Population-Based Frequency of Ophthalmic Adverse Events in Melanoma, Other Cancers, and After Immune Checkpoint Inhibitor Treatment

Abstract

To examine the frequency of ophthalmic immune-related adverse events (OirAEs) in melanoma, other cancers, and after immune checkpoint inhibitor (ICI) treatment. Retrospective clinical cohort study. This study identified patients diagnosed with OirAEs between January 1, 2011, and December 31, 2018, in the Kaiser Permanente Southern California electronic health records. The primary exposures of interest were prior initiation of ICIs and underlying cancer diagnosis. Risk-adjusted prevalence of OirAEs was evaluated in patients with melanoma, with nonmelanoma cancer, and without cancer. The 1-year incidence of OirAEs and recurrence of prior ophthalmic disease were identified in ICI-receiving patients with melanoma and nonmelanoma. Among 4,695,669 unique patients identified, 9.9% had a cancer diagnosis, of whom 2.8% had a diagnosis of melanoma. Overall prevalence for uveitis and selected neuro-ophthalmic diagnoses was 341.8/100,000 patient-years in patients with melanoma and 369.6/100,000 patient-years in patients with nonmelanoma cancer regardless of ICI treatment, compared with 142.2/100,000 patient-years in patients without cancer. A total of 2,911 unique patients received ICI therapy. Compared with patients with nonmelanoma cancer, patients with melanoma on any ICI had elevated 1-year incidence rates of uveitis (1.2% vs 0.2%; risk-adjusted odds ratio, 6.45). High 1-year recurrence rates for uveitis in ICI patients with a prior uveitis history were also observed. The prevalence of all OirAEs was substantially higher in patients with cancer, with ICI-related uveitis risk specifically increased in patients with melanoma compared with patients with nonmelanoma cancer. Evidence-based guidelines for ophthalmic monitoring of patients undergoing ICI treatment may require different risk stratifications based on underlying cancer diagnosis, specific ICI used, and prior history of uveitis.