Cardiovascular magnetic resonance derived risk predictors for arrhythmic events in confirmed myocarditis: a systematic review and meta-analysis

Background Although acute myocarditis is usually associated with a benign course, a subset of patients will develop ventricular arrhythmias (VA) or sudden cardiac death (SCD). Left ventricular ejection fraction (LVEF) and late gadolinium enhancement (LGE) on cardiovascular magnetic resonance (CMR) are associated with poor outcome. However,current knowledge with regards to risk prediction is hampered by inclusion of patients with suspected myocarditis and the use of combined endpoints, including heart failure sequelae, not necessarily relevant for guiding implantable cardioverter defibrillators (ICD) implantation. Purpose We aimed to summarize current knowledge and investigate the relationship between CMR-derived parameters and VA/SCD in confirmed myocarditis. Methods A systematic review with meta-analysis was conducted and reported following the PRISMA guidelines. Cohort studies and randomized controlled trials, describing associations of different risk predictors with arrhythmic events (AEs; VA or SCD) in patients with endomyocardial biopsy (EMB; Dallas or Marburg criteria) or CMR (original or revised Lake Louise criteria) confirmed myocarditis, with a mean time of follow-up > 6 months were included. Risk predictors of interest were LVEF, left ventricular end diastolic volume (LVEDV), and the presence, pattern, and extent of LGE, although other described demographic and clinical predictors were considered. Hazard ratios derived from univariate and multivariate analysis from individual studies were pooled using a random-effects model and visualized together with their 95% CIs in forest plots. Rating of evidence was performed using the GRADE tool for rating in evidence and recommendations. Results A total of 6165 records were screened and 11 studies were included (n=1591, age 43±7 years, 75% male, LVEF 50±9%). Anteroseptal LGE (HR 7.52, 95%CI 2.19-25.88, I2=32.2%) was the only predictor significantly associated with AEs, while LVEDV and LGE presence were not (see Figure 1 and 2). Pooled analysis showed that LVEF per se was not associated with AEs, although reduced LVEF was a predictor for outcome in two studies. LGE extent appeared to be a possible predictor, although pooled analysis was hampered by non-uniformity in definition. Other possible new predictors were CMR derived left ventricular circumferential strain and left ventricular intramyocardial fat. Conclusions Anteroseptal LGE is associated with AEs in patients with confirmed AM. Despite the low level of certainty of recommendations, this is the first systematic review assessing risk predictors in patients with EMB and CMR confirmed AM, providing the most robust evidence to date for risk stratification in these patients.