Immune checkpoint inhibitor-associated aortitis and classical large-vessel vasculitis: Clinical characterization and...

Immune checkpoint inhibitors (ICIs) may induce rare immune-related vascular toxicities, but evidence on ICI-associated aortitis and classical large-vessel vasculitis remains limited. This study characterized clinical features and disproportionality signals using the FDA Adverse Event Reporting System (FAERS).

FAERS reports from 2011 Q1 to 2025 Q3 were retrospectively analyzed. Cases were identified using the MedDRA Preferred Terms "Aortitis," "Giant cell arteritis," and "Takayasu's arteritis." ICI exposure was defined as an ICI recorded as the primary or secondary suspect drug. Clinical characteristics were summarized descriptively. Disproportionality was assessed using reporting odds ratio (ROR), proportional reporting ratio, information component, and empirical Bayesian geometric mean, with subgroup analyses by ICI class, individual agent, regimen, and event subtype.

We identified 83 reports of ICI-associated aortitis and classical large-vessel vasculitis. The median age was 68 years, and 57.8% of patients were male. Hospitalization and death were reported in 34.9% and 2.4% of cases, respectively. Among reports with available onset data, the median time to onset was 105.5 days. ICIs showed a positive disproportionality signal overall (ROR, 4.45; 95% CI, 3.57-5.55). Signals were most evident for anti-PD-1 agents (ROR, 4.32; 95% CI, 3.27-5.71), nivolumab, pembrolizumab, ipilimumab, and ipilimumab plus nivolumab. At the event-subtype level, aortitis and giant cell arteritis showed positive signals, whereas Takayasu's arteritis was rarely reported.

This FAERS analysis identified disproportionate reporting signals for ICI-associated aortitis and classical large-vessel vasculitis. Clinicians should consider this rare but potentially serious toxicity in patients with unexplained inflammatory symptoms or abnormal vascular imaging during ICI therapy. These findings are hypothesis-generating and require validation, and they cannot be used to infer incidence or causality.