Frequency of complex-to-manage, difficult-to-manage and treatment-refractory psoriatic arthritis: applying the 2025 GRAPPA and EULAR definitions in a real-world longitudinal PsA cohort Authors: Gonul Hazal Koc, Marc R Kok, Selinde V J Snoeck Henkemans, Jolanda J Luime, Ilja Tchetverikov, Lindy A Korswagen, Natasja H A M Denissen, Yvonne P M Goekoop-Ruiterman, Paul Baudoin, Petra Kok, Reinhard Bos, Radboud J E M Dolhain, Fabian Proft, Marijn Vis, CICERO Published: 2026 ABSTRACT AbstractObjectivesTo determine the frequency and baseline characteristic differences of complex-to-manage (C2M), difficult-to-manage (D2M) and treatment-refractory (TR) psoriatic arthritis (PsA) in a real-world longitudinal cohort, using the recently proposed 2025 Group for Research and Assessment of Psoriasis and Psoriatic Arthritis (GRAPPA) and the European Alliance of Associations for Rheumatology (EULAR) definitions.MethodsWe analysed data from newly diagnosed, disease-modifying antirheumatic drug-naïve patients with PsA enrolled in the Dutch Southwest Early Psoriatic Arthritis cohort between 2013 and 2023. Patients were classified using the 2025 GRAPPA (C2M, TR) and EULAR (D2M, TR) PsA definitions. Frequencies were calculated for each definition. Time-to-event analyses using Kaplan-Meier methods assessed the timing of definition fulfilment and individual components. Longitudinal differences in patient-reported outcomes (PROs) were assessed using linear mixed-effects models.ResultsAmong 885 newly diagnosed patients with PsA, 20% (179/885) fulfilled the GRAPPA-C2M definition, 4.2% (37/885) fulfilled the EULAR-D2M definition, 3.8% (34/885) met the EULAR-TR and 1.2% (11/885) fulfilled the GRAPPA-TR criteria definition during follow-up. Female sex, higher body mass index (BMI), longer symptom duration, presence of enthesitis and greater skin involvement differed with meeting definitions at baseline. Patients classified as GRAPPA-C2M, GRAPPA-TR, EULAR-D2M and EULAR-TR consistently reported worse long-term PROs (Health Assessment Questionnaire-Disability Index scores in C2M (β=0.17, 95% CI 0.10 to 0.24), GRAPPA-TR (β=0.37, 95% CI 0.22 to 0.51), EULAR-D2M (β=0.36, 95% CI 0.20 to 0.48) and EULAR-TR (β=0.40, 95% CI 0.25 to 0.55)).ConclusionsApproximately 20% of patients with early PsA fulfilled the GRAPPA-C2M criteria, representing the earliest and broadest subgroup, while 4% met EULAR-D2M/EULAR-TR and 1.2% GRAPPA-TR definitions. These latter subgroups were more likely to be female, had higher BMI, greater skin involvement and enthesitis and experienced persistently worse PROs, underscoring the need for tailored management strategies.Keywords: Psoriatic Arthritis; Treatment; Outcome Assessment, Health Care; Patient Reported Outcome Measures WHAT IS ALREADY KNOWN ON THIS TOPIC A subset of patients with psoriatic arthritis (PsA) have persistent disease despite multiple therapies. A subset of patients with psoriatic arthritis (PsA) have persistent disease despite multiple therapies. The Group for Research and Assessment of Psoriasis and Psoriatic Arthritis (GRAPPA) and the European Alliance of Associations for Rheumatology (EULAR) recently proposed definitions for complex-to-manage (C2M), difficult-to-manage (D2M) and treatment-refractory (TR) PsA, but longitudinal real-world data are lacking. The Group for Research and Assessment of Psoriasis and Psoriatic Arthritis (GRAPPA) and the European Alliance of Associations for Rheumatology (EULAR) recently proposed definitions for complex-to-manage (C2M), difficult-to-manage (D2M) and treatment-refractory (TR) PsA, but longitudinal real-world data are lacking. WHAT THIS STUDY ADDS This is the first longitudinal real-world application of the 2025 GRAPPA and EULAR definitions in PsA, using a large cohort. This is the first longitudinal real-world application of the 2025 GRAPPA and EULAR definitions in PsA, using a large cohort. GRAPPA-C2M identified a larger and earlier subgroup, whereas EULAR-D2M and EULAR-TR and GRAPPA-TR captured smaller, later-emerging populations with greater disease burden. GRAPPA-C2M identified a larger and earlier subgroup, whereas EULAR-D2M and EULAR-TR and GRAPPA-TR captured smaller, later-emerging populations with greater disease burden. At baseline, definitions descriptively included higher proportions of female patients, individuals with higher body mass index and patients with greater skin involvement and enthesitis. At baseline, definitions descriptively included higher proportions of female patients, individuals with higher body mass index and patients with greater skin involvement and enthesitis. Minimal Disease Activity best captured these definitions as a composite disease outcome. Minimal Disease Activity best captured these definitions as a composite disease outcome. Over time, these subgroups experienced persistently worse patient-reported outcomes, indicating a sustained disease burden. Over time, these subgroups experienced persistently worse patient-reported outcomes, indicating a sustained disease burden. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY These definitions provide a graded classification of disease states in PsA, with increasing stringency in temporal and clinical profile criteria. These definitions provide a graded classification of disease states in PsA, with increasing stringency in temporal and clinical profile criteria. Application of these definitions in routine care is currently limited by their complexity and the requirement for detailed, concurrently available clinical, patient-reported and treatment data, reflecting the heterogeneous and multidomain nature of PsA. Application of these definitions in routine care is currently limited by their complexity and the requirement for detailed, concurrently available clinical, patient-reported and treatment data, reflecting the heterogeneous and multidomain nature of PsA. Introduction Psoriatic arthritis (PsA) is a chronic inflammatory disease with a highly heterogeneous phenotype, including musculoskeletal, skin and systemic manifestations, often accompanied by comorbidities that complicate its clinical management.1 Despite the development of multiple biologic and targeted synthetic disease-modifying antirheumatic drugs (b/tsDMARDs), a considerable proportion of patients fail to achieve sustained low disease activity or remission across all disease domains.2 3 In real-world cohorts, treatment persistence remains suboptimal, and patients often cycle through multiple therapies with diminishing clinical response rates.4 5 These therapeutic challenges have led to growing interest in a distinct subset of patients referred to as ‘difficult-to-treat’ (D2T) PsA.6 7 While the concept has long been established in rheumatoid arthritis, only recently have formal definitions been proposed for axial spondyloarthritis. To address this unmet need in PsA, both the Group for Research and Assessment of Psoriasis and Psoriatic Arthritis (GRAPPA) and the European Alliance of Associations for Rheumatology (EULAR) independently introduced definitions for patients with persistent disease burden despite advanced therapies.8 9 In current literature, emerging studies using proxy definitions suggest that 16%–34% of patients with PsA may meet criteria for D2T disease, typically defined by failure of ≥1 conventional synthetic (cs)DMARD and ≥2 b/tsDMARDs with ongoing active symptoms.10,12 These patients are more often female, have higher body mass index (BMI) or body surface area (BSA) and frequently present with comorbidities such as fibromyalgia, depression or axial disease. However, these estimates are based on retrospective or registry-based data and rely on non-standardised definitions. The availability of formal GRAPPA and EULAR D2T definitions now enables a more systematic and standardised assessment of this complex patient subgroup. However, these definitions have not yet been applied in longitudinal real-world PsA cohorts, and key questions identified by the EULAR task force remain unanswered, including the prevalence of difficult-to-manage (D2M) and treatment-refractory (TR) PsA, the outcome measures that best capture these states and their temporal evolution over time.9 In this study, we applied the newly proposed 2025 GRAPPA-complex-to-manage (C2M), GRAPPA-TR, EULAR-D2M and EULAR-TR definitions to a cohort of newly diagnosed patients with PsA to determine their frequency, baseline characteristics, temporal evolution and long-term disease burden. This approach provides a valuable opportunity to evaluate the applicability and real-world frequency of these definitions and to characterise baseline differences and longitudinal disease trajectories in comparison with the overall PsA population. These insights may support earlier recognition and more tailored management in PsA. Methods Patient selection For this study, we included all patients with PsA consecutively enrolled in the Dutch Southwest Early Psoriatic Arthritis (DEPAR) cohort between July 2013 and March 2023, irrespective of follow-up duration, provided they had at least one follow-up visit after baseline, excluding those participating in Strategy Treatments Aiming at Minimal Disease Activity in PsA (STAMP) clinical trials. Follow-up time was variable due to the ongoing cohort. DEPAR is a multicentre, prospective observational cohort comprising newly diagnosed, DMARD-naïve patients with PsA from 14 centres (one academic hospital, 12 general hospitals and one specialised treatment centre).3 13 PsA was diagnosed by the treating rheumatologist based on clinical judgement, without requiring fulfilment of the CASPAR criteria. Although patients were routinely engaged in DEPAR research projects, they were not involved in the design or reporting of this particular study. Data collection Patients were assessed every 3 months during the first year, every 6 months in the second year and annually thereafter. At each visit, research nurses conducted a physical examination and collected data on medication use, laboratory parameters and patient-reported outcomes (PROs). Clinical assessments included tender and swollen joint counts (tender joint count-68 (TJC68), swollen joint count-66 (SJC66)), presence and extent of psoriasis (BSA: 0%–100%),14 enthesitis (Leeds Enthesitis Index (LEI): 0–6) and dactylitis.15 Laboratory measurements included C reactive protein (CRP). PROs included functional ability (Health Assessment Questionnaire-Disability Index (HAQ-DI)),16 patient and physician global health and pain assessments (both on 0–100 mm visual analogue scale (VAS)),17 fatigue (numeric rating scale, 0–10)18 and Psoriatic Arthritis Impact of Disease 12-item questionnaire.19 Composite disease activity measures were used to comprehensively assess PsA activity across multiple domains. These included the Disease Activity index for PSoriatic Arthritis,20 the Psoriatic Arthritis Disease Activity Score21 and Minimal Disease Activity (MDA).22 Treatment Treatment initiation and modifications were made through shared decision-making between the patient and the treating rheumatologist, based on national and international guidelines. There was no predefined treatment protocol.23 24 Outcomes Patients were classified according to the 2025 GRAPPA definitions of C2M PsA and TR PsA and the 2025 EULAR definitions of D2M PsA and TR PsA. All definitions are based on domains and proxy criteria listed in figure 1, table 1. Proxy measures for the domains were derived from published GRAPPA8 and EULAR9 definitions and operationalised using available clinical outcomes and PROs, with final selection refined by research team consensus. As these proxies were designed to reflect the original definitions, some overlap between domains (eg, persistent disease activity and objective inflammation) is inherent to the definitions, with certain proxies (eg, dactylitis, enthesitis, SJC ≥1 and Psoriasis Area and Severity Index (PASI)/BSA; see table 1) contributing to both domains. Definitions of GRAPPA and EULAR criteria. (A) Overview of the GRAPPA and EULAR 2025 criteria, categorised by broad and narrow criteria. The ‘broad criteria’ include the GRAPPA-C2M and EULAR-D2M definitions, which are less restrictive and include a broader range of patients. The ‘narrow criteria’ include the GRAPPA-TR and EULAR-TR definitions, which are stricter and require additional factors such as objective inflammation and exclusion of alternative explanations. (B) Illustration of the progression of D2T patients over time. Key factors (treatment failure, disease perception, persistent disease symptoms and objective signs of inflammation) are tracked from baseline to 7 years. b/tsDMARD, biologic and targeted synthetic disease-modifying antirheumatic drug; D2T, difficult-to-treat; EULAR-D2M, European Alliance of Associations for Rheumatology difficult-to-manage; EULAR-TR, EULAR treatment refractory; GRAPPA-C2M, Group for Research and Assessment of Psoriasis and Psoriatic Arthritis complex-to-manage; GRAPPA-TR, GRAPPA treatment refractory; MoA, mechanism of action. Table 1C2M/D2M/TR definitions according to GRAPPA and EULAR 2025 criteriaGRAPPA-C2MGRAPPA-TREULAR-D2MEULAR-TR1. Treatment failureFailure of ≥1 b/tsDMARD≥3 treatments (with different MoAs, including at least 2 b/tsDMARDs)≥2 b/tsDMARDs with different MoAs≥2 b/tsDMARDs with different MoAs2. Disease state perceived as problematicby the rheumatologist and/or the patientby the rheumatologist and the patientby the rheumatologist and/or the patientby the rheumatologist and/or the patient2.1.1 PsAID-12 >4.0++++2.1.2 Patient global VAS ≥40++++2.2 Physician global VAS ≥40++++3. Persistent PsA-related symptomsas defined by ≥1 of the below pointsas defined by ≥1 of the below points (3 and/or 4)as defined by ≥1 of the below points3.1 Composite-level (any one)and/orand/or3.1.1 Not achieving MDA++++3.1.2 DAPSA-68/66 >14++++3.1.3 PASDAS >3.2++++3.2 Domain specific (any one)and/orand/or3.2.1 TJC ≥3++++ 3.2.2 Dactylitis ( count ≥ 1 ) ++++ 3.2.3 Enthesitis (LEI ≥ 1 or MASES ≥ 1 ) ++++ 3.2.4 PASI > 1 or BSA > 1% ++++ 3.2.5 SJC ≥ 1 ++++4. Objective inflammationas defined by ≥1 of the below pointsas defined by ≥1 of the below pointsObjective signs of inflammation 3.2.2 Dactylitis ( count ≥ 1 ) +++ 3.2.3 Enthesitis ( LEI ≥ 1 or MASES ≥ 1 ) +++ 3.2.5 SJC ≥ 1 +++4.1 CRP >5 mg/dL+++Active extramusculoskeletal manifestations 3.2.4 PASI > 1 or BSA > 1% +++5. Exclusion of alternative explanation++In this table, each column represents a different definition: GRAPPA-C2M, EULAR-D2M, EULAR-TR and GRAPPA-TR. The rows show the criteria for treatment failure, disease perception, persistent disease symptoms and objective signs of inflammation. The red-coloured proxies in criterion (3) reflected overlap with the proxies included in criterion (4). For GRAPPA-TR and EULAR-TR, overlapping proxies were counted once and ≥2 distinct criteria were required.BSA, body surface area; b/tsDMARD, biologic and targeted synthetic disease-modifying antirheumatic drug; CRP, C reactive protein (>10 mg/dL); DAPSA, Disease Activity index for PSoriatic Arthritis; EULAR-D2M, European Alliance of Associations for Rheumatology difficult-to-manage; EULAR-TR, EULAR treatment refractory; GRAPPA-C2M, Group for Research and Assessment of Psoriasis and Psoriatic Arthritis complex-to-manage; GRAPPA-TR, GRAPPA treatment refractory; LEI, Leeds Enthesitis Index (>1); MASES, Maastricht Ankylosing Spondylitis Enthesitis Score; MDA, Minimal Disease Activity; MoA, mechanism of action; PASDAS, Psoriatic Arthritis Disease Activity Score; PASI, Psoriasis Area and Severity Index; PsA, psoriatic arthritis; PsAID-12, Psoriatic Arthritis Impact of Disease 12-item questionnaire; SJC, swollen joint count; TJC, tender joint count; VAS, visual analogue scale. C2M/D2M/TR definitions according to GRAPPA and EULAR 2025 criteria In this table, each column represents a different definition: GRAPPA-C2M, EULAR-D2M, EULAR-TR and GRAPPA-TR. The rows show the criteria for treatment failure, disease perception, persistent disease symptoms and objective signs of inflammation. The red-coloured proxies in criterion (3) reflected overlap with the proxies included in criterion (4). For GRAPPA-TR and EULAR-TR, overlapping proxies were counted once and ≥2 distinct criteria were required. BSA, body surface area; b/tsDMARD, biologic and targeted synthetic disease-modifying antirheumatic drug; CRP, C reactive protein (>10 mg/dL); DAPSA, Disease Activity index for PSoriatic Arthritis; EULAR-D2M, European Alliance of Associations for Rheumatology difficult-to-manage; EULAR-TR, EULAR treatment refractory; GRAPPA-C2M, Group for Research and Assessment of Psoriasis and Psoriatic Arthritis complex-to-manage; GRAPPA-TR, GRAPPA treatment refractory; LEI, Leeds Enthesitis Index (>1); MASES, Maastricht Ankylosing Spondylitis Enthesitis Score; MDA, Minimal Disease Activity; MoA, mechanism of action; PASDAS, Psoriatic Arthritis Disease Activity Score; PASI, Psoriasis Area and Severity Index; PsA, psoriatic arthritis; PsAID-12, Psoriatic Arthritis Impact of Disease 12-item questionnaire; SJC, swollen joint count; TJC, tender joint count; VAS, visual analogue scale. GRAPPA-C2M PsA was defined by the presence of both: Treatment failure, defined as failure of ≥1 PsA-recommended b/tsDMARD at an adequate dose and interval with ≥12 weeks of exposure (unless discontinued earlier due to adverse events or intolerance). Treatment failure, defined as failure of ≥1 PsA-recommended b/tsDMARD at an adequate dose and interval with ≥12 weeks of exposure (unless discontinued earlier due to adverse events or intolerance). Perceived as problematic by the rheumatologist and/or the patient, defined by the presence of ≥1 criterion in item (2). Perceived as problematic by the rheumatologist and/or the patient, defined by the presence of ≥1 criterion in item (2). Evidence of persistent disease, defined by the presence of ≥1 criterion in item (3). Evidence of persistent disease, defined by the presence of ≥1 criterion in item (3). GRAPPA-TR PsA was defined as a subset of C2M meeting all of the following: Treatment failure of ≥3 therapies with different mechanisms of action (MoAs), including at least 2 b/tsDMARDs. Intolerance or discontinuation due to adverse events should not be counted as failures. Treatment failure of ≥3 therapies with different mechanisms of action (MoAs), including at least 2 b/tsDMARDs. Intolerance or discontinuation due to adverse events should not be counted as failures. Perceived as problematic by the rheumatologist and the patient, as defined by the presence of ≥1 criterion in item (2). Perceived as problematic by the rheumatologist and the patient, as defined by the presence of ≥1 criterion in item (2). Evidence of persistent disease, defined by the presence of ≥1 criterion in item (3). Evidence of persistent disease, defined by the presence of ≥1 criterion in item (3). Objective signs of ongoing inflammation, in addition to (3) persistent disease symptoms, the presence of at least one of the following (4): SJC66 ≥1, active extramusculoskeletal manifestations, PASI >1 and/or BSA >3%, enthesitis (LEI ≥1 or MASES ≥1), dactylitis (count ≥1) or CRP >10 mg/dL. Objective signs of ongoing inflammation, in addition to (3) persistent disease symptoms, the presence of at least one of the following (4): SJC66 ≥1, active extramusculoskeletal manifestations, PASI >1 and/or BSA >3%, enthesitis (LEI ≥1 or MASES ≥1), dactylitis (count ≥1) or CRP >10 mg/dL. Exclusion of alternative explanation as main driver of systemic inflammation, comorbidities such as infection, obesity-related CRP, fibromyalgia, central sensitisation, generalised hypermobility or mechanical/degenerative pain should be considered as potential contributors but do not preclude TR-PsA classification if PsA remains the primary driver. Exclusion of alternative explanation as main driver of systemic inflammation, comorbidities such as infection, obesity-related CRP, fibromyalgia, central sensitisation, generalised hypermobility or mechanical/degenerative pain should be considered as potential contributors but do not preclude TR-PsA classification if PsA remains the primary driver. EULAR-D2M PsA was defined by the presence of all three of the following: Treatment failure of ≥2 b/tsDMARDs with different MoAs, according to EULAR recommendations for PsA management. Failures may include primary or secondary inefficacy, or discontinuation due to side effects, intolerance or contraindications. Treatment failure of ≥2 b/tsDMARDs with different MoAs, according to EULAR recommendations for PsA management. Failures may include primary or secondary inefficacy, or discontinuation due to side effects, intolerance or contraindications. Perceived as problematic by the rheumatologist and/or the patient, defined by the presence of ≥1 criterion in item (2). Perceived as problematic by the rheumatologist and/or the patient, defined by the presence of ≥1 criterion in item (2). Evidence of persistent disease, by the presence of ≥1 criterion in item (3). Evidence of persistent disease, by the presence of ≥1 criterion in item (3). EULAR-TR PsA was defined by meeting all D2M criteria plus: Treatment failure to achieve or maintain response to ≥2 b/tsDMARDs with different MoA, specifically due to primary or secondary inefficacy (ie, not due to discontinuation for side effects, intolerance or contraindications). Treatment failure to achieve or maintain response to ≥2 b/tsDMARDs with different MoA, specifically due to primary or secondary inefficacy (ie, not due to discontinuation for side effects, intolerance or contraindications). Perceived as problematic by the rheumatologist and/or the patient, defined by the presence of ≥1 criterion in item (2). Perceived as problematic by the rheumatologist and/or the patient, defined by the presence of ≥1 criterion in item (2). Evidence of persistent disease, by the presence of ≥1 criterion in item (3). Evidence of persistent disease, by the presence of ≥1 criterion in item (3). Objective signs of ongoing inflammation, in addition to persistent disease symptoms (3), the presence of at least one of the following: SJC66 ≥1, active extramusculoskeletal manifestations, PASI >1 and/or BSA >3%, enthesitis (LEI ≥1 or MASES ≥1), dactylitis (count ≥1) or CRP >10 mg/dL. Objective signs of ongoing inflammation, in addition to persistent disease symptoms (3), the presence of at least one of the following: SJC66 ≥1, active extramusculoskeletal manifestations, PASI >1 and/or BSA >3%, enthesitis (LEI ≥1 or MASES ≥1), dactylitis (count ≥1) or CRP >10 mg/dL. Exclusion of other drivers of symptoms, including comorbidities, psychosocial or other non-inflammatory factors that may explain ongoing symptoms. Exclusion of other drivers of symptoms, including comorbidities, psychosocial or other non-inflammatory factors that may explain ongoing symptoms. Statistical analysis Descriptive statistics were used to report the frequency of GRAPPA-C2M and GRAPPA-TR PsA and EULAR-D2M and EULAR-TR PsA. Frequency was calculated as the number of patients fulfilling each definition divided by the total study population. Baseline characteristics were compared between patients across these subgroups using appropriate summary measures: mean and SD for normally distributed continuous variables, median and IQR for non-normally distributed variables and frequencies and percentages for categorical variables. Time-to-event analyses were performed to evaluate the cumulative incidence and timing of fulfilling the GRAPPA and EULAR definitions. Kaplan-Meier methods were used to estimate the probability of fulfilling each definition during follow-up, with time calculated from baseline to the first visit at which the respective criteria or individual components were met. Patients who did not fulfil a definition were censored at their last available visit. Median times and IQRs were reported, and Kaplan-Meier curves were used to visually compare temporal patterns across definitions. Longitudinal differences in disease burden between groups were evaluated using linear mixed-effects models to assess changes in HAQ-DI, pain, fatigue and patient global health VAS over time. The models included random intercepts and random slopes to account for within-patient variability across repeated visits. All models were adjusted for time, age and sex, as well as CRP, SJC and TJC, which were incorporated as time-varying covariates to account for disease activity parameters (CRP, SJC, TJC) at each visit. Variables used to define GRAPPA-C2M and GRAPPA-TR PsA and EULAR-D2M and EULAR-TR PsA status were based entirely on observed data. No imputation of missing data was performed. All statistical analyses were conducted using Stata V.18.0 (StataCorp, College Station, Texas, USA). Results Patients The DEPAR cohort comprised 934 patients at baseline. After excluding five patients due to missing medication data and 44 patients who participated in the STAMP, a nested randomised controlled trial within the DEPAR cohort, 885 patients with all PsA clinical phenotypes were included in the current analysis (online supplemental figure 1). In the total eligible DEPAR population (n=885), the mean follow-up duration was 40.4 months (SD 27.2), with a median of 36 months (IQR 18–72). Data were censored at March 2023 and treatment outcomes were assessed within the available follow-up time for each patient (table 2, online supplemental figure 2). The mean age was 49.7 years (SD 13.8) and 51.0% were female (451/885). The median symptom duration at baseline was 10 months (IQR 4–32) and the mean BMI was 28.6 kg/m² (SD 9.7). Enthesitis was present in 8.0% (70/885) and dactylitis in 8.0% (68/885). The median TJC was 3 (IQR 1–7) and the SJC was 2 (IQR 1–5). Mean BSA affected by psoriasis was 3.5% (SD 5.2) and a history of psoriasis was recorded in 90.0% (794/885) of patients. Mean CRP was 8.8 mg/L (SD 15.8). PROs indicated moderate disease burden, with a mean HAQ-DI of 0.8 (SD 0.5) (table 2). Table 2Baseline characteristicsGRAPPAEULAREULARGRAPPAAll populationC2MD2MTRTRPrevalence, n (%)885 (100)179 (20)37 (4.2)34 (3.8)11 (1.2)Demographics Age (years), mean (SD)49.7 (13.8)47.7 (13.2)47.7 (12.8)46.9 (12.9)46.1 (12.9) Sex (female), n (%)451 (51)108 (60)26 (70)24 (71)8 (73) BMI, mean (SD)28.6 (9.7)28.8 (5.4)30.7 (4.2)30.8 (4.3)33.6 (4.2) BMI ≥30, n (%)290 (33)66 (37)18 (49)18 (53)7 (64) Symptom duration (months), median (IQR)10 (4–32)10 (4–36)13 (5–45)23 (4–45)16 (5–33) Follow-up duration (months), median (IQR)36 (18–72)48 (36–72)60 (48–72)60 (48–72)60 (36–72) The presence of PsO, n (%)794 (91)163 (91)34 (92)31 (91)10 (91)Disease activity outcomes Tender joint count, median (IQR)3 (1–7)5 (2–11)4 (2–12)4 (2–12)8 (2–17) Swollen joint count, median (IQR)2 (1–5)2 (1–5)2 (0–5)2 (0–5)4 (2–6) BSA, mean (SD)3.5 (5.2)4.8 (8.3)5.4 (8.9)5.5 (9.2)4.9 (9.3) The presence of enthesitis (LEI >1), n (%)198 (22)60 (34)16 (43)16 (47)3 (27) LEI, median (IQR)3 (2–4)2 (2–3)4 (3–4)4 (3–4)4 (4–4) The presence of dactylitis, n (%)164 (19)34 (19)6 (16)6 (18)3 (27) CRP, mean (SD)8.8 (15.8)9.7 (24.3)8.1 (8.9)8.7 (8.2)14.3 (6.7)Patient-reported outcomes HAQ-DI, mean (SD)8 (0.5)0.9 (0.5)0.9 (0.6)1.0 (0.6)1.0 (0.6)BMI, body mass index; BSA, body surface area; C2M, complex-to-manage; CRP, C reactive protein; D2M, difficult-to-manage; ESR, erythrocyte sedimentation rate; EULAR, European Alliance of Associations for Rheumatology; GRAPPA, Group for Research and Assessment of Psoriasis and Psoriatic Arthritis; HAQ-DI, Health Assessment Questionnaire-Disability Index; LEI, Leeds Enthesitis Index; PsO, psoriasis; TR, treatment refractory. Baseline characteristics BMI, body mass index; BSA, body surface area; C2M, complex-to-manage; CRP, C reactive protein; D2M, difficult-to-manage; ESR, erythrocyte sedimentation rate; EULAR, European Alliance of Associations for Rheumatology; GRAPPA, Group for Research and Assessment of Psoriasis and Psoriatic Arthritis; HAQ-DI, Health Assessment Questionnaire-Disability Index; LEI, Leeds Enthesitis Index; PsO, psoriasis; TR, treatment refractory. Based on 7-year follow-up data on medication use, patients were categorised into three main treatment groups: 134 patients (15.1%) were treated with NSAIDs only, 431 patients (48.7%) received only csDMARDs, 320 patients (36.2%) were treated with at least one b/tsDMARD. Among the b/tsDMARD users, 187 patients (21.1% of the total cohort) received only one b/tsDMARD, 133 patients (15.0% of the total cohort) received ≥2 b/tsDMARDs (online supplemental figure 1). Frequency Treatment failure GRAPPA-C2M PsA From the 320 patients who received at least one b/tsDMARD during follow-up, 34 were excluded due to lack of follow-up data. This resulted in 286 patients eligible for GRAPPA-C2M assessment based on treatment failure. Of these, 179 patients (63%) fulfilled the GRAPPA-C2M PsA definition at consecutive visits, corresponding to 20% of the total study population (179 of 885 patients) (figure 2, online supplemental figure 1). Distribution of GRAPPA and EULAR criteria among DEPAR cohort. This Venn diagram illustrates the overlap of patients in the DEPAR cohort (n=885) who meet the criteria according to the GRAPPA and EULAR 2025 definitions. Each circle represents a distinct definition: GRAPPA-C2M (20%): 179 patients meet the broad GRAPPA-C2M criteria; EULAR-D2M (4.2%): 37 patients meet the EULAR-D2M criteria; EULAR-TR (3.8%): 34 patients meet the narrower EULAR-TR criteria; GRAPPA-TR (1.2%): 11 patients meet the strictest GRAPPA-TR criteria. DEPAR, Dutch Southwest Early Psoriatic Arthritis; EULAR-D2M, European Alliance of Associations for Rheumatology difficult-to-manage; EULAR-TR, EULAR treatment refractory; GRAPPA-C2M, Group for Research and Assessment of Psoriasis and Psoriatic Arthritis complex-to-manage; GRAPPA-TR, GRAPPA treatment refractory. EULAR-D2M and EULAR-TR PsA EULAR criteria were applied to the same initial group of 133 patients treated with ≥2 b/tsDMARDs. Of these, 49 were excluded because both therapies had the same mechanism of action and 22 due to lack of follow-up data, leaving 62 patients eligible for EULAR-D2M and EULAR-TR assessment. Among these, 37 patients (60%) fulfilled the EULAR-D2M definition, corresponding to 4.2% of the total study population. The stricter EULAR-TR definition was met by 34 patients (55% of those eligible), after excluding six whose treatment failure was not classified as primary or secondary, corresponding to 3.8% of the total population (figure 2, online supplemental figure 1). GRAPPA-TR PsA Among the 133 patients treated with ≥2 b/tsDMARDs, 49 were excluded because both therapies had the same mechanism of action. This resulted in 84 patients eligible for GRAPPA-TR assessment based on treatment failure. Of these, 28 were excluded due to lack of follow-up data and after excluding six whose treatment failure was not classified as primary or secondary, leaving 50 patients for full evaluation. In total, 11 patients (20%) fulfilled the GRAPPA-TR definition at consecutive visits, corresponding to 1.2% of the total study population (11/885) (figure 2, online supplemental figure 1). The Kaplan-Meier curves illustrate the time to fulfilment of the GRAPPA-C2M, EULAR-D2M, EULAR-TR and GRAPPA-TR definitions, with the event time defined as the first follow-up visit at which all three required criteria were met: treatment failure occurring first, followed by disease perception at the subsequent visit, while persistent disease activity and signs of inflammation could be fulfilled at the same or later visits (figure 3). Consistent with the component-level timing shown in table 3, GRAPPA-C2M was characterised by early treatment failure, occurring at a median of 9 months (IQR 3–12), with completion of all criteria by 18 months (IQR 9–36). EULAR-D2M and EULAR-TR showed later treatment failure at a median of 24 months (IQR 12–36), with subsequent fulfilment of disease perception and persistent disease activity and signs of inflammation at 36 months (IQR 18–48). GRAPPA-TR demonstrated the latest trajectory, with treatment failure at 18 months (IQR 12–36) and completion of the remaining criteria at 48 months (IQR 36–60). Cumulative time-to-event plot for GRAPPA and EULAR criteria. This cumulative Kaplan-Meier plot illustrates the probability over time of first fulfilling all required criteria for the GRAPPA-C2M, GRAPPA-TR, EULAR-D2M and EULAR-TR definitions, calculated from baseline. The event was defined as the first study visit at which all criteria for a given definition were simultaneously met. EULAR-D2M, European Alliance of Associations for Rheumatology difficult-to-manage; EULAR-TR, EULAR treatment refractory; GRAPPA-C2M, Group for Research and Assessment of Psoriasis and Psoriatic Arthritis complex-to-manage; GRAPPA-TR, GRAPPA treatment refractory. Table 3Timing of key components of GRAPPA and EULAR C2M/D2M/TR definitionsGRAPPA-C2M(n=179)EULAR-D2M(n=37)EULAR-TR(n=34)GRAPPA-TR(n=11)Treatment failure (months), median (IQR)9 (3–12)24 (12–36)24 (12–36)18 (12–36)Disease perception (months), median (IQR)18 (9–36)36 (18–48)36 (18–48)48 (36–60)Persistent disease and signs of inflammation (months), median (IQR)18 (9–36)36 (18–48)36 (18–48)48 (36–60)Timing (in months) of fulfilment of individual components of the GRAPPA-C2M, EULAR-D2M, EULAR-TR and GRAPPA-TR definitions in a real-world longitudinal psoriatic arthritis cohort. For each definition, the time to treatment failure, disease perception and persistent disease activity and signs of inflammation are presented as median (IQR). Timing was calculated from baseline and reflects the first visit at which the respective component criteria were met.EULAR-D2M, European Alliance of Associations for Rheumatology difficult-to-manage; EULAR-TR, EULAR treatment refractory; GRAPPA-C2M, Group for Research and Assessment of Psoriasis and Psoriatic Arthritis complex-to-manage; GRAPPA-TR, GRAPPA treatment refractory. Timing of key components of GRAPPA and EULAR C2M/D2M/TR definitions Timing (in months) of fulfilment of individual components of the GRAPPA-C2M, EULAR-D2M, EULAR-TR and GRAPPA-TR definitions in a real-world longitudinal psoriatic arthritis cohort. For each definition, the time to treatment failure, disease perception and persistent disease activity and signs of inflammation are presented as median (IQR). Timing was calculated from baseline and reflects the first visit at which the respective component criteria were met. EULAR-D2M, European Alliance of Associations for Rheumatology difficult-to-manage; EULAR-TR, EULAR treatment refractory; GRAPPA-C2M, Group for Research and Assessment of Psoriasis and Psoriatic Arthritis complex-to-manage; GRAPPA-TR, GRAPPA treatment refractory. Criteria after treatment failure During the observation period, among the 885 included patients, 286 (32%) fulfilled criterion (1) of the GRAPPA-C2M definition (treatment failure). For the EULAR definitions, 62 patients (7%) met the corresponding D2M criterion, while 52 patients (6%) met the initial criterion for EULAR-TR. For GRAPPA-TR, 50 patients (6%) fulfilled criterion (1) (figure 4, online supplemental figure 3). Percentages of patients meeting criteria across GRAPPA and EULAR definitions. The proportions presented correspond to sequential (conditional) proportions derived from the stepwise application of each definition, as shown in online supplemental figure 2. (1) Treatment failure, (2) disease perception, (3) persistent disease symptoms, (4) objective signs of inflammation. BSA, body surface area; CRP, C reactive protein (>10 mg/dL); DAPSA, Disease Activity index for PSoriatic Arthritis; EULAR-D2M, European Alliance of Associations for Rheumatology difficult-to-manage; EULAR-TR, EULAR treatment refractory; GRAPPA-C2M, Group for Research and Assessment of Psoriasis and Psoriatic Arthritis complex-to-manage; GRAPPA-TR, GRAPPA treatment refractory; MDA, Minimal Disease Activity; PASDAS, Psoriatic Arthritis Disease Activity Score; PASI, Psoriasis Area and Severity Index; PsAID-12, Psoriatic Arthritis Impact of Disease 12-item questionnaire; SJC, swollen joint count; TJC, tender joint count; VAS, visual analogue scale. Criterion (2), disease perception, defined by ≥1 of PSAID >4.0, patient global VAS ≥40 or physician global VAS ≥40, was met by 187 of the 286 GRAPPA-C2M candidates (65%), 45 of 62 EULAR-D2M candidates (73%), 44 of 52 EULAR-TR candidates (85%) and only 11 of 50 GRAPPA-TR candidates (22%), which is the only definition that mandates both patient global VAS ≥40 and physician global VAS ≥40 simultaneously (figure 4, online supplemental figure 3). Among these, criterion (3), persistent disease symptoms, consisted of both a composite-level indicator and domain-specific components. At the composite level, not achieving MDA was met by 164 of 187 GRAPPA-C2M patients (88%), 37 of 45 EULAR-D2M patients (82%), 34 of 44 EULAR-TR patients (77%) and 11 of 11 GRAPPA-TR patients (100%). This made MDA non-achievement the dominant indicator of persistent disease classification across all definitions (figure 4, online supplemental figure 3). Within the domain-specific components, TJC ≥3 was present in 81% of GRAPPA-C2M, 56% of EULAR-D2M, 57% of EULAR-TR and 91% of GRAPPA-TR patients who reached criterion (3). Other objective inflammatory manifestations, including enthesitis, dactylitis, skin involvement (PASI >1 or BSA >3%) and swollen joints (SJC ≥1), were also frequently present across groups. These features constituted the main areas of overlap between criterion (3) and criterion (4) (objective signs of inflammation). Detailed distributions of domain-specific components are provided in figure 4, online supplemental figure 3. Consequently, for criterion (4), CRP >10 mg/dL was the last purely objective marker of inflammation and was identified in 41 of 161 GRAPPA-C2M patients (22%), 12 of 45 EULAR-D2M patients (27%), 11 of 44 EULAR-TR patients (25%) and five of 11 GRAPPA-TR patients (46%) (figure 4, online supplemental figure 3). Comparison of baseline characteristics At baseline, female sex was more common in all four groups, with 60% in GRAPPA-C2M, 70% in EULAR-D2M, 71% in EULAR-TR and 73% in GRAPPA-TR, compared with 51% in the total population. Patients fulfilling these definitions also had a higher BMI (≥30), which was more prevalent in stricter groups: 37% in GRAPPA-C2M, 49% in EULAR-D2M, 53% in EULAR-TR and 64% in GRAPPA-TR versus 33% in the overall cohort. Symptom duration was also longer, with median durations of 10 months in GRAPPA-C2M, 15 months in EULAR-D2M, 23 months in EULAR-TR and 16 months in GRAPPA-TR, compared with 10 months in the total population (table 2). Skin severity (BSA) was also greater in these groups, with mean values of 4.8% (SD 8.3) in GRAPPA-C2M, 5.4% (8.9) in EULAR-D2M, 5.5% (9.2) in EULAR-TR and 4.9% (9.3) in GRAPPA-TR, compared with 3.5% (6.2) in the total population. The prevalence of enthesitis (LEI >1) was higher, occurring in 22% of GRAPPA-C2M, 31% of EULAR-D2M, 24% of EULAR-TR and 46% of GRAPPA-TR patients, compared with 19% overall. Median LEI scores followed the same pattern: 3 (IQR 2–4) in GRAPPA-C2M, 3 (2–4) in EULAR-D2M, 3 (1–4) in EULAR-TR and 4 (3–4) in GRAPPA-TR, versus 3 (2–4) in the overall cohort. PROs of patients over time PROs over time were compared between patients who fulfilled and those who did not fulfil the GRAPPA-C2M, EULAR-D2M, EULAR-TR and GRAPPA-TR definitions in figure 5. Across all four groups, patients consistently reported worse HAQ-DI, global health, pain and fatigue scores throughout follow-up. Effect plots showing longitudinal trajectories of patient-reported outcomes by GRAPPA and EULAR definitions. Longitudinal effect plots displaying trajectories of HAQ-DI, VAS-GH, VAS-pain and fatigue over time for patients classified as definitions. D2T, difficult-to-treat; EULAR-D2M, European Alliance of Associations for Rheumatology difficult-to-manage; EULAR-TR, EULAR treatment refractory; GRAPPA-C2M, Group for Research and Assessment of Psoriasis and Psoriatic Arthritis complex-to-manage; GRAPPA-TR, GRAPPA treatment refractory; HAQ-DI, Health Assessment Questionnaire-Disability Index; VAS-GH, visual analogue scale general health. For GRAPPA-C2M, the HAQ-DI was significantly higher, with a coefficient of 0.26 (95% CI 0.18 to 0.34, p<0.001) after adjustment for time, age, sex and time-varying disease activity parameters. Patients classified as EULAR-D2M, EULAR-TR and GRAPPA-TR also demonstrated a significantly greater HAQ-DI burden, with coefficients of 0.39 (95% CI 0.20 to 0.48, p=0.001), 0.42 (95% CI 0.25 to 0.55, p<0.001) and 0.52 (95% CI 0.22 to 0.51, p<0.001), respectively (figure 5, online supplemental table 1). Consistent trends were seen for VAS general health, pain and fatigue (figure 5, online supplemental table 1). Discussion In this large, prospective real-world PsA cohort, we systematically applied the 2025 GRAPPA and EULAR definitions to identify C2M, D2M and TR PsA and to characterise their frequency, clinical features and longitudinal disease burden. This study represents the first longitudinal application of these definitions in a real-world dataset. Our findings demonstrate that, although the strictest TR categories are relatively uncommon, patients fulfilling any D2M definition experience a persistently higher disease burden over time. Using the definitions,8 9 20% of patients fulfilled the GRAPPA-C2M criteria, while smaller proportions fulfilled the stricter criteria: 4.2% for EULAR-D2M, 3.8% for EULAR-TR and 1.2% for GRAPPA-TR. These estimates are lower than previously reported rates of D2T PsA, which range from 16% to 34% in registry-based studies using proxy definitions.10,1225 This discrepancy likely reflects important methodological differences. Prior studies relied largely on retrospective data and simplified treatment-failure thresholds, whereas the present analysis applied multicomponent consensus definitions incorporating disease perception, persistent symptoms and, for TR disease, objective evidence of ongoing inflammation. A key finding of this study is the distinct temporal pattern with which the GRAPPA and EULAR definitions were fulfilled. GRAPPA-C2M was characterised by early treatment failure, typically occurring within the first year after baseline, with completion of all criteria by approximately 18 months. In contrast, EULAR-D2M and EULAR-TR, as well as GRAPPA-TR, were fulfilled later, with treatment failure occurring around 2 years and subsequent criteria extending to 3–4 years of follow-up. These findings indicate that GRAPPA-C2M captures disease complexity at an earlier stage, whereas the EULAR-TR and GRAPPA-TR definitions identify smaller subsets of patients with more prolonged and persistent disease trajectories. This temporal gradient closely reflects the relative stringency of the definitions. In particular, the GRAPPA-TR definition was the most restrictive, largely driven by the requirement that both patient-reported and physician-reported disease perception exceed predefined thresholds. This resulted in substantial attrition at the disease-perception step and highlights how subjective assessments can precede and, in practice, outweigh objective measures of disease activity in determining case identification. Across all definitions, failure to achieve MDA was the dominant indicator of persistent disease classification. This finding reinforces the central role of MDA as a clinically meaningful composite outcome that captures multidomain disease burden in PsA.2 Domain-specific contributors, particularly tender joint burden, enthesitis and skin involvement, were highly prevalent and often overlapped with objective inflammatory criteria. By contrast, elevated CRP was observed in only a minority of patients compared with other criteria, even in the strictest groups, occurring in fewer than 50% of patients, emphasising that persistent disease burden in PsA is frequently not reflected by systemic inflammatory markers alone.26 Patients fulfilling the definitions shared a consistent baseline profile characterised by female predominance, higher BMI, longer symptom duration and greater skin and entheseal involvement. These findings align with earlier reports linking female sex, obesity and enthesitis to poorer treatment response and higher disease impact in PsA.510,12 25 Notably, these adverse characteristics were already present at diagnosis, suggesting that difficult disease trajectories will be identifiable early in the disease course rather than being solely a consequence of treatment exposure. Longitudinal analyses demonstrated that patients meeting any of the D2M definitions experienced a persistently higher burden of disability, pain, fatigue and impaired global health throughout follow-up, even after adjustment for time-varying disease activity indicators. These findings indicate these definitions are not merely administrative or treatment-based labels, but represent clinically meaningful states associated with sustained impairment from the patient’s perspective. Our study has several strengths, including its large sample size, long follow-up and detailed, standardised data collection within a real-world PsA cohort. Moreover, this study also has several limitations. First, we applied the 2025 GRAPPA and EULAR definitions retrospectively; therefore, the classification of C2M, D2M and TR PsA relied on observational data rather than a protocol-driven approach, which may have introduced variability in treatment decisions and outcome reporting. In particular, key components such as the ‘perceived problematic disease state’ had to be approximated using proxy measures, which may not fully capture the intended multidimensional and subjective aspects of disease perception. We were also unable to apply the fifth criterion of the GRAPPA-TR and EULAR-TR definition as originally specified, the exclusion of alternative explanations for systemic inflammation. This limitation particularly affects the EULAR-TR definition, as it may have reduced its intended stringency and resulted in greater apparent overlap between the EULAR-D2M and EULAR-TR groups. The annual scheduling of follow-up visits after the second year may have limited the capture of treatment changes or switches occurring between visits. Despite long-term follow-up, not all patients had complete longitudinal data, particularly for PROs, which could bias our results. We addressed this issue through mixed-effects modelling, but we cannot fully exclude residual bias. Generalisability may be limited, as all patients were enrolled from Dutch clinical centres; treatment patterns and access to b/tsDMARDs may differ in other healthcare settings. Finally, although we adjusted for disease activity over time, we could not fully disentangle inflammatory and non-inflammatory contributors (eg, fibromyalgia, structural damage or mood disorders), which may have influenced classification according to these definitions. In this longitudinal real-world cohort, the GRAPPA-C2M definition identified a larger and earlier subgroup of patients. In contrast, the EULAR-D2M and EULAR-TR definitions appeared to delineate more advanced disease states, in which patient-reported and physician-reported perceptions were more closely aligned with objective inflammatory activity, reflecting a more balanced integration of subjective and objective disease domains, consistent with truly D2M or refractory PsA. The GRAPPA-TR definition was the most restrictive, largely driven by the dominance of disease-perception criteria, which may limit its ability to fully capture biological treatment refractoriness in some patients. Although these definitions provide a structured framework to characterise D2M PsA, their application in routine practice remains challenging due to the need for detailed and concurrently available clinical, patient-reported and treatment data, reflecting the heterogeneous and multidomain nature of PsA. Further refinement and simplification of these criteria may improve the feasibility and consistency of identification in real-world settings, thereby supporting earlier recognition and more tailored management of patients at risk of complex or TR disease. Supplementary material Acknowledgements We thank all participating patients, rheumatologists and research nurses. Footnotes Funding: The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors. Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: This study was conducted within DEPAR, a multicentre, prospective observational cohort approved by the Medical Ethics Committee of Erasmus Medical Center, the Netherlands (MEC-2012-549). The study was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants before enrolment. Data availability free text: Data are available from the corresponding author on reasonable request. Patient and public involvement statement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Data availability statement Data are available on reasonable request. References Associated Data Supplementary Materials Data Availability Statement Data are available on reasonable request.