Abstract
BACKGROUND: Cardiovascular magnetic resonance (CMR) is the test of choice for diagnosis and risk stratification of myocardial inflammation in acute viral myocarditis. The objective of this study was to assess patterns of CMR inflammation in a cohort of acute myocarditis patients from Northern Africa, Asia, and the Middle East using unsupervised machine learning. METHODS: A total of 169 racially and ethnically diverse adults ( ≥ 18 years of age) with CMR confirmed acute myocarditis were studied. The primary outcome was a combined clinical endpoint of cardiac death, arrhythmia, and dilated cardiomyopathy. Machine learning was used for exploratory analysis to identify patterns of CMR inflammation. RESULTS: Our cohort was diverse with 25% from Northern Africa, 33% from Southern Asia, and 28% from Western Asia/the Middle East. Twelve patients met the combined clinical endpoint - 3 had arrythmia, 8 had dilated cardiomyopathy, and 1 died. Patients who met the combined endpoint had increased anterior (p = 0.034) and septal (p = 0.042) late gadolinium enhancement (LGE). Multivariable logistic regression, adjusted for age, gender, and BMI, found that patients from Southern Asia (p = 0.041) and the Middle East (p = 0.043) were independently associated with lateral LGE. Unsupervised machine learning and factor analysis identified two distinct CMR patterns of inflammation, one with increased LGE and the other with increased myocardial T1/T2. CONCLUSIONS: We found that anteroseptal inflammation is associated with worsened outcomes. Using machine learning, we identified two patterns of myocardial inflammation in acute myocarditis from CMR in a racially and ethnically diverse group of patients from Southern Asia, Northern Africa, and the Middle East.