OA41 Joint involvement patterns across stages of rheumatoid arthritis: a multiple correspondence analysis of early, non-difficult-to-treat, and difficult-to-treat

Background/Aims Joint involvement patterns in RA are heterogeneous. However, little is known about how these may differ between early RA (ERA), non-difficult-to-treat RA (nonD2T-RA), and D2T-RA and how clinical covariates may influence them. We aim to identify and compare patterns of joint involvement across ERA, nonD2T-RA, and D2T-RA using multiple correspondence analysis (MCA), and to assess the influence of clinical characteristics on these patterns. Methods Patients with RA from a single centre were stratified into three groups: (1) ERA: new diagnosis with symptom duration <12 months; (2) nonD2T-RA: treated with conventional synthetic DMARDs (csDMARDs) and/or up to a second targeted therapy; (3) D2T-RA: failed ≥2 classes of targeted therapies. Both nonD2T-RA and D2T have minimum disease duration of 5 years. All patients had active disease, defined as DAS28-ESR >3.2. Swollen and tender joint counts (28-joint assessment) were analysed using MCA to identify clustering structures within and across groups. Regression models were then applied to assess the influence of sex, seropositivity, erosions, osteoarthritis, and smoking status on clustering. All analyses were performed using R (version 4.3.0). Results 298 patients were included: 113 ERA, 114 nonD2T-RA and 71 D2T-RA. The study population was predominantly female [69.9% (79/113) ERA group, 71.1% (81/114) nonD2T-RA group and 83.1% (59/71) D2T-RA group]. Median age [52 vs 57.34 vs 54.93 years] and DAS28-ESR scores [5.77 vs 5.56 vs 5.55] were comparable across the groups. Evaluation for joint symmetry revealed strong bilateral joint correlations in ERA for tenderness and swelling across all 13 paired joints, apart from shoulder swelling. NonD2T-RA also demonstrated symmetrical correlations for tenderness and for most swollen joints, except PIP5. In contrast, D2T-RA patients exhibited weaker symmetry: no significant bilateral correlation with swelling of the shoulders, MCP5, and PIP5, as well as tenderness at MCP3. In all RA groups, three clusters were identified: ERA: (1) the elbow, (2) all PIP joints together with MCP1, (3) a mixed cluster combining large, medium, and small joints; nonD2T-RA: (1) the shoulder and elbow, (2) PIP5 and MCP5, (3) the remaining hand joints grouped with the knees; D2T-RA: (1) large joints including the shoulders, elbows, and knees (2), PIP1 and PIP5 distinct cluster (3), all other joints grouped together. Multivariable analysis showed erosions and osteoarthritis were significantly associated with swelling in the elbow, knee, PIP1 and PIP3 (all p < 0.001). Osteoarthritis was also associated with involvement of the elbow, MCP1, MCP3, MCP4, PIP3, PIP4, and PIP5 (all p < 0.001). Conclusion Joint clustering in RA evolves from symmetrical small-joint patterns in ERA to asymmetrical and heterogeneous clusters in D2T-RA - distinct mechanisms may underlie these. Erosions and osteoarthritis significantly influence joint involvement distribution; incorporating these into the interpretation of joint patterns may enhance clinical assessment and help identify patients at risk of developing difficult-to-treat disease.