Session S64.4

Dynamically-Induced Spatial Dispersion of Repolarization and the Development of Conduction Block and VF in an Animal Model of Sudden Death

ARM Gelzer, NF Otani, ML Koller, MW Enyeart,
NS Moise, RF Gilmour Jr*

Cornell University
Ithaca, NY, USA

Both intrinsic and dynamic heterogeneity of refractoriness have been implicated in causing the wavebreak that initiates ventricular fibrillation (VF). Based on a 1-dimensional computer model, we developed a stimulation protocol that maximizes dynamic heterogeneity and precipitates conduction block. We showed previously that in normal beagle dogs (B) these stimulation protocols (CLVF-B) predictably induced VF. In this study we determined whether this protocol induces wavebreak and VF in German shepherd (GS) dogs having an inherited predisposition to the development of ventricular tachycardia and sudden death. Despite having >6000 PVCs per 24 hour Holter, only about 15% of GS develop VF, the reason for this rare occurrence being unknown.
Methods: Monophasic action potential duration (MAPD) was measured in the right (RV) and left ventricular (LV) endocardium during stimulation from the RV and LV in anesthetized, closed-chest affected GS (n=6). MAPD restitution for each individual ventricle was determined using a dynamic restitution protocol. The computer model used these restitution relations to calculate sequences of 4 appropriately timed premature stimuli (CLVF-GS) predicted to produce block. The LV (5 trials) and RV (6 trials) were paced using CLVF-B and CLVF-GS following a train of 20 stimuli at a constant S1S1 interval (400 ms). Percentage of stimulus combination categories that did or did not produce VF in the experiment were calculated relative to whether they were predicted to yield block using a chi-square test.
Results: CLVF-B induced VF in 4 of 12 induction trials. Of the 8 failed trials CLVF-GS predicted the inability to induce VF in 6, while of the 4 successful trials CLVF-GS predicted successful VF induction in 3. Linear regression analysis of the percentage of pacing combinations that induced VF vs. likelihood of action potential block predicted by the theory yielded a significantly positive slope (0.64 ± 0.12 (SEM)). The distribution itself was also highly significant (p=4.2 x 10-7), suggesting a strong positive correlation between theoretical predictions and experimental results.
Conclusions: CLVF based on GS APD restitution was a better predictor of VF induction than CLVF based on beagle APD restitution. The agreement between theory and experiment demonstrates that APD restitution in different animal models may be used to generate different predictions regarding sequences of cycle lengths that maximize dynamic heterogeneity and induce VF. These findings may suggest new approaches for establishing risk for the development of VF.

(Abstract Control Number: 130)