Entropy of cardiac repolarization predicts ventricular arrhythmias and mortality in patients receiving an implantable cardioverter-defibrillator for primary prevention of sudden death

Aims: The need for a readily available, inexpensive, non-invasive method for improved risk stratification of heart failure (HF) patients is paramount. Prior studies have proposed that distinct fluctuation patterns underlying the variability of physiological signals have unique prognostic value. We tested this hypothesis in an extensively phenotyped cohort of HF patients using EntropyX_QT, a novel non-linear measure of cardiac repolarization dynamics. Methods and results: In a prospective, multicentre, observational study of 852 patients in sinus rhythm undergoing clinically indicated primary prevention implantable cardioverter-defibrillator (ICD) implantation (2003-10), exposures included demographics, history, physical examination, medications, laboratory results, serum biomarkers, ejection fraction, conventional electrocardiographic (ECG) analyses of heart rate and QT variability, and EntropyX_QT. The primary outcome was first 'appropriate' ICD shock for ventricular arrhythmias. The secondary outcome was composite events (appropriate ICD shock and all-cause mortality). After exclusions, the cohort (n = 816) had a mean age of 60±13 years, 28% women, 36% African Americans, 56% ischaemic cardiomyopathy, and 29±16% Seattle HF risk score (SHFS) 5-year predicted mortality. Over 45±24 months, there were 134 appropriate shocks and 166 deaths. After adjusting for 30 exposures, the hazard ratios (comparing the 5th to 1st quintile of EntropyX_QT) for primary and secondary outcomes were 3.29 (95% CI 1.74-6.21) and 2.28 (1.53-3.41), respectively. Addition of EntropyX_QT to a model comprised of the exposures or SHFS significantly increased net reclassification and the ROC curve area. Conclusions: EntropyX_QT measured during ICD implantation strongly and independently predicts appropriate shock and all-cause mortality over follow-up. EntropyX_QT complements conventional risk predictors and has the potential for broad clinical application.