Background/Purpose: Immune checkpoint inhibitor (ICI)–associated arthritis has been classified into distinct phenotypes, including inflammatory arthritis (IA), polymyalgia rheumatica (PMR)-like syndrome, activated crystalline arthritis, and activated osteoarthritis (aOA). However, the boundaries between these phenotypes are often challenging to discern, and differences in their presentation, natural history and outcomes remain poorly defined1.

Methods: We retrospectively analyzed 85 patients with ICI-associated arthritis, classified as IA (n=58), crystalline (n=9), PMR-like (n=7), and aOA (n=23), referred to rheumatology at a tertiary center (Jan 2014–Feb 2020). We compared clinical features, joint involvement, serologies, synovial fluid, treatment, symptom duration, and DMARD use. Chi-square tests were used for proportions and Kruskal-Wallis for medians.

Results: IA typically presented as polyarthritis (43%) with frequent knee (79%), PIP (43%), and MCP (41%) involvement. PMR-like cases had predominant shoulder (71%) and MCP (71%) involvement. Crystalline arthritis favored elbows (33%) and hips (33%). aOA was more often monoarthritis (26%) with a degenerative pattern (wrists 30%, PIPs 35%, DIPs 4%) (p=0.7). Inflammatory synovial fluid was most frequent in IA (31%), and rare in PMR-like (0%) and aOA (4%) (p=0.1). Peripheral involvement in both IA and PMR-like suggests classification may hinge on shoulder involvement, despite overlap with IA features.PMR-like patients had the longest symptom duration (65 months, IQR 63–70) but the shortest DMARD exposure (12 months, IQR 12–34). Crystalline (35 months, IQR 26–72) and aOA (31 months, IQR 21–47) had shorter durations but comparable DMARD use (13 months, IQR 6–51; 20 months, IQR 6–39) (p=0.2). IA had the shortest duration (14 months, IQR 6–14) but longest DMARD use (33 months, IQR 9–63) (p=0.3), reflecting higher inflammatory activity. DMARDs—methotrexate (16%), sulfasalazine (43%), hydroxychloroquine (40%), and TNFi (7%)—were used almost exclusively in IA (p=0.9). Among aOA patients, 48% had concurrent inflammatory arthritis, affecting interpretation of DMARD use. Limited DMARD use in PMR-like cases may reflect historical steroid-based treatment; however, small sample size (n=7) limits interpretation.

Conclusion: ICI-associated arthritis includes phenotypes with differing inflammatory burden, duration, and treatment. IA is more active and treated more aggressively; others persist longer but receive less therapy. Whether these differences reflect biology or treatment variation is unclear. This historical cohort had low biologic use, possibly contributing to symptom duration. Findings support phenotype-based approaches that balance arthritis and cancer therapy, but small subgroup sizes warrant further study.

Table 1. Baseline Demographics, Cancer Characteristics, ICI Regimen, and Concomitant irAE

Table 2: Clinical Features, Joint Distribution, Serologic Markers, and Treatment Patterns by ICI-Arthritis Phenotype

Table 3: Symptom Duration and DMARD Exposure by ICI-Arthritis Phenotype

« Back to ACR Convergence 2025

ACR Meeting Abstracts - https://acrabstracts.org/abstract/distinct-clinical-phenotypes-of-immune-checkpoint-inhibitor-associated-arthritis-comparative-analysis-of-inflammatory-arthritis-polymyalgia-rheumatica-like-activated-crystalline-and-activa/