SAT0506 MULTIPLE PATTERNS OF ‘RESPONSE’ TO METHOTHREXATE IDENTIFIED IN A NATIONAL JUVENILE IDIOPATHIC ARTHRITIS COHORT

Abstract

Background:Disease activity following treatment for JIA is currently understood in terms of ‘response’ or ‘non-response’. This state is usually defined using composite measures such as the ACR Pedi scores or cut-offs on the Juvenile Arthritis Disease Activity Scores (JADAS). However, response is a complex state and it is likely that separate, identifiable clusters of children and young people (CYP) have different, varying levels of response across the individual measures of JIA disease activity. Identifying these clusters may facilitate stratified medicine in JIA.Objectives:To identify clusters of CYP with distinct patterns of change across the individual JADAS components following MTX initiation for JIA.Methods:MTX-naïve CYP enrolled into the MTX cohorts of the British Society for Paediatric and Adolescent Rheumatology Etanercept Cohort Study or the UK Biologics for Children with Rheumatic Diseases registers before January 2018 were selected. JADAS components (active joint count to 71, physician global assessment (0-100mm), parent global evaluation (0-100mm) and ESR (mm/hr)) were collected at MTX start and at (approximately) 6- and 12-month follow-ups. Outcomes were Log1p transformed for analysis and all outcome data were censored following start of a biologic. CYP were excluded if they had clinically inactive disease at MTX initiation, initiated a biologic within a month of MTX or had no available JADAS data at any time point.Multivariate group-based trajectory models explored MTX response clusters over the first year following MTX initiation using censored-normal models. Linear, quadratic and cubic polynomials were tested, with one to ten trajectories tested within each polynomial group. Optimal models within each polynomial group were selected using Bayesian Information Criteria, after excluding those with groups representing <1% of the cohort, average posterior probability for assigned group <70% or relative entropy <0.5.Results:Of 657 CYP, the majority were female (69%) and of white ethnicity (85%), with RF-negative polyarticular JIA the most common disease category (33%).The optimal model identified multiple patterns of disease activity following MTX initiation, with greater complexity than the traditional ‘response’ or ‘non-response’ paradigm. Although there were no substantial differences in ESR trajectories between the groups, there were differences in initial disease severity and speeds of improvement across active joint counts, physician and parental global assessments over time. In addition, individual JADAS components did not always change in parallel over time, even within the same cluster of CYP.Conclusion:There are multiple patterns of disease activity following MTX initiation for CYP with JIA. This suggests that a simple response/non-response analysis at a single time point may be inadequate. Understanding clinical or biological factors associated with these clusters could facilitate stratified medicine in JIA.Acknowledgments:The CLUSTER consortium is supported by contributions of its industry partners, currently Pfizer, AbbVie UCB, GSK, and SobiDisclosure of Interests:Stephanie Shoop-Worrall: None declared, Kimme Hyrich Grant/research support from: Pfizer, UCB, BMS, Speakers bureau: Abbvie, Lucy Wedderburn Speakers bureau: Pfizer, Wendy Thomson: None declared, Nophar Geifman: None declared