Occurrence and clinical course of immune checkpoint inhibitor-associated inflammatory arthritis and polymyalgia rheumatica-like syndromes: a systematic review and meta-analysis

Abstract

Immune checkpoint inhibitor-associated inflammatory arthritis (ICI-IA) and polymyalgia rheumatica (PMR)-like syndromes are clinically important rheumatic immune-related adverse events, but definitions vary across studies. We estimated their occurrence in real-world cohorts and summarized clinical course and management. PubMed, Ovid MEDLINE, Embase, Web of Science Core Collection, and Cochrane CENTRAL were searched from inception through July 25, 2026. Two reviewers independently screened studies, extracted data, and assessed risk of bias, with disagreements resolved by a third reviewer. Compatible single-group proportions were pooled using random-intercept binomial generalized linear mixed models; non-poolable clinical-course evidence was synthesized using the Synthesis Without Meta-analysis framework. Sixty-five reports representing 63 studies were included. Five independent ICI-treated cohorts reported 140 study-defined de novo ICI-IA cases among 4,644 patients, yielding a pooled crude occurrence proportion of 2.87% (95% CI, 2.02%-4.07%; 95% prediction interval, 1.06%-7.53%; I 2 = 71.7%). Three of the five primary cohorts were at high risk of bias. Definitions varied; excluding the largest cohort, which used a broader clinician-suspected joint-symptom definition, yielded a pooled proportion of 2.61% (95% CI, 1.50%-4.79%). Three studies reported 19 PMR/PMR-like events among 1,998 patients (1.08%; 95% CI, 0.32%-3.62%), with marked imprecision and heterogeneity. A 2026 cohort reporting 12 PMR cases among 734 patients was retained as contextual evidence because de novo timing could not be verified and possible overlap with an earlier report could not be excluded. Descriptive evidence showed heterogeneous phenotypes and disease courses, with persistence in some cohorts and remission in others. Glucocorticoids and disease-modifying antirheumatic drugs were frequently used, but comparative treatment effects could not be determined. In denominator-valid studies, approximately 3% of ICI-treated patients were classified as having de novo ICI-IA during variable study-specific follow-up. This is a crude study-period proportion, not a fixed-time incidence risk. Standardized definitions and prospective denominator-valid cohorts are needed. https://www.crd.york.ac.uk/PROSPERO/view/CRD420261460334 , identifier CRD420261460334.