Introduction
Rheumatoid arthritis (RA) is a chronic inflammatory musculoskeletal disease that affects ~0.8% of the adult population. RA also associates with an increased risk for cardiovascular disease (CVD) [
1], which is only partially explained by traditional CVD risk factors [
2]. The inflammatory processes of RA and CVD are remarkably similar, suggesting that RA disease-related inflammation might contribute to the excess CVD risk [
3].
The vascular endothelium is responsible for maintaining an atheroprotective environment through the release of vasoactive factors, particularly nitric oxide (NO). Noninvasive assessments of peripheral endothelial function in the microcirculation and the macrocirculation may predict adverse cardiovascular outcomes in patients at risk of developing, or with prevalent CVD [
4]. Endothelial cells differ in structure and phenotype depending on vessel type [
5], and heterogeneous responses to
in vitro stimulation are displayed in different vascular beds [
6], suggesting that endothelial dysfunction (ED) may occur differentially in different vascular beds [
6]. Evidence suggests that coronary microvascular disease is apparent in the absence of macrovascular disease in RA [
7], and we previously showed that microvascular and macrovascular endothelial function are independent of each other in this population [
8]. These findings highlight the importance of assessing endothelial function in different vascular beds.
A number of studies have reported the presence of microvascular and macrovascular ED in RA patients relative to age- and sex-matched healthy controls, which can be improved after treatment with antiinflammatory medications (such as anti-tumor necrosis factor-alpha (anti-TNF-α) [
9]. Even though an association between endothelial function and inflammation is often assumed, surprisingly few studies have actually examined this [
9]. Cross-sectional studies report equivocal findings; some find an association between inflammatory markers and endothelial function [
10], whereas others do not [
11]. Longitudinal studies revealed no correlation between changes in inflammatory markers and changes in endothelial function [
12,
13], and the improvement in microvascular endothelial function after treatment occurred without any change in inflammatory markers [
14]. Thus, it remains possible that endothelial function in RA is determined by factors other than systemic inflammation, such as CVD risk factors [
9].
The prevalence of classical CVD risk factors is increased in patients with RA [
2], and their control is worse [
15], compared with that of the general population [
16]. At present, no studies have examined associations between classical CVD risk factors in the microcirculation, and only a very few studies have been conducted in the macrocirculation, with inconsistent findings [
9]. CVD risk can be calculated by incorporating individual CVD risk factors into algorithms to yield "global" CVD risk scores. Examples of such algorithms are the Framingham Risk Score (FRS) [
17] and the Systematic Coronary Risk Evaluation (SCORE) [
18]. However, these algorithms do not account for systemic inflammation, whereas the more recent Reynolds risk score does, as it includes C-reactive protein (CRP) in the algorithm. In addition, QRISK2 is the only CVD-risk algorithm that incorporates the presence of RA, as well as socioeconomic status and ethnicity [
19]. To our knowledge, the relation between these risk algorithms and microvascular and macrovascular endothelial function has not been investigated in RA.
The objectives of the present studies were to examine determinants of microvascular and macrovascular endothelial function in RA, with a specific focus on disease-related markers of inflammation and classical CVD risk.
Discussion
These findings revealed that, in the cross-sectional study, disease-related markers of systemic inflammation were not associated with microvascular or macrovascular endothelial function in RA. Similarly, changes in inflammatory markers did not consistently relate to change in vascular function after anti-TNF-α treatment, despite a transient improvement in microvascular endothelial function. Global CVD risk, assessed by using several different algorithms, inversely correlated with microvascular endothelium-dependent function and macrovascular endothelium-independent function in the cross-sectional study, and some classical CVD risk factors improved after anti-TNF-α treatment.
RA is characterized by increased systemic inflammation, which has been hypothesized to affect the vasculature and contribute to accelerated atherosclerosis [
30]. However, the present findings suggest that systemic markers of inflammation (ESR, CRP, and DAS28) do not relate to endothelial function in two separate vascular beds, in a cross-sectional or longitudinal study. It is worth noting that a trend was seen for an association between microvascular endothelium-dependent function and CRP in the cross-sectional study; however, given the number of associations that were examined, this could be a chance finding. In the longitudinal study, ΔCRP was associated with change in macrovascular endothelium-dependent function at 2 weeks, but because macrovascular endothelial function did not significantly improve with anti-TNF-α treatment, it is difficult to interpret the implication of this association. In the microvasculature, conflicting associations between endothelial function and disease-related markers of inflammation are reported: one study found an association with TNF-α but not with ESR or CRP [
31]; in another, endothelial function was associated with CRP only [
10]; and yet another study reported no associations at all [
11]. Similarly, a number of studies in the macrovasculature reported no associations between disease-related markers of inflammation and macrovascular endothelium-dependent function [
32,
33]. Studies that have reported associations present an inconsistent picture, with some finding an association with CRP but not ESR [
34], or with DAS28 only [
35]. Likewise, the majority of longitudinal studies found no correlations between change in disease-related markers of inflammation and change in microvascular and macrovascular endothelial function in response to antiinflammatory treatment, with only one report of an association for macrovascular endothelium-dependent function [
9]. Collectively, these findings suggest that the relation between endothelial function and disease-related markers of inflammation may not be as strong as previously suggested.
The transient improvement in microvascular endothelium-dependent function after treatment with anti-TNF-α resonates with the findings of a previous study. Komai and colleagues [
36] found improvements in microvascular endothelium-dependent function after 2 weeks of treatment with anti-TNF-α when assessing microvascular endothelium-dependent function. In addition, although lower than that at 2 weeks, endothelial function was still increased after 6 weeks of treatment [
36]. Therefore, it is possible that after an initial improvement, a gradual decrease in microvascular endothelial function occurs between 6 weeks and 12 weeks. Without a 6-week assessment in the current study, this must remain speculation. In contrast, Hansel and colleagues [
37] observed no change in microvascular endothelium-dependent function at 2 weeks in RA patients; however, these patients exhibited consistently lower disease-related inflammation than did the present cohort. Given that elevated baseline inflammation was associated with greater change in endothelial function, it is possible that change in endothelial function after treatment is unlikely in patients with low disease-related inflammation, and this may explain the seemingly contrasting findings. In line with our findings, in patients with high baseline disease-related inflammation, short-term improvement in microvascular endothelial function after anti-inflammatory treatment has been reported [
38]. Although no long-term follow-up period was applied, the findings support the notion that, in patients with elevated inflammatory markers, acute improvement in microvascular endothelial function is more likely than longer-term improvements after treatment with anti-TNF-α.
Macrovascular endothelium-dependent function did not change in response to 3 months of anti-TNF-α treatment, which is in contrast to previous studies [
39‐
41]. Comparison of the patients included in these studies revealed that average baseline macrovascular endothelium-dependent function was better in the current study than in previous RA patient samples (9.4% versus 2.8% to 7.0%, respectively). Studies that included a healthy control group showed lower baseline macrovascular endothelium-dependent function in the RA patients [
39,
42]. However, in the present study, the baseline macrovascular endothelium-dependent function of the RA patients was similar to that of healthy control participants (data not shown), therefore making an improvement in response to anti-TNF-α unlikely. Thus, it is possible that treatment with anti-TNF-α may have less impact in RA patients who have similar endothelial function than in healthy individuals.
The majority of individual classical CVD risk factors did not associate with endothelium-dependent function in RA in the cross-sectional study, with age being an exception. In the general population, the association between classical CVD risk factors and endothelial function has been well characterized [
43]. To our knowledge, no study of RA patients has examined the impact of classical CVD risk factors on microvascular endothelial function, and only a few studies have examined the effects of classical CVD risk factors on macrovascular endothelial function [
32,
34,
44]. Associations between endothelium-dependent function and lipid levels were present in some [
34,
44], but not all [
32] studies. In the longitudinal study, classical CVD risk factors (SBP, DBP, and HDL cholesterol) improved after reducing RA disease-related inflammation, which is in agreement with other studies [
45]. This suggests that inflammation might contribute to the development of classical CVD risk factors in patients with RA. Despite the improvement in classical CVD risk factors with anti-TNF-α therapy, it should be noted that no direct associations were found between the improvement in disease activity and risk factors.
The observation that microvascular but not macrovascular endothelium-dependent function most commonly associated with global CVD risk algorithms and that only microvascular endothelium-dependent function changed after treatment with anti-TNF-α highlights the importance of examining endothelial function in more than one vascular bed. Microvessels make up a much larger proportion of the vasculature than do macrovessels, and may therefore have greater exposure to injurious stimuli [
46]. Consequently, it is possible that even small changes in global CVD risk could have a greater effect on microvascular endothelium-dependent function. Therefore, assessments that examine both vascular beds may provide more meaningful clinical information on vascular risk in RA. Even though ED has been associated with clinical end points in patients with cardiovascular disease [
4], only one such study with a small sample has reported that high levels of carotid artery intima-media thickness (a subclinical measure of atherosclerosis) are predictive of hard cardiac end points in RA [
47]. Therefore, further research is necessary to explore which vascular assessment is the best predictor of cardiac end points in RA patients.
In the macrocirculation, endothelium-independent function was associated with the FRS, Reynolds Risk Score, QRISK 2, insulin resistance, SBP, presence of high cholesterol levels, and hypertension. Evidence indicates that, in healthy individuals with CVD risk factors, abnormalities in endothelium-independent function may occur in the absence of abnormalities in endothelium-dependent function [
48]. Further, macrovascular endothelium-independent function, but not macrovascular endothelium-dependent function, was related to a reduction in SBP after 12 and 24 weeks of treatment in patients with hypertension [
49]. These findings indicate that CVD risk factors may differentially affect endothelial and smooth muscle cell function, particularly in the macrovasculature. In the current study, endothelium-independent function was significantly lower in patients with hypertension, and some evidence suggests that CVD risk factors like hypertension degrade cyclic guanosine monophosphate (cGMP) [
50], a second messenger responsible for the relaxation of vascular smooth muscle cells. In addition,
in vitro studies have shown that soluble guanylyl cyclase, an enzyme responsible for activating cGMP, has reduced sensitivity to NO in hypertensive rats [
51]. This means that, even if adequate NO is released from the endothelial cells, abnormalities in smooth muscle cell signalling could still lead to a reduced vasodilatory response. Thus, examination of vascular function should include combined assessments of endothelial function and smooth muscle cell function.
The strength of the present work was the inclusion of a large cohort of RA patients at baseline, and the longitudinal assessment of patients newly starting anti-TNF-α treatment. Such a study design allowed the investigation of specific predictors of endothelial function at baseline and over a protracted timescale. Another strength of the study was the extensive characterization of CVD risk by using multiple global CVD risk scores (FRS, TCSCORE, TC:HDL SCORE, Reynolds risk score, and QRISK2), along with comprehensive assessment of CVD risk factors, including HOMA and QUICKI. Unfortunately, the longitudinal study did not have a no-treatment RA control group, for obvious ethical reasons. It is acknowledged that the inclusion of a patient group taking stable medication could have strengthened the design of the study, as it would have allowed exploration of potential fluctuations in endothelial function. However, such a control group is likely to have lower baseline levels of disease-related inflammation, making comparisons to patients with active inflammation difficult. Additional limitations include the inability to examine the effects of other disease-modifying antirheumatic drugs, as well as treatment with corticosteroids. Further research is necessary to establish whether alternative anti-inflammatory medications may exert different effects on endothelial function. Furthermore, the use of other biomarkers of inflammation (for example, CD40, interleukin 6) may help to unravel important pathways that may contribute to vascular pathology in RA. Finally, the etiology of RA has a considerable genetic component in particular; certain human leukocyte antigen alleles appear to associate with worse macrovascular endothelium-dependent function in this group of patients [
52]. More recent studies have reported that deletion of certain protective genes (such as
CCR5Δ
32) can result in lower macrovascular endothelium-dependent function in RA [
53]. Thus, further work examining the genetic influence on endothelial function is required.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AS participated in the design of the study, recruited patients, performed the vascular assessments, conducted data analysis, and drafted the manuscript. GK participated in the design of the study, helped with data analysis and in drafting the manuscript. DC participated in the design of the study and helped with data analysis. JVvZ participated in the design of the study, and helped with data analysis and in drafting the manuscript. All authors read and approved the final manuscript.