Jun Won Park

Objectives To develop a model for predicting flares after tapering the dose of tumour necrosis factor inhibitors (TNFi) in patients with axial spondyloarthritis (axSpA). Methods Data were obtained from the Korean College of Rheumatology Biologics and Targeted Therapy Registry. In total, 526 patients who received the standard-dose TNFi for at least 1 year and tapered their dose were included in the derivation cohort. The main outcome was a flare occurrence defined as an Ankylosing Spondylitis Disease Activity Score with C-reactive protein (ASDAS-CRP) score of ≥2.1 after 1 year of TNFi tapering. The final prediction model was validated using an independent cohort. Results Among 526 patients, 127 (24.1%) experienced flares. The final prediction model included negative human leucocyte antigen B27 (β = 1.088), inflammatory back pain (β = 1.072), psoriasis (β = 1.567), family history of SpA (β = 0.623), diabetes mellitus (β = 1.092), TNFi tapering by ≥50% of the standard-dose (β = 0.435), ASDAS-CRP at tapering (β = 1.029), and Bath Ankylosing Spondylitis Functional Index score at tapering (β = 0.194) as covariates. It showed an excellent discrimination performance (AUC = 0.828). According to the predictive risk, patients were classified into three groups (low-, intermediate- and high-risk). The probabilities of flares in these groups were 4.5%, 18.1% and 61.8%, respectively. The performance of the model in the validation cohort was also comparable. Conclusion The established prediction model accurately predicted the risk of flares after TNFi dose tapering in patients with axSpA using eight simple clinical parameters, which could be helpful to select appropriate patients for tapering their TNFi without flare in daily clinical practice.