Advances in the treatment of rheumatoid arthritis in recent years have profoundly muted the destructive potential of this disease. The safety profile of methotrexate has made earlier treatment possible, and biologic therapies have opened the door to combination rather than sequential therapy, with the result that treatment recommendations now aim for disease remission.1,2 However, the evidence base guiding the treatment that should be given after a patient has an inadequate response to methotrexate monotherapy is weak, despite the armamentarium of efficacious agents.

O’Dell et al. now report in the Journal the results of their randomized clinical trial evaluating the safety and efficacy of adding sulfasalazine and hydroxychloroquine to methotrexate (triple therapy) as compared with adding etanercept (a tumor-necrosis-factor [TNF] receptor) to methotrexate in patients with rheumatoid arthritis who have active disease.3 If the Disease Activity Score for 28-joint counts (DAS28, with scores ranging from 2 to 10 and higher scores indicating greater disease activity) had not decreased by at least 1.2 units at 24 weeks, patients were switched in a blinded fashion to the alternative regimen. The primary outcome — the difference between the two groups in the change in the DAS28 at 48 weeks — showed the noninferiority of triple therapy to etanercept–methotrexate. The progression of radiographic joint damage (as assessed with the use of the van der Heijde modification of the Sharp score, which ranges from 0 to 380, with higher scores indicating more extensive disease) over 48 weeks also did not differ significantly between the groups (an increase of 0.54 Sharp score units in the triple-therapy group and 0.29 Sharp score units in the etanercept–methotrexate group, P=0.43). These results are certainly encouraging, but several caveats must be considered.

The noninferiority margin chosen by O’Dell et al. (0.6) is half the minimum clinically meaningful improvement of 1.2 in the DAS28, but justification that this margin is acceptable to clinicians and patients is not provided. This may be a moot concern, since improvement with triple therapy as compared with improvement with etanercept–methotrexate was much closer than 0.6. Furthermore, Table S2 in the Supplementary Appendix of the article (available with the full text of the article at NEJM.org) shows each component of the DAS28 improving similarly in the two regimens, suggesting equivalent efficacy across the DAS28 domains. Figure S2 in the Supplementary Appendix of the article indicates that the proportions of patients who continued with the originally assigned regimen and of patients who switched to the alternative regimen were similar in the two groups, as were the proportions of patients who withdrew or were lost to follow-up and of patients who completed the study — further reducing concerns that missing data or poor implementation biased the results toward a finding of noninferiority. Since a noninferiority margin was not specified for the secondary outcomes, the reader cannot determine whether P values for these outcomes, such as radiographic progression, identify triple therapy as noninferior to etanercept–methotrexate.

The study by O’Dell et al. is not the first to compare these treatment regimens. In the Treatment of Early Aggressive Rheumatoid Arthritis (TEAR) study, patients were randomly assigned to methotrexate monotherapy, triple therapy, or methotrexate plus etanercept.4 No significant difference in the primary outcome (the change in DAS28 scores from week 48 to week 102) was observed between the two combination-treatment groups — similar to the results in the study by O’Dell et al.; however, the Sharp scores for radiographic progression increased more in the triple-therapy group than in the methotrexate–etanercept group (1.69 vs. 0.64, P=0.046). In the Swedish Pharmacotherapy (Swefot) study, in which patients who had not had an adequate response to methotrexate were randomly assigned to triple therapy or to infliximab (an anti-TNF antibody) plus methotrexate, the superiority of infliximab–methotrexate with respect to a good response according to European League against Rheumatism (EULAR) criteria was observed at 12 months but not at 18 or 24 months.5,6 The Sharp scores for radiographic progression increased more in the triple-therapy group than in the infliximab–methotrexate group (7.23 vs. 4.00, P=0.009) — a finding consistent with that in the TEAR study.

There are considerable methodologic differences across the studies. The study by O’Dell et al. and the TEAR study were double-blind studies that included observed data without imputation of missing data and used etanercept as the biologic therapy, whereas the Swefot study was not a blinded study, included persons who, for any reason, did not have a measurable response, and used infliximab as the alternative therapy to triple therapy. The Swefot and TEAR studies targeted early disease, whereas O’Dell et al. targeted established disease. The inclusion and exclusion criteria also varied among the studies. Even so, the results were remarkably consistent. The efficacy rates at the extended time points, as well as the frequencies of total and serious adverse events, were relatively similar with the two treatment regimens in all the studies. The TNF inhibitor–methotrexate regimen showed modest superiority over triple therapy in slowing radiographic progression, although the trend in the study by O’Dell et al. was not significant.

We have to consider, however, whether these findings have arrived too late to influence modern practice, in which arguably a TNF inhibitor is the preferred next step when methotrexate alone is inadequate.7,8 Whether third-party payers who currently require failure of methotrexate monotherapy before prescription of expensive biologic therapy will change this policy to require failure of the cheaper nonbiologic combination is an interesting question. The development of more affordable biosimilar agents may change the treatment choices yet again,9 potentially rendering the studies with nonbiologic agents cited above irrelevant. We hope that with the ever-increasing number of effective treatments for rheumatoid arthritis, future recommendations for treatment will be guided by additional comparative-effectiveness studies such as the study by O’Dell et al. In addition, future identification of biomarkers to identify the patients who are most likely to have a response to, or are least likely to have side effects with, specific therapies will be the next great leap in the treatment of rheumatoid arthritis.

Source: Editorial, NEJM