SPEED trial post hoc analysis: Predictors of 1-year PASDAS response in early PsA

A post hoc analysis of the multicenter, open-label SPEED trial (NCT03739853), investigated predictors of 1-year Psoriatic Arthritis Disease Activity Score (PASDAS) response using machine learning in 140 patients with early psoriatic arthritis (PsA) and poor prognostic factors, treated with standard step-up conventional synthetic disease-modifying antirheumatic drug (csDMARD) therapy, combination csDMARD therapy, or early tumor necrosis factor inhibitor (TNFi) induction. Results were presented by Alexandre Garaïman during the European Alliance of Associations for Rheumatology (EULAR) 2026 Congress. 

Key data: Recursive partitioning (RPART) modeling identified body mass index (BMI) as the most influential baseline predictor of lower PASDAS at Week 48 (predictor importance, 32.7%), ranking above treatment type (sixth; predictor importance, 4.7%). Patients with BMI <25 kg/m² and lower baseline disease activity (PASDAS <5.4) achieved the most favorable outcomes (predicted Week 48 PASDAS, 2.1) independent of treatment. Among patients with baseline PASDAS ≥5.4 and BMI <27 kg/m², early TNFi therapy was associated with better outcomes vs csDMARD therapy (predicted Week 48 PASDAS, 2.6 vs 4.0). BMI ≥27 kg/m², disease duration ≥12 months, and polyarticular disease were factors associated with the poorest outcomes (predicted Week 48 PASDAS up to 6.0). 

Key learning: BMI was a stronger predictor of 1-year PASDAS response than treatment type in patients with early PsA, underscoring weight management as a key modifiable component of PsA care.