Objectives To assess the efficacy and safety of synthetic disease-modifying antirheumatic drugs (DMARDs) in adults with rheumatoid arthritis (RA)—a first step in a European League Against Rheumatism (EULAR) initiative to produce recommendations for the management of RA.
Methods A systematic review of the literature using PubMed, Embase and the Cochrane library was performed up to January 2009. All randomised controlled trials (RCTs) reporting the efficacy of synthetic DMARDs (vs placebo or other synthetic DMARDs) on signs and symptoms, disability and/or radiographic structural damage in patients with RA were selected. Studies of biological agents or glucocorticoids were excluded. A pooled effect size (ES) was calculated by meta-analysis. Safety and the occurrence of infections and neoplasia was also assessed.
Results 97 RCTs (14 159 patients) were analysed for efficacy. The pooled analysis indicated that methotrexate (MTX) was more efficacious in reducing signs and symptoms, disability and radiographic structural damage than other synthetic DMARDs pooled: ES for swollen joint count (SJC) versus pooled DMARDs=1.42 (95% CI 0.65 to 2.18). Leflunomide appeared to be as effective as MTX. Sulfasalazine and injectable gold were efficacious in reducing signs and symptoms and structural damage. Ciclosporin, minocycline, tacrolimus and hydroxychloroquine showed some efficacy in reducing SJC. Auranofin and D-penicillamine showed no significant superiority over placebo. The risks of cancer and of infection were increased with cyclophosphamide and azathioprine.
Conclusions MTX was well-tolerated and effective in reducing signs and symptoms, disability and structural damage. A comparison with other synthetic DMARDs was in favour of MTX, though at the tested doses MTX and leflunomide were equally effective.
Statistics from Altmetric.com
Despite extensive research, rheumatoid arthritis (RA) continues to have a profound impact on the life of the patients, their families and society. To avoid long-term disability and premature death, therapeutic approaches call for early intervention with disease-modifying antirheumatic drugs (DMARDs), either as monotherapy or combination therapy. In 2009, a European League Against Rheumatism (EULAR) Task Force aimed at aggregating available information on disease modification in RA into practical recommendations.1 To this end, several systematic literature reviews (SLRs) were necessary for the provision of current states of evidence. The objectives of this report were to analyse the literature on the efficacy on signs and symptoms, disability and structure, of all synthetic DMARDs available in RA, in published randomised controlled trials and to assess the safety, with a special focus on cancers and infections. For efficacy, the focus was on mono- and combination therapy but disregarding addition of biological agents or glucocorticoids; the latter will be discussed elsewhere.2 ,3
Methods are described briefly here; more details are given in the online supplementary material.
A systematic literature search was performed in PubMed Medline, Embase and Cochrane library databases up to January 2009 without limitation of years of publication or journal, using the followings keywords for articles in English: (arthritis, rheumatoid [MESH] OR rheumatoid arthritis [tiab]) AND (‘name of drug’ [substance name] OR ‘name of drug’ [tiab]) AND (clinical trial OR randomised controlled trial). Research questions defined in the subgroup were: (a) the relative efficacy of methotrexate (MTX) versus other synthetic DMARDs; (b) the efficacy of other synthetic DMARDs versus placebo; (c) the efficacy of synthetic DMARDs in monotherapy versus in combination. It should be noted that when a drug A or placebo was added to the treatment of patients who were inadequate responders to drug B, the trial was analysed as a comparison of drug A versus placebo, even if drug B was continued.
To examine safety, a systematic literature search was performed in the PubMed Medline database up to March 2009 without limitation of year of publication or journal, using the followings keywords for articles in English: (drug name) AND arthritis, rheumatoid [MESH] AND (tolerability OR toxicity OR infection OR malignancy) NOT (controlled trial OR randomised controlled trial).
Efficacy was assessed by the change in signs and symptoms or disability status between baseline and week 24 or closest time point, and by the change in radiographic joint damage between baseline and week 48 or closest time point in both groups.
In each trial the effect size (ES) or the standardised response mean (SRM) was determined to assess the magnitude of the treatment effect. Detailed explanations are presented in the online material. A pooled ES and pooled SRM were calculated by meta-analysis, using the inverse variance method. RevMan version 5.0 (Review Manager, Copenhagen, The Nordic Cochrane centre, The Cochrane Collaboration, 2008) statistical software was used. Statistical heterogeneity was tested by the Q test and I2 test. All meta-analyses were carried out with a random-effects model in case of significant heterogeneity.
Literature search results and trials characteristics
Initially, 759 potentially relevant articles were screened. Finally, 97 randomised trials were included (selection process shown in online supplementary figure A), comprising a total of 14 159 patients (72.9% women). Their mean age was 51.9±4.6 years and mean disease duration 5.2±3.9 years. Patients' characteristics according to the drug used are detailed in table 1; 57 (59%) trials reported clear intention-to-treat analyses.
Initially, 821 potentially relevant articles were screened. Finally, 39 articles were included (selection process shown in online supplementary figure B).
Efficacy of MTX versus monotherapy with other synthetic DMARDs
MTX versus pooled monotherapy with other synthetic DMARDs
Seventeen trials reported comparisons of MTX monotherapy and monotherapy with other synthetic DMARDs. The meta-analysis showed MTX to be better than the other DMARDs pooled: the pooled SRM for swollen joint count (SJC) versus all other synthetic DMARDs, whatever the dose, was in favour of MTX: 1.42 (95% CI 0.65 to 2.18) (figure 1). It should be noticed that MTX was used at relatively lower doses than the optimal doses currently recommended.4
MTX versus leflunomide
The results were not significantly different for leflunomide and MTX (at the tested doses) on SJC (four studies, 1889 patients, SRM=0.09 (95% CI −0.12 to 0.30)),5,–,8 pain (four studies, 1475 patients, SRM=−0.04 (−0.33 to 0.26)),5,–,8 disability (four studies, 1465 patients, SRM=−0.09 (−0.30 to 0.11)),5,–,7 ,9 ACR20 response criteria (four studies, 1889 patients, OR=1.04 (0.60 to 1.79))5,–,8 and structure (two studies, 895 patients, SRM= 0.03 (−0.10 to 0.16)).6 ,8 The mean dose of MTX, reported in three studies, was 12.5 mg/week. MTX could be increased to ≥15 mg/week in all studies, but was not as high as currently recommended.4
MTX versus sulfasalazine
The results were not significantly different for sulfasalazine versus MTX (at the tested doses) on SJC (two studies, 206 patients, SRM=0.59 (95% CI −1.96 to 3.15)),10 ,11 disability (two studies, 208 patients, SRM=0.62 (−0.86 to 2.10))10 ,11 and on ACR50 response criteria (two studies, 193 patients, OR=1.57 (0.82 to 3.00)).12 ,13 Mean dose of MTX, reported in four studies, was 14.2 mg/week and mean dose of sulfasalazine was 2.6 g/day. MTX could be increased to ≥15 mg/week and sulfasalazine could be increased to ≥3 g/day in all studies.
MTX versus intramuscular gold
MTX was not significantly different from intramuscular gold on SJC (one study, 174 patients, ES=−0.03 (−0.33 to 0.27)),14 disability (one study, 138 patients, ES=0.00 (−0.33 to 0.33))15 but was less efficacious on structural changes in one study than intramuscular gold (174 patients, ES=−1.15 (−1.47 to −0.83))14 though MTX patients had higher radiographic scores and more eroded joints at baseline.
MTX versus auranofin
MTX was more efficacious than auranofin on SJC (one study, 197 patients, SRM=6.02 (95% CI 5.35 to 6.68)) and on structure (one study, 167 patients, SRM=0.63 (0.32 to 0.94)).16 However, in another study with 144 patients, difference for SJC was not statistically significant: ES=0.0 (−0.33 to 0.33) but MTX dose in that study was only 7.5 mg/week.17
MTX versus ciclosporin
MTX appeared more efficacious than ciclosporin on SJC in one study (103 patients): but the difference did not reach statistical significance: ES=0.35 (95% CI −0.04 to 0.74)18; for ACR50 response criteria the difference was significant in one study (84 patients): OR=2.97 (1.22 to 7.28).13
MTX versus tetracyclines
There was no difference in the effect on SJC between doxycycline and MTX in one study but the dose of MTX was only 7.5 mg/week.19 There were no interpretable data comparing minocycline and MTX.
Efficacy of monotherapy with other synthetic DMARDs versus placebo
The research question was the efficacy of synthetic DMARDs versus placebo. Leflunomide, sulfasalazine, intramuscular gold, hydroxychloroquine, ciclosporin, minocycline, azathioprine and tacrolimus were all shown to be efficacious (details are given in online supplementary file 3). A Cochrane review indicated that cyclophosphamide was efficacious on SJC and structure.22
Synthetic DMARD monotherapy versus combination therapy with synthetic DMARDs
MTX monotherapy versus MTX combination therapy with other synthetic DMARDs
For this question, we used data from a previously published SLR,23 in which a meta-analysis of 19 trials (2025 patients) comparing MTX in combination versus MTX monotherapy had been performed.
DMARD naive patients
In DMARD naive patients, the trials showed no significant advantage of the combination treatments with MTX versus MTX monotherapy for pain, Health Assessment Questionnaire, ACR20, 50, 70 response criteria, and withdrawals due to lack of efficacy or toxicity were similar in both groups (relative risk, RR=1.16 (95% CI 0.70 to 1.93)).
DMARD inadequate responders
In DMARD inadequate responders, results were less conclusive. In MTX inadequate responders (three trials), MTX combination was more efficacious than continued MTX monotherapy on SJC (three studies, SRM= −0.78 (95% CI −1.30 to −0.25)), pain (three studies, SRM=−0.64 (−1.01 to −0.28)), Health Assessment Questionnaire (three studies, SRM= −1.21 (−2.07 to −0.36)) and ACR response (two studies, RR ACR20=0.26 (0.16, 0.40)) (figure 2). However, these results may be balanced by toxicity aspects, as indicated by a previously published meta-analysis23: eight trials performed in DMARD inadequate responder patient populations did not show significant differences in the number of withdrawals between MTX combination and MTX monotherapy. One exception was O’Dell's study of 102 DMARD inadequate responders, in which the combination of MTX with sulfasalazine and hydroxychloroquine showed a smaller number of withdrawals than MTX alone with a RR of 0.30 (0.14 to 0.65).24
Sulfasalazine monotherapy versus sulfasalazine combination
In six trials (657 patients), there was no difference for SJC, function, ACR20, 50 and 70 response criteria. The only significant result was for structural damage in one trial favouring the combination MTX+sulfasalazine+hydroxychloroquine: SRM=−1.70 (95% CI −2.03 to −1.37)25 and for pain in one trial favouring sulfasalazine monotherapy: SRM=4.10 (2.91 to 5.29)10 (online supplementary file table K). Other combinations are detailedin the online supplementary material.
Long-term safety of synthetic DMARDs
Smitten et al presented a retrospective study from 1999 to 2006 including 24 530 patients with RA and 500 000 controls, in which patients with RA were observed to have an increased risk of being hospitalised for an infection.26 In a nested case–control analysis performed within the RA cohort, oral corticosteroid use was associated with a dose-related increase, and biological treatments were associated with only a slightly increased risk. Patients treated with hydroxychloroquine and MTX had a decreased risk of hospitalisations for infections, whereas for sulfasalazine, leflunomide and other traditional DMARDs there was no association.
A case–control design nested within a cohort of 23 733 patients with RA studied between 1980 and 2003 was reported by Bernatsky et al.27 The risk for all infections requiring hospitalisation appeared to be most increased with current exposures to cyclophosphamide: RR=3.26, 95% CI 2.28 to 4.67 and systemic glucocorticoid agents: RR=2.56, 95% CI 2.29 to 2.85. Azathioprine was associated with a moderately increased risk: RR=1.52, 95% CI 1.18 to 1.97. There was a suggestion of increased risk of pneumonia due to MTX: RR=1.16, 95% CI 1.02 to 1.33.
Lacaille et al performed a retrospective, longitudinal study of a population-based RA cohort followed up from 1996 to 2003.28 A total of 27 710 patients with RA provided 162 710 person-years of follow-up. Use of DMARDs without corticosteroids was associated with a small decrease in mild infection risk: adjusted rate ratio=0.90, 95% CI 0.88 to 0.93 relative to no corticosteroid or DMARD use. Use of DMARDs without corticosteroids was not associated with increased serious infection risk: adjusted rate ratio=0.92, 95% CI 0.85 to 1.0. Use of corticosteroids increased the risk of mild and serious infections.
Salliot and van der Heijde have performed a SLR of the long-term safety of MTX in RA.29 They concluded that long-term use of MTX treatment does not appear as a risk factor for infections in general, or for serious infections including herpes zoster.
A role for azathioprine in the development of lymphomas30 ,31 and a role for cyclophosphamide32,–,34 in cancers, particularly bladder cancer, have been suggested. However, no studies have shown that MTX increases the risk of cancer in patients with RA.29 Studies that showed an increased risk of cancer associated with gold or ciclosporin treatment in patients with RA were inconclusive as they used cancer incidence in the general population as the reference.35 ,36 Two studies measured the RR of cancer in a group of ciclosporin-treated patients with RA (46 months and 1.6 years on average) using patients with RA as a control and indicated no enhanced risk.37 ,38
This meta-analysis of synthetic DMARDs in monotherapy or in combination with other synthetic DMARDs indicated that MTX monotherapy is more efficacious in RA on signs and symptoms, disability and structural damage than other non-biological DMARDs. Results were comparable for MTX and leflunomide at the MTX doses assessed. There was also no clear evidence for better efficacy of MTX (at tested doses) versus sulfasalazine or intramuscular gold. It should be noted that MTX was used at relatively lower doses than the optimal doses currently recommended in a recent meta-analysis—namely, starting treatment with MTX 15 mg/week orally, escalating by 5 mg/month to 25–30 mg/week or the highest tolerable dose.4 Ciclosporin, minocycline, tacrolimus and hydroxychloroquine have shown some efficacy on SJCs versus placebo; hydroxychloroquine and minocycline had only marginal effects and auranofin and D-penicillamine showed no significant advantage versus placebo. Few data were available for azathioprine and cyclophosphamide and no data for chlorambucil and mycophenolate. In DMARD naïve patients, a meta-analysis found no significant advantage of the combination of MTX with other non-biological DMARDs over MTX monotherapy, whereas the results were inconclusive in DMARD inadequate responders when taking into account the balance between efficacy and toxicity.23 A combination of MTX with sulfasalazine, ciclosporin or azathioprine showed some beneficial effect versus the same drug given in de novo monotherapy (sulfasalazine, ciclosporin or azathioprine). Data on safety indicated that the risk of infections was increased with cyclophosphamide and azathioprine. Cyclophosphamide and, potentially, azathioprine increased the risk of malignancies in patients with RA.
This systematic literature research with meta-analysis is quite substantial: 97 studies were analysed encompassing the effect of 14 DMARDs used as monotherapy but also in combination with other non-biological DMARDs on several outcomes—namely, on signs and symptoms, on disability and on structural damage. However, there are limitations to our analyses; to interpret the results on so many outcomes of so many different drugs, it was necessary to pool the results on efficacy. Some studies could not be included in this meta-analysis because we needed at least one measure of variability such as SD. Some studies used lower doses of MTX than in current practice. Most of the combinations were studied in only one or two trials. Also, outcome measures were inconsistently reported. Studies including high doses of corticosteroids or biological agents were analysed in different subgroups of the task force.2 ,3 However, the current analysis is based on a large number of studies enabling provision of the evidence for the respective management recommendations.1
Competing interests None. Francis Berenbaum was the handling editor.
Provenance and peer review Not commissioned; externally peer reviewed.
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.