Abstract
INTRODUCTION: Cardiac sarcoidosis (CS) is associated with a significant risk of ventricular arrhythmias (VAs). Recently, late gadolinium enhancement (LGE) phenotypes have been described; the 'pathology-frequent' phenotype has been shown to be strongly associated with future VA risk. The aim of our study was to compare the relative predictability of quantitative LGE burden and LGE phenotype. Secondary aims were to contrast the reproducibility and speed of LGE burden quantification and LGE phenotyping. METHODS AND RESULTS: This is a two-centre sub-study of the Cardiac Sarcoidosis Multi-Center Prospective Cohort Study (CHASM-CS; NCT01477359). All patients underwent CMR at baseline. A total of 206 patients (112/206 (54.4%) with CS and 94/206 (45.6%) with extra-cardiac sarcoidosis) were included in the study. Pathology-frequent LGE phenotype occurred in 85/206 (41.3%) and 22/206 (10.7%) patients had sustained VA during a mean follow-up period of 5.1 ± 2.8 years. All events occurred in patients with a pathology-frequent phenotype. LGE phenotype and categorical LGE% had similar high discriminative accuracy in predicting future VA; however, only LGE phenotyping had 100% negative predictive value (NPV). Interobserver reproducibility of LGE phenotype was very high (Cohen's κ = 0.97, P < 0.001) and much better than LGE categorical quantification (Cohen's κ = 0.41, P = 0.04). LGE phenotyping was on average five-fold faster (1.75 ± 0.9 min compared to 10 ± 2.2 min, P = 0.012). CONCLUSION: Our study confirms, for the first time in a fully prospective cohort, the prognostic importance of the pathology-frequent phenotype of LGE. This phenotype had 100% NPV, indicating the absence of VA events among patients without a pathology frequent phenotype. LGE phenotyping showed much higher interobserver reproducibility than LGE quantification and was five-fold faster.