Current evidence on the diagnosis and management of immune checkpoint inhibitor associated myocarditis

Immune checkpoint inhibitor-associated myocarditis (ICI-M) is a rare but life-threatening immune-related adverse event. Reported mortality ranges from 27 to 50% in earlier cohorts, with a more recent pooled estimate of approximately 26% in systematic reviews incorporating 2021 to 2022 data. Incidence is approximately 1%, rising to 2 to 4% with combination regimens. We performed a narrative review of the epidemiology, pathophysiology, overlap syndromes, risk stratification, diagnostic criteria, and immunosuppressive management of ICI-M. Diagnosis integrates an obligate rise in high-sensitivity troponin with multimodal imaging, framed by the International Cardio-Oncology Society (IC-OS) 2021 criteria and the 2024 American College of Cardiology (ACC) Consensus Pathway, after exclusion of acute coronary syndrome and infectious myocarditis. Risk stratification is discussed using a recently reported multivariable risk score. High-dose intravenous methylprednisolone is the accepted first-line therapy, with observational data associating higher initial dosing with more favourable biomarkers and survival trends. Up to half of severe cases fail first-line therapy and require escalation. Evidence for second-line agents, including mycophenolate mofetil, tacrolimus, abatacept, anti-thymocyte globulin, infliximab, tocilizumab, and ruxolitinib, derives from case series and small cohorts rather than randomised trials, and is presented with that limitation stated. ICI-M requires a structured, multidisciplinary approach integrating early recognition, systematic risk stratification, prompt high-dose immunosuppression, and a structured escalation approach for steroid-refractory disease, recognising that second-line evidence remains limited. This review synthesises the current, largely observational evidence to support individualised multidisciplinary management, and distinguishes guideline-endorsed recommendations from areas resting on limited or emerging data.