Article Text


Extended report
Remission is the goal for cardiovascular risk management in patients with rheumatoid arthritis: a cross-sectional comparative study
  1. Sella A Provan1,
  2. Anne Grete Semb1,
  3. Jonny Hisdal2,
  4. Einar Stranden2,
  5. Stefan Agewall3,
  6. Hanne Dagfinrud1,
  7. Kristin Angel3,
  8. Dan Atar4,
  9. Tore K Kvien1
  1. 1Department of Rheumatology, Diakonhjemmet Hospital, Oslo, Norway
  2. 2Section of Vascular Investigations, Oslo University Hospital Aker and Faculty of Medicine, University of Oslo, Norway
  3. 3Department of Cardiology B, Oslo University Hospital Aker, Oslo, Norway
  4. 4Department of Cardiology B, Oslo University Hospital Ullevål and Faculty of Medicine, University of Oslo, Norway
  1. Correspondence to Sella Aarrestad Provan, Department of Rheumatology, Diakonhjemmet Hospital, PB.23 Vindern, N-0319 Oslo, Norway; sellaprovan{at}


Objectives To compare markers of cardiovascular disease (CVD) risk between patients with rheumatoid arthritis (RA) in an active disease state and those with RA in remission, and to compare both groups with community controls.

Methods 113 patients with RA and 86 community controls were assessed across a panel of biomarkers for CVD. RA in remission was defined as Clinical Disease Activity Index ≤2.8. Community controls were selected at random by Statistics Norway, and controls were matched with patients in the cohorts in strata using details of age, sex and residential area. A panel of biomarkers (N-terminal pro-brain natriuretic peptide (NT-proBNP), total cholesterol, reactive hyperaemia index (RHI), pressure measurements, measures of arterial stiffness and intima-media thickness) were compared between patients with active RA and those with RA in remission. Both groups were compared with controls. In addition, biomarker levels were compared across subgroups based on anticyclic citrullinated peptide status, level of joint destruction and presence of extra-articular manifestations.

Results Patients with active RA had significantly higher levels of NT-proBNP, brachial systolic pressure, augmentation index and central systolic pressure but lower cholesterol than patients in remission and controls. In addition, patients with active RA had significantly higher levels of pulse wave velocity and worse RHI than patients in remission. Comparison across other subgroups gave less consistent differentiations in levels of CVD risk markers.

Conclusion Patients with active RA, but not those in remission, had significantly increased levels of CVD risk markers. These results link inflammatory activity to markers of CVD risk in patients with RA and may indirectly support the notion that remission in RA confers diminished cardiovascular morbidity.

Statistics from


Patients with rheumatoid arthritis (RA) have an increased risk of cardiovascular disease (CVD) and a reduced life expectancy.1 2 The increased occurrence of CVD is partly explained by an adverse combination of traditional risk factors and factors related to RA disease activity such as inflammation.

The individual risk of CVD can be assessed by biomarkers that reflect many facets of CVD-related pathology and morbidity.3 The pathogenesis of CVD involves endothelial dysfunction, arterial stiffening and atherosclerosis which may ultimately lead to clinical disease, evident as hypertension, myocardial infarction, stroke or heart failure.4 5 A huge array of biomarkers is becoming available to audit the CVD risk at the level of the individual.

The primary aim of the current analyses was to compare levels of CVD risk markers between patients with clinically active RA and those with RA in remission and to compare these two groups with community controls. We also compared CVD risk markers between subgroups of RA patients categorised according to anticyclic citrullinated peptide (anti-CCP) status, level of joint destruction and presence of extra-articular manifestations.


One hundred and thirteen patients with RA and 86 community controls aged 30–70 years were assessed across a panel of biomarkers for CVD. The data for the patients with RA were derived from the 15-year follow-up of the EURIDISS cohort (n=80)6 and the 10-year follow-up of the Oslo RA register (n=33).7 The patients had been diagnosed according to the 1987 American College of Rheumatology criteria.8 The patients were stratified according to the following groups: female versus male, age in decades and residential area. Two hundred community controls were randomly selected by Statistics Norway from the inhabitants of Oslo to match these stratifications. The only exclusion criterion was a history of inflammatory arthritis. Demographic and health status variables were collected by a self-reported questionnaire, but the subjects could request help from attending staff. Body mass index (BMI) was calculated. The patients were requested to abstain from food, drink (except for water) and smoking for at least 3 h prior to examination.

RA disease activity

A trained study nurse blinded to the CVD risk profile of the patients assessed the number of swollen and tender joints (28 joint counts), and disease activity was assessed by the Clinical Disease Activity Index (CDAI) as the sum of the number of swollen joints + number of tender joints + patient global VAS (in cm) + investigator global VAS (in cm).9 Patients with a CDAI ≤2.8 were considered to be in remission.10 Joint destruction was measured by radiographic damage of the hands and scored according to the van der Heijde-modified Sharp criteria (vdHSS).11 The presence of extra-articular manifestations was self-reported using a modification of the AIMS 2 questionnaire.12

Soluble biomarkers

Biomarkers were examined consecutively: erythrocyte sedimentation rate by the Westergren method, C reactive protein (CRP) and total cholesterol by COBAS 6000 (Roche Diagnostics, Basel, Switzerland), N-terminal pro-brain natriuretic peptide (NT-proBNP) by a Modular E170 device (Roche Diagnostics, Basel Switzerland; coefficient of variation (CV) 7.6–9.7%) and anti-CCP using an ELISA method (Inova Diagnostics, San Diego, California, USA).

Markers of CVD risk

Brachial blood pressure was measured after a 5 min rest in a supine position using the OMRON M7 (Kyoto, Japan). Several measurements were made until two measurements differed by ≤5 mm Hg in both systolic and diastolic mm Hg as well as heart rate, and averages were then calculated. We performed pulse wave analysis assessments using the Sphygmocor apparatus (Atcor, West Ryde, Australia). The apparatus uses applanation tonometry, equalising the circumferential pressure by slightly flattening the artery to obtain accurate pressure waveforms. Patients suffering from atrial fibrillations were excluded from the analysis. Several recordings were made in each patient and the recordings considered to have the highest quality according to predetermined requirements were selected for further analyses.13

The reactive hyperaemia index (RHI), also called the digital hyperaemic response, was measured using the Itamar apparatus (Caesarea, Israel).14 Eighteen controls did not participate in this examination for logistical reasons. The patients were asked to recline on a bed in a comfortable position and probes were attached to bilateral index fingers. A sphygmomanometer cuff was wrapped around the upper arm. After a 5 min recording of the digital pulse, the cuff was inflated to 200 mm Hg and the recording examined for any sign of pulsatile activity on the occluded side, in which case the cuff was further inflated, although not above 300 mm Hg. After a 5 min occlusion the cuff was released and a further 5 min of recording was made during which the hyperaemic phase occurred. The RHI was calculated by the software installed on the Itamar.

Pulse wave velocity (PWV) was calculated from knowing both the transit time for the pulse wave travelling from the heart to two sites and the distance between these sites.15 We chose to estimate the carotid-femoral PWV between the site on the carotid and on the femoral artery where the pulse was most strongly palpated. The recordings of the foot of the pressure wave at the carotid and femoral artery were gated to an ECG as a measure of transit time. The distance between the suprasternal notch and the measurement site on the carotid artery was subtracted from the distance between the suprasternal notch and the site on the femoral artery.

Central systolic pressure and the augmentation index (AIx) were derived by applying a validated transfer system to recordings of the arterial pressure waves at the radial artery. Central systolic pressure is an estimation of the systolic pressure in the ascending aorta. AIx is defined as the change in pressure between the second and first systolic peaks as a percentage of the pulse pressure and was standardised to a heart rate of 75 bpm.16

B-mode ultrasonographic examinations were performed on bilateral common carotid arteries using a GE Vivid 7 ultrasound scanner (GE Vingmed Ultrasound, Horten, Norway) with a 12 Mhz probelinear matrix array transducer. 9,,14 The images were analysed offline from JPG images as single readings performed by two of the authors (ES and JH). The intima-media thickness (IMT) of a 5 mm long section of the far wall, 10 mm proximal to the carotid bulb, was obtained by the AMS analysis software (Artery Measurement System, Tomas Gustavsson, Gothenburg, Sweden). Each 5 mm section generated approximately 50 IMT calculations and median values were used when comparing study groups.

The reliability of all biomarkers was tested by repeated examinations in a subset of patients. The inter-reader correlation coefficient was then calculated. For all CVD risk markers the inter-reader correlation coefficient was found to be good (see online supplement for further information).

The Framingham 10-year risk of coronary heart disease and the European Society of Cardiology SCORE 10-year risk of CVD-related death were calculated using standardised risk calculators.17 18

Statistical analyses

Bivariate comparisons were made using the Student t test, χ² test or Mann–Whitney U test as appropriate.

The level of CVD risk markers was compared across the following groups using analysis of covariance (ANCOVA): active RA, RA in remission versus community controls; anti-CCP status positive versus negative; vdsHss joint destruction ≤ median versus > median; and presence versus absence of extra-articular manifestations. All analyses were adjusted for age and sex. Skewed variables were log transformed. The validity of the ANCOVA models was explored by examining the following variables as possible confounders in the final model: smoking (now/never), BMI, education, presence of relevant comorbidities (prior myocardial infarction, cerebral insult and diabetes) and current use of CVD-related medication (statins and antihypertensive medication). In addition, the heart rate was explored as a possible confounder of brachial and central systolic pressure and PWV, and height as a confounder of AIx. Disease duration and current RA disease treatments (methotrexate, prednisolone, tumour necrosis factor α (TNFα) inhibitor) were entered into the models which compared patients across categories of RA disease activity. The results were further verified by constructing a case-control cohort matching a patient with active RA to a patient in remission of the same age and sex. In a mixed model procedure we compared levels of CVD risk markers between these groups. A variable denoting each pair was included in the model as a random effect.

A logistic regression model was constructed with active RA versus RA in remission as the dependent variable in order to assess the independent contribution of each CVD risk marker. Individual CVD risk markers were entered consecutively into models that were adjusted for age and sex. CRP was omitted from these analyses at it is a marker of both RA disease activity and CVD risk. Variables that were significant at p≤0.10 were entered into the multivariate model and removed according to level of significance. Due to multicollinearity, separate models were constructed, each of which only included one of the following variables— brachial systolic pressure, central systolic pressure and AIx—and the model with the greatest Nagelkerke value and accuracy was chosen. Rejected variables were re-entered into the final model to check for possible confounding.

The statistical analyses were performed using SPSS Version 14.


Eighty-two patients had active RA (CDAI >2.8) and 31 were in remission (CDAI ≤2.8). The demographic data are presented in table 1 and show some differences between the groups, especially for active RA versus the controls. The participation rate of the community controls was 43%, while 57% of the surviving participants of the EURIDISS and Oslo RA register cohorts agreed to participate at the 15- and 10-year follow-up, respectively.

Table 1

Cross-sectional demographic data compared across groups

The results of the adjusted comparisons of CVD risk markers (estimated marginal means) are shown in table 2. Compared with patients in remission, those with active disease had higher CRP, NT-proBNP, PWV, brachial systolic pressure, AIx and central systolic pressure, whereas total cholesterol and RHI were lower in patients with active RA (p<0.05 for all comparisons). The CVD risk profile in patients with active RA remained significantly disadvantageous when compared with community controls, although the total cholesterol was lowest in the patients with RA. In contrast, patients in remission only had significantly higher CRP levels when compared with the controls, and PWV and RHI levels were more favourable in these 31 patients than in the controls (table 2). We did not identify any variable that altered the consistency of our findings by confounding several models.

Table 2

Cardiovascular risk markers compared across groups

Thirty pairs were matched in the case–control study. The comparisons confirmed the results of lower total cholesterol (p=0.02), worse RHI (p=0.001), higher CRP (p=0.02), PWV (p=0.01) and AIx (p=0.01) in patients with active RA compared with patients in remission.

The logistic regression analyses with active RA versus RA in remission as the dependent variable found that a model composed of the following variables best differentiated between these patients: RHI (OR 0.15 (95% CI 0.06 to 0.39)), AIx (OR 1.10 (95% CI 1.02 to 1.18)) and PWV (OR 1.72 (95% CI 1.10 to 2.68)), with adjustments for age (OR 0.09 (95% CI 0.84 to 0.98)) and sex (female sex OR 2.98 (95% CI 0.66 to 13.50)). The accuracy of the final model was 84%, sensitivity 96%, while the Nagelkerke value (pseudo R²)19 was 0.42. Entering use of antihypertensive medication into the model (OR 11.46 (95% CI 1.76 to 74.80)) removed PWV from the model and gave an improvement in the Nagelkerke value of 0.07, a slight decrease in the sensitivity of 1.3%, without altering the accuracy of the model.

Fifty-five patients were anti-CCP positive, the median vdHSS was 10 (range 0–152) and 13 patients (11.5%) had current or a history of extra-articular manifestations. The distribution of CVD risk markers was similar overall across these categories (table 3).

Table 3

Comparison of cardiovascular risk markers across three patient categorisations†


This study shows that patients in RA disease remission score significantly better than patients with active RA across a range of biomarkers predicting CVD. Importantly, the level of CVD risk in this group of RA patients in remission, as judged by these biomarkers, is comparable to that of community controls whereas patients with active RA had worse scores than controls (table 2). Furthermore, we found that a multivariate logistic regression model which included measures of endothelial dysfunction, arterial stiffness and AIx gave the best differentiation between patients with active RA and those with RA in remission. To our knowledge, this is the first study to examine CVD risk in patients in remission according to the CDAI criteria, and the first to compare levels of RHI between patients with RA and community controls.

In this study, systolic pressure in patients with active RA was an average of 8 mm Hg higher than patients in remission. Previous studies on brachial systolic pressure in patients with RA have given conflicting results.20,,24 A recent interesting post hoc analysis of data from the BeSt trial showed that systolic pressure in patients with RA was correlated to disease activity and reduced by infliximab.25 Our results further showed that patients with active RA had lower total cholesterol than patients with RA in remission and community controls, a finding that has obvious implications for the validity of the calculation of the Framingham score and European Society of Cardiology SCORE in patients with active RA. Indeed, there were no statistically significant differences in the Framingham score between patient groups and controls, although SCORE was higher in patients with active RA (table 2). The use of SCORE may be advantageous for our patients for two reasons: it can be calculated using the atherogenic index (dividing the total cholesterol by the high density lipoprotein) and it is based on European populations. Heterogeneous findings concerning the lipid profiles of patients with RA have been reported. Adverse lipid profiles have been found in cohorts across a range of disease durations and activity states.26 27 However, RA and chronic inflammation have also been shown to correlate with reduced total cholesterol levels (high density, low density and very low density lipoproteins), possibly due to increased clearance of low density lipoprotein cholesterol by the reticuloendothelial system or decreased lipoprotein lipase activity.22 28,,31

The RHI is a measure of endothelial function which correlates with the gold standard of coronary endothelial function measurements and was recently shown to predict cardiovascular events in a prospective study.14 32 Flow-mediated dilation, the classic method of non-invasive endothelial function assessment, has been found to be decreased in patients with RA.33 34 Endothelial dysfunction may impact on arterial stiffness through nitric oxide, which is important in arterial stiffness regulation.35 Inflammatory cytokines may also act through upregulation of angiotensin type 1 receptors to cause vasoconstriction and hypertension.24 36 Furthermore, an injury to the endothelium can proceed to intimal thickening with a decrease in vascular wall contractile elements as smooth muscle cells migrate to the intima, multiply and lay down extracellular matrix.37 The resultant arterial stiffening is characterised by increased vascular collagen formation, calcification and breakdown of elastin.15

Carotid-femoral PWV pulse wave velocity, AIx and central systolic pressures are measures of arterial stiffening and wave reflection that have been used to predict cardiovascular morbidity and mortality in numerous studies.15 38,,41 The AIx and central pressure estimations quantify the pressure amplification caused by wave reflection and have been shown to be superior to brachial pressure in predicting CVD.38 AIx42,,44 and PWV45 46 are both reported to be increased in patients with high levels of inflammation. AIx has also been shown to be predicted by cumulative scores of disease activity,43 44 whereas PWV—rather than AIx—is reduced by treatment with TNFα inhibitors.45 47 48 These heterogeneous findings may partly be explained by the peripheral vasodilation that accompanies acute inflammation, reducing pulse wave reflection and thereby the AIx and central systolic pressures.49

NT-proBNP is an independent predictor of cardiovascular morbidity and also general mortality in several populations including patients with RA.50 51 NT-proBNP levels reflect the stretching of the atria and ventricles and are associated with measures of arterial stiffness,52 but this biomarker is also upregulated by proinflammatory cytokines.53 In patients with RA, NT-proBNP levels correlate with CRP and are reduced by TNFα inhibitors.54 55

The IMT is a direct measure of the atherosclerotic disease process,56 and an increased IMT predicts a greater risk of cardiovascular events.57 A recent meta-analysis of IMT in patients with RA included 25 studies of which 19 reported increased IMT in patients with RA, and it is thus surprising that our study did not find that patients with active RA had significantly higher IMT compared with the other groups.58

We chose to categorise RA disease activity according to the CDAI criteria for active RA versus remission as it is a more stringent categorisation than the DAS criteria.10 Since CRP is a marker of both RA disease activity and CVD risk, we found it advantageous to categorise patients by a composite score that does not include CRP as a component. Verification of RA disease remission by long-term radiographic outcome data would, however, have added validity to our findings. A weakness of our study is the low participation rate among the invited controls, which may challenge our findings through a selection bias. The controls participating in our study had a poorer CVD risk profile with regard to RHI and PWV compared with patients in remission, which may suggest that some controls chose to participate as they felt the need for a CVD risk assessment. Such a bias would make a type I statistical error unlikely, but a type II error may have occurred. This is a cross-sectional study and our findings should be confirmed in a longitudinal study, preferably with incident CVD or mortality as the outcome.

This study indicates that classification of RA disease activity state according to the CDAI gives better differentiation between the groups than a comparison of CVD risk according to anti-CCP status, level of joint destruction or extra-articular manifestations (tables 2 and 3). Our results indicate that patients with active RA should receive extra attention from clinicians with regard to CVD prevention.

In conclusion, patients with active RA—but not those in remission—have significantly increased levels of biomarkers for CVD. These results support the association of inflammatory activity and markers of CVD risk in patients with RA and indirectly support the finding that remission in RA may also confer diminished cardiovascular morbidity.


The authors thank Petter Mowinckel for providing statistical expertise.


View Abstract


  • Funding This study was supported by grants from the Eastern Norway Regional Health Authority.

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the Norwegian Regional Committee for Research Ethics and patients gave their informed consent.

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

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.