Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Numerous reports of the increased incidence of cardiovascular disease (CVD) among patients with rheumatoid arthritis (RA) have been published during the past decade.1 2 In addition, the increased risk of CVD in patients with RA cannot be explained by traditional CV risk factors alone.3 4 While these reports have increased awareness of CV morbidity and mortality among patients with RA, the lack of CV risk assessment tools and evidence-based practice guidelines developed specifically for patients with RA has slowed the translation of this knowledge into clinical decision making. The following two case studies illustrate the need for accurate RA-specific CV risk assessment tools.
The first patient, a woman, was diagnosed as having RA at age 40 with rheumatoid factor (RF) and anti-citrullinated protein antibody (ACPA) positivity and rheumatoid nodules present at diagnosis. She was a non-smoker and non-diabetic with total cholesterol of 6.7 mmol/litre (258 mg/dl), high-density lipoprotein cholesterol (HDL-c) of 1.4 mmol/litre (54 mg/dl) and a systolic blood pressure of 120 mm Hg. The SCORE (for 'Systematic COronary Risk Evaluation') risk assessment for populations at high CVD risk categorised her 10-year risk of fatal CVD as <1% and the general Framingham Risk Score (FRS) estimated her 10-year risk for any CVD event as 4%.5 6 Despite these low risk assessments, she experienced a myocardial infarction (MI) at age 45. Thus, the SCORE and FRS did not accurately predict CV risk for this patient.
The second patient, a man, was diagnosed as having RA at age 46 with RF/ACPA positivity and rheumatoid vasculitis present at diagnosis. He was a non-smoker and non-diabetic with total cholesterol of 8.8 mmol/litre (341 mg/dl), HDL-c of 2.0 mmol/litre (76 mg/dl), a systolic blood pressure of 140 mm Hg and normal body mass index (26 kg/m2). His SCORE risk was 2.6% and FRS was 9.2%. Despite these moderate risk assessments, he was given antihypertensive medication and statins, but he still experienced a MI 6 months later. Thus, the SCORE and FRS did not accurately predict CV risk for this patient.
The European League Against Rheumatism (EULAR) recommendations for CV risk management constitute the first attempt to provide recommendations to improve CVD management in patients with RA.7 A multiplier of 1.5, applied to a conventional CV risk assessment (eg, SCORE or FRS), is recommended for patients with RA with two of the following three criteria: disease duration >10 years, RF or ACPA positivity and presence of severe extra-articular manifestations (ExRA). In the case studies, our first patient would not meet these criteria. Our second patient was RF positive and had a severe ExRA, so the multiplier would apply. This would increase his SCORE risk to 3.9% and his FRS to 13.8%, neither of which would meet recommended thresholds for intensified risk modification efforts of 10% for SCORE and 20% for FRS. Thus, application of these recommendations had little impact on the risk assessments for both of our patients. Also, while these recommendations are evidence based, some of the evidence is weak. Indeed, among patients with RA, the accuracy of CV risk assessment using tools for the general population and the ability of a simple multiplier to improve the accuracy of risk assessment are unknown. While the EULAR task force should be commended for addressing this important issue, the use of this multiplier and the choice of the criteria defining the patients with RA requiring the multiplier deserve a closer inspection.
First, the evidence of increased CVD risk associated with longer (>10 years) disease duration is lacking as neither of the two works cited in the recommendations support the inclusion of this criterion.8 9 Naz et al8 reported increased CVD mortality was present at RA symptom onset and throughout follow-up. Gabriel et al9 showed the excess all-cause mortality among patients with RA was not apparent until 8–10 years after diagnosis of RA. Neither of these reports examined the risk of CVD events, which could be increased earlier in the disease course than mortality. In contrast, several recent reports noted an increased CVD risk early in the RA disease course.10,–,14 Maradit-Kremers et al10 found an increased risk of MI prior to diagnosis of RA. In reports emerging after publication of the EULAR recommendations, Holmqvist et al reported an increased risk of MI within 1 year after diagnosis of RA, Franklin et al reported an increased risk of hospital admissions for CVD within 7 years of diagnosis of RA, and Södergren et al reported significant increases in intima media thickness among patients with RA by 18 months after diagnosis of RA.12,–,14 Thus, most of the evidence now supports an increased risk of CVD early in the RA disease course.
Second, evidence for the association between CVD risk and RF or ACPA positivity is wanting. While the increased risk of CVD among patients with RA who are RF positive is well established, the lack of an increased risk of CVD among patients with RA who are RF negative is unclear. Previous reports showed no increased risk of all-cause and CVD mortality and of heart failure among patients with RA who are RF negative compared to persons without RA.2 8 15 However, Holmqvist et al12 recently reported similarly increased risks of hospitalisation for MI among patients with RA who are RF positive and patients with RA who are RF negative compared to persons without RA. Also, few studies have examined CVD risk among patients who are ACPA positive or negative. Farragher et al16 reported a higher risk of all-cause and CVD mortality among patients with RA who are ACPA positive compared to patients with RA who are ACPA negative. Since most patients with RA who are ACPA positive are also RF positive, it seems likely that patients who are ACPA positive have an increased risk of CVD. However, whether patients with RA who are ACPA negative have a higher risk of CVD than the general population is unknown. Hence, the possibility of an increased risk of CVD for either patients with RA who are RF negative or patients with RA who are ACPA negative requires more investigation.
Third, while the increased risk of CVD among patients with severe ExRA of RA is substantial (eg, greater than threefold increased risk), the rarity of these ExRA (eg, the cumulative incidence at 10 years of RA disease duration is <10%) limits their clinical utility in the context of the EULAR recommendations, which require two of three criteria to be met.17,–,19 For example, 30% of the prevalent cases of RA in Olmsted County, Minnesota on 1 January 2000 satisfied two of the three criteria, but only 4% used the presence of severe ExRA to fulfil the criteria. The other 3.5% with ExRA also had >10 years duration of RA and RF positivity. In addition, the inclusion of other risk factors associated with increased risk of CVD in patients with RA, such as rheumatoid nodules or abnormal C reactive protein, should be considered.8 16 Also, if each of the criteria is indicative of increased risk, why not apply the multiplier to patients with RA with only one criterion?
Furthermore, the multiplier value of 1.5 was based on many reports of the increased risk of CVD among patients with RA compared to those without RA, but the increased risk among the subset of patients with RA that meet two of the three criteria is unknown. If these patients are at higher risk than other patients with RA, their increased risk of CVD may be much higher than 1.5 times the general population risk. Conversely, if this subset of patients with RA also have more CV risk factors, the general population risk assessment tools may already account for some of their increased risk, so the 1.5 multiplier could lead to overestimation of the risk of CVD. Another complexity is that traditional CV risk factors appear to have differing, even paradoxical, effects on the risk of CVD for patients with RA.3 4 20 21 For example, Peters et al reported no association between total cholesterol and CVD events, and Myasoedova et al reported increased CVD risk for patients with RA with low total cholesterol.20 22 Others have reported low body mass index was associated with increased risks of all-cause and CVD mortality in patients with RA.21 23 These risk factors may require different weights than those used in the general population in order to accurately predict CV risk in patients with RA.
In addition, calibration and discrimination should be assessed to determine the accuracy of CVD risk prediction using the multiplier.24 Calibration refers to the ability to predict the absolute risk level. While applying a multiplier may improve the average level of CVD risk estimated for patients with RA, it is unclear whether the resulting risk estimates are accurate for individual patients with RA. Perhaps the multiplier increases the estimated risk too much for some patients and not enough for others. Discrimination refers to the ability to accurately rank risk levels in order to distinguish low from high risk. The use of a multiplier would change their ranking relative to other patients with RA, but it would have little impact on patients with low risk estimates (eg, 1.5 times 1% yields 1.5%). Hence, if patients with RA with high levels of systemic inflammation but few or no traditional CV risk factors have high risk for CVD, then the use of a multiplier would not substantially improve discrimination. Thus, while a simple multiplier applied to existing CV risk tools may improve average calibration, it may not be accurate for individual patients and it may not improve discrimination, raising questions regarding the clinical utility of this approach to accurately predict risk.
Instead, the incorporation of non-traditional risk factors, such as measures of inflammation and perhaps genetic markers extensively associated with CV risk, such as the presence of human leucocyte antigen (HLA)-DRB1* shared epitope, into a risk score would be needed in order to correctly rank the CV risk in patients with RA.16 25 26 In addition to single measures of inflammation, inflammatory burden may play a crucial role in CV risk, and suppression of inflammation may be a pivotal requirement for improving CV risk management.20 In addition, non-invasive surrogate markers of subclinical CV disease, such as carotid intimamedial thickness, should be considered and possibly used in the assessment of the CV risk of patients with RA.27 28 However, careful consideration of the clinical utility, healthcare costs and recommended frequency of assessment should be performed before including these measures in a risk assessment tool for CV disease among patients with RA.
In conclusion, CV risk assessment tools designed for the general population may not accurately estimate CV risk for individual patients with RA, even if a multiplier is applied. In addition, while the EULAR recommendations aim to improve CV risk management in patients with RA, rheumatologists should be cautioned to avoid overinterpreting them. Patients satisfying the criteria for use of the multiplier likely do have increased risk of CVD, but those who do not meet these criteria may also have increased risk. The disease duration criterion is particularly worrisome as the increased risk of CVD likely begins early, and the inclusion of this criterion may inadvertently lead clinicians to underestimate the early risk of CVD, perhaps leading to less diligence regarding CV risk assessment at diagnosis of RA. Further research is needed to derive a tool for accurately assessing the risk of CVD in patients with RA.
Funding This work was partially funded by the National Institutes of Health (R01 AR46849).
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.