Efficacy and Safety of Cyclophosphamide Treatment in Severe Juvenile Dermatomyositis Shown by Marginal Structural Modeling

Abstract

Objective: In patients with severe or refractory juvenile dermatomyositis (DM), second-line treatments may be required. Cyclophosphamide (CYC) is used to treat some connective tissue diseases, but evidence of its efficacy in juvenile DM is limited. This study was undertaken to describe clinical improvement in juvenile DM patients treated with CYC and model the efficacy of CYC treatment compared to no CYC treatment.

Methods: Clinical data on skin, global, and muscle disease for patients recruited to the Juvenile DM Cohort and Biomarker Study were analyzed. Clinical improvement following CYC treatment was described using unadjusted analysis. Marginal structural models (MSMs) were used to model treatment efficacy and adjust for confounding by indication.

Results: Compared to the start of CYC treatment, there were reductions at 6, 12, and 24 months in skin disease (P = 1.3 × 10^-10 ), global disease (P = 2.4 × 10^-8 ), and muscle disease (P = 8.0 × 10^-10 ) for 56 patients treated with CYC in unadjusted analysis. Limited evidence suggested a reduction in glucocorticoid dose (P = 0.047) in patients treated with CYC. MSM analysis showed reduced global disease and skin disease in patients who started an ~6-month course of CYC treatment >12 months ago compared to patients never or not yet treated with CYC. In the treated patients, the modified skin Disease Activity Score for juvenile DM was 1.19 units lower (P = 0.0085) and the physician's global assessment was 0.66 units lower (P = 0.027). Minor adverse events were reported in 3 patients within 1 year of stopping CYC.

Conclusion: Our findings indicate that CYC is efficacious with no short-term side effects. Improvements in skin, global, and muscle disease were observed. Further studies are required to evaluate longer-term side effects.

Figure 1

Improvement in A, modified Disease Activity Score (DAS) for juvenile dermatomyositis, B, physician's global assessment (PGA), C, Childhood Myositis Assessment Scale (CMAS), and D, glucocorticoid dose within 24 months of starting cyclophosphamide (CYC) treatment. Friedman's test for nonparametric repeated‐measures analysis of variance showed that overall, disease activity improved over the time points analyzed, with improvements in modified DAS for juvenile DM (χ²[3] = 49.0, P = 1.3 × 10⁻¹⁰), PGA (χ²[3] = 38.3, P = 2.4 × 10^‐8), CMAS (χ²[3] = 45.3, P = 8.0 × 10⁻¹⁰), and glucocorticoid dose (χ²[3] = 7.9, P = 0.047). Percentages of missing data at 0, 6, 12, and 24 months, respectively, were as follows: 0%, 0%, 0%, and 0% for modified DAS for juvenile DM; 17.9%, 21.4%, 30.4%, and 28.6% for PGA; 28.6%, 25.0%, 23.2%, and 28.6% for CMAS; and 78.6%, 57.1%, 41.1%, and 44.6% for glucocorticoid dose. Circles represent individual patients; horizontal lines and boxes show the median (interquartile range). Bars above and below the boxes show the range.

Figure 2

Longitudinal marginal structural model (MSM) analysis of cyclophosphamide (CYC) efficacy for improvement in juvenile dermatomyositis (DM). Forest plots depict estimated regression coefficients (Coef) from final analytical models where modified Disease Activity Score (DAS) for juvenile DM (A), physician's global assessment (PGA) (B), and Childhood Myositis Assessment Scale (CMAS) (C) were modeled as outcomes using data from 56 patients treated with CYC and 144 patients not treated with CYC. These final analytical models were weighted using inverse propensity score values generated by an MSM, in order to balance confounding differences between patients who were and those who were not treated with CYC. Asterisks indicate significant P values. 95% CI = 95% confidence interval.