Challenges in linking interferon signatures to clinical trajectories in juvenile idiopathic arthritis

Abstract Background Based on the recent finding that higher baseline expression of interferon (IFN)-driven gene signature in blood positively correlates with good response to methotrexate (MTX) in Juvenile Idiopathic Arthritis (JIA), this study tested the hypothesis that baseline IFN scores differ significantly between trajectory groups of response to MTX, defined by machine learning models enabling clinicians to predict not just the likelihood, but the specific longitudinal course of therapeutic response. Findings Longitudinal clinical data from 699 children and young people (CYP) with active JIA initiating MTX and a nested sub-cohort of 102 CYP with previously available baseline blood transcriptomic data were analysed. Clinical disease courses were classified using multivariate group-based trajectory modelling, while baseline interferon activity was quantified using established 51-gene and 5-gene expression signatures derived from bulk RNASeq. There was no association between pre-treatment 51-gene and 5-gene IFN response scores and distinct clinical disease trajectories following methotrexate initiation by regression modelling. Post-hoc calculations indicated that the study was underpowered (44–58%) to detect medium effects, demonstrating that the increased clinical resolution provided by trajectory analysis significantly compromises statistical power in size-limited paediatric cohorts. Conclusions Clinical trajectories accurately describe patient heterogeneity, but the IFN score does not represent a marker for a specific trajectory group?. Future stratification strategies require larger multi-centre cohorts to identify non-responders early and exploit the window of biological opportunity.