Depolarization spatial variance as a cardiac dyssynchrony descriptor

Ventricular depolarization dispersion refers mainly to heterogeneity in interlead QRS durations. Here, we propose the spatial variance (SVd) to describe depolarization dispersion about a mean QRS morphology. We hypothesize that waveform changes can be more accurate than interval changes at measuring QRS heterogeneity, and less sensitive to delineation errors. To prove this, SVd was computed on 36 dyssynchrony patients either in sinus rhythm (SVBd) or under nHB (SVHBd), and on 32 control subjects with native conduction (SVNd). In the normal ECG, there are interlead sets that produce maximal (SVNmaxd) and minimal (SVNmind) spatial variance. In Baseline patients, SVBd significantly increased from controls in the minimal variance situation (p < 0.005), deviating one or more leads from the normal morphology (similar in all the ensemble) and consequently increasing the interlead distance. The opposite held for the maximal variance situation, where it was expected a wide span of morphologies in health, there was a tendency to stacking in morphologies at baseline (p<0.005). In both cases nHB induced QRS normalization, demonstrated by the return of SVHRd towards controls SVNd (p = NS). However, QRS narrowing not always accompanied morphological changes. The average rate of QRS normalization was 84% and 89% for SVNmaxd and SVNmind respectively while the rate of QRS narrowing equaled 71% in a multilead approach and performed worse than SVd at baseline-nHB separation. Ensembles with minimal interlead distance produced the best AUC values (aV(R)-V-1, I-aV(F)-V6,I-II-aV(F)-V-6). In conclusion, SVd may complement QRS width in the electrocardiographic assessment of dyssynchrony. (C) 2018 Elsevier Ltd. All rights reserved.