Abstract
Objective. The prevalence of heart failure is increased 2-fold in patients with rheumatoid arthritis (RA); this is not explained by ischemic heart disease or other risk factors for heart failure. We hypothesized that in patients with RA without known heart disease, cardiac magnetic resonance imaging (cMRI) would detect altered cardiac structure, function, and fibrosis.
Methods. We performed 1.5-T cMRI in 59 patients with RA and 56 controls frequency-matched for age, race, and sex, and compared cMRI indices of structure, function, and fibrosis [late gadolinium enhancement (LGE), native T1 mapping, and extracellular volume (ECV)] using Mann-Whitney U tests and linear regression, adjusting for age, race, and sex.
Results. Most patients with RA had low to moderate disease activity [28-joint count Disease Activity Score–C-reactive protein median 3.16, interquartile range (IQR) 2.03–4.05], and 49% were receiving anti-tumor necrosis factor agents. Left ventricular (LV) mass, LV end-diastolic and -systolic volumes indexed to body surface area, and LV ejection fraction and left atrial size were not altered in RA compared to controls (all p > 0.05). Measures of fibrosis were not increased in RA: LGE was present in 2 patients with RA and 1 control subject; native T1 mapping was similar comparing RA and control subjects, and ECV (median, IQR) was lower (26.6%, 24.7–28.5%) in patients with RA compared to control subjects (27.5%, 25.4–30.4%, p = 0.03).
Conclusion. cMRI measures of cardiac structure and function were not significantly altered, and measures of fibrosis were similar or lower in RA patients with low to moderate disease activity compared to a matched control group.
Rheumatoid arthritis (RA) is a chronic inflammatory disease associated with premature mortality caused largely by cardiovascular (CV) disease1,2. Initial research efforts focused on the increased risk of ischemic heart disease in RA3,4,5; however, heart failure is a major understudied problem that accounts for a substantial portion of the increased CV mortality6. Prevalence of heart failure is increased in RA with a relative risk of 1.6 to 2.01,7,8, and the outcomes are worse9.
The pathogenesis of heart failure in RA is not known. Although the prevalence of some heart failure risk factors (e.g., ischemic heart disease)10 is increased in RA, increased risk of heart failure is independent of conventional risk factors7,11. Thus, factors more specific to RA, such as inflammation, have been implicated11. In keeping with this idea, indicators of RA severity and disease activity are associated with increased risk of heart failure2,7,12.
Concentrations of aminoterminal prohormone brain-type natriuretic peptide (NT-proBNP), a sensitive marker of myocardial stretch, for which even minimally increased concentrations are associated with future risk of heart failure and CV events13,14, were significantly higher in RA compared with controls15,16. Moreover, NT-proBNP concentrations were associated with RA disease activity, anti-tumor necrosis factor (anti-TNF), interleukin 6, and C-reactive protein (CRP) concentrations, but not coronary atherosclerosis15. Further, concentrations of troponin I, a marker of myocardial necrosis measured by a high-sensitivity assay, were also significantly higher in patients with RA than control subjects16,17.
The findings of increased concentrations of markers of myocardial stress and damage in cross-sectional studies of patients with RA suggest the possibility of ongoing subtle subclinical structural and functional myocardial dysfunction. Echocardiographic findings in RA have generally shown preserved systolic function but increased prevalence of diastolic dysfunction1,18. Cardiac magnetic resonance imaging (cMRI), however, provides additional information about myocardial structure, as well as fibrosis and inflammation. An early cMRI study suggested smaller left ventricular (LV) size in RA19, and several other small studies suggested that some patients with RA have focal scarring detected by late gadolinium enhancement (LGE), and diffuse fibrosis detected by T1 mapping and increased extracellular volume (ECV)20,21,22,23,24. However, these studies have been small and included few patients receiving biologic therapy.
We hypothesized that patients with RA have altered cardiac structure and function, and increased myocardial fibrosis detectable by cMRI, and that this is related to underlying inflammation.
MATERIALS AND METHODS
Study population
We performed a cross-sectional study of 59 patients with RA and 56 controls, frequency-matched for age (within 2 yrs), race, and sex. Inclusion criteria consisted of age ≥ 18 years, the ability to provide informed consent, meeting of ACR classification criteria for RA25 (RA subjects), and no inflammatory disease (control subjects). Exclusion criteria included previous or current heart failure or ischemic CV disease (e.g., stroke, myocardial infarction, angina, prior coronary artery bypass grafting, or percutaneous coronary intervention), atrial fibrillation, known structural or functional cardiac abnormality including pulmonary hypertension, an estimated glomerular filtration rate < 60 ml/min, pregnancy or breastfeeding, inability to undergo cMRI, and hypersensitivity to gadolinium. Participants were recruited from the Vanderbilt Clinic Rheumatology practice, responses to advertisements, and word of mouth. This study was approved by the Vanderbilt Institutional Review Board (IRB#120314) and registered with ClinicalTrials.gov (#NCT01589770). All subjects gave written informed consent.
Clinical and laboratory data
Clinical details and a cumulative medication history with particular attention to RA therapies and CV drugs were obtained from participants and the electronic medical record, as we have previously done4. Tender and swollen joint counts were measured in patients with RA. Fasting venous blood was drawn, and erythrocyte sedimentation rate (ESR) and high sensitivity CRP were measured in the hospital clinical laboratory. RA disease activity was measured by 28-joint count Disease Activity Score (DAS28)26.
Details of cMRI scanning
Study participants underwent cMRI using a 1.5 T Siemens Magnetom Avanto scanner (Siemens Healthcare Sector). Subjects were scanned using a phased array torso receiver coil, and imaging protocols included cine imaging for ventricular structure and function, native (without contrast) and postcontrast modified Look-Locker (MOLLI) imaging for T1 mapping, and LGE imaging for myocardial fibrosis and inflammation. Cine imaging was performed using steady-state free-precession sequences aligned to the horizontal and vertical long axis of the heart. Typical acquisition settings for cine images were the following: field of view 300 × 340 mm, matrix 156 × 192, slice thickness 8 mm, flip angle 80°, echo time 1.1 ms, and usually 30 phases per cardiac cycle to maintain repetition time below 50 ms. Parallel imaging was done using the generalized autocalibrating partially parallel acquisition technique with an acceleration factor of 2. For T1 mapping, MOLLI images were obtained in 3 LV short axis (base, mid, and apical) planes before contrast injection, and 12 and 25 min after contrast injection. Gadolinium [0.15 mmol/kg body weight gadopentetate dimeglumine (Magnevist, Bayer HealthCare Pharmaceuticals)] was injected intravenously through an antecubital vein. The timing of the contrast injection and MOLLI sequences were chosen to be comparable with the Multi-Ethnic Study of Atherosclerosis27.
Phase-velocity encoded flow imaging was performed through-plane in the ascending aorta and the main pulmonary arteries as a second measure of LV and right ventricular (RV) output, and quantification of any valvular regurgitation. Twelve minutes after gadolinium injection, short- and long-axis myocardial LGE imaging was performed using both single-shot inversion recovery (IR) and phase-sensitive IR true fast imaging with steady-state precession imaging. The third set of MOLLI images were obtained at 25 min postcontrast, as detailed above.
Individuals specialized in cMRI (WB and JHS) performed the cMRI analyses of structure and function and were blinded to disease status. LV and RV measurements from cMRI were calculated from manually traced endocardial and epicardial end-diastolic and end-systolic contours from a stack of contiguous short-axis images from the apex to the base of the LV. Left and right atrial dimensions were obtained by caliper measurements of the major axis of the atria and calculated on a Leonardo workstation using Argus software version VB17 (Siemens). Measures of focal and diffuse fibrosis/inflammation, including the presence or absence of LGE, myocardial T1 mapping, and ECV calculations, were determined using commercial software (CVI42 version 5.3, Circle Cardiovascular Imaging) and by a third blinded specialist in cMRI (DAB, with the assistance of NKB and CL), who has led efforts to standardize image acquisition and developed techniques to minimize inter- and intraobserver variation for many large cMRI studies27,28,29,30. Additionally, LGE was assessed by WB, with consensus of all reads. ECV was calculated as ECV = (1-hematocrit) × (1/T1 myocardium postcontrast) − (1/T1 myocardium precontrast) / (1/T1 blood postcontrast) − (1/T1 blood precontrast)31.
Statistics
Our study was powered to detect differences in LV mass index and postcontrast T1 mapping comparing RA to control subjects, based on previous data from published studies32,33. Based on previously reported mean ± SD of 67.6 ± 12.6 g/m2 for LV mass index32 and 564 ± 103 ms for postcontrast T1 mapping33, ≥ 55 subjects in each group would provide about 80% power to detect a difference of ≥ 10% in both measures with a 2-sided significance of 5%. Data are expressed as median [interquartile range (IQR)].
The cMRI indices of structure, function, and fibrosis (LGE, T1 values, and ECV) were compared between RA and control subjects using Mann-Whitney U tests and linear regression, adjusting for age, race, and sex. Skewed variables were log-transformed to normalize residuals. Spearman correlation (r) was used to assess the relationship between cMRI indices and RA-specific variables.
RESULTS
Subject characteristics
Patients with RA (median age 53 yrs) and control subjects (median age 52 yrs) were of similar age, race (98% white in both), and sex (76% and 79% female in RA and controls, respectively; Table 1). Patients with RA had low to moderate disease activity (DAS28-CRP 3.16 units, IQR 2.03–4.05 units), and established disease (disease duration 10 yrs, IQR 5–15 yrs, range < 1 mo to 47 yrs). About 63% of patients with RA were taking methotrexate and 49% were taking an anti-TNF agent (Table 1).
Patient demographics. Values are n (%) or median (IQR) unless otherwise specified.
Myocardial structure and function
LV mass indexed to body surface area (BSA) was similar in patients with RA (43.8 g/m2, 40.0–49.5 g/m2) and control subjects (42.2 g/m2, 36.4–48.5 g/m2, p = 0.19; Table 2). Right atrial major axis dimension was significantly smaller in RA (30 mm, 25–34 mm) compared to control subjects (34 mm, 30–38 mm, p = 0.001). Heart rate was significantly higher in RA (72 bpm; 66–79 bpm) compared to control subjects (68 bpm, 59–73 bpm, p = 0.01). LV ejection fraction, end-diastolic volume, and end-systolic volume indexed to BSA did not differ significantly in RA and control subjects (Table 2). Patients with RA had lower RV end-diastolic volume indexed to BSA (58.3 ml/m2, 50.7–69.5 ml/m2) compared to control subjects (63.3 ml/m2, 59.1–72.3 ml/m2, p = 0.004; Table 2). Similarly, patients with RA had lower RV end-systolic volume indexed to BSA compared to control subjects [20.4 ml/m2 (15.7–27.9 ml/m2) vs 26.6 ml/m2 (20.8–33.6 ml/m2), p = 0.002; Table 2].
Cardiac MRI findings in RA and control subjects. Values are median (IQR) unless otherwise specified.
Myocardial fibrosis or inflammation
Assessing focal fibrosis or inflammation, 2 patients with RA (3%) and 1 control subject (2%) had LGE (Table 2). Among RA, 1 had subepicardial patchy LGE and the other had inferior RV insertion patchy LGE. The control subject had a subendocardial scar.
Assessing diffuse fibrosis or inflammation of the heart, native T1 mapping (p = 0.45) and postcontrast T1 values (25-min; p = 0.37) were not significantly different in patients with RA and control subjects (Table 2). ECV was significantly lower in patients with RA (26.6%, 24.7–28.5%) compared to control subjects (27.5%, 25.4–30.4%, p = 0.03), contrary to our hypothesis.
Native T1 value was weakly correlated with tender joint count (ρ = 0.29, p = 0.03), but not swollen joint count, CRP, patient-reported global health, or overall DAS28-CRP score (all p > 0.05). ECV was not correlated with tender or swollen joint counts, CRP, global health, or DAS28-CRP score (all p > 0.05). There was a trend for an inverse association between age and native T1 time in patients with RA (r = −0.25, p = 0.06); the opposite of what was observed in control subjects (r = 0.20, p = 0.15). Native T1 time did not correlate with duration of RA (r = −0.05, p = 0.74). Patients taking anti-TNF agents had lower native myocardial T1 and ECV than nonusers, but differences were not significant (Table 3).
Cardiac MRI findings in patients with RA based on anti-TNF use. Values are median (IQR) unless otherwise specified.
DISCUSSION
To our knowledge, our study represents one of the largest cMRI studies in RA and matched controls to date. Our findings were unexpected. Based on previous reports, we hypothesized there would be structural abnormalities and increased fibrosis or inflammation detected on cMRI by LGE (representing focal fibrosis or inflammation) or by increased ECV and native T1 time (representing diffuse fibrosis or inflammation) in RA. However, we found little LGE and no evidence of increased diffuse fibrosis or inflammation in patients with RA.
LGE cMRI can be used to assess myocardial tissue for scarring, fibrosis, and inflammation34 because the residence time for gadolinium in expanded interstitial space is prolonged compared to the intravascular space. LGE cMRI is helpful for identifying focal fibrosis rather than diffuse fibrosis, in which the myocardium may be uniformly abnormal with a lack of normal myocardial segments for comparison35. LGE cMRI can also detect inflammation of the heart, as in myocarditis, where it is present in active disease and resolves over time34, which is of interest in RA, where some have observed high rates of myopericarditis among patients with cardiac symptoms36.
Myocardial T1 before (native) and after gadolinium has emerged as a technique to measure diffuse fibrosis33. ECV adjusted for hematocrit is thought to more accurately represent the ratio of interstitial space to total myocardial volume, independent of field strength and gadolinium dose and clearance. ECV is increased in infiltrative states such as cardiac amyloid and with both interstitial fibrosis and replacement fibrosis37. ECV correlates with the collagen volume measured histologically37 and increases with age in healthy individuals, a trend in controls but not in RA. The significance of the small reduction in ECV observed in patients with RA is not known; although different from controls, the ECV of patients with RA was within normal limits32 and thus could reflect chance. Alternatively, lower ECV in RA could also represent loss of cardiac collagen, dense myocardial space or perhaps failure of typical repair mechanisms, or other uncharacterized alterations in the myocardium.
As also reported by others, heart rate was higher in patients with RA than controls, a finding that may be due to deconditioning38. We also observed a nonsignificant decrease in LV end-diastolic and end-systolic volumes, and significantly decreased RV end-diastolic and end-systolic volumes. Faster heart rates may be compensating for smaller hearts in RA to maintain cardiac output, which was preserved in RA.
The absence of major cardiac structural and functional alterations in RA and no increase in LGE or ECV suggest that the myocardium is not markedly abnormal in the setting of low to moderate disease activity. If the availability of more effective therapies and tighter control of disease activity prevent myocardial fibrosis or treat myocardial inflammation in RA, the effects of such therapy on the incidence of heart failure will be of great significance. This is consistent with recent work in atherosclerosis demonstrating that patients with RA (n = 139) with remission or low disease activity for ≥ 75% of the time over a 3-year period of followup had no acceleration of carotid intima-media thickness compared to matched control subjects (n = 139)39.
TNF may be particularly important in heart failure. In animal models of heart failure, circulating TNF levels were elevated, and blocking TNF was beneficial40,41,42. However, although humans with heart failure have high circulating TNF levels43, anti-TNF therapy had no benefit or even resulted in increased mortality44,45. Conversely, in some8,46,47,48,49 but not all50 large RA observational studies, anti-TNF therapy was associated with improved CV outcomes, including decreased heart failure. This suggests that although anti-TNF agents can worsen existing heart failure, they might decrease risk of heart failure in RA. Studies with other biologic agents will be of interest.
Differences between the findings of our current study, suggesting that the myocardium is not markedly abnormal in patients with low to moderate disease activity, many of whom were receiving an anti-TNF or another biologic agent, and the findings of other cMRI studies are informative (Table 4). Ntusi, et al performed cMRI in patients with RA and matched controls (n = 39 each)20, and found LGE in 46% of patients with RA and in no controls; ECV and native T1 time were also higher in RA, suggesting an increase in both focal and diffuse fibrosis. Patients were similar to those in our current study with regard to age, prevalence of rheumatoid factor, and DAS28 scores. However, median CRP concentrations were significantly higher in RA (9 mg/l) compared to controls (1 mg/l), whereas in our current study, CRP was similar in RA and controls (1.7 mg/l in both). Moreover, no anti-TNF use was reported, whereas 49% of RA patients in our current study were taking an anti-TNF agent. Thus, differences in findings between the 2 studies may be related to earlier and more aggressive therapy of RA. We examined this idea by stratifying cMRI measures based on any biologic use and anti-TNF use; patients using anti-TNF agents had lower native T1 and ECV, but these differences were not significant.
Comparison between current study and prior cardiac MRI studies.
Holmstrom, et al performed cMRI in 60 patients with RA (n = 31 newly diagnosed RA and n = 29 established RA about to start biologic therapy), 11 healthy controls, and 10 patients with fibromyalgia (FM)21. Patients had a median of 8 and 6 swollen joints in the early and chronic RA groups, respectively; this compares to a median of 1 swollen joint in our current study. LGE was present in 55% of patients with RA (68% in newly diagnosed RA and 41% in established RA). LGE was not measured in the healthy controls and was absent in FM. In RA, LGE was associated with higher DAS28-CRP. More patients with early RA (duration 0.4 yrs) than established RA (duration 13 yrs) had LGE, suggesting that it may not be due to fibrosis, which accumulates over time, but more likely represents inflammation or edema of the myocardium.
In an early study, 7 of 18 RA patients (39%) had LGE22. Compared to our current study, patients were slightly older and had shorter disease duration, higher DAS28-CRP scores, and higher ESR and CRP. Those with LGE had higher CRP, ESR, and DAS28-CRP scores. Of the 7 patients with LGE, 1 was taking an anti-TNF agent, whereas 6 of 11 patients without LGE were taking an anti-TNF. In another early study, LGE was higher in RA patients (n = 24, all with DAS28 > 7, and none on anti-TNF drugs) compared to healthy controls and patients with myocarditis23. Recently, another study showed that 19 of 60 (32%) patients with RA had LGE24. These patients had similar demographics compared to the earlier study, with slightly older patients who had shorter disease duration and disease activity compared to our current study. The authors found that those with LGE had higher disease activity (median DAS28 score 5.1) compared to those without (median DAS28 score 3.5). Comparing our current study and its findings with previous work suggests that key differences in patient populations (low RA disease activity, low concentration of CRP, low swollen joint count, and high prevalence of biologic therapy use) contributed to the low prevalence of LGE. Population differences in rate of focal fibrosis (LGE) by cMRI may also be important, because the rate is > 2-fold lower in the United States29 than in Iceland51 and Sweden52, which is probably a result of underlying risk factors in the population. Reports of LGE being particularly prominent in early RA suggest that inflammation may be more important than permanent fibrosis. Thus, patients with RA may have cardiac inflammation and LGE in the setting of high disease activity, and this could resolve with aggressive treatment. Further studies will be necessary to test this hypothesis.
Our study did have limitations. The cross-sectional features of the study preclude us from knowing whether aggressive treatment and control of disease explains the differences between our current study and some previous studies. Moreover, given that we studied primarily RA patients with low to moderate disease activity, we cannot extrapolate the findings of our study to patients with uncontrolled RA. Similarly, small numbers precluded us from evaluating whether anti-TNF use in the presence of persistently active disease was associated with abnormal cMRI findings.
In RA patients with low to moderate disease activity compared to a matched control group, cMRI measures of cardiac structure and function and fibrosis were not significantly altered.
Footnotes
This study was funded by the Arthritis Foundation Innovative Research Grant (Stein), Veterans Administration CDA IK2CX001269 (Ormseth), US National Institutes of Health (NIH) Grants: US National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS) K23 AR068443 (Ormseth), NIAMS P60 056116 (Stein), US National Heart, Lung and Blood Institute K23 HL123938 (Soslow), and Clinical and Translational Science Awards UL1TR000445 from the National Center for Advancing Translational Sciences (NCATS) and the NIH Intramural Research Program. The contents of this article are solely the responsibility of the authors and do not necessarily represent official views of the NCATS or the NIH.
- Accepted for publication January 3, 2018.
