Objective To investigate repeat revascularisation outcomes in patients with rheumatoid arthritis(RA) after percutaneous coronary intervention (PCI).
Methods We performed a single-centre, retrospective matched cohort study of patients with RA matched to non-RA patients post PCI. Primary endpoints were time to target lesion revascularisation (TLR) and target vessel revascularisation (TVR) analysed by Cox proportional hazard shared frailty models.
Results A total of 228 lesions (143 patients) were identified in the RA cohort and matched to 677 control lesions (541 patients). TLR occurred in 33% (n=75) of RA lesions versus 25% (n=166) of control lesions (adjusted HR 1.3; 95% CI 0.97 to 1.8). TVR occurred in 39% (n=89) of RA lesions versus 31% (n=213) of control lesions (adjusted HR 1.15; 95% CI 0.82 to 1.6). There was a significant hazard for TLR (adjusted HR 1.48; 95% CI 1.03 to 2.13) and TVR (adjusted HR 1.55; 95% CI 1.12 to 2.14) when excluding lesions with revascularisation events or follow-up less than 1 year. When stratified by treatment with methotrexate or tumour necrosis factor (TNF) α inhibitors or both at discharge, lesions from patients with RA treated with these agents had similar TVR and TLR as control lesions, whereas lesions from patients with RA not treated with these agents had significantly more TLR and TVR (TLR adjusted HR 1.48; 95% CI 1.08 to 2.03; TVR adjusted HR 1.38; 95% CI 1.04 to 1.84).
Conclusions RA predisposes to repeat revascularisation, specifically in patients followed after the 1-year landmark. In the absence of RA treatments including methotrexate and/or TNFα inhibitors, RA is associated with a 50% increased relative risk of repeat revascularisation following PCI. These findings emphasise the adverse effects of chronic inflammation on the durability of PCI and provide further support for aggressive anti-inflammatory treatment in patients with RA.
Statistics from Altmetric.com
Rheumatoid arthritis (RA) is a chronic inflammatory condition that affects approximately 1% of the population. It is a significant risk factor for cardiovascular disease with a 50% increase in cardiovascular mortality,1 a twofold increase in sudden death2 and a threefold increase in myocardial infarction compared with normal subjects.2–4 For cardiovascular events including myocardial infarction, heart failure, revascularisation and death, absolute risk in patients with RA is equivalent to that in non-RA subjects who are 5–10 years older.5 Most telling, the risk of myocardial infarction in patients with RA is equivalent to that in patients with diabetes mellitus.6
Given the high frequency of cardiovascular disease and atherosclerosis in RA, it is reasonable to expect a substantial need for coronary revascularisation. A small, retrospective case–control study involving 43 patients with RA undergoing percutaneous coronary intervention (PCI) noted acute myocardial infarction as the most common indication with no significant differences in short-term outcomes or clinical characteristics when compared with non-RA patients.7 Outside of this single study, little is known about the outcomes of revascularisation in this population. Specifically, the impact of chronic inflammation in RA on progression of atherosclerosis and restenosis after PCI is unknown.
Using a retrospective matched cohort study, we sought to investigate repeat revascularisation outcomes in patients with RA and hypothesised that the increased inflammation of RA would be associated with increased restenosis after PCI and further progression of atherosclerosis requiring repeat interventions.
We performed a single-centre, retrospective matched cohort study to evaluate repeat revascularisation in patients with and without RA. The Washington University in St. Louis Institutional Review Board approved the study with a waiver of consent. Cases were selected using a searchable electronic medical record database (Clinical Investigation Data Exploration Repository) maintained by Washington University in St. Louis Center for Biomedical Informatics. The Barnes Jewish Hospital (BJH) electronic medical record was queried using ICD-9 codes for RA (714.0 and 714.2) and the presence of cardiac catheterisation reports between the years 1996 and 2012. This population was further refined by manual review to include only those patients who had undergone PCI (n=143). Lesions were identified for inclusion if they met the following criteria: first lesion in an epicardial coronary artery and no prior history of PCI in that vessel. Each RA lesion was matched to a control lesion, and only one lesion was matched for each epicardial coronary artery or branch. Bifurcation lesions were treated individually if PCI was performed on the side branch. RA lesions were matched to control lesions that also met the above inclusion requirements based on the following variables: drug eluting stent (DES), bare metal stent (BMS), presence of diabetes mellitus at the time of PCI, age at PCI within 10 years, date of index PCI within 2 years and index PCI preformed on a saphenous vein graft lesion. Lesions were matched by manual review of the medical record by the primary author (MAS). If multiple control lesions met the criteria for matching, gender and specific DES type were prioritised. Conservative power calculations for repeat revascularisation were performed using event rates in patients with diabetes mellitus who had PCI performed with a second-generation DES.8 We determined that three control lesions were required per RA lesion to have appropriate statistical power.
Each lesion was followed until a repeat revascularisation event or the last clinical encounter in the medical record up to 31 December 2013. The primary endpoints were target lesion revascularisation (TLR) and target vessel revascularisation (TVR). TLR was defined as any revascularisation procedure (PCI or coronary artery bypass surgery) for stenosis at the site of index PCI. TVR was defined as any revascularisation procedure (PCI or coronary artery bypass surgery) performed on the same epicardial vessel as the index PCI. Patients who met a TLR endpoint were also considered to have met a TVR endpoint. To evaluate the effects of exposure to chronic inflammation and reduce procedure-related confounders, we examined the association of RA with revascularisation occurring more than 1 year post index PCI; lesions were excluded if they had a revascularisation event prior to 1 year or were followed for less than 1 year. All data was extracted by manual review of the medical record (MAS and CTS) and recorded directly in Research Electronic Data Capture (RedCaps) database maintained at Washington University in St. Louis on a case-by-case basis. Data quality was monitored and checked by the primary author (MAS).
Baseline demographics and medical history were compared on a per-patient basis between groups with χ2 testing for categorical variables and independent Student's t test when appropriate. Procedural variables and medications at discharge were analysed on a per lesion basis. Logistic and normal linear models for dichotomous and continuous variables, respectively, were created using generalised estimating equations to account for clustered data. Robust SE estimates were used to test for differences between groups. Number of stents was analysed using a Poisson distribution scaled to account for overdispersion. Kaplan–Meier analysis was performed for the primary endpoints but given the clustered nature of the data (ie, patients having more than one lesion), shared frailty multivariable Cox proportional hazards models were created to evaluate the significance of primary endpoints. Multivariable models were constructed to account for procedural differences or known risk factors for restenosis after PCI including stent length, overlapping stents, bifurcation stenting, BMS use and intervention on a saphenous vein graft. Finally, using separate Cox proportional hazards models for TVR only, interaction analysis was performed to examine the interaction of various categorical variables on TVR between RA and control lesions. All tests were two sided with an accepted type I error rate of 5%.
A total of 228 lesions were identified in 143 patients with RA and were matched to 677 lesions in 541 control patients. All RA lesions were matched to at least one control lesion. Thirty-four patients with administrative coding of RA did not have medical record confirmation of clinical RA, and 24 had incomplete or missing discharge medication information. Twelve RA lesions had an unknown stent type and were matched to DES as all unknown lesions were after 2003. Demographic characteristics between RA and control patients were not significantly different (table 1).
Patients included had PCI performed between 1996 and 2012 (median year, 2005; see online supplementary figure S1). There were no differences in PCI indication between RA and control lesions or in cardiovascular medications or lipid measurements at discharge (table 2). More than 50% of the cohort had a repeat cardiac catheterisation at some point after the index PCI, with the RA cohort having less repeat catheterisations (table 2). Sixty per cent of the lesions were treated with DES (table 3); 65% of all DES were first-generation DES, and there was no significant difference between RA or control lesions with respect to the type of stent used (first-generation stent: RA 72% vs control 66%; p=0.28). Both groups were well matched for procedural variables including number of stents, total stent length and lesion complexity (table 3).
TLR occurred in 33% (n=75) of RA lesions versus 25% (n=166) of control lesions (unadjusted HR 1.3; 95% CI 0.88 to 2.0). TVR occurred in 39% (n=89) of RA lesions versus 31% (n=213) of control lesions (unadjusted HR 1.2; 95% CI 0.78 to 1.7). Time to event analysis and shared frailty Cox proportional hazard models showed a trend towards increased TLR and TVR in lesions in patients with RA compared with controls, although these differences did not reach statistical significance (figure 1). To determine if any predictors of TVR were more important in RA lesions, an interaction analysis was performed. Interaction analysis for TVR showed lesions in female patients with RA were at significantly elevated risk of TVR compared with lesions in males (figure 2). Diabetes, BMS use or bifurcation lesions did not have an important interaction in RA lesions. In an attempt to focus on repeat revascularisations related to chronic inflammation and to reduce procedure-related confounders that could predispose to TVR or TLR, a secondary analysis was performed. This landmark analysis excluded any control or RA lesion followed for less than 1 year or any lesion that had a TLR or TVR event prior to 1 year. After excluding these lesions, RA lesions were associated with a significant hazard for TLR and TVR compared with control lesions, and this hazard persisted after adjusting for stent length, overlapping stents, bifurcation stenting, BMS use and intervention on a saphenous vein graft (table 4). Diabetes mellitus in the control cohort was associated with a significant increased hazard for both TVR (adjusted HR 1.53; 95% CI 1.1 to 2.1) and TLR (adjusted HR 1.33; 95% CI 1.0 to 1.77). In the landmark analysis, this hazard was no longer significant in diabetic lesions (TVR adjusted HR1.25; 95% CI 0.8 to 1.95; TLR adjusted HR 1.06; 95% CI 0.72 to 1.55).
Treatment effects of RA medications at discharge were then explored for lesions in patients with RA exposed to methotrexate (MTX), tumour necrosis factor (TNF) α inhibitor treatments (anti-TNF) or both compared with other RA lesions and control lesions (see online supplement table S1). Both drugs have been associated with improved cardiovascular outcomes in patients with RA.9–13 Lesions exposed to MTX or anti-TNF at discharge had TLR performed in 19% (n=11) compared with 25% (n=166) of controls and 38% (n=64) of other RA lesions (p=0.02). Lesions exposed to MTX or anti-TNF at discharge had TVR performed in 19% (n=11) compared with 31% (n=78) of controls and 46% (n=213) of other RA lesions (p<0.01). Time to event analysis also showed significant differences between the three groups (figure 3). After adjusting for procedural differences between groups including stent length, overlapping stents, bifurcation stenting, BMS use and intervention on saphenous vein graft, a significant hazard remained for TLR and TVR in RA lesions not treated with MTX or with anti-TNF (figure 3). After the 1-year landmark, the trend for increased TLR and TVR remained, and the rate of TVR was significantly elevated in those RA lesions not treated with MTX or with anti-TNF compared with control lesions (adjusted HR 1.6; 95% CI 1.13 to 2.28; see online supplement table S2).
While RA is known to be a potent risk factor for the development of atherosclerosis, our study is the first to examine the impact of RA on repeat revascularisation outcomes after PCI. Lesions present in patients with RA showed a strong trend towards an overall increased risk of TLR and TVR despite these patients being less likely to have repeat angiography. When lesions were analysed after the 1-year landmark, there was a significant hazard for TLR and TVR in those with RA. Patients with RA treated with the potent anti-inflammatory drugs MTX or anti-TNF at discharge had a hazard for repeat revascularisation similar to matched controls, whereas patients with RA who did not receive these anti-inflammatory treatments at discharge had a significantly elevated hazard for repeat revascularisation. Together these findings suggest that there are adverse long-term outcomes after PCI in patients with RA, and these adverse outcomes are likely the result of chronic inflammation.
RA predisposes to atherosclerosis, possibly due to the pro-atherogenic effect of chronic exposure to inflammatory cytokines.14 In support of this hypothesis, RA disease severity and inflammatory biomarkers have been associated with atherosclerosis and cardiovascular events independent of traditional cardiovascular risk factors.15–18 Models that predict cardiovascular events are markedly improved by including measures of RA disease severity and activity, both likely representing surrogates for levels of inflammation.18 Treatment of inflammation has been associated with regression in carotid atherosclerosis19 and improved cardiovascular outcomes in patients with RA.9 ,10 A meta-analysis of MTX treatment in prospective studies totalling nearly 66 000 patients with RA was associated with a 21% relative risk reduction in cardiovascular events.10 Although not as robust, TNFα inhibitors may favourably impact cardiovascular events,11 ,12 with the most significant impact noted on stroke.13 The combination of both agents is commonly used in the management of severe RA and may also reduce cardiovascular events.17 Here we present new data to support additional benefits of intense anti-inflammatory therapy in patients with RA. Specifically, the use of MTX and TNFα inhibitors was associated with a reduced risk of repeat revascularisation after PCI. These findings also highlight the need to understand systemic disease processes and their potential impact on the durability of PCI.
Restenosis after PCI has been demonstrated to be an inflammation mediated process20 ,21 with the resultant neointimal scar22 causing narrowing at the site of the previously treated stenosis. Prior studies have shown inhibition of inflammation after BMS implantation with systemic steroids23 and coronary stents covered with MTX can reduce neointimal proliferation.23 Inhibition of inflammation by the use of stents coated with potent antiproliferative medications (ie, first and second-generation DES) has made the most significant impact on restenosis, but certain populations still remain at risk. Patients with diabetes mellitus carry an increased risk of revascularisation even with use of second-generation DES,8 and many have attributed this risk to the chronic inflammatory state of diabetes mellitus.14 Diabetes was significantly associated with repeat revascularisation in the control arm and provides external validity for our data. Other risk factors for restenosis include multiple stents, a small luminal diameter after PCI, increased lesion length, increased lesion complexity and prior bypass surgery.24 ,25 Additionally, a late restenosis ‘catch up’ phenomenon has been observed with DES,26 and speculative causes cited include hypersensitivity to drug polymers26 and neoatherosclerosis.27 Restenosis and repeat revascularisation usually occur within the first year after PCI.28 ,29 Procedural factors associated with repeat revascularisation such as poor stent apposition, geographical miss and incomplete revascularisation largely manifest within this 1-year time frame. In our secondary 1-year landmark analysis, we excluded these potentially confounding events to more uniformly study the effect of chronic inflammation. In this analysis, RA was associated with a significant increase in repeat revascularisation which persisted after adjustment for other procedural factors associated with restenosis. Interestingly, diabetes was not associated with an increased hazard for revascularisation after the year landmark, suggesting that there may be different aetiologies for repeat revascularisation among different groups. This data suggests an alternative explanation for late restenosis events; chronic inflammation may contribute to the progression of neointima or neoatherosclerosis. This secondary analysis also further highlights the risk for adverse outcomes after PCI in patients with RA.
Women with RA appeared to have a higher risk of TVR in our interaction analysis. When compared with men, women more frequently have positive remodelling in the presence of atherosclerosis30 and smaller coronary vessel size. Positive remodelling has been associated with increased risk of repeat revascularisation.31 Female gender has been a known risk factor for poor response to RA treatments including MTX32 and TNFα inhibitors.33 The mechanism behind this poor response is unclear, but a poor response to anti-inflammatory therapies may contribute to poor disease control and increased inflammation. Predisposition to positive remodelling and elevated inflammation contributing to neointimal proliferation provide a speculative explanation for the observed results in this study and lend further support to the hypothesis that inflammation after PCI leads to increased repeat revascularisation.
Interest in anti-inflammatory therapy for the prevention of atherosclerosis has intensified substantially. In the Justification for the Use of Statins in Primary Prevention: An Intervention Trial Evaluating Rosuvastatin trial (JUPITER), published in 2008 by Ridker et al,34 highlighted the potential importance of inflammation in mediating cardiovascular events. The Cardiovascular Inflammation Reduction Trial (CIRT) is currently enrolling high-risk non-RA patients to receive very low dose MTX or placebo with the primary endpoint being secondary prevention of myocardial infarction, stroke or cardiovascular death. Our study provides further evidence that MTX may be beneficial in reducing cardiovascular events, specifically in patients with RA after PCI. Patients with systemic lupus erythematosus, psoriatic arthritis and inflammatory bowel disease, or populations with chronically elevated biomarkers of inflammation may also benefit from similar therapeutic strategies to reduce inflammation. Inflammation is important in increasing the risk of de-novo cardiovascular events and appears to be important in prognosticating the outcomes of revascularisation therapies, namely PCI. Both have substantial clinical implications for maximising the durability of PCI.
These conclusions are limited by the single-centre, retrospective study design, and we do not have access to repeat revascularisations preformed at outside institutions. RA was defined by administrative database codes, and as such we are limited in the assessment of disease severity or the authenticity of this diagnosis, although a majority were treated with at least one medication known to be used in RA management. We cannot rule out the possibility that disease misclassification, missing information and undertreatment of RA may bias our results toward the null, suggesting that the effects seen in this study likely represent conservative risk estimates. Patients treated with MTX or TNFα inhibitors were different, and although we attempted to control for differences known to affect restenosis, there may be residual confounding. Finally, the longitudinal use of medication, assessment of RA disease activity and biomarkers of inflammation, were not available for analysis; data that will be informative to examine in future efforts. A prospective trial of patients with RA after PCI should be considered to more specifically address the benefits of anti-inflammatory therapies and cardiovascular outcomes.
Patients with RA have an increased need for revascularisation after PCI, especially female patients and patients followed after the 1-year landmark. Treatment with MTX or TNFα inhibitors at discharge from the initial PCI was associated with a reduced risk of repeat revascularisation compared with non-treated patients with RA and was similar to matched controls. In patients with RA not receiving these therapies, the risk of repeat revascularisation was significantly higher than in matched controls. These findings emphasise the adverse effects of chronic inflammation on outcomes after PCI and provide further support for aggressive anti-inflammatory treatment in patients with RA with atherosclerosis.
What is already known on this subject?
There are no studies examining the long-term durability of percutaneous coronary interventions (PCIs) in patients with a chronic inflammatory disease such as rheumatoid arthritis (RA).
What might this study add?
This study highlights the increased risk of repeat revascularisation in patients with RA and suggests inflammation is the major contributor to this increased risk. RA lesions were associated with an increased risk of revascularisation compared with control, especially when excluding lesions with revascularisation events or follow-up less than 1 year and in those patients not treated with potent anti-inflammatory medications at discharge.
How might this impact on clinical practice?
RA is a known risk factor for cardiovascular disease, and here we show that it is also a risk factor for adverse outcomes after PCI, namely, an increased risk of repeat revascularisation. Clinicians treating this population should be aware of this risk when evaluating patients after PCI.
Clinical Investigation Data Exploration Repository (CIDER) maintained by Washington University in St. Louis Center for Biomedical Informatics and funded by NIH CTSA Grant # UL1 TR000. NIH Clinical Training Grant 5T32HL007081-39.
▸ Additional material is published online only. To view, please visit the journal online (http://dx.doi.org/10.1136/heartjnl-2015-308634).
Contributors All authors have contributed substantively to this work, have read the manuscript and approved its content.
Funding This work was supported by NIH grants 5T32HL007081-39 (MAS).
Competing interests None declared.
Ethics approval Institutional Review Board.
Provenance and peer review Not commissioned; externally peer reviewed.