Real-world performance of EULAR and GRAPPA definitions of difficult-to-treat psoriatic arthritis within the ORCHESTRA cohort: an observational cohort study.

Background: Psoriatic arthritis (PsA) is a complex inflammatory disease with significant musculoskeletal, dermatological, and comorbidity burden. Despite therapeutic advances, many patients remain difficult to treat (D2T). Conventional definitions often overlook distinctions between inflammatory and non-inflammatory drivers, risking inappropriate treatment escalation. Two consensus frameworks-the European Alliance of Associations for Rheumatology (EULAR) and the Group for Research and Assessment of Psoriasis and Psoriatic Arthritis (GRAPPA)-have recently been proposed to better capture this complexity based on disease drivers. Objectives: To evaluate the applicability, overlap, and clinical utility of EULAR and GRAPPA definitions for difficult/complex-to-manage (D2M/C2M) and treatment-refractory PsA (TR-PsA) in a real-world cohort. Methods: We conducted a retrospective analysis within the prospective Ottawa Rheumatology CompreHEnSive TReatment and Assessment (ORCHESTRA) cohort, including consecutive PsA patients initiating or switching advanced therapy between April 2022 and September 2025. Patients were classified according to EULAR and GRAPPA frameworks into the broader D2M-PsA (EULAR) and C2M-PsA (GRAPPA) groups, and narrower inflammation-restricted TR-PsA subsets. Prevalence, overlap, and agreement were assessed using descriptive statistics and kappa analysis. Baseline clinical, comorbidity, and ultrasound characteristics were compared across groups. Results: Seventy-six PsA patients were included. GRAPPA identified more patients as C2M-PsA (35/76; 46%) than EULAR's D2M-PsA (16/76; 21%). All EULAR D2M-PsA patients were included in C2M-PsA, whereas 19 additional patients were uniquely identified by GRAPPA due to less stringent treatment failure thresholds. Agreement between EULAR D2M-PsA and GRAPPA C2M-PsA was moderate (κ = 0.48). TR-PsA classification showed perfect agreement (κ = 1.000), with identical patients identified by both frameworks. D2M/C2M patients had longer disease duration and greater prior therapy exposure than non-D2T patients, while objective inflammatory measures were largely comparable. Conclusion: EULAR and GRAPPA D2T PsA frameworks share conceptual intent but differ in real-world clinical capture. EULAR preferentially enriches for multidomain involvement and inflammatory refractoriness after multiple therapy failures, whereas GRAPPA identifies a broader population reflecting earlier disease complexity. While both frameworks identified the same patients using inflammation-restricted criteria, differences in operationalization limit assumptions of interchangeability.