Article Text

Extended report
A randomised controlled trial of etanercept and methotrexate to induce remission in early inflammatory arthritis: the EMPIRE trial
  1. J L Nam1,
  2. E Villeneuve1,2,
  3. E M A Hensor1,
  4. R J Wakefield1,
  5. P G Conaghan1,
  6. M J Green3,4,
  7. A Gough4,
  8. M Quinn4,
  9. R Reece1,5,
  10. S R Cox6,
  11. M H Buch1,
  12. D M van der Heijde7,
  13. P Emery1
  1. 1Leeds Institute of Rheumatic and Musculoskeletal Medicine, University of Leeds, and NIHR Leeds Musculoskeletal Biomedical Research Unit, Leeds Teaching Hospitals NHS Trust, Leeds, UK
  2. 2Rheumatology department, Centre Hospitalier de l'Université de Montréal, Hôpital Notre-Dame, Montréal, Canada
  3. 3Rheumatology Department, Harrogate District Foundation Trust, UK
  4. 4The York Hospital, York Hospitals NHS Foundation Trust, UK
  5. 5Department of Rheumatology, Darlington Memorial Hospital, County Durham and Darlington NHS Foundation Trust, UK
  6. 6Flinders Medical Centre, Adelaide, South Australia, Australia
  7. 7Department of Rheumatology, Leiden University Medical Center, Leiden, The Netherlands
  1. Correspondence to Professor P Emery, Leeds Institute of Rheumatic and Musculoskeletal Medicine, University of Leeds, and NIHR Leeds Musculoskeletal Biomedical Research Unit, Leeds Teaching Hospitals NHS Trust, Leeds LS7 4SA, UK; p.emery{at}


Objective To compare the efficacy of etanercept (ETN) and methotrexate (MTX) versus MTX monotherapy for remission induction in patients with early inflammatory arthritis.

Methods In a 78-week multicentre randomised placebo-controlled superiority trial, 110 DMARD-naïve patients with early clinical synovitis (≥1 tender and swollen joint, and within 3 months of diagnosis) and either rheumatoid factor, anticitrullinated protein antibodies or shared epitope positive were randomised 1:1 to receive MTX+ETN or MTX+placebo (PBO) for 52 weeks. Injections (ETN or PBO) were stopped in all patients at week 52. In those with no tender or swollen joints (NTSJ) for >26 weeks, injections were stopped early. If patients had NTSJ >12 weeks after stopping the injections, MTX was weaned. The primary endpoint was NTSJ at week 52.

Results No statistically significant difference was seen for the primary endpoint (NTSJ at week 52 (32.5% vs 28.1% [adjusted OR 1.32 (0.56 to 3.09), p=0.522]) in the MTX+ETN and MTX+PBO groups, respectively). The secondary endpoints did not differ between groups at week 52 or 78. Exploratory analyses showed a higher proportions of patients with DAS28-CRP<2.6 in the MTX+ETN group at week 2 (38.5% vs 9.2%, adjusted OR 8.87 (2.53 to 31.17), p=0.001) and week 12 (65.1% vs 43.8%, adjusted OR 2.49 (1.12 to 5.54), p=0.026).

Conclusions In this group of patients with early inflammatory arthritis, almost a third had no tender, swollen joints after 1 year. MTX+ETN was not superior to MTX monotherapy in achieving this outcome. Clinical responses, however, including DAS28-CRP<2.6, were achieved earlier with MTX+ETN combination therapy.

Trial registration number The EMPIRE trial is registered on the following trial registries: Eudract-2005-005467-29; ISRCTN 55428162 ( The full trial protocol can be obtained from the corresponding author.

  • Anti-TNF
  • DMARDs (biologic)
  • Early Rheumatoid Arthritis
  • Methotrexate

Statistics from

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.

Remission has become the primary goal of therapy for patients with rheumatoid arthritis (RA).1 While this may be achieved at any time during the course of the disease, there is evidence that early disease modifying antirheumatic drug (DMARD) therapy is associated with better patient outcomes.2 ,3

Identifying appropriate patients early is therefore paramount. In patients with early inflammatory arthritis (IA), autoantibodies including rheumatoid factor (RF) and anticitrullinated protein antibodies (ACPA)—both of which are now included in the 2010 American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) RA classification criteria4 ,5–9—and genetic markers, in particular shared epitope (SE),10–12 have been shown to be markers of persistence of inflammation and progression to RA.

Although there is debate regarding optimal induction therapy, methotrexate (MTX) is generally part of the first-line treatment strategy in RA,13 and in undifferentiated arthritis (UA) with positive ACPA has also been shown to temporarily delay progression to RA.14 The combination of MTX and etanercept (ETN) has proven to be effective for remission induction,15 with superior efficacy to MTX monotherapy, particularly in those with early RA (≤4 months disease duration);2 however, the role of this combination has not been explored in early IA.

The Etanercept and Methotrexate in Patients to Induce Remission in Early Arthritis (EMPIRE) trial aimed to investigate clinical, radiographic and functional outcomes, to determine whether combination therapy with MTX+ETN was superior to MTX monotherapy in subjects with DMARD-naïve early inflammatory arthritis with at least one joint with clinical synovitis.

Patients and methods


Eligible patients were 18–80 years old, DMARD-naïve with at least one tender and swollen joint, within 3 months of a diagnosis of inflammatory arthritis, and RF, ACPA or SE positive. Exclusion criteria included previous treatment with any synthetic or biologic DMARD, corticosteroid within 28 days of screen, current crystal or infective arthritis and important concurrent medical diseases or relevant comorbidities.

Treatment allocation and intervention

This was an 18-month multicentre, randomised, double-blind, placebo-controlled, parallel-group, phase 3, superiority trial conducted at four sites in West Yorkshire, UK (Leeds, Huddersfield, Harrogate and York).

Patients were randomised (1:1) according to a computer-generated list in blocks of four with no stratification, and treatment was assigned by the central pharmacy. Clinicians, nurses, local pharmacists, patients and assessors were blinded to treatment allocation throughout the study. Data were unblinded only if needed for medical management. The study was performed in accordance with the ethical principles of the Declaration of Helsinki. The protocol and amendments received independent ethics committee and regulatory review and approval before site initiation and recruitment of patients. Informed consent was obtained prior to enrolment into the study.

Patients received MTX plus ETN or MTX plus placebo (PBO). Etanercept 50 mg subcutaneous (SC) or visually identical placebo (sterile lyophilised powder of similar appearance to ETN) SC injections was administered once weekly. Study injections were continued until patients had no tender and no swollen joints (NTSJ) (Ritchie articular index (RAI)=0 and 44-swollen joint count (SJC44)=0) for 26 weeks or until 52 weeks, whichever was reached first. In both groups, MTX was started at 10 mg weekly and increased by 5 mg every 4 weeks up to 20 mg. If NTSJ was not achieved, the MTX dose could be increased up to 25 mg/week at week 12 or thereafter. Where NTSJ was achieved by week 4 of the study, patients were maintained on 15 mg weekly MTX. All study subjects received folic acid 5 mg tablets at least twice a week. If patients achieved sustained NTSJ for 12 weeks after stopping ETN or placebo injections, MTX was weaned (figure 1). Wherever possible clinical data collection continued and included patients who withdrew from the study but were able to attend for follow-up.

Figure 1

Trial design. ETN, etanercept; MTX, methotrexate; ow, once a week, SC, subcutaneous; 1°, primary endpoint; 2°, secondary endpoint.

Non-steroidal anti-inflammatory drugs were permitted. Patients were allowed to receive an intramuscular or intra-articular corticosteroid injection to a maximum dose of 120 mg of methylprednisolone on one occasion within the first 9 months of the study and as clinically indicated thereafter. Oral steroids were not permitted during this period. The addition of other DMARDs was allowed after week 52 at the discretion of the investigator if there was ongoing disease activity.

Efficacy and safety outcomes

The primary outcome was NTSJ at week 52. Secondary outcomes included NTSJ at week 78, pain and fatigue visual analogue scores (VAS), physician global assessment of disease activity, DAS44-CRP, remission by 2010 ACR/EULAR criteria, HAQ, EQ-5D-3L and SF-36 at weeks 52 and 78, the proportion of patients with drug-free NTSJ and etanercept-free NTSJ (for those in the MTX+ETN arm).

Secondary radiographic outcomes were the change in total van der Heijde modified Sharp score (mTSS),16 erosion (ERO) and joint space narrowing (JSN) scores and proportions of patients achieving radiographic non-progression (change not exceeding the calculated smallest detectable change (SDC) for this study, or change ≤0.5 units17) at weeks 52 and 78. Radiographs were scored independently by two experienced readers, in time order, blinded to treatment allocation.

Exploratory outcomes were NTSJ at weeks 2, 12 and 26; clinical responses (DAS44-CRP<1.6, DAS28-CRP<2.6, DAS44-CRP≤2.4, DAS28-CRP≤3.2) at weeks 2, 12, 26, 52 and 78; proportions of patients achieving a minimal clinical important difference (MCID) in HAQ (a decrease of≥0.22 units from baseline)18; normal HAQ status (≤0.519); and treatment changes (synthetic DMARD and biologic DMARD drug changes, additional steroid requirement and proportion of patients requiring biologic DMARDs at week 78). Among patients able to stop injections ± MTX early, duration of biologic DMARD-free and drug-free NTSJ was determined (loss of this state was defined by RAI>0 or SJC44>0 or any change in disease activity requiring an increase in therapy).20 Additionally, we sought to identify baseline variables that were predictive of clinical and radiographic responses, while also testing for potential moderators of the treatment effect.

The subgroup of patients attending the Leeds site had ultrasound assessments for synovitis and bone erosion at baseline, weeks 52 and 78. Assessments were performed (Philips HDI 5000, employing a 15–8 MHz transducer) by three experienced rheumatologists blinded to treatment allocation, trained in musculoskeletal ultrasound, with good inter-reader and intra-reader reliability (data not shown). Findings in 20 joints (bilateral wrists, proximal inter-phalangeal joints 2 and 3, metacarpophalangeal joints 2 and 3, and metatarsophalangeal joints 1 to 5) were compared between groups. Synovitis was defined using the OMERACT definitions.21 The EULAR-OMERACT scoring system, using a 0–3 semiquantitative scale, was used to assess grey scale (GS) and power Doppler (PD) synovitis.22 The number of ultrasound detected erosions per joint was determined.

At each visit, patients were asked to report new adverse events. Clinical and laboratory adverse events were recording using standard medical terms.

Sample size

The primary outcome of this study was the proportion of patients with NTSJ at 12 months, defined as the absence of symptoms and signs of inflammatory arthritis. The study was designed to test the null hypothesis that the difference between the proportions of patients with NTSJ in each of the treatment groups was zero. It was anticipated that 30% of patients receiving MTX monotherapy would achieve NTSJ at 12 months compared with 60% of patients receiving MTX+ETN based on previous data.23 The sample size required to demonstrate a significant difference between these two proportions, at the 5% significance level with 80% power, was 50 per group. Allowing for 10% drop-out brought the number of patients required to 55 per group (110 in total).

Statistical methods

Both efficacy and safety analyses were performed on the set of subjects who received at least one dose of study drug. The primary analysis assigned patients to the groups to which they were originally randomised (intention-to-treat) and used multiple imputation (MI) by chained equations to account for missing data (details in the online supplementary section). Sensitivity analyses that either imputed non-response for those who withdrew or were lost to follow-up, or included only the observed data available at each timepoint were performed (see online supplementary section). In addition, a per-protocol analysis was performed that excluded patients who withdrew, were lost or in some other way deviated from the study protocol to a degree deemed likely to affect the outcome. Analyses were conducted using Stata 12.1. Smallest detectable change in mTSS was calculated according to the Bland–Altman 95% limits of agreement method based on the change scores of both readers.24 Continuous interval outcomes were analysed using linear regression. If problems of non-normality or heterogeneity of residuals were identified in the linear models, non-parametric quantile regression models with heteroscedasticity-robust SEs were used. Categorical outcomes were analysed using binary logistic regression. In each analysis, we controlled for baseline values and study site. As proportions estimated in the combined MI datasets cannot be summarised in terms of the number of patients exhibiting each characteristic (because these vary between the imputed sets), numerators and denominators for proportions are only presented for the observed data in the online summary material. The threshold for statistical significance was adjusted for multiple comparisons among the secondary outcomes using the Holm method. No adjustments were made for the exploratory analyses.

Descriptive results are presented for harms. For events that recurred over time, only the most severe occurrence was counted for each patient. Both absolute and relative frequencies are presented together with the number of occurrences per 100 patient-years of follow-up.


A total of 110 patients were recruited from four sites in West Yorkshire, UK, between October 2006 and May 2009 (figure 2). Demographics and baseline disease characteristics are presented in table 1. The mean (SD) age was 48.6 (13.34) years, 76% were female, median symptom duration was 7.0 months and mean (SD) baseline DAS44-CRP was 2.94 (0.91); 53% of patients were RF-positive, 77% ACPA-positive and 82% SE-positive; 41% (45/110) fulfilled the 1987 ACR RA classification criteria, and 94% (103/110) the 2010 ACR EULAR criteria.4

Table 1

Baseline characteristics of patients randomised to receive methotrexate+placebo or methotrexate+etanercept

Figure 2

Patient disposition.

The median (IQR) maximum MTX dose achieved in each group during the study was 25 (20 to 25) mg/week in both groups. Only three patients in the MTX+ETN group (5.5%) and five patients in the MTX+PBO group (9.1%) were not prescribed at least 20 mg/week. Three patients (two in the MTX+PBO group and one in the MTX+ETN group) withdrew from study treatment during the escalation phase, two patients (one in the MTX+PBO group and one in the MTX+ETN group) experienced adverse events, which limited the tolerated dose, and three patients (two in the MTX+PBO group and one in the MTX+ETN group) achieved NTSJ before reaching 20 mg/week.

Clinical outcomes

Primary clinical outcome: NTSJ

At week 52, the proportions of patients that achieved NTSJ were similar between the two groups (32.5% vs 28.1%, adjusted OR (95% CI) 1.32 (0.56 to 3.09), (p=0.522) in the MTX+ETN and MTX+PBO groups, respectively; table 2). Sensitivity analyses imputing non-response for patients who discontinued the intervention, using complete case data and/or per-protocol data, showed similar findings with no significant between-group differences (see online supplementary material table S1).

Table 2

Combined results from analyses of multiply imputed datasets of primary and secondary efficacy outcomes at weeks 52 and 78

Secondary clinical outcomes:

Remission according to the 2011 ACR/EULAR remission criteria

Remission using the 2011 ACR/EULAR Boolean-based definition at week 52 was 26.7% in the MTX+ETN and 22.5% in the MTX+PBO groups.25 In both groups, remission was achieved in a greater proportion of patients when defined by SDAI≤3.3 (47.5% vs 37.0%). For both outcomes at week 52 and 78, there were not statistically significant between-group differences (table 2).

Drug-free remission

Of the MTX+ETN group, 7.3% (4/55) achieved sustained NTSJ (>26 weeks) by week 52 of whom two stopped ETN early; 9.1% (5/55) in the MTX+PBO stopped placebo injections after achieving sustained NTSJ. Two patients in the MTX+ETN group stopped injections prematurely—each had one swollen or tender joint. In total, 3.6% (2/55) in each group had NTSJ and were drug free at week 78.

Exploratory clinical outcomes: DAS28-CRP

Earlier DAS28-CRP clinical responses were seen with MTX+ETN combination therapy. At week 2, the proportion of patients achieving DAS28-CRP<2.6 and DAS28-CRP≤3.2 was significantly greater in the MTX+ETN versus MTX+PBO groups [38.5% versus 9.2% (adjusted OR (95% CI) 8.87 (2.53 to 31.17), p=0.001) and 55.5% vs 22.2% (adjusted OR (95% CI) 6.03 (2.22 to 16.36), p<0.001), respectively] (figure 3 and see online supplementary table S3). This difference was maintained for DAS28-CRP<2.6 at week 12 [65.1% versus 43.8% (adjusted OR (95% CI) 2.49 (1.12 to 5.54), p=0.026 for the MTX+ETN and MTX+PBO groups, respectively)]; however, there was no significant difference in proportions achieving DAS28-CRP≤3.2. By week 52, high proportions of patients in both groups achieved DAS28-CRP<2.6 and DAS28-CRP≤3.2 [68.8% vs 62.5% (adjusted OR (95% CI) 1.32 (0.58 to 3.04), p=0.508) and 84.5% versus 75.1% (adjusted OR (95% CI) 1.84 (0.65 to 5.24), p=0.250) in the MTX+ETN and MTX+PBO groups] (figure 3).

Figure 3

Proportions of patients with DAS28-CRP<2.6 and DAS44-CRP<1.6, and DAS28-CRP≤3.2 and DAS44-CRP≤2.4. Methotrexate+placebo (n=55) (— — —), methotrexate+etanercept (n=55) ( ……… ).

Of the patients in the MTX+ETN group, 41.9% maintained DAS28-CRP<2.6 between week 52 and week 78 and 57.7% maintained DAS28-CRP≤3.2. For comparison, these figures were 42.9% and 64.9%, respectively, in the MTX+PBO group.

Patient-reported outcomes

Secondary outcomes: HAQ and SF-36

Improvements in patient function at week 52, measured by the HAQ (−0.4 vs −0.31 (p=0.381) in the MTX+ETN and MTX+PBO groups, respectively) were seen in both groups. Change in physical function was greater than that in mental function (ΔSF-36 PCS 8.10 vs 6.69 and ΔSF-36 MCS 0.97 vs 2.99 in the in the MTX+ETN and MTX+PBO groups) (table 2).

Exploratory outcomes: MCID HAQ and normal HAQ

The proportion of patients achieving a reduction in HAQ≥MCID was similar between groups (HAQ≥0.22 units at week 52 [MTX+ETN 59.5% vs MTX+PBO 58.3%; unadjusted OR 1.05 (0.47 to 2.37); adjusted OR 1.08 (0.47 to 2.49), p=0.854] and week 78 [58.1% vs 62.5%; unadjusted OR 0.83 (0.38 to 1.84); adjusted OR 0.85 (0.38 to 1.88), p=0.682]). Similarly, no between-group differences were seen in the proportions achieving normal function (HAQ≤0.5 at week 52 [MTX+ETN 51.5% vs MTX+PBO 44.5%; unadjusted OR 1.32 (0.60 to 2.91); adjusted OR 1.37 (0.58 to 3.26), p=0.474] or week 78 [46.1% vs 48.6%; unadjusted OR 0.91 (0.42 to 1.97); adjusted OR 0.86 (0.37 to 2.02), p=0.735]). The site adjustment for these comparisons pooled sites 2–4.

Imaging outcomes

Secondary outcome: radiographic findings

Smallest detectable change in mTSS was calculated to be three units. High proportions in both groups achieved radiographic non-progression. At week 52, there was progression ≤SDC in 93.1% and 95.5% of the MTX+ETN group and MTX monotherapy group, respectively (table 2). At week 78, this was achieved in 87.1% and 80.0% of each group, respectively (p=0.598).

Secondary outcome: ultrasound findings

In the subset of patients with ultrasound data available (MTX+ETN, n=48; MTX+PBO, n=46), there were 28/45 (62.2%) with GS>1 and PD signal PD>0 in at least one joint at baseline in the MTX+ETN group and 30/45 (66.7%) in the MTX+PBO group. Median (IQR) total ultrasound scores at baseline did not differ substantively between the treatment groups (GS: MTX+ETN=14 (9 to 22) vs MTX+PBO=12 (7 to 17); PD: MTX+ETN=3 (0 to 9) vs MTX+PBO=3 (1 to 5); erosion score MTX+ETN=1.0 (0.0 to 2.5) (n=44) vs MTX+PBO=1.0 (0.0 to 2.0)).

Of these, 27.5% and 35.7% of patients in the MTX+ETN and MTX+PBO groups, respectively, showed US synovitis (GS>1 and PD>0) in at least one joint at week 52 and 41.6% and 38.0% at week 78. Neither total GS score nor total PD score differed between the treatment groups at week 52 or 78. The number of ultrasound detected bone erosions was low in both groups; there were no statistically significant differences at week 52 or 78 (table 2).

Changes in medication

Exploratory outcomes: changes in DMARD therapy and additional steroid requirement

In total, 30.9% (17/55) and 32.7% (18/55) patients in the MTX+ETN and MTX+PBO groups, respectively, required a change in DMARD therapy during the study period. From baseline to week 52, changes were made in 5% (3/55) and 18% (10/55) patients, and from week 52 to week 78 in 25% (14/55) and 16% (9/55) patients in the MTX+ETN and MTX+PBO groups. The number of additional synthetic DMARDs used was similar between the two groups. Biological DMARDs were added in three patients in the MTX+PBO group (see online supplementary tables S5A–C).

Approximately 50% of patients in each group received at least one steroid injection during the study period. The total doses received were similar in the two groups (see online supplementary figure S1).

Predictors of outcome

A post hoc analysis of baseline predictors of outcome suggested earlier reduction in DAS28-CRP in the ACPA-positive versus ACPA-negative patients. Controlling for treatment, study site and baseline DAS28-CRP, improvement in disease activity scores was higher in ACPA-positive patients at week 2 (adjusted difference −0.86 (−1.23 to −0.49), p<0.001) and week 26 (−0.92 (−1.46 to −0.38), p=0.001) but did not differ at week 12 (−0.36 (−0.89 to 0.17), p=0.179) or week 52 (−0.40 (−0.97 to 0.17), p=0.163). These trends did not differ significantly between treatment groups (interactions all p>0.2). At week 78, the trend was reversed to an extent in the MTX+PBO group (adjusted difference 0.48) compared with the MTX+ETN group (adjusted difference −0.97: interaction −1.45 (−2.53 to −0.37), p=0.009). It should be noted however that there were few ACPA-negative patients in each group; only patients whose ACPA status was known at baseline were included in these comparisons (MTX+ETN ACPA-negative n=15, ACPA-positive n=39; MTX+PBO ACPA-negative n=10, ACPA-positive n=42). Numbers in each group were too small to determine a difference using a definition of NTSJ.

In those fulfilling the 1987 ACR RA classification criteria, 68% and 66% of the MTX+ETN and MTX+PBO groups, respectively, had a DAS28-CRP<2.6 at week 52. Similar proportions of those not fulfilling the criteria achieved DAS28-CRP<2.6 (69% and 60%).

The majority of patients fulfilled the 2010 ACR/EULAR RA classification criteria. In both the MTX+ETN and MTX+PBO groups, all patients not fulfilling the 2010 criteria (3/3 (100%) and 4/4 (100%), respectively) achieved DAS28-CRP<2.6 at week 52. Of those who fulfilled the 2010 classification criteria, similar proportions in both groups achieved remission (67% and 60%).

We investigated whether DAS28-CRP<2.6 at 12 weeks was associated with outcome at 78 weeks. Controlling for treatment and study site, patients with DAS28-CRP<2.6 at 12 weeks were more likely to have a DAS28-CRP<2.6 at 78 weeks (OR 3.53 (1.46 to 8.53), p=0.005). There was no substantive difference in the change in total van der Heijde Sharpe score (adjusted difference in medians (95% CI) −0.06 (−0.58 to 0.47), p=0.825) according to DAS28-CRP<2.6. Those with DAS28-CRP<2.6 at 12 weeks had substantively lower HAQ values (mean difference −0.53 (−0.72 to −0.34), p<0.001).

Adverse events

Total numbers of adverse events were comparable between the two groups (events per 100 patient-years MTX+ETN=451.6; MTX+PBO=417.3). Although a greater number of SAEs occurred in the MTX+ETN group (16.4 per 100 patient-years compared with 3.7 in the MTX+PBO group), many were related to medical/surgical procedures. Only two were thought possibly related to study medication and of these only one was considered possibly related to ETN. This was in a patient who had ongoing weight loss and persistently elevated inflammatory markers despite good control of her arthritis. On further investigation 4 months after baseline she was diagnosed with metastatic breast cancer and study medication was discontinued. One malignancy was also described in the MTX+PBO arm—a case of non-Hodgkin's lymphoma (NHL). The patient completed the study, and no specific treatment was required for the NHL (table 3).

Table 3

Adverse events experienced by participants during follow-up

There were 195 infectious episodes recorded in total: 105 in 37 patients in the MTX+PBO group and 90 in 43 patients in the MTX+ETN group. The majority were due to pulmonary or upper respiratory tract infections (71.4% (75/105) and 77.8% (70/90) in the MTX+PBO and MTX+ETN groups, respectively). Two of the infections were reported as severe. Both were pulmonary infections and were in the MTX+ETN group. All other infections were reported to be of mild or moderate severity (see online supplementary table S6).


To our knowledge, the EMPIRE study is the first double-blind, placebo-controlled RCT comparing MTX monotherapy with MTX+ETN combination in patients with early IA. Although designed prior to the establishment of the 2010 ACR/EULAR RA classification criteria,4 the principles of the study embody those of these criteria—to identify and treat RA in the earliest stages. Indeed, while only 41% fulfilled the 1987 criteria, the majority of the patients in this study (94%) fulfilled the subsequent 2010 RA classification criteria.

The primary outcome at week 52, NTSJ, was similar (approximately 30%) between the two groups. Although DAS responses were as expected, the outcome of NTSJ was lower in the ETN group than initially predicted (60%). This primary endpoint was chosen as it was thought that complete normalisation could be achieved in this early population. It is likely that NTSJ was too strict a target and that the achievement of no swollen joints alone may have been a more realistic outcome. Using ultrasound, synovitis at week 52 (GS<1 and PD<0) was only seen in approximately 30% of patients.

Several studies have addressed the use of MTX+ETN in early RA according to the 1987 criteria.15 ,26 ,27 In the COMET study,15 significantly higher proportions of patients achieved DAS28-ESR remission (study primary endpoint) with combination therapy compared with MTX monotherapy (50% vs 28% (effect difference 22.05% (95% CI 13.96 to 30.15%), p<0.0001) at 1 year). Remission was higher in patients with shorter disease duration (≤4 months); however, between-group difference remained significant (69.8% vs 34.7%; p<0.05).23 In the PRIZE study (mean duration 6 months)28 in which MTX+ETN was used as induction therapy, DAS28-ESR remission was seen in 70.5%. In our study, in which DAS28-CRP was used as one of the outcome measures, high proportions of DAS28-CRP<2.6 and DAS28-CRP≤3.2 were achieved in both groups. At 1 year, DAS28-CRP<2.6 with MTX+ETN and MTX monotherapy was 68.8% and 62.5%, respectively. It is likely that the high response rates in this trial are partly due to the fact that these results are based on DAS28-CRP instead of DAS28-ESR, which is known to yield slightly lower values.29 ,30 However, other explanations for the relatively high clinical responses, in particular to MTX monotherapy, in this study may be that these were patients with DMARD-naïve early disease with a relatively low disease burden (although most fulfilled the 2010 ACR/EULAR RA classification criteria, the majority had not fulfilled the 1987 ACR criteria). It is also possible that these were a MTX responsive group given the high proportion of ACPA positivity. Results from the PROMPT study in which MTX delayed the development of RA in ACPA-positive patients would further support this.14 Although the study compared MTX with MTX+ETN, there was also a ‘treat to target’ component to the study in that MTX was increased at regular intervals up to 25 mg/week if patients did not achieve NTSJ. At week 52, although ETN was stopped in the combination therapy group, additional conventional DMARDs were allowed in both groups if required. Another possibility not addressed in this study is the immunological profile that may have played a role in their clinical outcomes. Higher proportions of naïve T cells at baseline have been shown to correlate with remission induction with MTX monotherapy.31

Radiographic and functional outcomes in both groups were also high, with 80–90% achieving radiographic non-progression and almost 50% achieving normal function by week 78. Of note, direct comparisons cannot be made between studies given the differences in population groups, study designs and clinical outcomes.30

Between weeks 52 and 78, the rates of clinical responses remained high, suggesting that remission induction with a biological DMARD and maintenance of remission with MTX monotherapy, as described by Quinn et al32 and shown in other studies including BeSt33 and OPTIMA,34 may be a potential treatment strategy for some patients. The slight reduction in rates of remission in the MTX+ETN groups on withdrawal of TNFi therapy however suggests that perhaps not all may be able to achieve this. In the HIT HARD study,35 the initial clinical improvement in DAS at week 24 with MTX+adalimumab combination therapy was lost on withdrawal of adalimumab. It may be that a proportion of these will require continuation of their biological therapy, with half-dose therapy as described in the PRIZE36 and PRESERVE37 studies being a possible option. The improvement in clinical responses in the MTX monotherapy group in our study is likely due to the additional DMARD therapy allowed after week 52.

The speed of response in the combination group was remarkable with DAS44-CRP<1.6 in almost 30%, and DAS28-CRP<2.6 in 40% at 2 weeks after a single injection. This level of early response was presumably a result of the relatively short symptom duration and mild disease. Our exploratory analyses found evidence that DAS28-CRP<2.6 at 12 weeks was associated with increased likelihood of achieving DAS28-CRP<2.6 and improved function at week 78. Evidence suggests that early achievement of low levels of disease early activity relates to significantly better long-term clinical38 and radiographic outcomes.39 The longer-term outcomes of these two treatments groups are to be seen.

MTX was commenced at 10 mg weekly and increased at regular interval to 25 mg weekly or the maximum tolerated dose. The impact of starting MTX at higher dose on speed of response would be of interest.

In general, both treatment strategies were well tolerated with no new safety signals, confirming the benefit of early MTX use with dose escalation. While earlier clinical responses were seen with MTX+ETN, with maintenance of clinical and imaging responses in the majority of patients when stopping ETN, the value of this combination, particularly in the context of cost effectiveness, needs to be considered. Currently, early diagnosis and treatment with conventional synthetic DMARDs and glucocorticoids, treating to target and escalating therapy remains the mainstay of treatment of patients with RA.13 Ideally, determining which patients would benefit significantly and in the longer term with initial intensive induction therapy with a biological DMARD with the possibility of biologic or even drug-free remission would be of importance.

Our study has its limitations. It has not addressed the role of other available therapies, for example, combination synthetic DMARDS and glucocorticoids, which has shown to be effective in this group of patients.40

Another limitation is one relating to missing data. Sample size calculation based on the primary outcome allowed for a 10% drop-out. Clinical data were available for 51/55 (93%) of patients in the MTX+PBO group but 48/55 (87%) in the MTX+ETN group, falling just below the 10% margin. Analysis of the data however using both multiple imputations to account for missing data was no different from that for the observed cases only.

In summary, in this group of patients with newly diagnosed early inflammatory arthritis, almost a third had no tender, swollen joints after 1 year with a high proportion achieving DAS28-CRP<2.6 and low ultrasound synovitis. In conclusion, although clinical responses were more rapid with MTX+ETN combination therapy, it was not superior to methotrexate monotherapy in achieving the primary outcome.


The authors would like to thank the research staff at all the study sites, Sarah Fahy for her administrative support and Ged Connolly-Thompson, Andrea Paterson and Jonathan Thompson for their IT support.


View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Handling editor Tore K Kvien

  • JLN and EV share joint first authorship.

  • Correction notice This article has been corrected since it was published Online First. Figure 3 has been corrected.

  • Funding Pfizer provided study drug and unrestricted grant funding. The authors had sole responsibility for data analysis and manuscript preparation.

  • Competing interests DVDH has received consultancy fees and/or research grants from AbbVie, Amgen, AstraZeneca, Augurex, BMS, Celgene, Centocor, Chugai, Covagen, Daiichi, Eli-Lilly, GSK, Janssen Biologics, Merck, Novartis, Novo-Nordisk, Otsuka, Pfizer, Roche, Sanofi-Aventis, Schering-Plough, UCB and Vertex. EV has received consultancy fees from Janssen, Bristol-Myers-Squibb, Abbvie, Amgen, UCB. JLN has received speakers bureau fees from UCB. MHB has received honoraria for talks and/or been on advisory boards for AbbVie, BMS, Chugai and Roche and has received grant funding from Pfizer. PE has received consultancy fees and done speakers bureaus for AbbVie, Roche, UCB, Pfizer, BMS and MSD. PGC has done speakers bureaus or been on advisory boards for BMS, Janssen, Novartis, Pfizer and Roche. RJW has received honoraria from AbbVie.

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