Characterization of the causal interactions between depolarization and repolarization temporal changes in unipolar electrograms

The causes of beat-to-beat cardiac repolarization variability (RV), a marker of electrical instability associated with increased risk of sudden cardiac death, are undetermined. An issue which is often overlooked is whether RV is entirely due to repolarization mechanisms or whether it is partially due to beat-to-beat depolarization variability (DV). To address this issue we propose a methodology to reveal the causal interactions between DV and RV, estimated from unipolar electrograms (EGMs). The methodology is based on the comparison between the coefficients of two autoregressive bivariate models: one describes the actual variabilities, while the other represents the variabilities of surrogate time-series in which directional coupling is selectively destroyed. A simulation study which involves synthetic EGMs generated by using a simplified biophysical model shows that the methodology is accurate in typical conditions. Data from high density, multielectrode, cardiac mapping of the in-vivo human heart recorded in one cardiac patient show that DV drove RV in about 28% of electrodes, suggesting that DV may contribute to RV.