Left ventricular structure and function in patients with rheumatoid arthritis, as assessed by cardiac magnetic resonance imaging

Abstract

Objective

Heart failure is a major contributor to cardiovascular morbidity and mortality in patients with rheumatoid arthritis (RA), but little is known about myocardial structure and function in this population. This study was undertaken to assess the factors associated with progression to heart failure in patients with RA.

Methods

With the use of cardiac magnetic resonance imaging, measures of myocardial structure and function were assessed in men and women with RA enrolled in the Evaluation of Subclinical Cardiovascular Disease and Predictors of Events in Rheumatoid Arthritis study, a cohort study of subclinical cardiovascular disease in patients with RA, in comparison with non-RA control subjects from a cohort enrolled in the Baltimore Multi-Ethnic Study of Atherosclerosis.

Results

Measures of myocardial structure and function were compared between 75 patients with RA and 225 frequency-matched controls. After adjustment for confounders, the mean left ventricular mass was found to be 26 gm lower in patients with RA compared with controls (P < 0.001), an 18% difference. In addition, the mean left ventricular ejection fraction, cardiac output, and stroke volume were modestly lower in the RA group compared with controls. The mean left ventricular end systolic and end diastolic volumes did not differ between the groups. In patients with RA, higher levels of anti–cyclic citrullinated peptide (anti-CCP) antibodies and current use of biologic agents, but not other measures of disease activity or severity, were associated with significantly lower adjusted mean values for the left ventricular mass, end diastolic volume, and stroke volume, but not with ejection fraction. The combined associations of anti-CCP antibody level and biologic agent use with myocardial measures were additive, without evidence of interaction.

Conclusion

These findings suggest that the progression to heart failure in RA may occur through reduced myocardial mass rather than hypertrophy. Both modifiable and nonmodifiable factors may contribute to lower levels of left ventricular mass and volume.

Rheumatoid arthritis (RA) is an inflammatory autoimmune disorder that affects ∼1% of adults and frequently results in significant joint deformity and disability. Premature mortality, primarily due to cardiovascular disease (1, 2), is a prominent feature of the disease. The average lifespan is reduced by 8–15 years in patients with RA compared with age-matched controls (1). Evidence to date suggests that RA is an independent risk factor for heart failure (HF) (3), and that HF is a major contributor to the overall cardiovascular disease–related mortality in RA patients (4). However, the natural history of myocardial dysfunction leading to HF in RA has received little investigation.

Although some established risk factors for HF, such as hypertension and diabetes mellitus, do not appear to be increased in RA patients (5), other risk factors, such as ischemic heart disease and insulin resistance, are more prevalent (6). However, these traditional risk factors do not appear to account for all of the excess risk of HF in patients with RA (3). In the general population, elevated levels of cytokines, such as tumor necrosis factor α (TNFα) and interleukin-6 (IL-6), in both the serum and myocardium are predictive of, and associated with, the severity of HF (7). Furthermore, cardiac-restricted overexpression of TNFα in mice is associated with the spontaneous development of inflammatory myocarditis, leading to HF and death (8). Although there are no reports of myocardial cytokine expression in humans with RA, autopsy studies confirm the presence of inflammatory myocarditis and arteritis in a substantial fraction of patients (9). Furthermore, serum levels of TNFα and IL-6 in patients with RA are considerably higher than those in patients with HF (10, 11). Taken together, these data suggest that continuously high levels of systemic and/or myocardial inflammation may contribute to the increased rate of HF in RA.

Echocardiography and cardiac magnetic resonance imaging (MRI) have been successfully applied in the general population to identify early structural changes in the left ventricle (LV) that predate the development of clinically overt HF. These studies have demonstrated that an increase in LV mass in asymptomatic individuals is a potent predictor of incident HF, both systolic and diastolic, even in those who remain free of obstructive coronary artery disease at followup (12, 13). We hypothesized that the prevalence of increased LV mass and of enlarged LV chamber volumes, which are myocardial parameters typically associated with progression to clinical HF, would be higher in RA patients compared with controls. We therefore designed a comparative cohort study of LV structure and function in patients with RA in comparison with a control group of non-RA subjects from the Multi-Ethnic Study of Atherosclerosis (MESA) cohort (14). We selected cardiac MRI as the method for measuring LV structure and function, due to its superiority over echocardiography in terms of resolution, sensitivity, and reproducibility (15–17).

PATIENTS AND METHODS

Participants and enrollment.

RA participants.

The Evaluation of Subclinical Cardiovascular Disease and Predictors of Events in Rheumatoid Arthritis (ESCAPE RA) study is an ongoing cohort study investigating the prevalence, progression, and risk factors of subclinical cardiovascular disease in men and women with RA (18). The ESCAPE RA cohort was assembled with the use of the same inclusion and exclusion criteria (except for the criteria for RA diagnosis) as used in the MESA study, a population-based cohort study of subclinical cardiovascular disease that had similar objectives. Enrollment in the ESCAPE RA study required a diagnosis of RA (according to the American College of Rheumatology [formerly, the American Rheumatism Association] criteria for the classification of RA [19]) for ≥6 months and having an age between 45 and 84 years. Individuals with prevalent cardiovascular disease, defined as a self-reported history of physician-diagnosed myocardial infarction, HF, coronary artery revascularization, angioplasty, peripheral arterial disease or procedures, pacemaker, or defibrillator devices and current atrial fibrillation, were excluded. Additional exclusions were having a body weight exceeding 300 pounds (due to imaging equipment restrictions) and having undergone computed tomography scanning of the chest within 6 months prior to enrollment (in order to limit radiation exposure). Participants were recruited from among the patients being followed up at the Johns Hopkins Arthritis Center and by referral from local rheumatologists. The study was approved by the Johns Hopkins Hospital Institutional Review Board. All participants provided their written consent to participate prior to enrollment, occurring from October 2004 through May 2006.

Control participants.

Non-RA controls were participants of the Baltimore MESA cohort. The MESA study design and methods have been described previously (14). Briefly, participants in the MESA study were part of a multiethnic cohort of 6,814 subjects from 6 US communities who were enrolled in the study between 2000 and 2002, among whom 1,086 were enrolled at the Johns Hopkins (Baltimore) Field Center. A total of 788 of these subjects underwent cardiac MRI at baseline, with the remainder not being scanned due to contraindications to cardiac MRI. Thirteen control subjects who reported use of disease-modifying antirheumatic drugs (DMARDs), which are typically prescribed for the treatment of RA, were excluded. Among the remaining 775 subjects, 3 control subjects were frequency-matched, at random, to each RA patient based on sex, age, and ethnicity.

Cardiac MRI assessments.

All consenting MESA participants without contraindications underwent cardiac MRI at the Johns Hopkins Cardiac Magnetic Resonance Research Suite. Contraindications to cardiac MRI included having a mobile implanted metallic device, presence of metallic particles in the eye, exposure to metalworking or welding in the past and absence of available skull radiographs, and claustrophobia. A subset of the ESCAPE RA participants (predefined as 40% of the RA patient sample) were also scanned with cardiac MRI, using the same equipment, protocols, and personnel as used in the MESA study. RA patients were selected at random for scanning, without consideration of clinical characteristics. The characteristics of the subset of patients undergoing cardiac MRI scanning were not significantly different from those of the full ESCAPE RA cohort.

Electrocardiography-triggered contiguous breath-hold gradient-echo cine imaging was performed on 1.5T-magnet scanners (GE Healthcare, Piscataway, NJ or Siemens Medical Systems, Issaquah, WA) as previously described (20). Calibration studies revealed no significant differences in the measurement of LV mass or volumes between scanners. Cine imaging was used to determine volumes (end systolic and end diastolic) and LV mass, with imaging data analyzed using commercially available software (MASS, version 4.2; Medis, Leiden, The Netherlands) by MESA-trained readers. Cardiac output was determined as follows: (end diastolic volume − end systolic volume) × heart rate. The ejection fraction was defined as the stroke volume divided by end diastolic volume. The MESA interreader intraclass coefficients for LV mass, end diastolic volume, and end systolic volume were 0.97, 0.98, and 0.95, respectively (21).

Covariate assessments.

Clinical covariates.

The ESCAPE RA study utilized questionnaires, equipment, methods, and quality control procedures identical to those used in the MESA study. The study coordinators were trained and certified by MESA trainers.

Information on demographics, smoking, and family history was collected from health questionnaires. Resting blood pressure (BP), determined with subjects placed in a seated position, was measured 3 times, and the mean value of the last 2 measurements was used in the present analysis. Hypertension was defined as a systolic BP of ≥140 mm Hg, diastolic BP of ≥90 mm Hg, or use of antihypertensive medications. Diabetes was defined as a fasting serum glucose level of ≥126 mg/dl or use of antidiabetic medications. Physical activity was assessed using a modified 7-Day Physical Activity Recall questionnaire (22). Body mass index (BMI) was calculated as the weight (in kg) divided by the height squared (in m²). Body surface area was calculated using the following formula: 0.20247 × ([height in meters]^0.725 × [weight in pounds × 0.4536]^0.425). Prescription and over-the-counter medications taken within the preceding 2 weeks were documented from containers supplied by the participant.

All participants underwent cardiac multidetector computed tomography scanning, using methods as described previously (23). Coronary calcification was quantified using the Agatston method (24), with a phantom of known calcium density scanned, along with the participant, to ensure standardization of the findings across scans.

RA-specific covariates.

In RA patients, 44 joints were examined by a single trained assessor for swelling, tenderness, deformity, and surgical replacement or fusion. RA disease duration was calculated based on self-reported duration of disease from the time of diagnosis by a physician. RA activity was calculated using the Disease Activity Score in 28 joints (DAS28) with C-reactive protein (CRP) level (25). Functional limitation was assessed with the Stanford Health Assessment Questionnaire (HAQ) (26). Current and past use of glucocorticoids and of biologic and nonbiologic DMARDs was ascertained by patient interview. Single-view, anteroposterior radiographs of the hands and feet were obtained and scored for radiographic damage of RA using the Sharp/van der Heijde method (27), performed by a single, trained radiologist who was blinded to the patients' characteristics. For 5 subjects with incomplete radiographic assessments, the missing score (hand or foot) was imputed from the available data, based on a regression equation that utilized data from the remaining subjects in the cohort.

Laboratory covariates.

Fasting samples of serum and plasma were separated by centrifugation and stored at −70°C. All assays (except those for RA autoantibodies) were performed at MESA-designated laboratories using the MESA quality control procedures. High-sensitivity CRP and IL-6 levels were measured as previously described (28). Low-density lipoprotein cholesterol was estimated in those plasma specimens having a triglyceride level of <400 mg/dl, with estimations made using the Friedewald equation. RA-specific autoantibodies and rheumatoid factor (RF) were defined as positive at a concentration of ≥60 units, while anti–cyclic citrullinated peptide (anti-CCP) antibodies were defined as positive at ≥40 units.

Statistical analysis.

Summary statistical data for outcome and predictor variables were calculated for the RA and control groups, with results expressed as the mean ± SD for normally distributed continuous variables, the median (interquartile range) for nonnormally distributed continuous variables, and the counts and percentages for categorical variables. Differences in mean values between the RA and control groups were compared using t-tests. When departures from normality were detected in the continuous variables, differences in distributions were compared with the Kruskal-Wallis test. Differences in proportions for the categorical variables were compared using the chi-square goodness-of-fit test or Fisher's exact test.

Linear regression was used to explore the association of MRI-derived cardiac measures with RA status. First, in unadjusted models, only RA status was included as a predictor variable. Next, in multivariate models, sociodemographic and cardiovascular risk factors associated with both the exposure and outcome (or with strong biologic plausibility in relation to the outcome) at a level of significance of P less than 0.15 in univariate analyses were included as potential confounders, with adjusted mean values, 95% confidence intervals, and P values calculated for the groups according to RA status. The Shapiro-Wilk test was used to check for normality of distribution for the MRI outcomes in relation to the predictor variables, and no departures from normality were found. Variance inflation factors were calculated to ensure that variables with excessive collinearity were not comodeled. Differences in cardiac MRI measures by RA status were explored in subgroups defined by sex and age. Analysis of covariance (ANCOVA) was used to test interactions between sex and RA status with regard to cardiac MRI outcome measures.

Within the RA group, the associations of RA characteristics with cardiac MRI outcomes were explored initially in simple linear regression models, with the individual RA characteristic of interest as the only variable in the model. Multivariate linear regression was then used to adjust for pertinent demographic and cardiovascular risk factors. Finally, a complex model including all of the RA characteristics identified as significant at P less than 0.15 from the adjusted models was constructed for each cardiac MRI outcome. The potential for interaction of RA characteristics with significant associations with the cardiac MRI outcome from the complex model was explored in stratified analyses. ANCOVA was used to test the statistical significance of potential interactions.

All statistical calculations were performed using Intercooled Stata 9 (StataCorp, College Station, TX). In all tests, a 2-tailed α value of 0.05 was defined as the level of statistical significance.

RESULTS

Participant characteristics.

Seventy-five RA patients and 225 frequency-matched non-RA controls were studied. The characteristics of the participants according to RA status are summarized in Table 1. Compared with non-RA controls, the mean systolic BP, diastolic BP, and heart rate were higher in RA patients. The frequency of diabetes and current smoking did not significantly differ by RA status. Although the median levels of CRP did not differ between the groups, the median levels of IL-6 were significantly higher in RA patients. Although the prevalence of any coronary calcium did not differ between the groups, the median Agatston calcium scores among participants with any coronary calcium were higher in RA patients than in controls (P = 0.023), which is consistent with the results previously reported for the complete ESCAPE RA cohort (18).

Table 1. Characteristics of the participants according to rheumatoid arthritis (RA) status*

Characteristic	RA (n = 75)	Control (n = 225)	P
* Except where indicated otherwise, values are the mean ± SD. HDL = high-density lipoprotein; LDL = low-density lipoprotein; IQR = interquartile range; anti-CCP = anti–cyclic citrullinated peptide; CRP = C-reactive protein; HAQ = Health Assessment Questionnaire; NSAID = nonsteroidal antiinflammatory drug; DMARD = disease-modifying antirheumatic drug; TNF = tumor necrosis factor. † For patients with a coronary calcium score >0 only. ‡ For current or past users only.
Demographic
Age, years	59 ± 9	59 ± 9	0.87
Female sex, no. (%)	39 (52)	117 (52)	1.00
Caucasian, no. (%)	66 (88)	198 (88)	1.00
Body surface area, m^2	1.88 ± 0.22	1.91 ± 0.22	0.37
Cardiovascular risk factor
Diabetes, no. (%)	3 (4.0)	13 (5.8)	0.77
Hypertension, no. (%)	41 (54.7)	94 (41.8)	0.052
Systolic blood pressure, mm Hg	129 ± 17	123 ± 18	0.012
Diastolic blood pressure, mm Hg	77 ± 9	71 ± 10	<0.001
Use of antihypertensive medicine, no. (%)	27 (36.0)	72 (32.0)	0.52
Heart rate, beats per minute	70 ± 11	65 ± 11	0.003
Current smoking, no. (%)	12 (16.0)	28 (12.4)	0.43
HDL cholesterol, mg/dl	55 ± 20	51 ± 14	0.12
LDL cholesterol, mg/dl	116 ± 33	120 ± 28	0.33
Triglycerides, mg/dl	113 ± 63	127 ± 75	0.11
Current statin use, no. (%)	10 (13.3)	37 (16.4)	0.52
Intentional exercise, median (IQR) minutes/day	21 (0–77)	32 (9–79)	0.081
CRP, median (IQR) mg/liter	2.0 (1.0–4.6)	1.7 (0.8–4.1)	0.35
Interleukin-6, median (IQR) pg/ml	2.4 (1.5–6.0)	1.1 (0.7–1.8)	<0.001
Coronary calcium score >0, no. (%)	42 (56.0)	114 (50.7)	0.42
Coronary calcium score, median (IQR)†	190 (104–540)	93 (22–279)	0.023
RA-specific feature
RA duration, median (IQR) years	7 (4–17)	–	–
Rheumatoid factor seropositivity, no. (%)	50 (66.7)	–	–
Anti-CCP seropositivity, no. (%)	53 (70.7)	–	–
Any HLA–DRB1 shared epitope alleles, no. (%)	58 (77.3)	–	–
Disease Activity Score in 28 joints, with CRP	3.51 ± 1.10	–	–
HAQ score, median (IQR)	0.63 (0–1.25)	–	–
Total modified Sharp score, median (IQR)	49 (19–110)	–	–
Current rheumatoid nodules, no. (%)	25 (33.3)	–	–
Current prednisone use, no. (%)	29 (38.7)	–	–
Cumulative prednisone, median (IQR) gm‡	5.0 (1.8–10.4)	–	–
Current NSAID use, no. (%)	49 (65.3)	–	–
Any current nonbiologic DMARD use, no. (%)	66 (88.0)	–	–
Current methotrexate, no. (%)	53 (70.7)	–	–
Any current biologic DMARD use, no. (%)	37 (49.3)	–	–
TNF inhibitors, no. (%)	36 (48.0)	–	–
Rituximab, no. (%)	1 (1.3)	–	–

Within the RA group, the median disease duration was 7 years, with 25% of patients having a disease duration of <4 years. The mean current RA disease activity, based on the DAS28 score, was in the moderate range. All but 1 RA patient had evidence of radiographic damage (i.e., a total modified Sharp score >0). Among the 66 patients (88%) treated with non-biologic DMARDs, 53 (80%) were receiving methotrexate. Among the 37 patients receiving current biologic treatment, 36 (97%) were receiving TNF inhibitors, 19 (51%) were receiving etanercept, 11 (30%) were receiving adalimumab, and 6 (16%) were receiving infliximab. The remaining patient had received a single treatment course of rituximab but had been treated with TNF inhibitors prior to this course.

Association of RA status with cardiac MRI measures of structure and function.

In univariate analyses of the combined RA and control groups, LV mass was positively and significantly associated with male sex, increasing BMI and body surface area, diabetes, hypertension, current smoking, increasing frequency of exercise, and prevalent coronary calcification and was inversely associated with increasing age, heart rate, and levels of high-density lipoprotein (HDL) cholesterol (summary data are available from the corresponding author upon request). There was no statistical interaction of RA status in relation to the association of any of these variables with LV mass (results not shown). Furthermore, in unadjusted analyses, the mean LV mass was 26 gm lower in the RA group compared with the control group (P < 0.001), representing an 18% difference (Table 2).

Table 2. Crude and adjusted cardiac magnetic resonance imaging (MRI) measures of cardiac structure and function according to rheumatoid arthritis (RA) status*

Cardiac MRI measure	Unadjusted			Adjusted†
	RA	Control	P	RA	Control	% difference‡	P
* Values are the mean (95% confidence interval). LVM = left ventricular mass; LVMI = left ventricular mass index; BSA = body surface area; LVEDV = left ventricular end diastolic volume; LVEF = left ventricular ejection fraction; LVSV = left ventricular stroke volume; LVSVI = left ventricular stroke volume index; LVEDVI = left ventricular end diastolic volume index; LVESV = left ventricular end systolic volume; LVESVI = left ventricular end systolic volume index. † Adjusted for systolic and diastolic blood pressures, heart rate, levels of high-density lipoprotein cholesterol and triglycerides, habitual exercise, and coronary calcium score. ‡ Percent difference in the adjusted mean values of the indicated variable for the RA group compared with the control group.
Mass
LVM, gm	120 (111–129)	146 (141–150)	<0.001	119 (112–127)	145 (141–149)	−17.9	<0.001
LVMI, gm/m^2 BSA	63 (60–67)	76 (74–78)	<0.001	64 (61–67)	75 (74–77)	−14.7	<0.001
LVM/LVEDV, gm/ml	0.99 (0.93–1.06)	1.16 (1.13–1.19)	<0.001	0.98 (0.92–1.04)	1.16 (1.13–1.19)	−15.5	<0.001
Ejection fraction
LVEF, %	66 (64–67)	68 (67–69)	0.006	65 (64–67)	68 (67–69)	−4.4	0.003
Cardiac output, liter/minute	5.5 (5.2–5.8)	5.6 (5.4–5.8)	0.58	5.3 (5.0–5.6)	5.7 (5.5–5.9)	−7.7	0.034
Stroke volume
LVSV, ml	80 (75–84)	87 (84–89)	0.012	80 (75–85)	86 (84–89)	−7.0	0.019
LVSVI, ml/m^2 BSA	42 (40–44)	46 (44–47)	0.012	43 (41–45)	45 (44–47)	−5.6	0.041
End diastolic volume
LVEDV, ml	122 (115–130)	128 (124–132)	0.18	123 (116–131)	127 (124–131)	−3.2	0.34
LVEDVI, ml/m^2 BSA	65 (62–68)	67 (65–69)	0.23	66 (63–69)	67 (65–68)	−1.2	0.67
End systolic volume
LVESV, ml	40 (36–44)	42 (39–44)	0.66	41 (38–45)	41 (39–43)	–	0.92
LVESVI, ml/m^2 BSA	21 (20–23)	22 (21–23)	0.92	22 (20–24)	22 (20–22)	–	0.56

After adjusting for the clinical characteristics that differed by RA status (systolic and diastolic BPs, heart rate, HDL cholesterol and triglyceride levels, exercise frequency, and coronary calcium scores), the mean LV mass remained 26 gm lower in the RA group (P < 0.001 versus controls). Further adjustment for differences in IL-6 or CRP levels did not change this association of LV mass with RA (results not shown). Similar differences between the groups were observed for LV mass indexed by body surface area and LV mass indexed by LV end diastolic volume (Table 2). Furthermore, a significantly lower mean LV mass was observed in both men and women with RA compared with non-RA male and female controls (Figure 1), although the relative reduction in mean LV mass was greater in men with RA than in women with RA when compared with their respective controls (20% and 14% lower than in male and female controls, respectively) (P for heterogeneity = 0.028). In both sexes, the mean LV mass declined with advancing age, regardless of RA status (Figure 2).

Figure 1. Adjusted mean left ventricular mass as determined by cardiac magnetic resonance imaging in patients with rheumatoid arthritis (RA) compared with non-RA controls, in the combined group and by sex. Bars show the calculated adjusted mean values and 95% confidence intervals after adjustment for systolic and diastolic blood pressure, heart rate, high-density lipoprotein cholesterol levels, triglyceride levels, frequency of exercise, and coronary calcium score. Values beside the symbols are the mean, while percentage values are the percent difference in adjusted mean values between the corresponding groups.

Figure 2. Linear association of left ventricular mass with age in patients with rheumatoid arthritis (RA) (solid red line) compared with non-RA controls (solid blue line) among men (A) and women (B). Broken lines indicate the 95% confidence intervals around the least-squares estimates of the linear association of left ventricular mass with age.

After adjustment for the same confounders as noted above, the mean LV ejection fraction, cardiac output, and LV stroke volume were modestly, but significantly, lower in the RA group than in the control group. In contrast, the adjusted mean LV end diastolic and end systolic volumes (as well as these measures indexed by body surface area) did not differ significantly according to RA status (Table 2).

Association of RA characteristics with measures of cardiac structure and function.

In analyses adjusted only for body surface area (model 1 in Table 3), the RA characteristics of increasing anti-CCP antibody level and use of biologic DMARDs showed a significant inverse association with measures of cardiac structure, particularly with lower mean LV stroke and end diastolic volumes. The association with anti-CCP antibody titer remained significant after adjusting for the demographic and cardiovascular risk factors that had the strongest associations with the outcomes in RA patients (age, sex, systolic blood pressure, use of antihypertensive medications, and current smoking) (models 2 and 3 in Table 3). However, after these adjustments were made, the association of biologic DMARD use with lower mean LV mass was only of borderline significance.

Table 3. Crude and adjusted associations of selected rheumatoid arthritis (RA) characteristics with cardiac magnetic resonance imaging (MRI) measures of left ventricular structure and function*

Outcome, characteristic	Adjusted model†
	Model 1		Model 2		Model 3
	β	95% CI	β	95% CI	β	95% CI
* The RA characteristics tested and found to have no significant associations with cardiac MRI outcomes included RA duration, rheumatoid factor seropositivity, presence of the HLA–DRB1 shared epitope, the Disease Activity Score in 28 joints, the Health Assessment Questionnaire global health score, the total modified Sharp score, physician global assessment of disease activity, presence of rheumatoid nodules, log C-reactive protein and interleukin-6 levels, current prednisone use, current prednisone dose, cumulative prednisone dose, use of nonbiologic disease-modifying antirheumatic drugs (DMARDs), and use of nonsteroidal antiinflammatory drugs (nonselective and cyclooxygenase 2 selective). 95% CI = 95% confidence interval; anti-CCP = anti–cyclic citrullinated peptide (see Table 2 for other definitions). † Model 1 = RA characteristics individually adjusted only for body surface area, except for the outcome of LVEF, in which only crude associations are shown; model 2 = RA characteristics individually adjusted for age, sex, body surface area, systolic blood pressure, antihypertensive medication use, and current smoking; model 3 = RA characteristics modeled simultaneously and adjusted for model 2 demographic and cardiovascular risk covariates. ‡ P < 0.05. § 0.10 > P < 0.05.
LVM
Anti-CCP, per 10 units	−0.35	−0.89, 0.20	−0.52	−0.97, −0.06‡	−0.46	−0.91, −0.01‡
Any biologic DMARD	−6.54	−14.69, 1.62	−6.69	−13.48, 0.10§	−5.75	−12.44, 0.95§
LVEF
Anti-CCP, per 10 units	−0.005	−0.02, 0.02	−0.02	−0.22, 0.19	−0.01	−0.33, 0.20
Any biologic DMARD	−0.06	−3.15, 3.04	−0.58	−3.60, 2.44	−0.23	−3.35, 2.90
LVSV
Anti-CCP, per 10 units	−0.54	−0.97, −0.10‡	−0.55	−1.00, −0.10‡	−0.48	−0.92, −0.04‡
Any biologic DMARD	−7.32	−13.98, −0.67‡	−8.37	−15.07, −1.68‡	−7.39	−13.97, −0.82‡
LVEDV
Anti-CCP, per 10 units	−0.87	−1.52, −0.21‡	−0.84	−1.53, −0.16‡	−0.75	−1.42, −0.07‡
Any biologic DMARD	−11.04	−21.05, −1.02‡	−11.80	−22.02, −1.58‡	−10.28	−20.31, −0.24‡

Importantly, neither RF seropositivity, increasing RF titer, nor the presence of the HLA–DRB1 shared epitope was associated with decreases in measures of cardiac structure (results not shown). Likewise, other measures of RA disease activity (the DAS28, physician global assessment), disease severity (e.g., total modified Sharp score, HAQ score), and systemic inflammation (log CRP and IL-6 levels) and use of other treatments (nonbiologic DMARDs, current or cumulative use of glucocorticoids, and nonsteroidal antiinflammatory drugs) were not significantly associated with the mean values for LV structure (results not shown). Despite their associations with measures of cardiac structure, anti-CCP antibody level and biologic DMARD use were not significantly associated with measures of cardiac function (i.e., the LV ejection fraction or cardiac output), in either crude or adjusted analyses.

Because of the potential association of biologic DMARD use with anti-CCP antibody level, models that simultaneously included both of these variables were constructed, with adjustment for demographic and cardiovascular risk factors (model 3 in Table 3). No substantial differences in the associations were seen in this combined model when compared with the model including these variables separately (i.e., model 2 in Table 3). Models assessing the combined associations of these variables in mutually exclusive groups that were defined by current use of biologic agents and anti-CCP antibody levels above or below the median value (128 units) demonstrated additive effects, without evidence of significant interactions (i.e., P > 0.05 for all tests of statistical interaction) between current biologic DMARD use and anti-CCP antibody levels (Figure 3). The number of RA patients with former, but not current, exposure to biologic agents (n = 4) was not sufficient to analyze this group separately. However, excluding these 4 patients from the analyses did not significantly alter the associations of current biologic use with measures of LV structure (results not shown).

Figure 3. Adjusted individual and combined associations of anti–cyclic citrullinated peptide (anti-CCP) antibody level and use of biologic disease-modifying antirheumatic drugs (DMARDs) with cardiac magnetic resonance imaging measures of left ventricular (LV) structure and function in patients with rheumatoid arthritis. Adjustment was made for body surface area, age, sex, systolic blood pressure, antihypertensive medication use, and current smoking. Bars show the mean and 95% confidence intervals for the adjusted mean values of LV mass (in gm) (A), stroke volume (in ml) (B), and end diastolic volume (in ml) (C) in patients with anti-CCP titers below the median value (set at 128 units) and no current use of biologic DMARDs (open squares), anti-CCP titers below the median value and current use of biologic DMARDs (solid squares), anti-CCP titers above the median value and no current use of biologic DMARDs (open diamonds), or anti-CCP titers above the median value and current use of biologic DMARDs (solid diamonds). The solid horizontal line indicates the adjusted mean value in the control group, while the broken lines indicate the 95% confidence limits for the adjusted mean value in the control group. Percentage values indicate the percent difference between individual subgroups.

DISCUSSION

To our knowledge, this study is the first to utilize cardiac MRI to investigate cardiac structure and function in RA. We found that the mean LV mass was strikingly lower in RA patients compared with controls, even after accounting for demographic and cardiovascular risk factors, with differences more pronounced in men than in women. Two RA characteristics, increasing titer of anti-CCP antibodies and use of biologic DMARDs, were significantly and independently associated with a lower adjusted mean LV mass, stroke volume, and end diastolic volume, but not with ejection fraction. Other measures of RA disease activity and severity, systemic inflammation markers, and use of other RA therapies were not associated with cardiac measures.

Our finding of a lower LV mass in RA patients was unexpected and contrary to our original hypothesis that RA patients without clinical cardiovascular disease would exhibit myocardial features that antedate clinical HF in the general population, such as increased LV mass, LV chamber dilation, and decreased ejection fraction (29, 30). Although the etiology and implications of our finding of a lower LV mass in RA patients are not readily apparent, it is tempting to speculate that the lower LV mass observed in RA patients could be related to chronic myocarditis and/or microvascular hypoperfusion resulting in myocyte loss and/or fibrosis. In support of this notion, Raza et al (31) reported a case of microvascular dysfunction in an RA patient that was reversed with immunosuppression. Alternatively, a lower LV mass could reflect reduced physical conditioning in RA patients, although this is less likely, since adjustment for exercise frequency did not alter the relationship between RA status and LV mass. While macrovascular ischemia might contribute to myocardial abnormalities (32), adjustment for coronary calcification did not alter the differences in myocardial measures between RA patients and controls.

Our findings raise the hypothesis that RA-associated HF may progress along a different preclinical pathway than HF that is attributable to coronary artery disease or chronic hypertension. Longitudinal followup and histopathologic studies are needed to evaluate these etiologic possibilities. Our observation that age-associated declines in LV mass were exaggerated in RA patients compared with controls lends credence to the concept of the RA disease state as one of accelerated aging, in which the myocardia of RA patients leads to these subjects being effectively older than their non-RA counterparts.

Two categories of HF are recognized clinically. Systolic HF is characterized by contractile dysfunction with eccentric LV remodeling, ventricular dilation, reduced ejection fraction, and a decreased mass:volume ratio. In contrast, nonsystolic (or nondiastolic) HF is characterized by concentric remodeling, resulting in a stiff ventricle with relatively preserved ejection fraction and chamber volumes. It has been suggested that diastolic dysfunction is a precursor to both types of HF (33). Several 2-dimensional echocardiographic studies have explored LV function in RA, with the findings in most of these studies indicating increased diastolic dysfunction with preserved systolic function, while a minority of studies show systolic dysfunction (for review, see refs.34 and35). Although our cardiac MRI sequences did not allow for investigation of diastolic function, our results do suggest that systolic function may be very modestly reduced in asymptomatic patients with RA.

A recently reported study by Rudominer et al (36) evaluated LV mass in RA patients compared with controls. In contrast to our finding of a lower mean LV mass index, Rudominer et al detected a higher mean LV mass index in their RA group compared with a control group matched for demographic characteristics and hypertension status. There are several potential explanations for this discrepancy. First, the RA populations studied are quite dissimilar. The cohort described by Rudominer et al was considerably younger than ours (mean age 47 years), and their patients were almost exclusively female (99%). Because differences in LV mass between RA patients and controls in our study were more pronounced in men and increased with age in women, both the absence of men and the absence of older women in the study by Rudominer et al could have resulted in exclusion of these preferentially affected individuals.

However, the more likely explanation for the differences between studies is the fact that different imaging techniques were utilized. The formulas utilized by Rudominer et al for calculating LV mass using M-mode and 2-dimensional echocardiography extrapolate measures of myocardial thickness and chamber size obtained in 1 or 2 planes, which relies on assumptions of ventricular morphology and wall thickness. This method overestimates the LV mass in the setting of eccentric LV hypertrophy (37, 38). Thus, bias in comparisons might be introduced if eccentric morphology is more prominent in the RA group. This was noted in the study by Rudominer et al, in which 15 of 16 patients with LV hypertrophy were found to have eccentric remodeling. Use of cardiac MRI avoids this potential error, since measurement of ventricular mass is 3-dimensional and does not rely on assumptions of myocardial morphology. However, additional studies in a larger group of RA patients will be required to reconcile the divergent findings between our cardiac MRI study and the echocardiographic study by Rudominer et al.

Although our initial hypothesis was that long-term elevation of inflammatory cytokine levels would result in greater LV mass (hypertrophy) in the RA group, we found no significant relationship between current levels of inflammation and measures of LV structure or function. However, because of wide variations in the CRP and IL-6 levels over time in RA patients, particularly in association with treatment, current levels may not be reflective of cumulative exposure. Longitudinal studies will be needed to evaluate the cumulative effect of cytokines on the myocardium.

Interestingly, treatment with biologic agents (predominantly, anti-TNF therapy) and increasing titer of anti-CCP antibodies were associated with a lower LV mass and smaller LV volumes in the RA group. It is possible that these variables are a proxy for RA disease severity, and that disease severity accounts for the cardiac abnormalities observed. However, we did not observe similar associations with other RA severity measures, such as presence of RF, presence of the shared epitope, or degree of radiographic joint damage. There is biologic plausibility in the notion that both use of biologic agents and the anti-CCP antibody titer can mediate the observed outcomes. In experimental animal models, the augmentation of myocardial TNFα levels induces LV hypertrophy and dilation (39), and anti-TNF treatment limits these changes. Thus, anti-TNF therapy could be protective in RA by reducing the pathogenic effects of TNFα. However, a minimum level of TNFα may actually be required to maintain myocardial structural homeostasis, as suggested by studies in another mouse model in which myocardial TNFα is rendered completely inactive (40). Such mouse models demonstrate restrictive cardiomyopathy and dramatically reduced ventricular volumes.

Thus, anti-TNF therapy in RA patients may reduce TNFα levels below those required to maintain normal myocardial structure. The cross-sectional nature of our study does not enable us to distinguish between these 2 possibilities. Incidentally, studies in humans receiving anti-TNF therapy have yielded conflicting results. In subjects without RA in whom advanced HF developed, TNF inhibition was not effective in reducing morbidity or mortality (41, 42), and there was a suggestion of worsening in one trial (41). In RA patients, case reports and case series of new-onset or worsening HF after anti-TNF therapy have been presented (43). In contrast, population-based studies of RA patients have demonstrated lower rates (44), similar rates (45), and higher rates (46) of incident HF in anti-TNF–treated RA patients compared with RA patients who did not receive such treatment. Thus, definitive conclusions on the effects of TNF inhibitors on the myocardium will require a prospective interventional study. Furthermore, whether a reduction in LV mass and ventricular volumes is ultimately protective or pathogenic in RA myocardium will also require extended observations.

Anti-CCP antibodies were also a potent predictor of a lower LV mass and ventricular chamber volumes in our RA patients, suggesting a potential pathophysiologic link to HF. Anti-CCP antibodies are directed against citrullinated proteins such as vimentin and fibronectin. Interestingly, antivimentin and antifibronectin antibodies have been identified and associated with induction of inflammatory myocarditis and with cardiac transplant rejection in animals (47, 48) and were predictive of transplant rejection in humans (49). However, antibodies against the citrullinated homologs were not investigated. Citrullination of skeletal muscle has been demonstrated in patients with rheumatic disease (50), but we are not aware of any studies that have investigated citrullination in rheumatoid myocardium. It is tempting to speculate that anti-CCP antibodies may be directed against myocardial proteins, and that these immune processes are involved causally in the cardiac abnormalities observed in our study.

There are notable limitations to our study. Despite being the first study to utilize cardiac MRI in RA patients, the sample of RA patients was relatively small. However, because differences between RA patients and controls in the outcome measures were generally large, with limited variability, even small differences were detectable within the bounds of statistical significance. We were, however, limited in our ability to perform many subgroup analyses. We plan to incorporate additional patients into future studies in order to validate these findings. In addition, our cross-sectional design does not permit conclusive inference regarding cause and effect. Longitudinal followup with repeat cardiac MRI of these patients is under way to identify predictors of change in myocardial measures.

In summary, RA patients demonstrated a markedly lower LV mass and modest decreases in ejection fraction, cardiac output, and stroke volume compared with non-RA controls. Differences in LV mass between RA patients and controls were greater in men than in women. A higher titer of anti-CCP antibodies and the use of biologic agents were associated with a lower LV mass and with lower LV stroke and end diastolic volumes, after accounting for pertinent confounders. Further investigation is under way to better understand the implications of these findings in the development of clinical HF in RA.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Bathon had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Giles, Fernandes, Post, Bluemke, Szklo, Petri, Brumback, Lima, Bathon.

Acquisition of data. Giles, Malayeri, Fernandes, Bluemke, Szklo, Brumback, Lima, Bathon.

Analysis and interpretation of data. Giles, Malayeri, Fernandes, Post, Blumenthal, Bluemke, Vogel-Claussen, Szklo, Petri, Gelber, Brumback, Lima, Bathon.

Acknowledgements

We are indebted to the ESCAPE RA study staff, Marilyn Towns, Michelle Jones, Patricia Jones, Marissa Hildebrandt, and Shawn Franckowiak, for their dedication and hard work, and to the staffs of the field center of the Baltimore MESA cohort and the MESA Coordinating Center at the University of Washington, Seattle. We particularly thank Drs. Uzma Haque, Clifton Bingham III, Carol Ziminski, Jill Ratain, Ira Fine, Joyce Kopicky-Burd, David McGinnis, Andrea Marx, Howard Hauptman, Achini Perera, Peter Holt, Alan Matsumoto, Megan Clowse, and Gordon Lam, among others, for generously recommending their patients for this study.