Three-Dimensional Computer Modelling of Endo-Epicardial Dissociation and Transmural Conduction in Atrial Fibrillation
In this interview, we speak with Ulrich Schotten, MD, PhD, from the Department of Physiology, Cardiovascular Research Institute Maastricht, in Maastricht, The Netherlands.
Describe what causes endo-epicardial dissociation of electrical activity, and how it can lead to breakthrough waves affecting fibrillatory conduction and atrial fibrillation (AF) stability.
The steady and progressive process of structural remodeling due to structural heart disease or AF in itself leads to an increasing degree of electrical uncoupling between muscle bundles. This uncoupling occurs particularly within the thin subepicardial layer of the atrial wall, but also between the subepicardial layer and the endocardial bundle network. The increasing degree of electrical uncoupling leads to dyssynchrony in the electrical activation between the layers. These dyssynchronies can be so pronounced that the subepicardial layer, endocardial bundle network, and possibly additional layers of the atrial wall can be activated by separate fibrillation waves. In most cases, the electrical uncoupling between the layers is not complete, allowing for transmural conduction of fibrillation waves. The waves that propagate in a transmural way from one layer to the other appear on the surface of the atrial wall as breakthroughs. Approximately 30% of all fibrillation waves are of this kind, implying that transmural conduction significantly contributes to the increasing stability of AF in structurally remodeled atria.
Why is it important to study the role of fibrosis?
The role of atrial fibrosis for the occurrence of conduction disturbances was demonstrated many years ago; however, the spatial distribution of atrial fibrosis has only more recently been studied. Recent histological studies by our group show that AF-related atrial fibrosis primarily occurs in the subepicardial layer of the atrial wall. As a consequence, slowing of conduction and conduction block may preferentially occur in this layer, resulting in endo-epicardial dissociation of electrical activity.
Tell us about the development and design of an anatomical computer model of the atria to study human AF.
The unique feature of this model is that it contains very realistic 3D anatomy of the atrial wall. In large parts of the atria, it contains several layers or — where appropriate — a subepicardial layer and an endocardial bundle network allowing for simulation of wavefronts that propagate either in the endocardial bundles or in the epicardial layer. The gross anatomy of the atria was taken from an MRI of a healthy male volunteer; however, many detailed anatomical features from anatomical studies were added manually to the model, including pectinate muscles in the right atrial free wall as well as Bachmann’s bundle, interatrial bundles, the crista terminalis, and a LAA trabecular network. Furthermore, one to three layers of fiber orientations were added to the model based on several anatomical studies. Comparison with results of a recently published sub-millimeter diffusion tensor magnetic resonance imaging study showed excellent agreement of the prevailing local fibre orientations in our model with human atria.
How is creation of this computer model for AF significant?
Thirty percent of all simulation waves are breakthroughs, and this model is the first that simulates the occurrence of breakthrough in a quantitatively realistic way. We recently validated the model extensively against simultaneous endo-epicardial high-resolution, direct-contact mapping in patients with acutely induced or persistent AF. All main electrophysiological characteristics such as refractory period, conduction velocity, AF cycle length, and — most notably — degree of endo-epicardial dissociation and the typical correlation between breakthrough rate and degree of endo-epicardial dissociation, were very similar in the model as compared to the simultaneous endo-epicardial direct-contact mapping recordings. Thus, our model can be seen as one further step towards simulation of realistic fibrillation patterns during AF.
In the computational model of AF, how was fibrosis simulated and compared?
Fibrosis was implemented by reducing the transverse conductance of the subepicardial units to a variable extent. We took great care to mimic the localization and distribution of fibrosis of our histological studies.
How was the number of fibrillation waves and breakthroughs in both the endocardial and epicardial layers determined?
We quantified the fibrillation waves using the classical wave map procedure. Local activation times were determined from the fibrillation electrograms using a probabilistic activation annotation algorithm, and from these activation times, lines of conduction block surrounding fibrillation waves were determined.
Describe how endo-epicardial dissociation was calculated in simulations.
Essentially the same method was used as in the clinical data set. We calculated histograms of endo-epicardial activation time differences. These histograms usually show a narrow peak around 0 ms. These activations represent waves that propagate (nearly) simultaneously in the subepicardial layer and the endocardial bundle network. The histogram also shows a component with far larger endo-epicardial activation time differences. This component represents waves in the two layers of the atrial wall that are not related to each other and therefore are called “dissociated.” The percentage of the area under the curve of this component was taken as the degree of endo-epicardial electrical dissociation.
Discuss your methods for performing simultaneous endo-epicardial mapping of AF in the right atrium. How many patients were evaluated in this study?
We conducted a small proof-of-principle study with electrode arrays mounted on a clamp-like mapping electrode. One arm of the clamp could be introduced into the cavity of the right atrium, while the other was placed on the epicardial surface.
Where did most breakthroughs occur, and why?
Breakthroughs occurred throughout the entire mapping area. We could not identify any preferential sites of breakthrough occurrence. In our previous studies with larger electrodes, we found that in atria with mild structural remodeling, there appears to be some degree of clustering at specific sites; however, the more pronounced the structural remodeling process, the more widespread and uniform the distribution of breakthroughs becomes.
During recording and simulation, what was the degree of endo-epicardial dissociation in patients with persistent AF? What effect did fibrosis have on the degree of endo-epicardial dissociation?
The degree of endo-epicardial dissociation was between 20 and 60% in both the patients and simulations. In the simulations, we could show that with more fibrosis, the degree of endo-epicardial dissociation increased.
Tell us about the correlation between endo-epicardial dissociation and the incidence of breakthroughs in both recordings and simulations.
In both cases, the correlation was highly significant. This finding strongly supports the hypothesis that endo-epicardial dissociation is the main driving force of transmural conduction and breakthrough during AF.
How many breakthroughs per cycle actually occur during AF in patients?
Assuming that thirty percent of all fibrillation waves conduct from one layer of the atrial wall to the other, and extrapolating the number of waves per cycle we recorded to the entire atrial surface, there are likely dozens of breakthroughs during each AF cycle.
What are the key take-home points of this research?
AF is a three-dimensional process. Breakthroughs make up to 30% of all fibrillation waves. Atrial fibrosis, preferentially occurring in the subepicardial layer, is sufficient to explain endo-epicardial dissociation. Finally, the strong correlation between endo-epicardial dissociation and breakthrough suggests that transmural conduction, and not ectopic discharges, is the main explanation for fibrillation waves with radial spread of activation.
Why is it important to understand the 3D substrate of AF? How do these findings on the mechanisms of AF add to our understanding of the variety of conduction patterns present during AF?
Our study adds a lot to the knowledge of the diversity of fibrillation wavefront shapes and organization. It demonstrates that all simulations as well as imaging techniques neglecting transmural conduction do not take a quantitatively significant part of the fibrillation waves into account. This may have important implications for our understanding of the mechanisms driving AF, including the mode of action of antiarrhythmic compounds.
What is the potential impact of these findings on clinical practice?
Hopefully, our study can contribute to the acceptance of a high complexity and diversity in the mechanisms that drive AF. Theories that explain AF drivers based on two-dimensional mechanisms only likely miss crucial elements of AF perpetuation.
Will future research in this area include a larger cohort of AF patients?
The 3D nature of AF certainly contributes to the overall complexity of AF. Larger studies are certainly required to identify the main structural and mechanistic determinants of AF complexity. Whether this always requires the measurement of endo-epicardial dissociation has to be seen. So far, we found a very strong correlation between the number of fibrillation waves or the incidence of conduction block during AF and the degree of endo-epicardial dissociation.
Do you plan to cover more information on this at the upcoming AF Symposium in 2018?
Yes, I will be showing some more details of the fibrillation pattern in both clinical recordings and in the model.