Adaptation of cardiac action potential durations to stimulation history with random diastolic intervals

The restitution hypothesis proposes that adaptation of cardiac action potential duration (APD) to rate changes is a predictor of ventricular fibrillation (VF). Conventional restitution kinetics plots the APD of a premature beat as a function of the previous diastolic interval (DI), and VF vulnerability is related to how rapidly APD shortens with decreasing DI. However, APD depends not only on the previous DI but also on the history of previous APDs and DIs. For a comprehensive understanding of APD restitution, we developed a random stimulation protocol and curve fitted each APD with the previous DIs and APDs using multiple autoregressive analyses.

Guinea pig hearts (n = 5) were perfused and stained with di-4 ANEPPS to record optical APs from 252 sites. Activation and repolarization times were detected in real time from one pixel and hearts were stimulated at random DIs (range 0-50 or 0-100 ms). We found that the first, second, and third previous APDs and DIs are required to obtain the best curve fit, which provides the most significant feedback control to APD and up to six previous beats contributed to curve fits (R > 0.8). The coefficients relating the previous DI to APD increased systematically in going from apex to base reflecting the intrinsic gradient of APD across the epicardium.

Random restitution is more comprehensive than steady-state restitution, being based on random and dynamic DIs, and makes possible characterization of restitution in only 32 seconds to track changes in restitution during time-varying conditions such as ischemia/reperfusion.