Difficult-to-treat psoriatic arthritis: refining the definition using a statistical model in a real-life cohort.

Objectives: The study aims to evaluate the applicability of the D2T psoriatic arthritis (PsA) definition, adapted from rheumatoid arthritis, within a single-center observational cohort of PsA patients treated with b/tsDMARDs. In addition, we aimed to establish a numerical index defining D2T-PsA based on the ratio of observed to expected failed b/tsDMARDs and to develop a predictive model identifying features associated with the D2T condition. Methods: The study included 267 consecutive adult PsA patients receiving b/tsDMARDs, collecting demographic, clinical, and clinimetric data. The prevalence of D2T PsA patients was assessed using a proposed definition. We then developed a predictive model to assess treatment difficulty, utilizing PsA-normalized failed b/tsDMARDs. A generalized linear model was applied to identify clinical and demographic features associated with D2T PsA, employing a bagging procedure for robust variable selection, followed by univariate and multivariable analyses. Results: Among the 267 patients, only 8 of them (2.9%) met the proposed D2T PsA criteria. In a subset of 177 patients analyzed using the predictive model, 17.2% of them demonstrated higher treatment difficulty. Univariate analysis revealed associations between treatment difficulty and female sex, psoriasis pattern, fibromyalgia, and steroid therapy. Multivariate analysis confirmed significant associations between fibromyalgia, nail and pustular psoriasis, and steroid use. Conclusion: According to the predictive model, the proposed D2T PsA definition identified a small subset of patients with increased treatment difficulty. These findings highlight the need for refining the criteria to better define D2T PsA patients, providing valuable insights into managing complex treatment challenges in PsA.