According to current guidelines, five types of atrial fibrillation (AF) are distinguished: (1) first diagnosed, (2) paroxysmal, (3) persistent, (4) long persistent and (5) permanent. This is a symptom-based clinical classification. Recent studies indicate, however, that the transition between the classes is ill-defined and insufficiently correlated with the actual burden of AF and, thus, the progress of the underlying AF disease process. Nevertheless, this classification is used in clinical practice, such as for individual therapy planning (rhythm vs. rate control). Since the success rates of interventional or medical therapy are depending on the individual disease progression stage, a classification is required which better correlates with the severity of electrophysiological remodeling to improve the currently unsatisfactory success rates of AF treatment.

In the context of the previous grant (SE 1758/3-1, SCHO 1350/2-1), techniques for an electrophysiological characterization of tissue properties and arrhythmia mechanisms have been developed. On the one hand, computational models have been adjusted to reproduce genetic mutations in silico using experimental data. By this, patho-mechanisms leading to AF have been identified. On the other hand, we showed that the mode of action and the efficacy of different antiarrhythmic agents differ significantly between mutation and control models.

The developed techniques will now be used and enhanced in this project to perform an individualized characterization of the heterogeneous tissue substrate of AF. For this purpose, the models are developed further using a new method for invasive detection, in patients, of the effective refractory period as well as the heart rate dependence (restitution) of signal amplitude (voltage) and conduction velocity. Both individual and patient group-specific models will be generated from the atrial measurement data of up to 80 patients with mostly persistant AF. The arrhythmic potential (preclinical arrhythmia markers) will then be determined numerically in the models by simulations for the different groups. Through the identification of clusters in the data, a novel classification will be developed to better reflect the individual disease progression. Optimal medical or interventional therapy for AF will then be numerically calculated for the various stages. Most promising results will be evaluated in first experimental studies. By correlating the stages with additional, non-invasively measured parameters on each patient (e.g. ECG P-wave shape and ultrasound measures), an improved non-invasive, individual therapy recommendation should be possible in the long run. Thereby, success rates of AF cardioversion will be increased, combined with a more favorable relation between the benefits and the risks of therapy concepts.