Article Text

Download PDFPDF

Effect of a treat-to-target intervention of cardiovascular risk factors on subclinical and clinical atherosclerosis in rheumatoid arthritis: a randomised clinical trial
  1. Benjamin Burggraaf1,
  2. Deborah F van Breukelen-van der Stoep2,
  3. Marijke A de Vries1,
  4. Boudewijn Klop1,
  5. Anho H Liem3,
  6. Gert-Jan M van de Geijn4,
  7. Noelle van der Meulen1,
  8. Erwin Birnie5,
  9. Ellen M van der Zwan4,
  10. Jende van Zeben2,
  11. Manuel Castro Cabezas1
  1. 1 Department of Internal Medicine, Center for Diabetes and Vascular Medicine, Franciscus Gasthuis & Vlietland, Rotterdam, The Netherlands
  2. 2 Department of Rheumatology, Franciscus Gasthuis & Vlietland, Rotterdam, The Netherlands
  3. 3 Department of Cardiology, Franciscus Gasthuis & Vlietland, Rotterdam, The Netherlands
  4. 4 Department of Clinical Chemistry, Franciscus Gasthuis & Vlietland, Rotterdam, The Netherlands
  5. 5 Department of Statistics and Education, Franciscus Gasthuis & Vlietland, Rotterdam, The Netherlands
  1. Correspondence to Dr Benjamin Burggraaf, Department of Internal Medicine, Center for Diabetes and Vascular Medicine, Franciscus Gasthuis & Vlietland, Rotterdam, 3045 PM, The Netherlands; b.burggraaf{at}


Background Patients with rheumatoid arthritis (RA) have an increased risk for cardiovascular disease (CVD). No long-term intervention trials on CVD risk factors have been published, and a debate on the efficacy of controlling traditional risk factors in RA is ongoing. We aimed to evaluate a treat-to-target approach versus usual care regarding traditional CVD risk factors in patients with RA.

Methods In this open-label, randomised controlled trial, patients with RA aged <70 years without prior CVD or diabetes mellitus were randomised 1:1 to either a treat-to-target approach or usual care of traditional CVD risk factors. The primary outcome was defined as change in carotid intima media thickness (cIMT) over 5 years, and the secondary outcome was a composite of first occurrence of fatal and non-fatal cardiovascular events.

Results A total of 320 patients (mean age 52.4 years; 69.7% female) with RA underwent randomisation and 219 patients (68.4%) completed 5 years of follow-up. The mean cIMT progression was significantly reduced in the treat-to-target group compared with usual care (0.023 [95% CI 0.011 to 0.036] mm vs 0.045 [95% CI 0.030 to 0.059] mm; p=0.028). Cardiovascular events occurred in 2 (1.3%) of the patients in the treat-to-target group vs 7 (4.7%) in those receiving usual care (p=0.048 by log-rank test).

Conclusion This study provides evidence on the benefit of a treat-to-target approach of traditional CVD risk factors for primary prevention in patients with well-treated RA.

Trial registration number NTR3873.

  • atherosclerosis
  • cardiovascular risk
  • intervention
  • rheumatoid arthritis

Statistics from

Key messages

What is already known about this subject?

  • Patients with rheumatoid arthritis have an increased risk for cardiovascular disease; however, no studies on interventions aiming at traditional cardiovascular risk factors in these patients have been published so far.

What does this study add?

  • This randomised controlled trial in patients with rheumatoid arthritis suggests that a treat-to-target intervention of traditional cardiovascular risk factors leads to reduced subclinical and clinical atherosclerosis.

How might this impact on clinical practice or future developments?

  • These findings provide for the first time evidence on the efficacy of cardiovascular risk management in patients with rheumatoid arthritis.


Rheumatoid arthritis (RA) has been associated with an increased risk for cardiovascular disease (CVD).1–4 The presence of RA has been linked to a 50% increase in risk for CVD compared with the general population.4 5 Since atherosclerosis is, in part, an inflammatory process,6 the ongoing inflammation in RA may play an important role in the development of atherosclerosis.7 8 Several studies have shown that effective treatment of disease activity with immunomodulatory drugs in RA is associated with a reduced cardiovascular risk.9–14

Besides inflammation, traditional CVD risk factors contribute to the increased cardiovascular risk in RA. There is a higher prevalence of hypercholesterolaemia, hypertension, smoking and obesity,15 16 and these risk factors are underdiagnosed and undertreated in patients with RA.17 A recent study showed that 30% of the CVD events were attributed to RA characteristics and 49% to traditional CVD risk factors.18 The importance of these risk factors has been strengthened by guidelines recommending CVD risk estimation in patients with RA by modification of the traditional risk models.19 20 Due to these adaptations, more patients with RA will be identified and treated for their CVD risk. Limited data on the effect of primary prevention in patients with RA are available. Several population-based studies showed a relation between use of statins and a reduced cardiovascular risk.21 22 To date, however, the effect of a multifactorial intervention of traditional CVD risk factors on clinical and subclinical atherosclerosis in RA has not been reported.

In the present study, we aimed to evaluate the effect of a predefined treat-to-target intervention of cardiovascular risk factors on structural vascular changes and cardiovascular outcomes compared with usual care.

Materials and methods

Study design and population

As described previously,23 the Franciscus Rheumatoid Arthritis and Cardiovascular Intervention Study (FRANCIS)is an open-label, single-centre, randomised clinical trial.

Patients aged ≤70 years with RA (defined according to the American College of Rheumatology 1987 criteria24) attending the rheumatology outpatient clinic at Franciscus Gasthuis were consecutively invited to participate. Exclusion criteria were presence of diabetes mellitus (either a documented diagnosis or a haemoglobin A1c [HbA1c] >48 mmol/mol) or CVD. The latter was defined as a documented history of myocardial infarction, cerebrovascular disease, amputation due to peripheral artery disease, intermittent claudication, or a prior percutaneous transluminal coronary angioplasty or coronary artery bypass graft. In addition, kidney disease, defined as an estimated glomerular filtration rate <40 mL/min/1.73 m2, was an exclusion criterion.

Before randomisation, the CVD risk score was calculated according to the 2006 unadjusted Systematic Coronary Risk Evaluation (SCORE) risk assessment.25 Patients with a 10-year CVD risk <10% were randomised (1:1) to either treat-to-target intervention or usual care by means of opaque sealed envelopes. Patients with a 10-year CVD risk ≥10% were included in a separate ‘high risk cohort’.


All patients visited the diabetes and vascular centre outpatient clinic twice a year; a standardised set of non-fasting blood samples were obtained at these visits and anthropomorphic measurements (eg, blood pressure, weight and height) were recorded. Blood samples at baseline were obtained after an overnight fast. Carotid ultrasound scans were carried out once a year using the ART-LAB (Esaote, Italy) by a trained and experienced sonographer as described earlier.26 Two different assessors measured the carotid intima media thickness (cIMT) of the same patients throughout the trial (intraobserver intraclass correlation coefficient [ICC] 0.95, smallest detectable change [SDC] 0.0025 mm; interobserver ICC 0.91, SDC 0.053 mm). An atherosclerotic plaque was defined as a lesion with a focal cIMT of 1.0 mm or more, with a localised protrusion of the vessel wall into the lumen.

Treatment for RA for all participants was carried out by the patient’s own rheumatologist, aiming for RA disease remission, following state-of-the-art treatment as described in the most recent RA guidelines.27 28 No limitation was set on the type of treatment. RA disease activity was assessed by the Disease Activity Score (DAS28) with 28 joints evaluated and the erythrocyte sedimentation ratio (ESR) was scored at every visit. Remission was classified as either DAS-based or Boolean-based.29 Low disease activity was defined as DAS28 ≤3.3.

Concerning CVD risk management, patients with RA in the usual care group were only evaluated and their general physician received written advice based on the results of the visits; however, initiation of treatment was based on their own judgement and CVD risk guidelines.

Patients with RA within the treat-to-target group were treated according to a prespecified protocol. Antihypertensive treatment was initiated when hypertension was diagnosed. Hypertension was defined as a mean blood pressure of >140/90 mm Hg obtained from measurements over 15 min or, if necessary, from a 24-hour ambulant blood pressure measurement. The first choice of treatment was perindopril 4 mg once daily, and when the treatment target was not met the dosage was increased to 8 mg once daily. Other agents were added to reach the treatment target if necessary, first, with the addition of indapamide 2.5 mg once daily, followed by the addition of a dihydropyridine calcium channel blocker.

Treatment with lipid-lowering drugs was initiated when either low-density lipoprotein cholesterol (LDL-C) was >3.0 mmol/L, apolipoprotein B >0.90 g/L or triglycerides >2.20 mmol/L. Treatment was initiated with simvastatin 40 mg once daily, and if the targets were not met the dosage was increased to a maximum of 80 mg once daily or ezetimibe 10 mg once daily was added. In the case of an isolated hypertriglyceridaemia (triglycerides >2.20 mmol/L), bezafibrate 400 mg once daily was initiated.

Dietary and lifestyle advice was given in case of low high-density lipoprotein cholesterol (≤1.2 mmol/L for women and ≤1.0 mmol/L for men) and/or a body mass index >25 kg/m2. In the case of smoking, cessation was advised and a referral to the outpatient clinic for smoking cessation was offered.

In the case of HbA1c >48 mmol/mol, patients were treated initially with metformin, and when the response was insufficient additional drugs were added.


The primary outcome was progression of cIMT between baseline and 5 years of follow-up. The secondary outcome was a composite endpoint of first occurrence of death by cardiovascular causes, non-fatal myocardial infarction, non-fatal stroke, coronary artery bypass grafting, percutaneous coronary intervention, revascularisation of peripheral atherosclerotic arterial disease or amputation related to peripheral atherosclerotic arterial disease between baseline and 5 years of follow-up. If a patient experienced more than one cardiovascular event, only the first was included in the analyses.

Endpoints were assessed by the study team by reviewing the medical file and information from the general practitioner. Assessment of endpoints was blinded and by means of consensus.

Statistical analysis

This study was primarily designed to show a reduced progression of the cIMT in the treat-to-target group compared with usual care. Assuming an average increase of 0.025±0.015 mm by usual care and 0.020±0.015 mm by the treat-to-target intervention, with a 1:1 randomisation ratio, a type 1 error of 0.05 (two-sided) and a statistical power of 0.80, a total of 2×143 patients should be included in the analysis.

To account for the high dropout, all missing cIMT values after 5 years of follow-up were imputed with all available cIMT values (baseline, years 1–4) and treatment group as covariate. Assuming that unobserved cIMT values were missing at random, missing data were imputed with multiple imputation using the fully conditional specification30 method for seven cycles.

As part of the imputation procedure, a generalised linear model was used with progression of the cIMT as the dependent variable and treatment group, sex and cIMT assessor as adjusting covariates. The mean difference in cIMT was calculated by the least square method. Calculations were performed of each imputation cycle and the mean of the pooled results is presented.

To assess the robustness of the intervention effect on cIMT, a repeated measurements analysis (linear mixed-effect model) was used with cIMT as the dependent variable, time as repeated measure (covariance type: unstructured), and treatment group, sex, baseline cIMT, DAS28, methotrexate use, biological use, and the interaction between treatment group and time as covariates.

For the analyses of the secondary outcomes, Kaplan-Meier time-to-event analysis was used, with log-rank tests for the comparison of treatment groups. All endpoints were analysed according to the intention-to-treat principle, with the exception of patients in whom no data regarding cardiovascular outcomes could be recorded. These patients were reported as lost to follow-up.

Differences between groups were determined using the unpaired Student’s t-test, Mann-Whitney U test or χ2 test, where appropriate. Differences between baseline and follow-up were tested with either a paired t-test for continuous variables with normal distribution, Wilcoxon signed-rank test for continuous variables with skewed distributions or the McNemar test for dichotomous variables.

A p value <0.05 (two-sided) was considered a statistically significant difference. All statistical analyses were performed using PASW Statistics V.23.0 (IBM SPSS Statistics, New York, USA).


Between February 2011 and December 2012, 335 patients were assessed for eligibility and 320 patients were randomly assigned to the treat-to-target intervention (n=162) or usual care (n=158) (figure 1). The characteristics of the patients were similar at baseline except for female sex (table 1). Of the randomised patients, 219 patients (68.4%) completed the 5 years of follow-up, with 116 patients in the treat-to-target group vs 103 in the usual care group. There were no differences in baseline characteristics in those who did and did not complete follow-up (data not shown).

Figure 1

Flow diagram of participants through the trial. Data from patients who withdrew early from the study were included in the intention-to-treat analysis, and follow-up information with respect to cardiovascular events was obtained with permission. For those who were lost to follow-up, no information regarding cardiovascular outcomes could be obtained and these were excluded in the secondary outcome analysis.

Table 1

Clinical, demographic, biochemical and other characteristics of patients with rheumatoid arthritis

After 5 years, there was no difference in mean systolic blood pressure between those allocated to treat-to-target versus usual care (124.6 [15.8] mm Hg vs 124.7 [14.6] mm Hg; p=0.97) (figure 2A). Furthermore, there was no difference in the percentage of participants who reached treat-to-target treatment targets for blood pressure (72.4% for usual care vs 75.9% for treat-to-target; p=0.56) (figure 2E).

Figure 2

Changes in selected risk factors during follow-up. Shown are mean (SEM) of systolic blood pressure (A), haemoglobin A1c (B), low-density lipoprotein (LDL) cholesterol (C) and carotid intima media thickness (D) for the usual care (●) and treat-to-target intervention group (◼) during 5 years of follow-up. The total number of patients in both groups was 320 (baseline), 251 (year 1), 236 (year 2), 238 (year 3), 221 (year 4) and 216 (year 5). (E) The percentage of patients in each group who reached treatment goals for the treat-to-target intervention group at the end of the study. *P<0.05 for the comparison between usual care and the treat-to-target intervention group.

There was no difference in mean HbA1c between groups after 5 years (37.0 [4.6] mmol/mol vs 37.6 [5.5] mmol/mol; p=0.39) (figure 2B). In both groups, over 98% of participants had values below the HbA1c target for the treat-to-target group (p=0.80).

A significant and persistent lower LDL-C within 1 year in the treat-to-target group versus usual care was found. The mean LDL-C was 2.5 (0.8) mmol/L for treat-to-target vs 3.1 (0.9) mmol/L for usual care after 5 years (p<0.0001) (figure 2C). Significantly more patients in the treat-to-target group had LDL-C values below the treat-to-target group target (75.6% vs 48.6% for usual care; p<0.0001).

During 5 years of follow-up, RA was well controlled in both groups, reflected by comparable low values of both DAS28 and ESR between groups (online supplementary figure s1). More patients in the treat-to-target group were on statin treatment (48.3% vs 18.4%; p<0.0001). There were no significant differences in the use of other medications (see table 2).

Table 2

Medication use at baseline and after 5 years of follow-up

At baseline, there was no difference in mean cIMT between those allocated to treat-to-target and usual care (0.570 [0.108] mm versus 0.569 [0.113] mm; p=0.91). Subjects allocated to treat-to-target showed a reduced mean progression of cIMT after 5 years of follow-up compared with the usual care (0.023 [95% CI 0.011 to 0.036] mm vs 0.045 [95% CI 0.030 to 0.059] mm; p=0.028). At 5 years of follow-up, the cIMT values were not different between the groups (0.626 [0.131] mm vs 0.602 [0.108] mm; p=0.15) (figure 2D). Using a linear mixed model, treatment group was associated with a difference in cIMT, independent from the use of biologicals or methotrexate (p<0.001; see for details online supplementary table S1).

During 5 years of follow-up, seven patients (4.7%) in the usual care group and five patients (3.2%) in the treat-to-target group died (p=0.51, log-rank test) (table 3), with a between-group difference of 1.5% (95% CI −2.9% to 5.9%). Three patients (2.0%) in the usual care group died of cardiovascular cause and none in the treat-to-target group (p=0.07), with a between-group difference of 1.2% (95% CI −0.2% to 4.3%). A total of 11 cardiovascular events occurred during the study, with 9 first cardiovascular events. There were a total of 7 (4.7%) first cardiovascular events in the usual care group and 2 events (1.3%) in the treat-to-target group (p=0.048) (figure 3), with a between-group difference of 3.4% (95% CI −0.04% to 7.2%).

Figure 3

Kaplan-Meier estimates of the composite of first occurrence of fatal and non-fatal cardiovascular events during 5 years of follow-up. Shown is the cumulative incidence of the composite endpoint of first occurrence of fatal and non-fatal cardiovascular events, including death by cardiovascular cause, non-fatal myocardial infarction, non-fatal stroke, coronary artery bypass grafting, percutaneous coronary intervention, revascularisation of peripheral atherosclerotic arterial disease or amputation related to peripheral atherosclerotic arterial disease.

Table 3

Cardiovascular events and cardiovascular and all-cause mortality according to treatment groups after 5 years of follow-up

At baseline, 10 patients with a CVD risk >10% were identified and followed in the high-risk cohort. The mean age was 65.6 (2.9) years, 90.0% were male and the median DAS28 was 2.1 (1.6–2.6) (online supplementary table s2). Five years of treatment significantly reduced their blood pressure and LDL-C. At baseline, their cIMT was 0.751 (0.085) mm and the 5-year change of cIMT was −0.056 (0.056) mm (online supplementary figure s2).


This study evaluated prospectively an intervention regarding CVD risk factors in patients with well-treated RA with a low CVD risk. We show a significant reduction in the progression of subclinical atherosclerosis in patients with RA. Furthermore, a reduction in the first occurrence of cardiovascular events was found, although the number of events was low.

Multifactorial intervention trials in type 2 diabetes mellitus (T2DM) have shown benefit regarding CVD reduction,31 and our study confirms this finding in patients with well-treated RA. Although the difference in cIMT progression between the groups was relatively small in absolute terms, the relative reduction in progression was almost 50% in favour of the treat-to-target group. In light of the reduction of cardiovascular events, these effects are, in our opinion, clinically relevant. This is also supported by the observed effects in the high-risk patients who were treated according to the treat-to-target protocol resulting in stabilisation of cIMT during follow-up. The mixed model analysis further supported the beneficial effects of the intervention. It should be noted that our study included a relatively small number of participants and a small number of clinical events. Therefore, these data need to be confirmed in larger trials.

The cardiovascular risk in patients with RA has often been compared with the risk of patients with T2DM.1 3 32 Our data suggest that the overall risk for CVD in well-treated RA is not as high as previously thought,1 with a 5-year risk for first occurrence of cardiovascular events of 4.7% in those randomised to usual care. This number is considerably lower compared with data from cohort studies, in which the 5-year risk for first events ranges 7%–13%,1 33 or in patients with T2DM in whom the risk ranges 12%–15%.34 35 However, the risk in our cohort seems to be elevated compared with the general population, in which the prevalence of first CVD event ranges 2.3%–3.2%.36 37 The current CVD risk models either do not take the increased CVD risk into account in subjects with RA (eg, the Framingham38 and SCORE25 tables) or overestimate the risk (eg, the Dutch CVD risk management guidelines19). Therefore, the current European League Against Rheumatism guidelines,20 where the calculated risk is multiplied by a factor of 1.5, seems to be the best clinical option so far.

Atherosclerosis has a multifactorial genesis, with many factors involved including inflammation.6 The chronic inflammation in RA has been associated with the development of atherosclerosis,7 and therefore control of disease activity remains the primary goal. Advances in the treatment of RA might explain the reduced CVD risk found in our study. Effective treatment of disease activity leading to less systemic inflammation with methotrexate and biologicals is considered to be partly responsible for the reduction of CVD events.13 39 The CARdiovascular research and RhEumatoid arthritis (CARRÉ) study found a CVD event rate of 12.9%; however, only 10% of patients with RA were treated with a biological compared with 40% in our study. The median DAS28 in our trial was lower compared with the CARRÉ study (2.4 vs 3.8), which could explain, in part, the low incidence of subclinical and clinical atherosclerosis observed here.

Only patients with RA without clinical CVD and with an estimated 10-year risk for fatal CVD below 10%, according to the 2006 SCORE guidelines, were randomised. Therefore, the CVD risk found in our study may underestimate the CVD risk in a general RA cohort. Furthermore, even though the majority of our patients had a low disease activity, only a quarter of our patients reached Boolean-based RA remission.

It should be emphasised that our study has several limitations. First, the dropout in our study was considerable. Despite this dropout, we were able to construct a large database using multiple imputation. Differences at baseline and at time of dropout, between completers and non-completers, were comparable. Furthermore, we were also able to collect reliable information on clinical events. Second, cIMT is only a soft endpoint for modern cardiovascular trials. At the start of this study, this was not unusual and many cardiovascular intervention trials used cIMT changes at that time as their primary endpoint. Third, the measurement of the cIMT was not performed blinded, but it was carried out according to a strict protocol with fixed settings. Fourth, according to current standards, the assessment of clinical endpoints should have been performed by an independent endpoint committee. Finally, a stricter target for LDL-C should have been chosen, which may have resulted in even larger effects.

Of note, due to the study design, usual care patients visited our outpatient clinic for evaluation. Their primary care physicians received information and recommendations on cardiovascular risk management resembling the intervention applied to our treat-to-target group. This probably led to a more intensive treatment of usual care patients, and it may explain for example the fact that blood pressure was so well controlled in both groups. In our opinion, this may lead to an underestimation of the absolute risk found in these patients.

Our study shows a larger cIMT progression than estimated by our sample size calculation. Due to limited data on cIMT progression in patients with RA, data from other cohorts were used to perform the power calculation. For the effect of the intervention, we used data from a multifactorial intervention in T2DM which showed an almost 50% reduction in CVD events. We had doubts whether this high risk would apply to patients with RA and therefore used a more conservative target of 20% reduction for cIMT progression. In addition, we did not expect that this small number of subjects would result in a significantly lower number of events in such a relatively short follow-up period.

In conclusion, the present study provides support for a treat-to-target approach of traditional cardiovascular risk factors in subjects with well-controlled RA.


We are grateful to Rene Bakker, Department of Clinical Chemistry, Franciscus Gasthuis & Vlietland, for the development of the study database and technical support in data management, as well as Jessica van de Arend, Mathilde Kool, Sonja Belchior and Chantal Geel, specialised rheumatology nurses from the Department of Rheumatology, and Evelien Kars, specialised vascular nurse, Diabetes and Vascular Center, Franciscus Gasthuis & Vlietland, Rotterdam, for their excellent role in taking care of the patients.



  • Handling editor Professor Josef S Smolen

  • Contributors DFvB-vdS, JvZ and MCC conceived and designed the study. BB, MAdV, DFvB-vdS, BK and NvdM performed the study. BB, EB and MCC performed the statistical analysis. BB and MCC drafted the manuscript. All authors contributed to interpretation and edited the draft report. BB and MCC had full access to the data and take responsibility for the integrity of the data and the accuracy of the data analysis.

  • Funding This work was part of the FRANCIS study, which is supported by the board of directors of the Franciscus Gasthuis & Vlietland, the Foundation for Research and Development of the Department of Internal Medicine and the Coolsingel Foundation, Rotterdam. The latter is a public research fund with no involvement in the study or in the submitted manuscript.

  • Competing interests MCC served as a consultant for Merck and received lecture honoraria from Merck.

  • Patient consent for publication Obtained.

  • Ethics approval The study was approved by the Institutional Review Board of Franciscus Gasthuis, Rotterdam, the Netherlands and the regional independent medical research ethics committee TWOR, Rotterdam.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Original data are available upon reasonable request.

Request Permissions

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.