Background/Purpose: Psoriatic arthritis (PsA) shows the largest complexity in joint involvement among all inflammatory types of arthritis. In order to improve feasibility and to reduce the number of joints to be evaluated for a global PsA disease activity assessment, we aimed to identify clusters of joint involvement in the Swiss Clinical quality management (SCQM) PsA cohort.

Methods: From inclusion visits of all 1256 patients with PsA diagnosis, manifestations in the 66 swollen/68 tender joint score were clustered according to their dissimilarities in profiles by an iterative algorithm, at each stage joining the two most similar joint profiles into one cluster. The distances between clusters were recomputed by the Lance-Williams dissimilarity update formula. Dendrograms of hierarchical clustering were performed separately on the swollen and tender joint data. Distances between clusters were reported by the number of patients with differences in their state of joint involvement.

Results: Among 1256 individuals diagnosed of PsA by board certified rheumatologists, 749 patients (60%) fulfilled the CASPAR classification criteria. Numbers [in brackets] are from CASPAR+ patients. Patients had mean age 50 [48] years, mean disease duration 5.8 [5.9] years (IQR 2-12 for both) and 48%  [45%] female gender. Current skin involvement was moderate to severe in 21 [24] %, absent in 21 [14] %, and nail involvement reported in 29 [29] %. Skin involvement was reported in the patient’s history in 92 [98] %, arthritis in 84 [92] %, enthesitis in 62 [65] %, dactylitis in 57 [74] %, spine involvement in 45 [44] %, and a positive family history in 38 [41] % of the cases. Treatment at inclusion were: NSAIDs 54 [57] %, synthetic DMARDs 61 [65] % or biologics 50 [52] %. Mean SJC was 2;IQR 0-6 [3;IQR 0-7], and mean TJC 4; IQR 0-10 [5; IQR 1-11]. Favourable clustering and symmetry in 9 out of ten clusters was obtained only for tender joints. The level of agreement was likewise the highest for symmetric involvement of wrist and knee joints in all and in CASPAR+ patients. Other clusters, symmetric involvement of PIP 2-5, MCP 1, MCP 2-5 [-3],  MTP 1-5, shoulder joints, ankle joints,  IP 1 + DIP 2-3 [-5], and a very complex pattern in the 10^th cluster were the same, but varied in sequence of agreement between analyses in all or only in CASPAR+ patients. In contrast, the cluster resolution for swollen joints was far from satisfying even when applying the Ward’s method, which minimizes the increase of within-cluster variance in each merging step. With exception of the wrist, knee and ankle joint clusters, all other obtained SJ clusters differed substantially from TJ clusters. An oligoarticular pattern according to Moll & Wright was seen in 42% and polyarticular in 16%. In contrast to highly prevalent dactylitis, predominant DIP involvement was seen in only 3% of cases.

Conclusion: Swollen and tender joint patterns in PsA are hardly in agreement. The enormous complexity in type and pattern of joint involvement may limit the definition of joint involvement only on basis of clinical data. With regard to the many cases with oligoarticular disease, it appears that no joint from the 66/68 joint count should be omitted for a meaningful joint count based PsA disease activity assessment.

ACR Meeting Abstracts - https://acrabstracts.org/abstract/major-differences-in-the-pattern-of-joint-swelling-and-tenderness-in-a-large-psoriatic-arthritis-cohort-results-from-an-exploratory-hierarchical-cluster-analysis/