Objectives Combining disease-modifying antirheumatic drugs (DMARDs) with glucocorticoids (GCs) is an effective treatment strategy for early rheumatoid arthritis (ERA), yet the ideal schedule and feasibility in daily practice are debated. We evaluated different DMARD combinations and GC remission induction schemes in poor prognosis patients; and methotrexate (MTX) with or without GC remission induction in good prognosis patients, during the first treatment year.
Methods The Care in ERA (CareRA) trial is a 2-year investigator-initiated randomised pragmatic open-label superiority trial comparing remission induction regimens in a treat-to-target approach. DMARD-inexperienced patients with ERA were stratified into a high-risk or low-risk group based upon presence of erosions, disease activity, rheumatoid factor and anticitrullinated protein antibodies. High-risk patients were randomised to a COBRA Classic (MTX + sulfasalazine + prednisone step-down from 60 mg), COBRA Slim (MTX + prednisone step-down from 30 mg) or COBRA Avant Garde (MTX + leflunomide + prednisone step-down from 30 mg) scheme. Low-risk patients were randomised to MTX tight step-up (MTX-TSU) or COBRA Slim. Primary outcome was the proportion of patients in 28 joint disease activity score calculated with C-reactive protein remission at week 52 in an intention-to-treat analysis. Secondary outcomes were safety and effectiveness (ClinicalTrial.gov identifier NCT01172639).
Results 98 COBRA Classic, 98 COBRA Slim (high risk), 93 COBRA Avant Garde, 47 MTX-TSU and 43 COBRA Slim (low risk) patients were evaluated. Remission was achieved in 64.3% (63/98) COBRA Classic, 60.2% (59/98) COBRA Slim (high risk) and 62.4% (58/93) COBRA Avant Garde patients at W52 (p=0.840); and in 57.4% (27/47) MTX-TSU and 67.4% (29/43) COBRA Slim (low risk) patients (p=0.329). Less adverse events occurred per patient with COBRA Slim (high risk) compared with COBRA Classic or COBRA Avant Garde (p=0.038). Adverse events were similar in MTX-TSU and COBRA Slim (low risk) patients (p=0.871). At W52, 76.0% patients were on DMARD monotherapy, 5.2% used GCs and 7.5% biologicals.
Conclusions MTX with a moderate-dose GC remission induction scheme (COBRA Slim) seems an effective, safe, low-cost and feasible initial treatment strategy for patients with ERA regardless of their prognostic profile, provided a treat-to-target approach is followed.
Trial registration numbers EudraCT-number 2008-007225-39 and NCT01172639; Results.
- Early Rheumatoid Arthritis
- DMARDs (synthetic)
Statistics from Altmetric.com
In the last decades, new treatment strategies and potent new drugs have improved clinical prospects for patients with rheumatoid arthritis (RA). Current guidelines recommend treating a patient with early rheumatoid arthritis (ERA) rapidly, intensively and to target.1–3
The recommendations were prompted by trials on intensive treatment strategies with combinations of classical synthetic disease-modifying antirheumatic drugs (csDMARD) and glucocorticoids (GCs).4–11 These studies advocate the ‘window of opportunity’ theory stating that intensive treatment early in the disease process and rapid control of disease activity results in better radiographic outcomes and preserved functionality later on.12 However, there are still problems with the implementation of such strategies in daily practice, partly because of questions regarding the optimal dosage and combination of drugs.13 ,14 Indeed, methotrexate (MTX) monotherapy might be sufficient to control ERA in many patients, and in case of insufficient response, stepping-up to triple csDMARD or biological therapy could rescue them.15 ,16 However, delaying optimal disease control results in unnecessary further suffering and loss of participation.
Guidelines also advise adapting treatment to the patient's prognostic profile.1–3 Such recommendations are however mostly based on results from preselected trial populations, initiating a fixed treatment schedule without treating-to-target. Classical prognostic algorithms predicting structural damage proved to be unreliable in daily practice, causing undertreatment in patients with a better prognosis.9 ,17 ,18 We hypothesise that a combination of csDMARDs with rapid remission inducing agents like GCs or biologicals is the most effective approach for all patients with ERA, regardless of the prognostic profile, emphasising also the need to define the most feasible drug combination for daily practice.
GCs are commonly used to bridge the slow onset of csDMARD effect, but the optimal dosage, treatment duration and administration route are up for discussion.19 The role of GCs is perceived as ambiguous, both by patients and physicians.14 ,20
Biologicals with MTX are more efficacious than MTX monotherapy in ERA trials, but their temporary use as remission induction agents just as GCs is less well studied.21 Most trial setups led to persistent biological use after the induction phase and only recently the feasibility of dose tapering has been shown.22 So far, compared with GCs, induction with tumour necrosis factor-blocking agents did not demonstrate superior efficacy8 ,23 ,24 and their cost-effectiveness remains unproven in ERA.25
COBRA (Combination therapy for early Rheumatoid Arthritis)-like schemes (MTX ± sulfasalazine (SSZ) + GCs), triple therapy (MTX, SSZ and hydroxychloroquine) or other csDMARD combinations demonstrate excellent clinical efficacy compared with monotherapy, while studies comparing different combination regimens following a treat-to-target approach are scarce.
In view of determining the optimal treatment strategy for every patient with ERA, regardless of the prognostic profile, we compare in this study the effectiveness of different initial csDMARD combinations, with or without GC remission induction in high-risk and low-risk patients, 52 weeks after treatment initiation.
The CareRA trial (Care in ERA) is a 2-year prospective investigator-initiated multicenter pragmatic open-label randomised superiority trial. Thirteen Flemish rheumatology centres (2 academic centres, 7 general hospitals and 4 private practices) actively recruited patients.
The central Ethics Committee (EC) of the University Hospitals Leuven and the local ECs approved the study.
Patients with RA, as defined by the American College of Rheumatology 1987 revised criteria, were recruited between January 2009 and May 2013. Main inclusion criteria were having a disease duration ≤1 year and being DMARD/GC unexperienced. Patients having contraindications for intensive treatment with GCs were excluded. Online supplementary material 1 provides all exclusion criteria. Patients with specific comorbidities such as controlled diabetes, osteoporosis and previous malignancy were not excluded. All patients gave written informed consent.
Randomisation and procedures
Patients were stratified to a high-risk or low-risk group based on an algorithm including erosions, rheumatoid factor and/or anticitrullinated protein antibody and disease activity score calculated with C-reactive protein (CRP) (DAS28(CRP)) at screening (figure 1).
After risk allocation, patients were randomised via a digitally generated sequence.
High-risk patients were randomised into 1/3 treatment arms:
COBRA Classic: 15 mg MTX weekly, 2 g SSZ daily and a weekly step-down scheme of oral prednisone (60-40-25-20-15-10-7.5 mg).
COBRA Slim: 15 mg MTX weekly with a weekly step-down scheme of oral prednisone (30-20-12.5-10-7.5-5 mg).
COBRA Avant Garde: 15 mg MTX weekly, 10 mg leflunomide (LEF) daily and a weekly step-down scheme of oral prednisone (30-20-12.5-10-7.5-5 mg).
Low-risk patients were randomised into 1/2 treatment arms:
MTX tight step-up (MTX-TSU): 15 mg MTX weekly, no oral steroids allowed.
Prednisone was tapered down over 6 weeks to 7.5 mg in COBRA Classic and 5 mg in the other arms, maintained until week (W) 28 and then further tapered until complete discontinuation at W34. At W40, MTX monotherapy was aimed for in all arms, except COBRA Avant Garde, where patients were re-randomised to MTX or LEF.
Prophylactic treatment with oral folic acid, calcium and vitamin D was prescribed. Moreover, face-to-face education, medication schemes and standardised info-material (leaflet, DVD and website) were provided.
If patients failed to reach the target of DAS28(CRP) ≤3.2 in a tight control setting, treatment adjustments were made per protocol from week 8 onwards. Rheumatologists had the option not to adapt treatment once, but had to motivate their decision based on a predefined list of clinical conditions. Treatment adjustments not stated in the protocol were not allowed. The first adjustment in all treatment arms was an increase in the weekly MTX dose to 20 mg. If necessary a second adjustment could be made from 8 weeks after the first, depending on the treatment arm: a SSZ dose increase to 3 g daily in COBRA Classic, the addition of 10 mg LEF daily in COBRA Slim and MTX-TSU or a LEF dose increase to 20 mg daily in COBRA Avant Garde. Not reaching the target after two predefined treatment adjustments was considered a strategy failure for efficacy reasons (efficacy failure). In case of toxicity, the protocol predefined schemes for tapering/interrupting the assigned treatment. Persistent toxicity implied a strategy failure for safety reasons (safety failure).
Intramuscular/articular GC injections were allowed maximally every 8 weeks, but not within 4 weeks preceding W16, W28, W40 and W52.
Patients were assessed at screening, baseline, W4, W8, W16, W28, W40 and W52. Maximally 4 weeks were allowed between screening and baseline. If treatment adjustment was required, optional visits could be performed at W12, W20, W24, W32, W36, W44 and/or W48. Demographics were registered at screening. Clinical parameters including DAS28(CRP) and health assessment questionnaire (HAQ) were measured at every visit. X-rays of hands and feet were obtained at baseline, W28 and W52. Patients experiencing failure of the predefined treatment schedules were followed further and evaluated at W28 and W52.
As predefined in the CareRA protocol, the co-primary study end point at W52 was the proportion of patients in remission (DAS28(CRP) <2.6). The other two co-primary end points were remission at W16, published earlier26 ,27 and at W104. Secondary efficacy outcomes per protocol were radiographic evolution by the Sharp van der Heijde (SvdH) score, proportion of good European League Against Rheumatism (EULAR) responders (DAS28(CRP) change >1.2 and DAS28(CRP) ≤3.2), proportion of patients having a clinically meaningful improvement of the HAQ (HAQ change >0.22) and proportion of patients having a HAQ=0 at W52. Secondary effectiveness parameters included in the CareRA protocol were proportion of treatment failures, proportion of patients treated out of protocol, proportion lost to follow-up, total cumulative/mean daily steroid dose and proportion starting biologics in the first year.
At each visit, patients were questioned about any adverse events (AE). Each AE was registered and evaluated (relation to therapy, seriousness and severity) by the treating rheumatologist.
This superiority study compared COBRA Classic versus COBRA Slim and COBRA Avant Garde versus COBRA Slim in the high-risk arm. Sample size calculation was based upon the expected proportion of patients in remission at W16.27 Eighty-five patients per treatment arm were required for a power of 80% and significance level of 0.05, starting from an estimated clinically relevant difference in effect size of 20%. Analysis of the low-risk population was exploratory.
All patients were considered for analysis. An intention-to-treat (ITT), a per-protocol (PP) and a safety analysis were performed. The safety analysis used all available data, the PP analysis included only patients following the protocol until week 52 and the ITT analysis imputed data when missing or from the moment patients were considered treatment failures.
Screening variables were used to impute missing baseline variables and vice versa. To impute other data, the last observation carried forward method was applied.
Primary and secondary outcomes were examined by χ2 or Kruskal-Wallis test, when appropriate. Radiographic images were scored chronologically via the SvdH method by one experienced reader (DDC) and four participating trained medical students. Each X-ray was scored by three readers, including DDC and two students. The mean score was retained in case of differences between readers. All statistical analyses were carried out using SPSS V.22.0. A p value <0.05 was considered to be statistically significant.
In total, 400 patients were screened and 379 were included. The stratification resulted in 289 high-risk and 90 low-risk patients, 21 patients were excluded from analysis because of screening or randomisation errors. Randomisation was successful, with similar characteristics in all arms for both risk groups (table 1). Of the 289 high-risk patients, 98, 98 and 93 were randomised in the COBRA Classic, COBRA Slim and COBRA Avant Garde arm, respectively. Of the 90 low-risk patients, 47 and 43 were randomised in the MTX-TSU and COBRA Slim group, respectively. Figure 1 shows the patient disposition.
ITT analysis of the high-risk group did not show differences in the primary outcome between groups (tables 2 and 3). Remission was achieved in 64.3% (63/98) COBRA Classic, 60.2% (59/98) COBRA Slim and 62.4% (58/93) COBRA Avant Garde patients (p=0.840) at W52. A good EULAR response was reached in 67.3% COBRA Classic, 68.4% COBRA Slim and 67.7% COBRA Avant Garde patients (p=0.995). A clinically meaningful HAQ response was reached in 68.4% COBRA Classic, 70.4% COBRA Slim and 71.7% COBRA Avant Garde patients (p=0.877). HAQ=0 was achieved in 37.8% COBRA Classic, 36.7% COBRA Slim and 44.1% COBRA Avant Garde patients (p=0.533).
ITT analysis of the low-risk group showed remission in 57.4% (27/47) MTX-TSU and 67.4% (29/43) COBRA Slim patients (p=0.329). A good EULAR response was achieved in 57.4% MTX-TSU and 60.5% COBRA Slim patients (p=0.771). A clinically meaningful HAQ response was reached in 59.6% MTX-TSU and 55.8% COBRA Slim patients (p=0.876). HAQ=0 was accomplished in 29.8% MTX-TSU and in 48.8% COBRA Slim patients (p=0.064).
PP analysis showed similar results. Remission based on different criteria including DAS28(erythrocyte sedimentation rate), simplified disease activity index and Boolean remission showed proportionally comparable results (see online supplementary material 2).
Mean±SD area under the curve (AUC) for DAS28(CRP) from baseline to W52 was 35.0±11.6, 35.3±10.6 and 33.9±8.6 for COBRA Classic, COBRA Slim and COBRA Avant Garde, respectively (p=0.685). MTX-TSU patients had a higher AUC for DAS28(CRP) than COBRA Slim patients over 52 weeks of treatment in the low-risk group (42.0±13.1 vs 35.8±14.1; p=0.017). Figure 2 displays disease evolution from baseline until W52 in both risk groups.
Presence or absence of erosions was evaluated on site in view of risk group stratification and 80% of baseline X-ray images of hands and feet were also available for centralised scoring according to SvdH. Overall, baseline structural damage and radiographic progression with a smallest detectable change of 3.3 were minimal in all groups. Tables 2 and 3 and figure 2 give more insight in the radiographic evolution.
Starting from 98 COBRA Classic, 98 COBRA Slim and 93 COBRA Avant Garde patients, 62 (63.2%) COBRA Classic, 75 (76.5%) COBRA Slim and 60 (64.5%) COBRA Avant Garde patients were followed PP up to and including W52 (table 4). More treatment adaptations and GC injections were noted in COBRA Slim patients. Starting from 47 MTX-TSU and 43 COBRA Slim patients, 36 (76.6%) MTX-TSU and 30 (69.8%) COBRA Slim patients were followed PP until W52. Data on number of treatment changes and cumulative/daily GC use per treatment arm can be found in table 4.
In the COBRA Classic arm, 6 protocol-deviations, 12 efficacy-failures and 11 safety-failures occurred. One COBRA Classic patient died while out-of-protocol. In the COBRA Slim (high risk) arm, 4 protocol-deviations and 11 efficacy-failures occurred. One COBRA Slim (high risk) patient was lost-to-follow-up while out-of-protocol. In the COBRA Avant Garde arm, 4 protocol-deviations, 15 efficacy-failures and 9 safety-failures occurred. Four COBRA Avant Garde patients were lost-to-follow-up while out-of-protocol. In both the MTX-TSU and the COBRA Slim (low risk) arm, seven efficacy-failures and one protocol-deviation occurred.
Biologicals were started out-of-protocol in 26 patients, with 2 patients (1 COBRA Slim high risk and 1 MTX-TSU) receiving 2 different biologicals in year 1. In the high-risk group, biologicals were administered to 10 COBRA Classic, 2 COBRA Slim and 6 COBRA Avant Garde patients, respectively. In the low-risk group, biologicals were given to four MTX-TSU and four COBRA Slim patients. Remission was achieved in 26.9% (7/26) and low disease activity in 50.0% (13/26) of patients receiving biologicals. Table 4 gives full detail about the treatment of all patients at W52.
Overall, of all patients in follow-up, including protocol-failures reaching and attending the W52 visit, 65.6% (227/346) achieved remission and 80.3% (278/346) low disease activity, with 76.0% (263/346) on DMARD monotherapy and minimal additional GC (5.2%; 18/346) or biological therapy (7.5%; 26/346).
Safety analysis in the high-risk group showed that therapy-related AEs were reported in 67.3% COBRA Classic, 66.3% COBRA Slim and 78.5% COBRA Avant Garde patients (p=0.125) (table 5). Per patient 1.9±2.0, 1.3±1.4 and 1.9±1.6 AEs related to therapy occurred in the COBRA Classic, COBRA Slim and COBRA Avant Garde group, respectively (p=0.028). Two therapy-related serious adverse events (SAEs) were reported in each treatment arm. Most AEs related to gastrointestinal problems and a general unwell feeling.
In the low-risk group, therapy-related AEs were reported in 63.8%, and 51.2% of MTX-TSU and COBRA Slim patients, respectively (p=0.224). Per patient 1.2±1.2 and 1.2±1.5 AEs related to therapy occurred in the MTX-TSU and COBRA Slim group, respectively (p=0.737). Gastrointestinal problems and a general unwell feeling, the two most frequent AEs, were equally prevalent in both groups. Two therapy-related SAEs were registered in the COBRA Slim group and none in the MTX-TSU group. Online supplementary table S3 gives a detailed overview of AEs related to therapy.
In patients with markers of poor prognosis, regardless of concomitant prednisone dosing, csDMARD combinations were not superior to MTX monotherapy with moderate-dose prednisone remission induction, the so-called COBRA Slim regimen, after 52 weeks. As previously reported, time-to-remission was short, with 63% of patients in remission after 4 weeks of treatment27 Furthermore, COBRA Slim seemed safer. However, following the treat-to-target approach per protocol, more treatment adaptations and more GC injections were needed with COBRA Slim compared with COBRA Classic and COBRA Avant Garde.
Although on average COBRA Slim does not appear less efficacious, a subgroup of high-risk patients with ERA might benefit from remission induction with a DMARD combination and prednisone. In the absence of effective biomarkers, reliable treatment stratification is however not yet possible. Therefore, COBRA Slim seems the most feasible initial approach for all high-risk patients, avoiding ‘collateral damage’ in terms of toxicity and retention rate. Contrary to our study, the Treatment in the Rotterdam Early Arthritis CoHort (tREACH) trial demonstrated that triple DMARD therapy was more effective than MTX monotherapy, however, with a low-dose short-term GC bridging strategy.28 ,29
For remission induction, a high GC dose was not more advantageous than a moderate dose, regardless of the DMARD strategy. den Uyl et al30 described similar findings comparing the original COBRA regimen with an attenuated version. Unfortunately, this study lacked decisive evidence: MTX was suboptimally dosed in the Classical COBRA arm and cumulative GC dosages in both arms were comparable.
After 1 year, the three high-risk CareRA treatment strategies showed similar effectiveness. Although more treatment adaptations occurred in the COBRA Slim arm, compared with COBRA Classic or COBRA Avant Garde the safety profile and the retention rate were better, increasing feasibility in daily practice and probably partly explaining the numerically higher need for biologicals after induction with DMARD combinations.
In patients stratified as low risk, MTX with or without moderate-dose prednisone remission induction yielded similar efficacy and safety at W52. However, AUC and time-to-remission were in favour of MTX with prednisone, confirming our W16 results.26
As shown in other studies, therapeutic strategies applying the treat-to-target principle have converging efficacy over time.31 However, we showed a lower cumulative disease activity and faster treatment response on MTX with GCs over the first year, which seems more in accordance with patients' preferences.32 Moreover, the comparable safety profile in both low-risk arms provides important new evidence that GC remission induction is relatively safe in patients with so-called mild ERA.
A limitation of this study is that medication adherence was not measured. Second, no blinding was implemented. These two drawbacks relate to the pragmatic trial design aiming to reflect daily clinical practice. Another limitation is the a priori defined superiority design. Therefore, we were only capable to demonstrate non-superiority of COBRA Classic and COBRA Avant Garde versus COBRA Slim and not non-inferiority or equivalence. Comparable to our experience in a previous observational study, only 25% of our early RA population entered the low-risk arm after stratification. Given the small sample size, results in the low-risk group should be interpreted as explorative and in need of confirmation.
Finally, DAS28(CRP) remission was chosen as primary end point, which is less stringent than recently developed criteria.33 However, despite differences in scale, the main study outcome was comparable irrespective of the remission criteria used (see online supplementary material 2). Additionally, we used a DAS28(CRP) of 3.2 as cut-off for treatment escalation throughout the study. Use of other treatment goals could have influenced the results.
A strength of our study is that it provides an optimal initial treatment solution for all patients with ERA, which is of relevance for daily clinical practice. Classically, trials preselect subpopulations most responsive to the treatment strategy under investigation. However, traditional markers of disease prognosis appear ineffective for response prediction to the currently recommended intensive treatment strategies.18 ,34 The stratification model used for treatment allocation in the CareRA trial, dividing the population in so-called high-risk and low-risk patients, proved not to be predictive of the observed treatment response. Paradoxically enough, a larger proportion of patients receiving a COBRA Slim regimen achieved a EULAR response and a clinically meaningful change in HAQ scores in the high-risk compared with the low-risk group (see tables 2 and 3). It is, however, conceivable that risk stratification in early RA could be optimised by including other clinical and radiological prognostic parameters or more sophisticated biomarkers and this should certainly be the subject of future studies in this field. Pending the development of better stratification models, our findings suggest that patients conventionally perceived as having a lower risk of a severe disease course should be treated according to the same standards as high-risk patients. Overall in this study, remission was achieved in approximately 65% and low disease activity in 80% of patients completing the W52 visit, with a large majority of patients on DMARD monotherapy and with minimal long-term use of GCs or biologicals (see table 5). These results should be a benchmark for future cost-effectiveness studies of biologicals in ERA.
In conclusion, we propose an effective, safe, feasible and standardised initial treat-to-target strategy, COBRA Slim, yielding excellent results in all patients with ERA regardless of classical prognostic markers, with high remission rates and low numbers of patients remaining on chronic GC or biological therapy.
We thank Veerle Stouten, Sylvie Van Vlasselaer and all other study personnel, and show our gratitude to all participating patients. We would like to thank all participating patients and study staff at all sites. Leflunomide was made available for free by SANOFI Belgium without any influence on trial design. Funding was provided by the Flemish Governmental Agency for Innovation by Science and Technology (IWT), the Fund for Scientific Research in Rheumatology (FWRO) and PV holds the unrestricted Pfizer chair ‘Early RA management’ at the KU Leuven. We thank medical students Veerle Ide, Marie-Anne Noreillie, Vincent Van den Bosch and Stefanie Willems for their help in reading the x-rays.
Handling editor Tore K Kvien
PV and DDC are co-first authors.
Collaborators On behalf of the CareRA study group: Maeyaert B, De Brabanter G, Devinck M, Langenaken C, Lenaerts J, Corluy L, Remans J, Vander Cruyssen B, Ravelingien I, Van Essche E, Vandevyvere K, Durnez A, Verbruggen A, Geens E, Raeman F, Joos R, de Vlam K, Taelman V, Vanhoof J, Coppens M, Geusens P, Sileghem A, Volders P, Van Den Bosch F, Verschueren P, Westhovens R.
Contributors All authors have made substantial contributions to the conception or design of the work. PV, JJ and RW were responsible for drafting the original protocol. LC, RJ, CL, VT, FR, IR, KV, JR, JL, EG, PG, JV, AD, BVC, EVE, AS and GDB were consulted to further refine the protocol and were responsible for daily patient management together with PV and RW. DDC was responsible for data analysis. Furthermore, DDC and PV drafted the paper. LC, RJ, CL, VT, FR, IR, KV, JR, JL, EG, PG, JV, AD, BVC, EVE, AS, GDB, JJ, KVdE, SM and RW revised it critically for important intellectual content and all made a final approval of the version to be published. Finally, all authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Funding This RCT was made possible by a Flemish governmental grant (IWT—Innovatie door Wetenschap en Technologie). PV holds the Pfizer Chair for Early Rheumatoid Arthritis Management at the KU Leuven.
Competing interests None declared.
Ethics approval The study was approved by the Ethics Committee of the University Hospitals Leuven and all study patients gave their written informed consent.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Further analyses are available at request from DDC/PV, the corresponding authors at KU Leuven.
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.