Article Text

This article has a correction. Please see:

Extended report
Beneficial effect of 1-year etanercept treatment on the lipid profile in responding patients with rheumatoid arthritis: the ETRA study
  1. A Jamnitski1,
  2. I M Visman1,
  3. M J L Peters2,
  4. B A C Dijkmans1,2,
  5. A E Voskuyl2,
  6. M T Nurmohamed1,2
  1. 1Jan van Breemen Institute, Amsterdam, The Netherlands
  2. 2VU University Medical Center, Amsterdam, The Netherlands
  1. Correspondence to Dr M T Nurmohamed, Department of Rheumatology, Jan van Breemen Institute, Dr Jan van Breemenstraat 2, 1056 AB Amsterdam, The Netherlands; m.nurmohamed{at}


Background Effective anti-inflammatory treatment with tumour necrosis factor α (TNFα) inhibitors may have favourable effects on the lipid profile. Available evidence is derived from short-term studies, and it is not clear whether TNFα inhibitors have a similar effect on the lipid profile in responders and non-responders to the treatment.

Objectives To investigate the effect of long-term etanercept treatment on the lipid profile in a large sample of patients with rheumatoid arthritis (RA), stratified for European League Against Rheumatism (EULAR) response.

Methods Between 2004 and 2008, 292 consecutive patients with active RA (DAS28 >3.2) and a new etanercept prescription were included in an observational cohort. Clinical response variables and lipid samples were collected at baseline and after 4 months and 1 year of etanercept treatment. Generalised estimating equation analyses were used to investigate the longitudinal course of lipid levels in relation to clinical response variables.

Results According to the EULAR response criteria, 76% of the patients were good or moderate responders at 4 months, and 85% of the remainder at 1 year. Significant changes in apoA-I (increased by 3.5% (p=0.002) at 4 months and 3.1% (p=0.005) at 1 year) and apoB/apoA-I ratio (decreased by 6.2% (p<0.001) at 4 months and 3.6% (p=0.025) at 1 year) were observed in EULAR responders. No significant differences were observed in EULAR non-responders at all time points.

Conclusions Treatment with etanercept resulted in a significant and sustained decrease in the apoB/apoA-I ratio in patients with good or moderate EULAR response. This may have a beneficial effect on the cardiovascular risk in patients with RA.

View Full Text

Statistics from


The excess cardiovascular (CV) risk in patients with rheumatoid arthritis (RA) is increasingly acknowledged, particularly in patients with untreated severe disease.1 Although atherosclerosis is a multifactorial process, an atherogenic lipid profile—that is, high levels of total cholesterol (TC), low-density lipoprotein cholesterol (LDLc) and triglycerides (TG) and low levels of high-density lipoprotein cholesterol (HDLc)—is one of the pivotal factors in the development of CV disease.2 Furthermore, recent data indicate that high levels of apolipoprotein B (apoB) and low levels of apolipoprotein A-I (apoA-I) and a high apoB/apoA-I ratio are strong predictors of an increased CV risk.3 Recent research has shown reduced levels of HDLc and apoA-I in patients with active RA, which translates into a higher TC/HDLc ratio and apoB/apoA-I ratio.4 These findings may be clinically relevant, as both indices are strongly and independently associated with an increased risk of incident CV disease.1 3 4 Furthermore, there is increasing evidence that inflammation impairs the antiatherogenic function of HDLc by deranging the protein composition.5 6 The concentration of several HDLc-specific proteins (eg, paraoxonase and serum amyloid A (SAA)) is modified, and, as a consequence, the ability of HDLc to protect LDLc from oxidation may be impaired in the inflammatory state.7

An excess of tumour necrosis factor α (TNFα), a proinflammatory cytokine, plays a pivotal role in chronic inflammation in patients with RA and has also been shown to influence lipoprotein metabolism, which may accelerate the development of atherosclerotic lesions.8 9

Effective anti-inflammatory treatment with TNF-blocking agents may improve the lipid profile by increasing HDLc and apoA-I levels. Moreover, we have previously shown that etanercept treatment in another chronic inflammatory arthropathy (ankylosing spondylitis) restores the anti-inflammatory properties of HDLc, as concentrations of SAA decrease in HDL particles.10 However, the available evidence has been derived from short-term (<6 months) studies with a small number of patients.11,,13 The evidence supporting a long-term effect of TNF inhibitors on lipid profile is scarce and contradictory.14,,17 In addition, it is not clear whether TNFα inhibitors have a similar effect on the lipid profile in responders and non-responders. The aim of this study was to investigate the longitudinal course of lipid levels (TC, LDLc, TG, HDLc, apoA-I, apoB, the TC/HDLc ratio and the apoB/apoA-I ratio), and their relationship with clinical response, after 4 months and 1 year of etanercept treatment in a cohort of consecutive patients with RA (Etanercept Treatment in Rheumatoid Arthritis (ETRA) study).

Patients and methods

Study population

Between December 2004 and October 2008, 292 consecutive patients with RA who had a new etanercept prescription were included in an observational cohort at the Department of Rheumatology of the Jan van Breemen Institute. All patients fulfilled the American College of Rheumatology 1987 revised criteria for RA, and started anti-TNF therapy according to the Dutch consensus statement on the initiation of TNF blocking therapy—that is, an active disease indicated by a Disease Activity Score in 28 joints (DAS28) of >3.2 and failure to respond to at least two disease-modifying antirheumatic drugs (including methotrexate (MTX)) at maximal or tolerable dosage.18 Patients were treated with either concomitant medication, including MTX and prednisone, or etanercept monotherapy. The dose of etanercept was either 50 mg subcutaneously every week or 25 mg twice a week. The local medical ethics committee approved the study, and all patients gave written informed consent.

Study design

Disease activity was assessed at baseline and after 4 months and 1 year of etanercept treatment using DAS28, C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR). Clinical response was evaluated using the European League Against Rheumatism (EULAR) response criteria.19 A composite of EULAR moderate and good responders was used as responders. Information on disease-specific variables, current medication use and blood pressure was obtained at each visit. Length and weight were measured at baseline and 1 year. Body mass index was calculated (weight (kg)/length2 (m)). Non-fasting lipid samples were collected at baseline (n=266), 4 months (n=264) and 1 year (n=171).

Assessment of lipids

TC and TG were measured by an enzymatic method using clinical chemistry analysers. HDLc was determined enzymatically with poly(ethylene glycol)-modified enzymes. apoA-I and apoB were assessed by an immunoturbidimetric method, using assays supplied by Roche Diagnostics (Mannheim, Germany). LDLc was calculated according to the Friedewald formula when TG levels were lower than 4.5 mmol/l (4 g/l). TC/HDL and apoB/apoA-I ratios were calculated. All blood samples were determined batchwise.

Statistical analysis

All analyses were performed using SPSS version 16.0. The distribution of variables was tested for normality and transformed if necessary. Data are represented as mean±SD, median and interquartile range (IQR), or percentages. p<0.05 was considered significant. Generalised estimating equation (GEE) analyses were performed to investigate the longitudinal course of lipid levels and changes in lipid profile in relation to clinical response. The exchangeable correlation matrix was used to perform GEE analyses. All baseline variables were considered as potential confounders and included in the GEE analyses.


Patient's characteristics

Baseline characteristics are shown in table 1. Briefly, the study population consisted of 292 (238 women and 54 men) RA patients with a mean age of 53 years (SD 13 years). The median disease duration was 8 years (IQR 3–16 years), 207 patients (72%) were IgM rheumatoid factor positive, and 72% had erosions. Of the 292 patients included in the study, 89 had previously been treated with infliximab (n=30) or adalimumab (n=59). Adjustment for previous biological use in the GEE analysis did not alter lipid levels (data not shown).

Table 1

Demographic and clinical characteristics at baseline

Changes in clinical response

After 4 months of the study, 276 (95%) patients were still receiving etanercept treatment. DAS28 had decreased significantly from 5.21±1.32 to 3.54±1.47 at 4 months. According to the EULAR response criteria, 211 patients (76%) were either good or moderate responders, while 65 patients (24%) were non-responders at 4 months.

Of the 292 patients at baseline, 191 (65%) completed 1 year of the observation period, and DAS28 had decreased significantly to 3.07±1.28 compared with baseline and 4 months. At 1 year, 162 (85%) patients were good or moderate responders compared with 29 (15%) non-responders. Of the remaining 101 patients, 10 had not yet completed 1 year of observation, and 91 stopped etanercept treatment within 1 year (60 (66%) because of inefficacy, 17 (19%) because of an adverse event and 14 (15%) were lost to follow-up). Patients who discontinued the etanercept treatment within 1 year had higher DAS28 (p=0.035), higher apoA-I (p=0.014) and lower TC/HDLc ratio (p=0.031) at baseline than the patients who continued to receive the treatment. However, adjustment for this difference in the GEE analysis did not change the results (data not shown).

Concomitant treatment with either MTX, prednisone or statins occurred in 76%, 29% and 13%, respectively, of all the patients. The number of patients receiving MTX had decreased by 3% at 4 months and by 4% at 1 year, which is reflected by a significantly lower dose of MTX at both 4 months and 1 year (from 19.7±7.1 to 19.5±7.0 at 4 months and to 18.7±7.3 mg/week at 1 year; all p values <0.01). The use of prednisone decreased by 5% at 4 months and another 5% at 1 year, which also resulted in a lower prednisone dose: from 8.2±3.9 at baseline to 7.9±3.8 at 4 months and to 7.2±4.0 mg/day at 1 year (all p values <0.01). The number of patients with a statin prescription remained unchanged over the study period. Adjustment for changes in use and dose of MTX and prednisone in the GEE analysis did not change results (data not shown). Body mass index decreased from 26.0±5.6 to 25.8±5.4 (p=0.277).

Changes in the lipid levels

The GEE analysis showed significant changes in lipid levels over time. After 4 months of etanercept treatment, TC, TG, LDLc and apoA-I levels had increased significantly (p=0.003, p=0.028, p=0.046 and p=0.026, respectively). The reduction in apoB/apoA-I ratio was 4% after 4 months (0.571 to 0.548, p=0.002). The changes in HDLc, TC/HDLc ratio and apoB were not significant (p=0.720, p=0.773 and p=0.479, respectively).

After 1 year, the increase in levels of TC, TG and apoA-I was still significant (p=0.026, p<0.001, p=0.031). The changes in HDL, LDL, TC/HDL ratio and apoB over time compared with baseline were not significant (p=0.848, p=0.474, p=0.653, p=0.212). The apoB/apoA-I ratio had decreased from 0.571 to 0.568 (0.5% reduction) at 1 year, but this was not significant (p=0.064). Adjustment for all baseline characteristics in the GEE analysis did not change the results (data not shown). A table showing absolute changes in lipid levels is available as an online supplemental file.

Changes in lipid levels stratified for EULAR response

When stratified for EULAR response, significant changes in lipid levels were only observed in EULAR responders (table 2). TC had increased by 2.5% (p<0.004), apoA-I had increased by 3.5% (p<0.002), and the apoB/apoA-I ratio had decreased by 6.2% (p<0.001) after 4 months of etanercept treatment in EULAR responders. Similar results were observed after 1 year of etanercept treatment; again this was only true for EULAR responders.

Table 2

Influence of EULAR response on lipid levels at 4 months and 1 year

Relationship between changes in lipid profile and disease activity

Changes in clinical response variables—that is, DAS28, CRP and ESR—were significantly associated with changes in lipid levels: an improvement in DAS28 was associated with an increase in TC, HDLc and apoA-I (all p values <0.01) and a decrease in apoB/apoA-I ratio (p=0.01), with regression coefficients 0.071, 0.030, 0.029 and −0.010, respectively. CRP was inversely associated with TC, HDLc, apoA-I, TG and LDLc, with the following regression coefficients: −0.007 (p<0.001), −0.002 (p=0.01), −0.002 (p<0.001), −0.003 (p<0.01) and −0.004 (p=0.01), respectively. Similar associations were observed for ESR with TC (−0.006; p<0.01), HDLc (−0.003; p<0.01) and apoA-I (−0.002; p<0.01), while decreasing levels of ESR were also associated with a decrease in apoB/apoA-I ratio (0.001; p=0.04). A table showing all results is available as an online supplemental file.


This study is important for several reasons. Firstly, it reaffirms the beneficial effect of TNFα inhibitors on the lipid profile—that is, apoA-I and the apoB/apoA-I ratio in a large cohort of RA patients. Secondly, our data show, for the first time, that the favourable effect of the TNFα inhibitor, etanercept, on lipids only applies to RA patients with a good or moderate EULAR response. Thirdly, consistent with previous findings, we observe a close relationship between inflammation and lipid levels.

The guidelines for risk evaluation and treatment of dyslipidaemias focus on measurement of plasma lipoproteins—that is, LDLc, HDLc and TC/HDLc ratio. However, new evidence supports the use of apolipoproteins over lipoprotein measurement to predict CV risk, because determination of apolipoproteins is unaffected by a non-fasting or hypertriglyceridaemic state, and apolipoproteins are not derived from other measurements.20 apoA-I is the main apolipoprotein on the surface of all HDL particles, whereas apoB is a protein found on the surface of all atherogenic particles—that is, LDL and small dense LDL, very-low-density lipoprotein and intermediate-density lipoprotein. Hence, the apoB/apoA-I ratio represents the balance of all proatherogenic and antiatherogenic lipoproteins, which appear to be highly predictive in the evaluation of CV risk.3 The present study shows a significant increase in apoA-I levels, resulting in a significantly lower apoB/apoA-I ratio, a lipid profile considered to be protective against CV disease. Importantly, this observation is only true for EULAR responders, as lipid levels did not change in EULAR non-responders. These results are in agreement with another study that reported a reduced risk of myocardial infarction in patients who responded to anti-TNFα therapy.21 In addition, we observed that TC level increased significantly in EULAR responders and did not change in non-responders. A possible explanation is that a high RA disease activity is associated with rheumatoid cachexia with concomitant decreased cholesterol levels.22 Treatment with etanercept resulted in decreased disease activity in EULAR responders and consequently increased levels of TC.

Another important observation from this study is an inverse association between decreasing levels of acute-phase reactants, ESR and CRP, and also DAS28 and an increase in apoA-I. This finding underscores the possible causal relationship between inflammatory activity and a disturbed lipid profile in RA patients, which has also been observed in other reports.4 10 16 New data show that inflammatory activity is associated with proinflammatory changes in HDL particles.5 This is interesting, as it is known that SAA is able to replace antiatherogenic apoA-I in the HDL particle, which renders the HDL particle less protective.23 24 Recent data have shown that the anti-inflammatory properties of HDLc were restored during treatment with TNF inhibitors as the concentration of SAA in HDL particles decreased and antioxidative capacity improved.10 25

Obviously, and inherent to the observational cohort design, confounding by indication cannot be excluded. Other limitations of this study also merit consideration. Firstly, differences reported in this study were small, raising questions about the clinical relevance of these findings. However, even small changes over a prolonged period of time may be clinically relevant—for example, every 0.03 mmol/l increase in HDLc results in a 2–3% reduction in CV risk.26 Furthermore, atherosclerosis is a chronic and multifactorial process in which long-term changes in the lipid profile will affect CV risk.2 Given that RA is a chronic inflammatory disease associated with excess mortality largely attributable to CV disease, even the small changes in the risk profile could be considered clinically relevant.27 Moreover, antiatherogenic changes in the lipid profile were sustained after 1 year of etanercept treatment, which may increase the clinical implications of our observations. Secondly, the lipid profile was measured in non-fasting blood samples, which can influence conventional lipid measurements, but do not affect apolipoproteins. Thirdly, owing to the longitudinal cohort design, the missing data, and also drop out, may have led to bias. Therefore, we used GEE analysis, as it adjusts for dependency of several measurements within one subject and is capable of dealing with unequally spaced time intervals and with missing data.28 Fourthly, all baseline variables were considered as potential confounders; however, we were not able to exclude confounding by unmeasured factors—for instance, smoking, exercise and non-steroidal anti-inflammatory drug use.

The strength of this study is that it is a one-centre-based cohort design reflecting the heterogeneous population of the true daily practice and systematic prospective assessment of a complete lipid profile and a broad range of disease characteristics. The results of this study reveal additional insights into the influence of disease activity markers and EULAR response during long-term etanercept therapy on lipid profile, as they have a marked effect on apoA-I and the apoB/apoA-I ratio. The clinical implications of our findings on decreasing CV disease in patients with RA remain to be established by future investigations.


We are grateful to the Clinical Research Bureau of the Jan van Breemen Institute, which receives support from the Dutch Arthritis Foundation, for help with the conduct of the study. We thank the research nurses, Marga Kammeijer-Rippen, Anne-Marie Abrahams, Martine Kos and Nina Kivit, for performing clinical assessments. Finally, we also thank Margret de Koning and Truus de Gast for preparing and performing the assays.


View Abstract

Supplementary materials


  • Funding The clinical part of this study was partially financed by Wyeth Pharmaceuticals. In addition, this investigation was facilitated by the Clinical Research Bureau of the Jan van Breemen Institute. The study sponsors had no involvement in the study design, in the collection, analysis and interpretation of data, in the writing of the report, or in the decision to submit the paper for publication.

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the Medical Ethics Committee of the Slotervaart Hospital, BovenIJ Hospital, the Jan van Breemen Institute.

  • 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.

Linked Articles

  • Miscellaneous
    BMJ Publishing Group Ltd and European League Against Rheumatism