Introduction

The recognition that rheumatoid arthritis (RA) patients are at a heightened risk of cardiovascular disease (CVD) events and mortality now spans more than two decades. During this period, much has been learned about the magnitude of the problem. In a large meta-analysis of 24 cohort studies , CVD mortality was 50% higher in RA compared with non-RA populations. Similarly, the relative increase in the risk of myocardial infarction (MI) and stroke was 68% and 41%, respectively . Independent of atherosclerotic ischemic heart disease, RA patients are also at a heightened risk of myocardial dysfunction and overt heart failure . CVD is the primary cause of death in RA patients , and as CVD events, such as MI and stroke, tend to occur at an earlier age in RA compared with the general (non-RA) population, life expectancy is reduced.

Premature CVD is a contributor to the widened mortality gap observed between RA and non-RA populations . Although there are some indications that this gap may be closing with the widespread adoption of early and aggressive treatment strategies , not all recent studies demonstrate the same promising trend , suggesting that CVD in RA remains a significant public health problem.

Reducing CVD event rates and mortality in RA requires the identification of susceptible subgroups and high-impact causal determinants. A substantial number of observational epidemiologic studies have sought to identify the RA-specific characteristics associated with CVD risk factors (traditional, nontraditional, and inflammatory), intermediate measures (i.e., atherosclerosis and atherothrombosis), and CVD events in RA. However, it is important to understand that these studies identify associations only, and, despite the number of studies that have been conducted, an incomplete understanding of causal determinants remains. Consistency and temporality are among the primary criteria for arguing causality in observational studies. However, there is notable heterogeneity in the factors identified as associated with RA CVD across studies, which may in part be explained by differing outcomes, populations, and exposure periods studied. Few are longitudinal, and among those that are, follow-up times are often limited, making the assessment of the association of exposure to outcome problematic. In addition, many combine RA patients at varying stages of disease, limiting the ability to identify subgroups of RA patients at heightened risk. Most rely on single point in time assessments of CVD risk factors and articular/systemic inflammation, which is problematic in a disease in which substantial time variance in risk factors is expected, both as a natural function of disease fluctuation and with treatment. It is not surprising, then, that our understanding of the determinants of CVD in RA remains limited.

Even with consistency and temporality demonstrated across multiple observational cohorts, more is required to justify health policy recommendations. Here, trials are essential to provide evidence that specific interventions affect outcomes. Unfortunately, there are few trials evaluating the efficacy of specific interventions for either primary or secondary prevention of CVD in RA, and none evaluating the effectiveness of preventive strategies in the setting of the delivery of RA clinical care. The scarcity of these trials is understandable given that (1) causal determinants from observational studies on which to base trials are uncertain, (2) trials with CVD events as outcomes often require large numbers (i.e., thousands) of enrollees, and (3) although the most common autoimmune inflammatory arthritis, RA is still relatively uncommon in the general population, and many of the most severely affected (and thus potentially most at risk of adverse CVD events) are unable or unwilling to participate in clinical trials. The recent Trial of Atorvastatin for the Primary Prevention of Cardiovascular Events in RA (TRACE RA) Trial (discussed subsequently) is testament to these challenges.

Tasked with caring for RA patients in 2015, what must the practicing rheumatologist do? Most will find it unacceptable to await definitive trials of CVD screening and treatment strategies specifically developed in RA populations, which remain years to a decade or more in the future. Short of these, extrapolations from observational studies of RA patients and trials of non-RA populations with some applicability to RA may provide some insight into areas for intervention. However, any recommendations based on lower levels of evidence require the consideration of possible adverse effects, which include not only direct adverse effects of testing and treatments but also indirect adverse effects, such as diverting resources and attention to interventions that may have no value. Even in the setting of known increased risk, current practice for treating CVD risk factors in RA patients is generally poor .

This review focuses on studies exploring determinants of CVD risk factors, intermediates, and events in RA, and studies evaluating CVD screening and treatment strategies for clinical practice. Current expert-opinion-based recommendations are discussed, with the current level of evidence reviewed in the context of a research agenda targeting a more rigorous approach to primary and secondary CVD screening and prevention in RA. There is a paucity of literature regarding the screening and prevention of cerebrovascular disease and myocardial dysfunction in RA. As such, this review concentrates on coronary and extra-coronary atherosclerosis as it pertains to the prediction of ischemic heart disease events.

How do the coronary and peripheral arteries differ in RA compared with the non-RA population?

Higher CVD event rates in RA may be mediated by a number of mechanisms. For one, RA patients may have a greater burden of atherosclerosis throughout the arterial tree, manifesting as CVD events occurring at a younger than expected age. A second scenario involves the preferential accumulation in RA patients of atherosclerotic plaques in anatomic sites at a higher risk of events. In a third scenario, RA patients may have a similar burden of atherosclerosis as those without RA, but the composition of the plaques makes them more vulnerable and rupture prone. This, combined with a greater thrombotic tendency in RA , could explain higher event rates in RA even if the quantity of atherosclerosis does not differ from the non-RA population. Interestingly, there is evidence in the literature to support all three of these scenarios in RA.

Using cardiac computed tomography (CT), calcified coronary atherosclerotic plaques can be radiographically visualized, and the amount of coronary arterial calcium (CAC) quantified using standardized and validated methods . In studies of the general population, the amount of CAC is highly correlated with direct measures of plaque burden (i.e., from endovascular ultrasound or coronary catheterization) , and it is predictive of future CVD events . Multiple studies of RA patients have identified significantly higher CAC compared with non-RA populations , even after adjusting for unbalanced demographics and CVD risk factors, confirming a higher coronary plaque burden in RA. However, CAC assessment using non-contrasted cardiac CT can only image calcified plaque. Differing from calcified plaque, non-calcified and mixed plaques may be less stable and more rupture prone .

CT angiography is a newer technique that allows the visualization of both calcified and non-calcified coronary atherosclerotic plaques. In the only study using CT angiography in RA, RA patients had significantly higher numbers of calcified, mixed, and non-calcified coronary plaques compared with a matched non-RA control group , with inflammatory features correlated only with non-calcified and mixed plaque prevalence.

The carotid arteries are readily visualized using bedside ultrasound, and techniques for delineating the presence of plaque and thickening of the intima-medial layer are well established. Moreover, in the general population, intima-medial thickness (IMT) and plaque are modestly predictive of coronary atherosclerotic events . Many carotid ultrasound studies have been conducted in RA populations of various characteristics, with sample sizes ranging from a handful of patients to many hundreds . Although there is heterogeneity across studies in differences observed between RA and control groups, the trend across studies is that carotid atherosclerotic burden is also higher in RA. Importantly, the presence of carotid plaque was associated with a two-to fourfold increase in incident acute coronary syndromes, depending on whether unilateral versus bilateral carotid plaque was present, in RA , an important confirmation when considering possible imaging modalities appropriate for screening.

However, not all arterial beds may be at risk of more atherosclerosis in RA. Occlusive atherosclerosis in the peripheral arteries does not appear to be as prominent in RA as in the general population . Instead, RA patients tend to manifest noncompressible peripheral arteries, perhaps representing medial calcification. The clinical implication of this finding, if any, is unclear.

In addition to the quantity of atherosclerosis in RA, the quality of the atherosclerotic plaques may also play a role in which RA patients will have CVD events. The makeup of atherosclerotic plaques is complex, and plaque stability is a function of the size of the lipid/necrotic core, the cellular infiltrate within the plaque, and the inflammatory cytokines that induce remodeling of the fibrous cap . Remodeling causes thinning of the fibrous cap, which, under unfavorable hemodynamic conditions, can rupture, exposing tissue factor and other prothrombotic factors, resulting in vaso-occlusive thrombus . Plaques are especially vulnerable at the shoulder regions, in which a heavy infiltrate of macrophages and other inflammatory cells can be observed in rupture-prone plaques . Although RA-specific data are limited, histopathologic examination of autopsied coronary arteries from RA versus non-RA patients demonstrated more inflamed and vulnerable plaques in RA , even in the setting of similar overall plaque burden. Together, these studies suggest that cardioprotective interventions in RA patients may require both slowing of atherogenesis and stabilizing vulnerable plaques, conditions that may vary between RA patients according to the presence/absence of traditional risk factors, the stage of RA (i.e., early vs. late disease), and RA disease and treatment characteristics.

The radiotracer fluorodeoxyglucose (FDG) is readily taken up by macrophages, and it can be imaged on positron emission tomography (PET), making an attractive modality for imaging tissue inflammation. In particular, atherosclerotic plaques in larger arteries can be visualized, and FDG uptake correlates with histologic plaque inflammation . In a small study , RA patients with active articular inflammation and no known CVD had higher levels of aortic FDG uptake than otherwise similar non-RA patients with known coronary artery disease. This modality shows great promise for defining the inflammatory characteristics of RA atherosclerosis; however, there are no data regarding its application to clinical practice for screening and prevention in any population.

How do the coronary and peripheral arteries differ in RA compared with the non-RA population?

Higher CVD event rates in RA may be mediated by a number of mechanisms. For one, RA patients may have a greater burden of atherosclerosis throughout the arterial tree, manifesting as CVD events occurring at a younger than expected age. A second scenario involves the preferential accumulation in RA patients of atherosclerotic plaques in anatomic sites at a higher risk of events. In a third scenario, RA patients may have a similar burden of atherosclerosis as those without RA, but the composition of the plaques makes them more vulnerable and rupture prone. This, combined with a greater thrombotic tendency in RA , could explain higher event rates in RA even if the quantity of atherosclerosis does not differ from the non-RA population. Interestingly, there is evidence in the literature to support all three of these scenarios in RA.

Using cardiac computed tomography (CT), calcified coronary atherosclerotic plaques can be radiographically visualized, and the amount of coronary arterial calcium (CAC) quantified using standardized and validated methods . In studies of the general population, the amount of CAC is highly correlated with direct measures of plaque burden (i.e., from endovascular ultrasound or coronary catheterization) , and it is predictive of future CVD events . Multiple studies of RA patients have identified significantly higher CAC compared with non-RA populations , even after adjusting for unbalanced demographics and CVD risk factors, confirming a higher coronary plaque burden in RA. However, CAC assessment using non-contrasted cardiac CT can only image calcified plaque. Differing from calcified plaque, non-calcified and mixed plaques may be less stable and more rupture prone .

CT angiography is a newer technique that allows the visualization of both calcified and non-calcified coronary atherosclerotic plaques. In the only study using CT angiography in RA, RA patients had significantly higher numbers of calcified, mixed, and non-calcified coronary plaques compared with a matched non-RA control group , with inflammatory features correlated only with non-calcified and mixed plaque prevalence.

The carotid arteries are readily visualized using bedside ultrasound, and techniques for delineating the presence of plaque and thickening of the intima-medial layer are well established. Moreover, in the general population, intima-medial thickness (IMT) and plaque are modestly predictive of coronary atherosclerotic events . Many carotid ultrasound studies have been conducted in RA populations of various characteristics, with sample sizes ranging from a handful of patients to many hundreds . Although there is heterogeneity across studies in differences observed between RA and control groups, the trend across studies is that carotid atherosclerotic burden is also higher in RA. Importantly, the presence of carotid plaque was associated with a two-to fourfold increase in incident acute coronary syndromes, depending on whether unilateral versus bilateral carotid plaque was present, in RA , an important confirmation when considering possible imaging modalities appropriate for screening.

However, not all arterial beds may be at risk of more atherosclerosis in RA. Occlusive atherosclerosis in the peripheral arteries does not appear to be as prominent in RA as in the general population . Instead, RA patients tend to manifest noncompressible peripheral arteries, perhaps representing medial calcification. The clinical implication of this finding, if any, is unclear.

In addition to the quantity of atherosclerosis in RA, the quality of the atherosclerotic plaques may also play a role in which RA patients will have CVD events. The makeup of atherosclerotic plaques is complex, and plaque stability is a function of the size of the lipid/necrotic core, the cellular infiltrate within the plaque, and the inflammatory cytokines that induce remodeling of the fibrous cap . Remodeling causes thinning of the fibrous cap, which, under unfavorable hemodynamic conditions, can rupture, exposing tissue factor and other prothrombotic factors, resulting in vaso-occlusive thrombus . Plaques are especially vulnerable at the shoulder regions, in which a heavy infiltrate of macrophages and other inflammatory cells can be observed in rupture-prone plaques . Although RA-specific data are limited, histopathologic examination of autopsied coronary arteries from RA versus non-RA patients demonstrated more inflamed and vulnerable plaques in RA , even in the setting of similar overall plaque burden. Together, these studies suggest that cardioprotective interventions in RA patients may require both slowing of atherogenesis and stabilizing vulnerable plaques, conditions that may vary between RA patients according to the presence/absence of traditional risk factors, the stage of RA (i.e., early vs. late disease), and RA disease and treatment characteristics.

The radiotracer fluorodeoxyglucose (FDG) is readily taken up by macrophages, and it can be imaged on positron emission tomography (PET), making an attractive modality for imaging tissue inflammation. In particular, atherosclerotic plaques in larger arteries can be visualized, and FDG uptake correlates with histologic plaque inflammation . In a small study , RA patients with active articular inflammation and no known CVD had higher levels of aortic FDG uptake than otherwise similar non-RA patients with known coronary artery disease. This modality shows great promise for defining the inflammatory characteristics of RA atherosclerosis; however, there are no data regarding its application to clinical practice for screening and prevention in any population.

Traditional and inflammatory CVD risk factors in RA

Numerous factors have been identified that contribute to atherogenesis and plaque vulnerability in the general population. Whether there are differential effects in RA that account for the excess in observed risk is less studied. Conceptually, a risk factor may account for excess risk observed between one population versus another if (1) its prevalence is unbalanced between populations, (2) the risk factor exerts a larger magnitude of effect on the outcome in one population versus the other (i.e., effect modification), or (3) a combination of (1) and (2). In RA, most studies have explored differences in the prevalence of risk factors between RA and non-RA groups only, with many studies demonstrating that most traditional CVD risk factors (i.e., diabetes, hyperlipidemia, and hypertension) do not tend to be more prevalent in RA . The exception is smoking, which, as a risk factor for incident RA, tends to be overrepresented in RA compared with the background non-RA population. However, even when considering the extent of imbalance in smoking prevalence and possible differences in the magnitude of effect between populations, the excess risk of CVD in RA is unlikely to be fully accounted for simply by a higher prevalence of smoking . In a recent meta-analysis of 10 publications relating the impact of traditional CVD risk factors to MI and CVD morbidity/mortality encompassing >4000 RA patients , both hypertension and diabetes were associated with MI in RA, and these risk factors, along with smoking, hypercholesterolemia, and obesity, were associated with CV morbidity. However, compared with the non-RA population, traditional CVD risk factors may have smaller overall magnitudes of effect in RA . This is particularly notable for the relationship of total and low-density lipoprotein (LDL) cholesterol level with CVD events. In particular, higher total and LDL cholesterol levels are not as strongly linked to CVD in RA compared with the non-RA population. In fact, multiple observational studies have identified RA patients with the lowest total and LDL-C levels (i.e., an LDL-C of <75 mg/dL) as those at highest risk of CVD events . The etiology of this finding, the so-called “lipid paradox” in RA, is unclear, although inflammation-induced reduction in lipid levels has been postulated . By contrast, high-density lipoprotein (HDL)-C and triglyceride levels appear to predict CVD events similarly in RA as in the non-RA population.

Importantly, the effects of many traditional CVD risk factors may be increased in the setting of inflammation. For example, in several studies , higher levels of systemic inflammatory markers were only associated with the progression of carotid atherosclerosis among those with higher levels of traditional CVD risk factors, suggesting that inflammatory risk factors may require a context of traditional risk factors in order to exert effects.

The immunopathology of RA is complex, yet similarities between the synovial inflammatory infiltrate and that of atherosclerotic plaques are striking . In addition, both processes share a systemic inflammatory component characterized by circulating activated immune cells and elevated inflammatory cytokines. Animal studies of atherosclerosis in which single cytokines are either augmented or genetically eliminated demonstrate convincingly that atherogenesis is causally related to the upregulation of inflammatory cytokines, such as tumor necrosis factor (TNF)-α, interleukin (IL)1-β, IL-6, IL-17, and others, which are also key players in RA immunobiology . Perhaps more importantly, these studies demonstrate that although many cytokines appear to contribute to atherogenesis, interventions targeting single cytokines can result in substantial reductions in the burden of atherosclerosis and the inflammatory quality of the remaining plaques. Whether human RA atherogenesis is completely compatible with the findings of animal models is unclear, as few interventional studies of atherosclerosis in RA patients are available. Circumstantial evidence of the role of inflammation on atherosclerosis is available from a number of longitudinal cohort studies, in which RA patients with higher levels of systemic inflammation have been shown to have a more rapid rate of progression of atherosclerosis . However, it should be noted that “inflammation” (i.e., RA disease activity features and or circulating inflammatory markers or cytokines) per se has not been consistently identified as a strong predictor of atherosclerosis or its progression across all studies . Whether this indicates that RA inflammation is of secondary importance in RA atherogenesis or that causal inflammatory mediators were not captured or imprecisely measured in these studies is uncertain. Mechanistic studies exploring the impact on RA atherosclerosis of interactions between cellular and humoral inflammatory factors, RA-specific autoimmunity, and genetic and environmental risk modifiers across the disease course are warranted. Such studies are challenging to design and execute, and treatment effects are difficult to disentangle from the natural history of the disease. Nevertheless, “inflammation” as a driver of atherogenesis/atherothrombosis in RA remains the cornerstone of the current conceptualization of disease mechanism.

Assessing and stratifying CVD risk in RA for primary prevention: who is at risk?

Directly quantifying atherosclerosis, via angiography or intravascular ultrasound, would be a definitive way of detecting atherosclerosis and risk stratifying on an individual RA patient׳s risk of ischemic CVD events. However, for every RA patient to undergo such an evaluation would result in excess exposure to invasive procedures and/or radiation, and it would not be cost-effective, even in a condition in which CVD events are overrepresented, such as RA. Extrapolating from recommendations for macrovascular disease screening in type II diabetes, a condition with similar CVD event rates as RA , even using less invasive imaging, such as CAC scoring from non-contrasted cardiac CT or carotid screening with ultrasound, would likely pose excess risk and/or cost if applied to all RA patients . In the general population, current evidence-based guidelines do not advocate any strong evidence for vascular imaging or functional testing (i.e., stress testing) for individuals without cardiac symptoms or certain electrocardiographic (ECG) abnormalities . However, it is unclear whether specific subgroups of RA patients would benefit from additional imaging or functional screening, and no clinical trials of any particular imaging or CVD functional screening have been conducted in RA. Even if atherosclerosis screening were to show incremental prognostic benefit in RA, studies demonstrating that screening alters CVD management and outcomes would still be needed for a level A recommendation.

Short of direct imaging of atherosclerosis, or functional evidence of a coronary arterial abnormality consistent with subclinical arterial occlusion, primary CVD screening in the general population consists of assessing traditional CVD risk factors and risk stratifying based on normative equations developed in large populations followed over long periods for the development of CVD events. Numerous such risk calculators have been validated (see Appendix 3 in Ref. ), and their utility is based on the populations in which they were developed. The CVD outcomes predicted differ across the various calculators, but most predict an individual׳s 10-year risk of having a “hard” CVD outcome (i.e., sudden cardiac death, nonfatal MI, fatal and nonfatal cerebrovascular accident (CVA), and occasionally others). Some were developed including high-risk groups, such as diabetes, and others omit high-risk groups, a notable distinction when adapting their use for RA CVD screening. In the US, the most frequently utilized risk calculator is the Framingham 10-year hard CVD end-point estimator. More recently , the American College of Cardiology/American Heart Association (ACC/AHA) developed a new set of risk equations from cohort data pooled from multiple large CVD cohort studies (including the original Framingham cohort) that are both geographically and ethnically diverse, at least for the US-based non-Hispanic Whites and African Americans. Those identified as having elevated CVD risk (a 10-year risk of ≥7.5% among those 40–79 years of age for the ACC/AHA risk assessment tool) are intended for more intensive lifestyle and risk factor management, primarily focused on lipid lowering. In the ACC/AHA CVD assessment guideline, additional assessments, such as C-reactive protein (CRP) or CAC scoring, are only recommended for individuals in whom a risk-based treatment decision is uncertain. However, whether these recommendations are applicable to RA patients depends on how accurately CVD risk assessment calculations perform in RA.

Several studies of RA patients have shown that risk calculators, such as Framingham, tend to underestimate CVD risk in RA patients , and that underestimation may be conditioned on certain RA characteristics. In a retrospective study of 525 RA patients followed up for an average of 8.4 years , a total of 84 CVD events occurred among a cohort in which only 46 events were predicted based on baseline assessments of Framingham risk factors. Excess events above the number predicted were observed in both men and women, and excess events occurred primarily among older RA patients (age >55 years), those seropositive for rheumatoid factor, and those with a persistently elevated erythrocyte sedimentation rate (ESR) (defined as three or more measures of >40 mm/h within the first year of RA). Applying a 1.5 multiplication factor for patients with RA disease duration of 10 years or more, seropositivity, or extra-articular disease, as recommended by the European League Against Rheumatism (EULAR) guidelines published in 2010 , did not improve the predictability of the model. The Reynolds risk score is a newer risk calculator that includes CRP in addition to traditional CVD risk factors. In nondiabetic women, the Reynolds risk score showed better calibration and discrimination than Framingham . However, CVD events were also underestimated in RA using the Reynolds risk score , even when women were all hypothetically assigned a high CRP level of 50 mg/L. Underestimation of CVD events in RA was also demonstrated in a European early RA cohort , with up to a third of RA patients classified as “low CVD risk” having a CVD event during follow-up. The specific reasons for the lack of clinical utility for CVD risk calculators in RA are not precisely known. Owing to the importance of lipids in the prediction of CVD events in the general population, most of the risk scores are heavily weighted on these measures. As discussed in an earlier section of this review, the total and LDL cholesterol have been shown to be less predictive of CVD events in RA patients, with some indication that RA patients at greatest risk may actually have the lowest levels for these lipid measures. Thus, an RA-specific risk calculator may need to account for the apparent “lipid paradox” in order to improve prediction. Other reasons for the underestimation of CVD events in RA include the difficulty in estimating cumulative inflammation across the disease course of RA, the lack of inclusion of RA-specific risk factors (such as autoantibody status), and the potential for unmeasured nontraditional risk factors to contribute differently in RA to CVD risk.

Efforts are underway to develop and validate an RA-specific CVD risk calculator. Using data from the CORRONA cohort that included almost 16,000 RA patients and 437 CVD events , four RA-specific characteristics (moderate to high articular disease activity, modified Health Assessment Questionnaire (HAQ) of >0.5, prednisone use, and RA duration of ≥10 years) were significantly associated with short-term CVD events over and above traditional CVD risk factors. However, the incremental discrimination of the model including RA-specific factors over and above the model with traditional CVD risk factors alone, although statistically significant, was relatively modest (i.e., the area under the receiver operator curve was 0.761 vs. 0.726, respectively), suggesting that, at least in terms of prediction, traditional CVD risk factors remain the foundation of CVD risk prediction in RA. It should be noted that some factors, such as continuous measurements of lipids and blood pressure, were not part of CORRONA data collection, and some RA characteristics of note, such as markers of systemic inflammation, were not able to be factored in. Because follow-up was relatively short (median 2.2 years), it is unclear whether the model will be similarly predictive, or whether additional risk factors will be more robustly associated, with a longer time horizon. Notably, erosive disease, autoantibody status, rheumatoid nodules, and treatment with biologic and nonbiologic disease-modifying antirheumatic drugs (DMARDs) did not contribute to CVD risk prediction. Although a step in the right direction, the clinical utility of the CORRONA risk calculator for clinical practice is uncertain. Additional efforts are underway to adapt other risk calculators for RA, such as the SCORE algorithm .

Without a validated RA-specific screening algorithm, a rational approach may be adopted for clinical practice. Any RA patient identified as high risk using a CVD risk calculator appropriate for the background population of the geographic areas should be considered for intensive CVD risk factor and lifestyle management, with additional screening not needed. However, what to do with those at intermediate or apparent low risk is less clear. Intermediate CVD risk patients with high-risk RA characteristics (i.e., seropositivity and/or long-standing, treatment-resistant disease) may be appropriate for second-stage imaging for atherosclerosis. The most sensitive and specific modality for RA patients is not clear; however, in the general population, current ACC/AHA guidelines recommend cardiac CT scanning with the assessment of CAC over carotid ultrasonography . A firm recommendation for RA patients who appear low risk based on CVD risk calculators cannot be made, since while the majority of these patients will not have atherosclerosis, a critical few will. As with those with intermediate CVD risk, some discretion based on RA characteristics can be used, with those with more RA risk factors considered for secondary atherosclerosis imaging, including patients with very low lipid levels who are not treated with lipid-lowering therapy. However, it is unclear where the threshold lies between identifying RA patients with critical subclinical atherosclerosis and the subjugation of an undo number to the cost and radiation of an imaging procedure, combined with the uncertainty of how to best treat subcritical atherosclerosis detected in RA patients with few to no traditional CVD risk factors. Future studies will need to address this issue highly relevant to clinical practice.

Interventions: does control of RA disease activity reduce CVD risk?

vide strong evidence for systemic inflammation as a general contributor to atherogenesis and plaque instability, and they highlight pivotal roles for specific inflammatory cytokines (i.e., TNF-α, IL-β, IL-17, and others) , all of which have been implicated in the RA disease process. Therefore, it is not surprising that therapies aimed at reducing inflammatory burden have been associated with lower CVD event rates. What is less certain is whether any intervention that reduces systemic inflammation is adequate to reduce CVD risk, or whether specific therapeutic agents are preferred for CVD risk reduction. It is important to note that the data relating RA therapeutics to CVD outcomes are only observational, and thus the potential for the magnitude of any detected treatment effect to be affected by confounding is very real. In particular, as the choice of RA therapies in practice may be conditioned on factors that are directly or indirectly associated with an individual׳s CVD risk, confounding by indication must be strongly suspected when interpreting the current literature.

With at least 10 commonly utilized nonbiologic DMARDs and almost as many available biologics, RA therapy in clinical practice is complex. Combination therapies, treatment switches due to inefficacy and intolerance, and the acute and chronic use of both glucocorticoids (oral and injected) and nonsteroidal anti-inflammatory drugs (NSAIDs) affect RA signs and symptoms, but they also may affect CVD risk, with some effects possibly being detrimental even in the setting of improved RA disease activity.