Introduction: Decades of clinical trials have not yet provided satisfactory approaches for the treatment of atrial fibrillation (AF) beyond pulmonary vein isolation. Patients presenting with recurrent AF emphasize the need for identification of arrhythmic substrate as potential target for ablation. Due to the high and variable activation frequency during AF, the rate dependency of substrate properties is suspected to play a significant role in the maintenance of AF. We measured and quantified biatrial conduction velocity (CV) restitution in order to reveal intra- and inter-patient variability of CV slowing in response to increased activation frequency.
Methods: Patients underwent simultaneous biatrial electrogram acquisition with either two CONSTELLATION basket catheters or one CONSTELLATION and one IntellaMap Orion mini basket catheter. A decremental stimulation sequence was driven from the coronary sinus starting from a pacing cycle length (PCL) of 600ms gradually decreasing until the onset of AF. Unipolar electrograms were preprocessed by adequate filtering and ventricular blanking before detecting atrial activity. Post pacing intervals (PPIs) were then measured as the time interval between stimulation and local atrial activation as a proportional surrogate for reciprocal CV. Finally, restitution curves were parametrized by an exponential function PPI(PCL)= PPI_0/(1-p∙exp(-k∙PCL)) and non-linear constrained least squares optimization. PPI_0 describes the baseline PPI for long PCLs; p and k are the exponential coefficients.
Results: Curve fitting for 13 patients with a total of 1196 valid recording sites revealed a narrow distribution of k with a 5% and 95% quantile at 0.007/ms and 0.041/ms, respectively, and an expected value of 0.02/ms. The coefficient p followed a decaying distribution with 50% of all estimates satisfying 0 < p < 9.
Conclusion: Biatrial CV restitution showed significant intra- and inter-patient variability. Presented insights improve the understanding of AF arrhythmogenesis and provide a tool to individually parametrize patient-specific computational models.