摘要：Summary Background Most data for treatment of dermatomyositis and juvenile dermatomyositis are from anecdotal, non-randomised case series. We aimed to compare, in a randomised trial, the efficacy and safety of prednisone alone with that of prednisone plus either methotrexate or ciclosporin in children with new-onset juvenile dermatomyositis. Methods We did a randomised trial at 54 centres in 22 countries. We enrolled patients aged 18 years or younger with new-onset juvenile dermatomyositis who had received no previous treatment and did not have cutaneous or gastrointestinal ulceration. We randomly allocated 139 patients via a computer-based system to prednisone alone or in combination with either ciclosporin or methotrexate. We did not mask patients or investigators to treatment assignments. Our primary outcomes were the proportion of patients achieving a juvenile dermatomyositis PRINTO 20 level of improvement (20% improvement in three of six core set variables at 6 months), time to clinical remission, and time to treatment failure. We compared the three treatment groups with the Kruskal-Wallis test and Friedman's test, and we analysed survival with Kaplan-Meier curves and the log-rank test. Analysis was by intention to treat. Here, we present results after at least 2 years of treatment (induction and maintenance phases). This trial is registered with ClinicalTrials.gov, number NCT00323960. Findings Between May 31, 2006, and Nov 12, 2010, 47 patients were randomly assigned prednisone alone, 46 were allocated prednisone plus ciclosporin, and 46 were randomised prednisone plus methotrexate. Median duration of follow-up was 35·5 months. At month 6, 24 (51%) of 47 patients assigned prednisone, 32 (70%) of 46 allocated prednisone plus ciclosporin, and 33 (72%) of 46 administered prednisone plus methotrexate achieved a juvenile dermatomyositis PRINTO 20 improvement (p=0·0228). Median time to clinical remission was 41·9 months in patients assigned prednisone plus methotrexate but was not observable in the other two treatment groups (2·45 fold [95% CI 1·2–5·0] increase with prednisone plus methotrexate; p=0·012). Median time to treatment failure was 16·7 months in patients allocated prednisone, 53·3 months in those assigned prednisone plus ciclosporin, but was not observable in patients randomised to prednisone plus methotrexate (1·95 fold [95% CI 1·20–3·15] increase with prednisone; p=0·009). Median time to prednisone discontinuation was 35·8 months with prednisone alone compared with 29·4–29·7 months in the combination groups (p=0·002). A significantly greater proportion of patients assigned prednisone plus ciclosporin had adverse events, affecting the skin and subcutaneous tissues, gastrointestinal system, and general disorders. Infections and infestations were significantly increased in patients assigned prednisone plus ciclosporin and prednisone plus methotrexate. No patients died during the study. Interpretation Combined treatment with prednisone and either ciclosporin or methotrexate was more effective than prednisone alone. The safety profile and steroid-sparing effect favoured the combination of prednisone plus methotrexate. Funding Italian Agency of Drug Evaluation, Istituto Giannina Gaslini (Genoa, Italy), Myositis Association (USA).