Predictors of b/tsDMARDs discontinuation and treatment response in psoriatic arthritis: results from the British Society for Rheumatology Psoriatic Arthritis Register (BSR-PsA)
Ovidiu Rotariu, Stephanie Lembke, Philip S Helliwell, Stefan Siebert, Gary J Macfarlane, Gareth T JonesAbstract
Objectives
Despite advances in biologic and targeted synthetic DMARDs (b/tsDMARDs), many patients with psoriatic arthritis (PsA) experience suboptimal responses and discontinue therapy. We aimed, in a real-world registry, to identify predictors of b/tsDMARD discontinuation and treatment response.
Methods
Patients commencing b/tsDMARD therapy were identified from the British Society for Rheumatology Psoriatic Arthritis Register (BSR‑PsA). Clinical and patient‑reported measures were collected at treatment start. Cox regression identified predictors of drug discontinuation using multiple imputation with predictive mean matching to handle missing data, and logistic regression identified predictors of achieving 12-month Minimal Disease Activity (MDA), ACR50 and PsARC response.
Results
Eight-hundred and fifteen participants were included; 40% male, median age 51 years, with high disease activity (median DAPSA 33.5; PASDAS 5.9). At censoring (01‑Aug‑2025), 65% remained on their index therapy (median exposure 31 months) with discontinuation primarily due to inefficacy (55%) or intolerance (26%). Patients commencing PDE4 inhibition experienced more than a doubling in the risk of discontinuation, compared to those starting TNF inhibitors (Hazard Ratio: 2.25; 95%CI: 1.45-3.50), while no major differences were observed with other modes of action. At 12-months, 24% achieved MDA, 24% ACR50, and 49% PsARC. Based on imputed outcomes, men, younger individuals, and those who were employed were consistently more likely to achieve treatment response, while poorer baseline clinical or patient‑reported measures predicted non‑response.
Conclusions
Treatment persistence and response in PsA remain a problem and are strongly influenced by demographic factors and baseline disease severity. A better understanding of predictors of non-response may aid future risk stratification.