Cardiac magnetic resonance imaging for diagnosis and prognosis of patients with myocarditis from immune checkpoint inhibitor therapy: a meta-analysis

Abstract Background and aims Myocarditis from immune checkpoint inhibitor (ICI) therapy is a challenge in clinical practice, and the assessment of ICI-related myocarditis (ICI-M) is often complicated by a variable phenotype. Cardiac magnetic resonance imaging (CMR) is used frequently, but evidence on CMR-based parameters for reliant diagnosis or prognostic assessment of patients with ICI-M is conflicting. This meta-analysis aims to determine whether CMR-derived parameters including late gadolinium enhancement (LGE) are suitable as independent parameters to predict ICI-M diagnosis or prognosis. Methods Scientific databases were searched for studies that assessed CMR parameters in adult patients with suspected myocarditis from immune checkpoint inhibitor therapy (ICI-M). Data from 8 trials with 661 patients were included. Results In this meta-analysis, the mean age of included patients was 64 years, 28% were female patients and 475 patients had ICI-M as per final diagnosis. LGE in CMR was associated with myocarditis following immune checkpoint inhibitor therapy odds ratio (OR) 4.35, 95% confidence interval (CI) 1.59-11.91; n = 601; p=0.004. The most common LGE pattern observed were mid-myocardial (43%) and sub-epicardial (38%) distributions. Myo- or pericardial oedema, despite being associated with vial-induced myocarditis, could not be associated as modality for reliant ICI-M diagnosis OR 0.18, 95% CI 0.07-0.43; n = 433; p=0.06. Further, the available evidence did not consistently indicate pericardial effusion as independent CMR-associated marker for diagnosis of ICI-M OR 1.19, 95% CI 0.03-47.23; n = 315; p=0.93. Presence of LGE showed a trend for prediction of major adverse cardiovascular events (MACE), but failed to statistical significance OR 2.86, 95% CI 0.87-9.40; n = 225; p=0.08. Despite a robust association, sensitivity and specificity of LGE for the diagnosis of ICI-M were only 62% and 64% respectively. Conclusion The diagnostic assessment of ICI-M is challenged by low sensitivity of CMR, often requiring a multimodal approach. Presence of LGE in CMR was associated with ICI-M, but sensitivity and specificity are low. These data indicate that a multimodal diagnostic approach is required for diagnosis and prognostic assessment of ICI-M.