The Impact of Rheumatoid Arthritis Disease Characteristics on Heart Failure

Abstract

Objective

To examine the impact of rheumatoid arthritis (RA) characteristics and antirheumatic medications on the risk of heart failure (HF) in RA.

Methods

A population-based incidence cohort of RA patients aged ≥18 (1987 ACR criteria first met between 1/1/1980 and 1/1/2008) without a history of HF was followed until HF onset (defined by Framingham criteria), death, or 1/1/2008. We collected data on RA characteristics, antirheumatic medications and cardiovascular (CV) risk factors. Cox models adjusting for age, sex and calendar year were used to analyze the data.

Results

The study included 795 RA patients (mean age 55.3 years, 69% females, 66% rheumatoid factor [RF] positive). During the mean follow-up of 9.7 years, 92 patients developed HF. The risk of HF was associated with RF positivity (HR 1.6, 95%CI 1.0, 2.5), erythrocyte sedimentation rate (ESR) at RA incidence (HR 1.6, 95%CI 1.2, 2.0), repeatedly high ESR (HR 2.1, 95%CI 1.2, 3.5), severe extra-articular manifestations (HR 3.1, 95%CI 1.9, 5.1) and corticosteroid use (HR 2.0, 95%CI 1.3, 3.2), adjusting for CV risk factors and coronary heart disease (CHD). Methotrexate users were half as likely to have HF as non-users (HR 0.5, 95%CI 0.3, 0.9).

Conclusion

Several RA characteristics and the use of corticosteroids were associated with HF adjusting for CV risk factors and CHD. Methotrexate use appeared to be protective against HF. These findings suggest an independent impact of RA on HF which may be further modified by antirheumatic treatment.

INTRODUCTION

Heart failure (HF) is a multifactorial clinical syndrome with poor prognosis representing a universal final stage of nearly every form of heart disease (1, 2). Patients with rheumatoid arthritis (RA) are at approximately 2-fold increased risk of HF compared with persons without RA, and this increased risk is not fully explained by traditional cardiovascular (CV) risk factors (3-5). Furthermore, RA patients appear to have more subtle HF presentation, yet significantly higher early mortality following HF compared with non-RA subjects (6). These findings suggest a potential role for RA-specific mechanisms in HF development and emphasize the need for understanding the determinants of HF in RA. RA disease activity and severity have been previously linked to HF development in RA (5, 7-10). Furthermore, antirheumatic medications, particularly methotrexate and more recently biologic response modifiers, but not corticosteroids, appear to have a beneficial effect on risk of HF in RA (11-13). However, the nature of these associations is unclear and longitudinal population-based studies analyzing the relative impact of the major potential contributors (CV risk factors, RA disease characteristics and antirheumatic medications) for HF development in RA are lacking. We sought to examine the impact of RA disease characteristics and antirheumatic medications on the risk of HF in RA.

METHODS

Study setting and design

This study was conducted using the population-based resources of the Rochester Epidemiology Project (REP) medical records linkage system (14). This system facilitates ready access to the complete inpatient and outpatient medical records of residents of Olmsted County, Minnesota, from all community medical providers including Mayo Clinic, its affiliated hospitals and the Olmsted Medical Center. REP resources ensure virtually complete ascertainment of all clinically recognized cases of RA and associated complications among the residents of Olmsted County, Minnesota (14, 15).

The study included a retrospectively identified incidence cohort of RA patients who were Olmsted County residents ≥18 years of age and first met the 1987 American College of Rheumatology (ACR) criteria (16) between 1/1/1980 and 1/1/2008. RA incidence date was defined as the earliest date at which each patient fulfilled ≥4 ACR criteria for RA. The original and complete medical records of all RA patients were screened longitudinally until HF incidence, death, migration or last date of follow-up (1/1/2008) by trained nurse abstractors blinded to the study hypothesis. HF was defined based on the Framingham criteria (17, 18). HF diagnosis requires ≥2 of the major criteria (i.e. paroxysmal nocturnal dyspnea or orthopnea, neck vein distention, rales, radiographic cardiomegaly [i.e. increasing heart size on chest X-ray film], acute pulmonary edema, S3 gallop, increased central venous pressure ≥16 cm of water at the right atrium, circulation time ≥25 seconds, hepatojugular reflux, weight loss >4.5 kg in 5 days in response to treatment of congestive HF), or the presence of 1 major criterion and ≥2 minor criteria (i.e. bilateral ankle edema, nocturnal cough, dyspnea on ordinary exertion, hepatomegaly, pleural effusion, decrease in vital capacity by 33% from maximal value recorded, tachycardia rate ≥120 beats/min). Minor criteria were counted only if they could not be attributed to another medical condition. Ejection fraction (EF) was determined by echocardiography and classified as preserved EF (≥50%) or reduced EF (<50%).

CV risk factors

The following CV risk factors were abstracted from the medical records at baseline and longitudinally throughout the follow-up as previously described (5), and were defined according to standardized diagnostic criteria as follows. Hypertension was defined according to the criteria of the Joint National Committee on Detection, Evaluation and treatment of High Blood Pressure as ≥2 ambulatory blood pressure readings ≥140 mm Hg systolic and/or ≥90 mm Hg diastolic obtained during a 1-year period, physician diagnosis or documented use of antihypertensive medications (19). Dyslipidemia was defined in accordance with Adult Treatment Panel III guidelines (20) as elevated lipid values of total cholesterol ≥240 mg/dL (≥6.2 mmol/L), low-density cholesterol ≥160 mg/dL (≥4.1 mmol/L), triglycerides ≥200 mg/dL (≥2.3 mmol/L) or high-density cholesterol <40 mg/dL (<1.0 mmol/L), physician diagnosis or documented use of lipid lowering medications. Obesity was present if body mass index (BMI) was ≥30 kg/m² based on the guidelines on the Identification, Evaluation and Treatment of Overweight and Obesity in Adults (21). Diabetes mellitus was defined as fasting plasma glucose ≥126 mg/dl (≥7.0 mmol/L), physician diagnosis or documented use of insulin and/or oral hypoglycemic agents in accordance with the diagnostic criteria adopted by the American Diabetes Association (22, 23). Personal history of coronary heart disease (CHD) was defined at baseline and throughout the follow-up as the presence of one of the following: angina pectoris, coronary artery disease, myocardial infarction (MI; including silent events), and coronary revascularization procedures (i.e. coronary artery bypass graft, percutaneous angioplasty, insertion of stents and atherectomy). MI was defined using standardized epidemiologic criteria (24), and Minnesota coding (25) of the electrocardiogram (ECG). Silent MI was considered as present at the date of the first documentation of a characteristic ECG or a recorded physician’s diagnosis in a patient with no documented history of MI. The definition of alcohol abuse was based on physician diagnosis of chronic alcoholism documented in the medical records. Smoking status (categorized as never, current or former) and family history of premature CHD (defined as the presence of CHD in first degree relatives at age <65 females and <55 males) were collected only at baseline.

RA characteristics and medications

The information on RA characteristics included RF status, erythrocyte sedimentation rate (ESR) at RA incidence and repeatedly high ESR (i.e. ≥3 ESR measures of ≥60 mm/hr with a minimum interval of 30 days between 2 measurements), large joint swelling, joint erosions/destructive changes on radiographs, joint surgeries (i.e. arthroplasty and synovectomy) and extra-articular manifestations of RA (ExRA). ExRA were classified according to the criteria used in our previous studies (26). Severe ExRA were defined according to Malmö criteria (27) and included pericarditis, pleuritis, Felty’s syndrome, glomerulonephritis, vasculitis, peripheral neuropathy, scleritis and episcleritis. Data regarding start and stop dates for use of systemic corticosteroids (e.g., oral/parental/intraarticular forms of prednisone, prednisolone, methylprednisolone, hydrocortisone, dexamethasone), disease-modifying anthirheumatic drugs (DMARDs) (methotrexate, hydroxychloroquine, other DMARDs) and biologic response modifiers (anti tumor necrosis factor alpha [anti-TNFα] agents, anakinra, abatacept, rituximab) were collected in all patients. Data on the use of non-steroid anti-inflammatory drugs (NSAIDs) and coxibs were also recorded. Data on the use of acetylsalicylic acid (ASA) for arthritis were collected, i.e. the use of >6 tablets of ASA per day (>1,950 mg/day) for at least 3 months. The history of rheumatic fever was documented. The study protocol was approved by Review Boards from Mayo Clinic and Olmsted Medical Center.

Statistical Methods

Descriptive statistics were used to characterize the demographics, traditional CV risk factors and RA disease characteristics. Cox proportional hazard models were used to examine the association of HF with RA characteristics. Adjustment for age and sex was performed by using age as the time scale and stratifying by sex. The models were additionally adjusted for calendar year of RA incidence. Time-dependent covariates were used to represent factors that developed during follow-up. Each RA disease characteristic was examined individually in models adjusted for age, sex and calendar year. In addition, models adjusted for age, sex, calendar year, CV risk factors and CHD were examined. Medications were modeled two ways: 1) as use at any time where the time-dependent covariates reflected medication start dates only and 2) as current use where the time-dependent covariates represented the time each patient was taking each medication, beginning at start date and ending at stop date for each medication. For each analysis of medications, all medications were assessed simultaneously in a multivariable model.

RESULTS

Patients’ characteristics

Among 815 incident RA cases identified during the 1980-2007 period, 20 fulfilled the Framingham criteria for HF before RA incidence, and were excluded from the study. The remaining 795 patients comprised the study cohort. Baseline characteristics are reported in Table 1. Mean age at RA incidence was 55.3 years; 69% were female. The patients were followed up for a mean of 9.7 years (a total of 7,692 patient-years). Hypertension, dyslipidemia and obesity were the most common CV risk factors at baseline (65%, 57% and 41% of patients, respectively), and their prevalence increased during follow-up. At the time of RA incidence, 22% of patients were current smokers; and 33% were former smokers. Family history of premature CHD was noted in 22% of patients. At baseline, 10% of patients had a personal history of CHD; and 23% had CHD at any time during follow-up. The proportion of patients with diabetes mellitus increased from 11% at baseline to 20% at any time during follow-up. Alcohol abuse was documented in 7% of patients at baseline and in 8% of patients at any time during follow-up.

Table 1.

Characteristics of 795 incident RA patients (1980-2007, Olmsted County, Minnesota) at RA incidence and during the follow-up^*

Variable	At RA incidence	At any time during follow-up
Age, years, mean ± SD	55.3 ± 15.5	--
Female	546 (69)	--
Length of follow-up, years, mean ± SD	9.7 ± 6.9	--
Smoking
- current	171 (22)	--
- former	254 (33)	--
Alcohol abuse	54 (7)	63 (8)
Hypertension	490 (65)	676 (88)
Dyslipidemia	431 (57)	569 (74)
Obesity (BMI ≥30 kg/m2)	302 (41)	368 (49)
Family history of premature CHD	172 (22)	--
Personal history of CHD	72 (10)	170 (23)
Diabetes mellitus	77 (11)	147 (20)

^*The values are given as number (%) if not indicated otherwise.

Abbreviations: RA = rheumatoid arthritis; SD = standard deviation; BMI = body mass index; CHD = coronary heart disease

During the follow-up, 92 patients developed HF. Of these, 31 patients had HF with reduced EF and 36 patients had HF with preserved EF. For the remaining 25 patients data on EF were not available. Of the above mentioned CV risk factors and CV comorbidities, family history of CHD (HR 1.6, 95%CI 1.03, 2.6), personal history of CHD (HR 3.1, 95%CI 1.96, 4.9), particularly, angina (HR 2.3, 95%CI 1.4, 3.6), and revascularization procedures (HR 2.3, 95%CI 1.3, 3.97), and alcohol abuse (2.4, 95%CI 1.2, 4.8) were significantly associated with the risk of HF. No time trends in the risk of HF were found (p=0.5).

RA characteristics and medications and their impact on HF

Data on RA characteristics are summarized in Table 2. RF positivity was present in 66% of patients. Mean ESR at RA incidence was 24.2 mm/hr and 12% of RA patients had ≥3 ESR measures of ≥60 mm/hr. Most patients (78%) experienced large joint swelling at some time during follow-up. The vast majority of patients (95%) had at least one radiograph. Joint erosions/destructive changes were found in 53% of patients. Joint surgery, i.e. arthroplasty or synovectomy, was performed in 17% and 11% of patients, respectively. Severe ExRA developed in 11% of patients. The most common other ExRA were rheumatoid nodules (occurred in 33% of patients). History of rheumatic fever was noted in 3% of patients.

Table 2.

RA characteristics and their associations with the risk of HF in 795 incident RA patients

RA characteristics	Value*	Hazard ratio [HR]** (95% confidence interval [CI]) adjusted for age, sex and calendar year of RA incidence
RF positive	527 (66)	1.6 (1.0, 2.5)
ESR at RA incidence (mm/hr), mean ± SD	24.2 ± 19.9	1.6 (1.2, 2.0)#
≥3 ESR ≥60 mm/hr	99 (12%)	2.1 (1.2, 3.5)
Large joint swelling	624 (78)	1.1 (0.7, 1.8)
Joint erosions/destructive changes	424 (53)	1.1 (0.8, 1.7)
Joint surgery
Arthroplasty	134 (17)	1.5 (0.9, 2.5)
Synovectomy	86 (11)	1.1 (0.6, 2.2)
ExRA overall
Rheumatoid nodules	266 (33)	1.1 (0.7, 1.7)
Rheumatoid lung disease	44 (6)	4.0 (2.1, 7.7)
Sjögren’s syndrome	32 (4)	0.6 (0.1, 2.4)
Severe ExRAX	87 (11)	3.1 (1.9, 5.1)
History of rheumatic fever	22 (3)	1.0 (0.4, 2.7)

^*The values are given as number (%) if not indicated otherwise;

^**significant (p<0.05) HRs are shown in bold;

^Xdefined according to the Malmö criteria [32];

^#reported per 30 mm/hr increase.

Abbreviations: HF = heart failure; RA = rheumatoid arthritis; SD = standard deviation; RF = rheumatoid factor; ESR = erythrocyte sedimentation rate; ExRA = extra-articular manifestations of RA

RF positivity (HR 1.6, 95%CI 1.0, 2.5; p=0.049), increased ESR at RA incidence (HR 1.6 per 30 mm/hr increase, 95%CI 1.2, 2.0), repeatedly high ESR (HR 2.1, 95%CI 1.2, 3.5) and severe ExRA (HR 3.1, 95%CI 1.9, 5.1) were significantly associated with HF (Table 2). Patients with RA disease duration <1 year were twice as likely to develop HF compared to patients with RA duration ≥1 year (HR 2.0, 95%CI 1.1, 3.8). The results remained essentially unchanged after adjustment for CV risk factors and CHD (data not shown).

Table 3 shows the data on antirheumatic medications. During follow-up, 58% were treated with methotrexate, 60% were treated with hydroxychloroquine, 32% of patients were treated with other DMARDs and 17% of patients were treated with biologic response modifiers, of which the vast majority (95%) received anti-TNFα therapy. Most patients received corticosteroids at some time during follow-up (77%). The vast majority of patients (91%) used NSAIDs at some time during follow-up. About half of the patients were treated with coxibs (49%). Analysis of medications used at any time revealed no significant associations with HF. Additional analyses of current use of medications were also performed (Table 3). Patients currently using methotrexate were half as likely to develop HF as non-users (HR 0.5, 95%CI 0.3, 0.9). The association changed only minimally (HR 0.4, 95%CI 0.2, 0.8) following additional adjustment for RA characteristics including RF positivity, RA duration, ESR at RA incidence and severe ExRA. Patients currently using biologic response modifiers or other DMARDs were also somewhat less likely to develop HF than those who did not currently use these agents; however, these associations were not statistically significant. The association of anti-TNFα treatment with HF were identical to the effect of biologic response modifiers overall. Use of hydroxychloroquine did not appear to be associated with HF. Current use of corticosteroids was associated with 2-fold increased risk of HF (HR 2.0, 95%CI 1.3, 3.2). These associations remained unchanged after adjustment for CV risk factors and CHD (data not shown).

Table 3.

Antirheumatic medications and their associations with the risk of HF in 795 incident RA patients

Antirheumatic Medications	N (%)	Hazard ratio [HR]** (95% confidence interval [CI]) adjusted for age, sex and calendar year of RA incidence
Used at any time
Methotrexate	465 (58)	0.9 (0.6, 1.5)
Hydroxychloroquine	475 (60)	0.9 (0.6, 1.4)
Other DMARDs	255 (32)	0.9 (0.5, 1.6)
Biologic response modifiers	135 (17)	0.8 (0.2, 2.6)
Corticosteroids	614 (77)	1.2 (0.7, 1.9)
NSAIDs	732 (91)	1.0 (0.5, 1.9)
Coxibs	386 (49)	1.5 (0.9, 2.3)
ASA	330 (43)	1.0 (0.6, 1.7)
Current Use
Methotrexate		0.5 (0.3, 0.9)
Hydroxychloroquine		1.0 (0.5, 1.8)
Other DMARDs		0.5 (0.2, 1.5)
Biologic response modifiers		0.5 (0.1, 3.5)
Corticosteroids		2.0 (1.3, 3.2)

^*significant (p<0.05) hazard ratios are shown in bold.

Abbreviations: HF = heart failure; RA = rheumatoid arthritis; DMARDs = disease-modifying antirheumatic drugs; NSAIDs = non-steroid anti-inflammatory drugs; ASA = acetylsalicylic acid for arthritis (>6 tab. of ASA per day [>1950 mg/day] for at least 3 months).

Additional analyses were performed to elucidate the impact of concurrent methotrexate and corticosteroid use. Among methotrexate users, 76% were using corticosteroids concurrently, and no increased risk of HF was found among these concurrent users compared to patients using neither methotrexate or corticosteroids (HR 0.8, 95%CI 0.3, 2.0; p=0.67). An increased risk of HF persisted among patients taking corticosteroids without methotrexate (HR 2.2, 95%CI 1.3, 3.6; p=0.002). However, no significant difference in risk of HF was noted among patients taking methotrexate without corticosteroids (HR 0.6, 95%CI 0.3, 1.4; p=0.27).

We also investigated whether risk factors differ for HF with preserved EF and HF with reduced EF. We found that males were more likely to have HF with reduced EF than females (HR 3.7, 95%CI 1.8, 7.7). This association remained significant after adjustment for CV risk factors and CHD. No gender difference was found for HF with preserved EF (HR 0.9, 95%CI 0.5, 1.9 in males versus females). No other statistically significant differences were found between risk factors for HF with reduced EF and HF with preserved EF (data not shown).

DISCUSSION

Herein we report risk factors for HF in a population-based incident RA cohort. We have shown that several RA characteristics including RF positivity, increased ESR at RA incidence and repeatedly high ESR, as well as the presence of severe ExRA were significantly associated with HF after adjusting for CV risk factors and CHD. Patients with RA duration <1 year were twice as likely to develop HF as patients with RA duration ≥1 year.

Some measures of RA activity including RF positivity and increased ESR were previously linked to HF in RA in ours and other studies (5, 7, 8, 10). The association of increased ESR at RA incidence with the risk of HF in RA corroborates our earlier observations regarding ESR as a potential signal for the subsequent development of HF (8). The associations of RF positivity and repeatedly high ESR with HF support the hypothesis that chronic immune inflammation may promote the development and progression of HF in RA (28). The association of severe ExRA with HF is concordant with the literature describing poor CV prognosis and increased likelihood of myocardial function impairment in patients with ExRA (29-31).

Several earlier studies showed an association of myocardial dysfunction (32, 33) and/or HF (9) with RA duration, while others did not find this relationship (29). Our study extends these observations showing that the risk of HF is significantly increased during the first year of RA. Active inflammation in early RA could be one of the reasons for this finding. However, the role of confounding factors (including antirheumatic medications use and preexisting comorbidities) cannot be excluded. Thus, the clinical implications of this finding are unclear and require further investigation.

In our study, current use of methotrexate was associated with decreased likelihood of HF even after adjustment for RA characteristics. This finding is concordant with previous observations of a protective effect of methotrexate against CV disease, including HF (11, 34). The lower risk of HF in methotrexate users versus non-users may be secondary to the decrease in inflammatory activity following methotrexate treatment. However, alternative mechanisms including methotrexate-specific effects cannot be excluded (35). Current corticosteroid use appeared to have an adverse effect on the risk of HF, which is concordant with the literature (9, 12, 13). Our findings of somewhat lower likelihood of HF in current users of biologic response modifiers versus non-users were similar to those of others suggesting a beneficial effect of biologic response modifiers (particularly, anti-TNFα treatment) on the risk of HF (13). However, our results did not reach statistical significance, likely due to insufficient statistical power. Since the vast majority of biologic response modifier use in our patients was anti-TNFα treatment, we were unable to analyze the effects of different biologic response modifiers on the risk of HF. Analyses of medications used at any time revealed no significant associations with HF, while analyses of current use were significantly associated with the risk of HF. Thus past use appears to confer less risk than current use. The reasons for this are unclear, but may be due to a dilution of effect over time. Confounding by indication/contraindication when the initiation of an antirheumatic medication depends on prognostic expectations of a physician, particularly with regards to RA activity and CV risk may also play a role. More studies (preferably randomized controlled trials) are needed to better understand the nature of the effects of antirheumatic medications on HF in RA.

The association between HF with reduced EF and male gender in our study is concordant with the findings from the general population (36). In contrast to the findings from the general population, we did not find higher likelihood of HF with preserved EF in females (36, 37). This raises a possibility that the mechanisms of HF in RA patients differ from the general population. Except for the gender differences, there were no statistically significant differences in the risk factors for HF with preserved EF versus HF with reduced EF in our study. However, statistical power to draw definite conclusions regarding these associations was lacking.

Our study has several potential limitations. First, the population of Olmsted County, Minnesota is predominantly white, thus the generalizability of our findings to more ethnically diverse populations may be limited. Second, the retrospective study design requires that only information available from medical records was used to ascertain risk factors and outcomes. Thus, risk factors and outcomes were not measured at regular intervals, and were dependent on physician observation and documentation. However, the use of the comprehensive population-based resources of REP and standardized case ascertainment likely minimized this bias. Another limitation inherent to the use of medical records is the lack of echocardiography data (particularly, the EF measures) in 27% of patients, suggesting that the results of analyses of patients with reduced versus preserved EF should be interpreted with caution. However, this is not likely to significantly affect other findings of the study. The definition of alcohol abuse in our study was based on physician’s diagnosis and thus reliability may be somewhat limited. However, the magnitude and the direction of this association are suggestive of increased risk of HF in RA patients with alcohol abuse and this finding is consistent with observations in the general population (38). As in any observational study, there is a possibility of confounding by indication/contraindication for the associations of medication use (particularly, methotrexate and corticosteroid use) with the risk of HF. Although the association of methotrexate with lower likelihood of HF remained essentially unchanged after adjusting for RA characteristics, other potential confounders including unmeasured and unknown confounders cannot be excluded. Finally, the use of the over-the-counter medications (particularly, NSAIDs and coxibs) was recorded as present or absent without a specification of the drug. Therefore, we were unable to analyze the association of each individual NSAID/coxib with the risk of HF in this study.

Our study has several important strengths. It is a large longitudinal population-based cohort study of incident RA patients which captures the spectrum of RA disease in the community. For instance, the rates of joint erosions/destructive changes in population-based studies and other large inception RA cohorts recruited from primary care settings are somewhat lower than in referral-based cohorts, suggesting that these estimates are more representative (39-41). The availability of extensive data on the use of antirheumatic medications including the use of oral and intravenous corticosteroids is another strength of the study. Finally, we used well established and validated criteria to identify RA patients, CV risk factors and CV events.

In conclusion, several RA characteristics, including RF positivity, increased ESR at RA incidence, repeatedly high ESR and the presence of severe ExRA were significant determinants of HF beyond traditional CV risk factors and CHD. Current use of methotrexate was associated with decreased likelihood of HF, while current use of corticosteroids was associated with an increased risk of HF. Our findings suggest an independent impact of RA on HF development, which may be further modified by antirheumatic treatment. More research is needed to better understand the mechanisms of HF in RA.