Methods
Methotrexate and leflunomide nave polyarticular JRA patients 3 to 17 years of age were randomized to receive leflunomide 20mg per day (after a weight based loading regimen) or weekly oral methotrexate (dosed by weight). After randomization, patients and treating physicians were blinded to treatment status for the first 16 weeks of treatment. Thereafter, patients were offered continuation in an addition 32 weeks of blinded treatment (48 weeks total).
The two primary efficacy endpoints were the ACR Pedi 30 response (roughly equivalent to an adult RA ACR 30 response) and the Percent Improvement Index (a composite index of the core ACR data set of disease activity measures for JRA) measured at 16 weeks. Secondary outcomes included the ACR Pedi 50 and 70 responses, change in disability (measured with the Childhood Health Assessment Questionnaire (CHAQ)), and change in serum CRP. Adverse events were monitored throughout the study.
Results
94 subjects (69 female, 25 male) were evenly randomized to receive leflunomide (LEF) or methotrexate (MTX). Mean age at enrollment was 10 years. The median disease duration was 4 months. Disease activity measures were generally elevated (mean active joint count of 14 joints, ESR of 30 to 34 mm/hr, and CRP of 14 to 20 mg/L) and were balanced between the two treatment groups. 86 subjects completed the initial 16 weeks of the study. 70 of these entered the blinded extension phase, with 55 of these completing the entire 48 weeks of the study.
| Initial 16 weeks (n=86) | 32 week extension (n=70) | |||||
| MTX | LEF | p | MTX | LEF | p | |
| ACR Pedi 30 | 89% | 68% | 0.02 | 91% | 79% | – |
| ACR Pedi 50 | 77% | 60% | 0.10 | 86% | 76% | – |
| ACR Pedi 70 | 60% | 43% | 0.14 | 83% | 70% | – |
| Percent Improvement Index (mean) | -52.87 +4.4% |
-44.41 +4.5% |
0.18 | -65.5% | -55.4% | – |
| Change in CHAQ | -0.39 +0.1% |
-0.44 +0.08% |
0.61 | no change | no change | – |
| Change in CRP (mg/L) | -11.4 | -3.9 | 0.04 | no change | no change | — |
Treatment-related adverse events were similar between the two treatment groups. During the initial 16 weeks of treatment, liver function abnormalities (greater than 1.2 times the upper limit of normal) were reported in 7 (15%) of the leflunomide treated patients and 15 (32%) of the methotrexate treated group. Two of these patients withdrew from the study due to liver function abnormalities. Two other serious treatment-related adverse events occurred in leflunomide treated patients: a suspected case of salmonellosis and a case of lichenoid pityriasis.
Conclusions
Both treatment with methotrexate and leflunomide was associated with rapid improvement in children with active polyarticular JRA. However, methotrexate was associated with significantly greater numbers of patients with clinically meaningful responses. The favorable effects of both medications were generally sustained to 48 weeks.
Editorial Comment
Randomized clinical trials in JRA are rare, and this study fills a much needed gap. In particular, this is the first head-to-head study comparing two viable treatments in JRA. Presumably, as the authors comprise the Leflunomide in Juvenile Rheumatoid Arthritis Investigator Group, that the a priori hypothesis was that leflunomide would be more effective than methotrexate. Regardless, both medications were remarkably well tolerated and effective.
Two design issues are worth noting that may have affected the outcomes. First, the small sample sizes (particularly in the 32 week extension phase) are the principle reason that many of the differences in outcomes between the two treatment groups did not attain statistical significance, and should not be misread as implying equivalence between the two therapies. Second, without a placebo comparison group and many patients with a very short duration of disease (median of 4 months), it is possible that the apparent rapid and dramatic responses to treatment actually represent spontaneous improvements of disease. Needless to say, a non-treatment placebo arm of a pediatric study is not an ethically sound option, and thus one must factor in that spontaneous improvements in some patients may overestimate the efficacy of the treatments. However, one cannot be sure that the numbers of patients with spontaneous improvements were equally divided between the two treatment groups.
