ordinal logistic regression, a method that is applicable to a continuous ordered response variable(22,23).

We assessed for a differential relationship of mediator effect on coronary calcification according to disease status (RA or control) by including an interaction term (each mediator x disease status (RA / control)) and adjusting for Framingham risk score and the presence of diabetes. Odds ratios (OR) per IQR difference with 95% confidence intervals (95% CI) for the main effects were calculated to assess adjusted associations by disease status. Statistical significance was determined using 2-sided 5% significance level (i.e. p value <0.05). Statistical analyses were conducted with R 2.8.1 (http://www.r-project.org) and SAS version 9.1.

Results

Clinical and demographic characteristics of patients with RA and control subjects are shown in Table 1. The two groups were similar in age and sex. The RA group had a higher rate of smoking and higher coronary calcium scores, as reported previously (1,3).

Table 1.

Clinical Variables in Patients with RAand Control Subjects

Factor	Control (N=92)	RA (N=169)	P value*
Age (years)	53.2±11.6	54.2±11.8	0.41
Sex (Males)	37.0%	30.8%	0.31
BMI (kg/m2)	28.4± 5.9	29.2± 6.8	0.44
Hypertension	39.1%	53.3%	0.03
Diabetes	4.3%	11.2%	0.06
Current Smokers	8.7%	24.3%	0.002
SBP(mmHg)	128.9± 17.4	133.3± 20.3	0.07
DBP(mmHg)	72.6± 8.9	74.9±10.8	0.11
Glucose (mg/dl)	92.7±34.3	91.2±18.6	0.70
Cholesterol (mg/dl)	193.2± 35.8	185.9± 39.4	0.09
HDL (mg/dl)	46.8±12.7	46.6±13.9	0.65
LDL (mg/dl)	122.5± 31.4	112.7± 33.4	0.02
Triglycerides (mg/dl)	118.1± 62.2	144.6±178.5	0.22
ESR (mm/hr)	-	15.5 [7.0–36.0]	-
CRP (mg/l)	-	4.0 [1.2–11.0]	-
MHAQ	0.12±0.3	0.53±0.5	<0.001
Framingham Score	11.1± 5.5	12.2± 6.0	0.10
Agatston Score	0.0 [0.0–18.7]	1.9 [0.0–150.3]	0.02

Data are presented as mean±SD, median and [IQR] or percentages. BMI: Body Mass Index. SBP: Systolic Blood Pressure. DBP: Diastolic Blood Pressure. MHAQ: Modified Health Assessment Questionnaire. Agatston score: Coronary calcium measured as described by Agatston(19).

^*Wilcoxon rank sum test was used for comparing continuous variable. Percentages were compared using chi-square test.

Concentrations of inflammatory markers in patients with RA and in control subjects are shown in Table 2. Median serum concentrations of IL-6, SAA, ICAM-1, E-selectin, TNF-α and MPO and neutrophil count were significantly higher in patients with RA than in control subjects. These associations remained significant when adjusted for the Framingham score and the presence of diabetes.

Table 2.

Concentrations of Inflammatory Markers in Patients with RA and Control Subjects

Factor	Control (N=92)	RA (N=169)	P value*	Median difference	95%CI†	P value‡
IL-6 (pg/ml)	4.2 [1.2–18.2]	13.8 [4.4–43.0]	<0.001	9.69	5.07–14.31	<0.001
SAA (ug/ml)	1.7 [1.0–2.8]	3.2 [1.6–11.1]	<0.001	1.46	0.73–2.19	=0.001
ICAM-1 (ng/ml)	119.4 [98.3–161.8]	166.6 [125.6–201.9]	<0.001	49.08	32.94–65.23	<0.001
E-Selectin (ng/ml)	15.0 [10.0–21.0]	21.5 [15.1–25.9]	<0.001	6.22	3.39–9.06	<0.001
TNF-α (pg/ml)	3.4 [2.4–4.8]	5.5 [2.8–11.0]	<0.001	2.46	1.30–3.62	<0.001
Neutrophil count (x103/ul)	3.4 [2.8–4.4]	4.0 [2.9–5.4]	0.005	0.57	0.10–1.04	0.02
MPO (ng/ml)	17.2 [8.9–27.2]	22.9 [11.7–38.5]	0.01	5.87	0.78–10.95	0.02
IL-1α (pg/ml)	97.8 [33.0–398.4]	144.1 [48.4–380.2]	0.35	68.61	−10.48–147.70	0.09
MMP-9 (ng/ml)	81.3 [62.2–138.6]	87.1 [63.7–146.2]	0.74	9.21	−8.60–27.03	0.31
VEGF (pg/ml)	32.1 [15.1–55.0]	31.6 [17.1–61.5]	0.99	1.39	−9.41–12.20	0.80
VCAM-1 (ng/ml)	953.5[755.5–1107.7]	929.8 [738.3–1189.9]	0.90	22.26	−70.35–114.87	0.64

Data are shown as median with IQR. See main text for abbreviations.

^*Wilcoxon rank sum test.

^†95%CI for differences of medians.

^‡Median difference for each bio-marker between RA and control, CIs, and P values were obtained by median quantile regression while controlling for Framingham score and diabetes.

Associations between coronary calcium and levels of inflammatory mediators and the difference in relationships (i.e. interactions) for patients with RA and controls are shown in Figure 1. IL-6 and TNF-α were associated significantly with coronary calcium in RA (IL-6: main effect OR=1.72, 95%CI (1.12–2.66) and TNF-α: main effect OR=1.49, 95%CI (1.16–1.90)). These associations remained significant in an additional analysis restricted to patients with RA when further adjusted for the use of TNF blockers (TNF-α: OR 1.70, 95%CI (1.24–2.33) and IL-6: OR=1.78. 95%CI (1.12–2.83)). We also compared the relative and additive contributions of IL-6 and TNF-α in models predicting coronary calcium scores in patients with RA using the pseudo-R², which is analogous to the R² in linear regression. The pseudo-R² with TNF-α in the model is 0.32, with IL-6 the pseudo-R² is 0.30, and with both TNF-α and IL-6 in the model the pseudo-R² is 0.32, indicating no additional contribution.

Figure 1. Association between Inflammatory Mediators and Coronary Calcium in Patients with Rheumatoid Arthritis and Controls.

A proportional odds logistic model including an interaction term for disease status (RA or Control) and adjusted for Framingham score and diabetes was used to estimate the main effects of each mediator (log-transformed) by disease status on coronary calcium. The values for the main effect odds ratios (OR) with 95% confidence intervals (CI) per IQR increment of each mediator by disease status are shown on the left and depicted in the Figure. The p-value for interaction between each mediator and disease status is shown on the right.

Although not significant, other mediators such as SAA and VEGF showed a trend for an association with coronary calcium in patients with RA, and the interaction term was significant suggesting that responses differed in patients and control subjects. ICAM-1 was significantly associated with coronary calcium in controls (main effect OR=2.21, 95%CI (1.27–3.86)) but not in patients with RA (main effect OR = 1.0, 95%CI (0.67–1.47) with a significant interaction (P=0.02). IL-6 (main effect OR=0.55, 95%CI (0.33–0.92) and VEGF (main effect OR=0.53, 95%CI (0.31–0.92) were negatively associated with coronary calcium in controls.

We reported previously in a smaller subset of patients with RA that ESR was associated with coronary calcium but CRP was not(1). In the present cohort coronary calcium was associated with ESR (unadjusted OR 1.85 95%CI 1.24–2.6, p=0.002) but not with CRP (OR 1.19 95% CI 0.96–1.5, p=0.117) in RA. Thus, additional sensitivity analyses were performed to examine the association between TNF-α and IL-6 and coronary calcium after additional adjustment for ESR. The associations between both cytokines and coronary calcification in RA were somewhat attenuated (IL-6 OR=1.61 95%CI=0.98–2.63, p=0.06; TNF-α OR=1.56, 95%CI=1.14–2.15, p=0.006) when we adjusted for ESR in addition to Framingham risk score and diabetes.

Discussion

Concentrations of several inflammatory mediators known to be associated with atherosclerosis were increased in patients with RA, but only IL-6 and TNF-α were significantly associated with coronary atherosclerosis as measured by EBCT in patients with RA. These data are consistent with the concept that inflammation promotes atherogenesis, as suggested by animal models and epidemiologic studies, and suggest that in patients with RA, IL-6 and TNF-α are particularly important in this process.

Premature cardiovascular mortality in RA and recognition of the importance of inflammation in atherogenesis have led to the hypothesis that atherosclerosis is increased in patients with chronic inflammatory diseases. We and others have shown that this is indeed the case in patients with SLE and RA(1,24–26). However, specific mechanisms whereby inflammation promotes coronary atherosclerosis in RA remain largely speculative. Therefore, we studied inflammatory mediators that have been implicated in the process of atherogenesis(6–14). Several of these mediators, particularly TNF-α(27,28) and IL-6(29), are also implicated in the pathogenesis of RA, a prototypic chronic inflammatory disease.

In patients with RA, studies of carotid artery intima-media thickness (IMT) have suggested associations with elevated ESR(30), CRP(31), adhesion molecules(32) and neutrophil count(33), but little is known about inflammatory mediators and coronary atherosclerosis. Our previous observation that ESR, but not CRP, was associated with coronary calcium in RA(1) led to the present study to define the role of specific inflammatory mediators in the pathogenesis of atherosclerosis.

Concentrations of many inflammatory mediators are elevated in patients with RA. Our findings suggest that IL-6 and TNF-α appear to be linked significantly to atherosclerosis in RA, providing support for the idea that specific mediators rather than a non-specific inflammatory response, are important in the pathogenesis of atherosclerosis in this condition, and perhaps also in other inflammatory conditions associated with atherosclerosis, such as SLE(34), ankylosing spondylitis, psoriatic arthritis(35), periodontitis(36). There was a significant interaction between disease status (RA or control) and IL-6, TNF-α, SAA, VEGF and ICAM for coronary calcium, suggesting that these biomarkers may be particularly important in explaining differences in coronary calcification between patients with RA and control subjects.

Additional evidence suggests that the pathogenesis of atherosclerosis associated with inflammation may differ in various inflammatory diseases. We previously observed that the mediators of inflammation associated with coronary atherosclerosis in patients with SLE were IL-6, TNF-α and the adhesion molecules ICAM-1, VCAM-1 and E-selectin(15,16). In contrast, in the present study the adhesion molecules E-selectin, ICAM-1 and VCAM-1 were not significantly associated with coronary calcium. In fact, ICAM-1 was associated with coronary calcium in controls but not in patients with RA.

Thus, therapeutic targets for prevention or slowing of premature atherosclerosis may differ in different conditions, and in RA therapeutic strategies specifically targeted at IL-6 and TNF-α, and perhaps SAA and VEGF may be more effective than those directed at other mediators of inflammation.

A key role for TNF-α in atherosclerosis is supported by the observations that anti-TNF-α agents improve endothelial dysfunction(37) and reduce carotid IMT thickness(38). In addition, anti-TNF therapy may decrease cardiovascular morbidity and mortality in RA(39,40). The finding of an association between IL-6 concentrations and coronary artery calcification in RA is intriguing since a new drug for RA, tocilizumab, targets IL-6 receptors. There are no human studies of the effects of IL-6 inhibition on atherosclerosis or cardiovascular outcomes, and to the contrary, adverse effects on lipid profiles have been reported with tocilizumab(41). In animals, inhibition of IL-6 has been reported to suppress atherogenesis(42). Our study suggests that IL-6 may be a target in atherosclerosis in RA, but whether inhibition of IL-6 will affect the pathogenesis or progression of atherosclerosis and improve cardiovascular mortality in RA remains an important open question.

This study design has several strengths, notably analyses of a well-characterized cohort, and measurement of many inflammatory mediators reported to be associated with atherosclerosis. Also, a highly specific, quantitative measurement of coronary atherosclerosis coronary calcium that is predictive of coronary heart disease risk(43), was used. Thus, the study provides unique information directly relevant to the question of interest accelerated coronary atherosclerosis.

The study also has limitations, including the cross-sectional design and the relatively low number of subjects studied. Thus, causal relationships cannot be inferred, and we cannot exclude the possibility that markers which show weaker, non-significant associations with atherosclerosis, such as SAA and VEGF, may also contribute to a relationship between inflammation and atherosclerosis.

Conclusion

In conclusion, IL-6 and TNF-α are significantly associated with premature atherosclerosis in RA independent of traditional cardiovascular risk factors. Specific mediators, and not global inflammation, might contribute to the pathogenesis of atherosclerosis associated with RA.