Document resource
Objectives Anti-nuclear antibody–associated rheumatic diseases (ANA-RMDs) are clinically and immunologically heterogeneous. Within single diseases, interferon (IFN) and related gene expression scores have been linked to differential disease activity, yet their prognostic value across diagnoses remains uncertain. We aimed to determine whether IFN and associated molecular signatures predict clinical outcomes across multiple ANA-RMDs.Methods Adults with ANA-RMDs enrolled in the UK DEFINITION Cohort C study were reviewed at baseline, 3 and 6 months. Clinical outcomes included disease activity (BILAG-2004, ESSDAI, MITAX, and physician global assessment), treatment escalation, and healthcare utilisation. Whole-blood TaqMan assays generated IFN Scores A and B and modules for plasmablast, myeloid, inflammatory, and erythropoiesis pathways. Predictive performance was evaluated using multivariable logistic regression and causal-inference frameworks guided by directed acyclic graphs.Results Among 219 participants (SLE, pSS, SSc, IIM, MCTD, UCTD), IFN Scores A and B were the strongest baseline predictors of longitudinal disease activity and therapy escalation. IFN-high patients experienced higher odds of flare and greater healthcare utilisation. Models integrating IFN and plasmablast modules with clinical covariates demonstrated superior discrimination for flare prediction compared with diagnosis-based models with an AUROC of 0.84 at 3 months and 0.79 at 6 months follow up. Causal analysis suggested that IFN exerts both direct effects on flare risk and indirect effects through downstream immunological pathways, although precision was limited by sample size and follow-up duration.Abstract PT5:02 Figure 1Conclusions Gene-expression signatures, particularly interferon and plasmablast modules, outperform legacy diagnoses in predicting outcomes across the ANA-RMD spectrum. These findings reinforce a shift toward molecularly defined disease axes, supporting diagnosis-agnostic stratification in clinical trials and personalised therapeutic approaches. Integration of molecular classifiers with clinical features could enable earlier identification of patients at risk of flare and guide precision treatment strategies across connective-tissue diseases.