What is atrial fibrillation? A RETRO-spective analysis
Thomas Rostock, MD1, Alexander P. Benz, MD,
MSc1,2 and Raphael Spittler, MD,
MSc1
1 University Hospital Mainz, Center for Cardiology,
Department of Cardiology II / Electrophysiology, Mainz, Germany
2 Population Health Research Institute, McMaster
University, Hamilton, Ontario, Canada
Running title: RETRO mapping of AF
Word count: 1435
Corresponding author:
Thomas Rostock, MD
University Hospital Mainz
Center for Cardiology
Cardiology II / Electrophysiology
Langenbeckstr. 1
55131 Mainz, Germany
Email:
throstock@gmail.com
Disclosures: None declared.
Funding: None.
Keywords: Atrial fibrillation, mapping, mechanisms, wavefront
Ideally, before implementing any treatment as standard of care,
pathophysiological mechanisms need to be identified in order to
elaborate and define potential therapeutic targets. Philosophy of
science described the fundamental role of categories for components of
biological mechanisms: entities and activities. In this regard,
activities between different entities are considered to establish a
specifically organized biological phenomenon (1).
More than six decades ago, pioneer research on mechanisms of atrial
fibrillation (AF) by Moe and co-workers introduced the “multiple
wavelet hypothesis” (2). It was not until the mid 1980s that a group of
researchers led by Allessie performed high-density atrial mapping in the
canine heart and demonstrated multiple wandering wavelets with a
continuous beat-to-beat change in activation pattern (3). Based on this
concept, the surgical MAZE procedure developed by Cox et al. aimed at
creating anatomical barriers for multiple reentrant wavefronts by means
of surgical incisions transecting the atria into separated areas was
introduced (4). The identification of focal repetitive electrical
discharges arising from the pulmonary veins by Haissaguerre and
colleagues in the late 1990s ushered in a new era in the interventional
treatment of AF (5). This new pathophysiological concept not only
provided a fundamentally new target for the interventional treatment of
AF, it also shifted the focus of research from the perpetuating
substrate to initiating triggers. However, while patients with
“short-lived” episodes of AF typically respond well to pulmonary vein
isolation, efficacy of catheter ablation is limited in many patients
with persistent AF. This led researchers to re-appraise mechanisms of
the atrial arrhythmogenic substrate. Using computational mapping of AF
models, several mathematical methods for signal processing (e.g.,
dominant frequency, Hilbert transformation, etc.), optical mapping,
non-invasive panoramic mapping and rotational activity mapping were
investigated in the pre-clinical and clinical setting for
characterization of electrophysiological mechanisms during electrically
established AF (6). Interestingly, there is one specific key feature
collectively described: activation patterns of consecutive phases during
AF demonstrate spatiotemporal organization with varying stability. Thus,
accumulating data derived from contemporary AF mapping studies challenge
the concept of a global randomly meandering atrial activation causing
different phenotypes of AF.
In this issue of the Journal , Smith and colleagues (9) report
from a prospective, observational study on a novel, computationally
acquired and algorithm-based AF mapping technology developed in
conjunction with a proprietary software (MATLAB). This mapping tool has
been previously introduced by the same group and is called RETRO-Mapping
(7). RETRO-Mapping is a technological expansion of RIPPLE-Mapping, an
algorithm that is independent of time-window settings, also developed by
the group of Linton et al (8). The basic principle of RETRO-Mapping is
an analysis of cross-referenced electrode activation derived from a
multi-polar catheter in the surface-covered area during AF. The current
study was designed as a proof-of-concept investigation aimed at the
evaluation of patterns of AF activation. A 20-pole spiral double loop
catheter (AF Focus II, St. Jude Medical) was placed at the posterior
left atrium. After stabilization, local electrograms during AF
>30 seconds were obtained. Anatomical positions of the
catheters electrodes were visualized with the NavX Ensite X system and
exported into the RETRO-Mapping software along with their corresponding
electrophysiological data. The algorithm calculated three parameter of
wavefront propagation: cycle length, conduction velocity, and wavefront
direction. The midpoint between two adjacent electrodes of the loop
catheter corresponds to the position of bipolar electrograms. Through a
process called triangulation, midpoints were used to define corners for
the segmentation of the left atrial posterior wall into triangles. The
idea of triangulation of the covered area is to obtain the most
appropriate position of two adjacent midpoints for further comparison of
two local electrograms connected by the edge of a triangle. The
geometric and activation properties of connected electrograms allowed
for further analysis used for calculation of cycle length. A
dichotomization of tissue excitability (depolarized myocardial tissue
was coded as 1, and repolarized tissue was coded as 0) was used to
create binary images. The Euclidean distance transform algorithm
calculated the distance of pixels with a value of 0 to the nearest pixel
with a value of 1. Incorporating distance information and the sampling
frequency of the mapping algorithm as a marker of time, conduction
velocity is calculated. Smith and colleagues used the Sobel operator for
image processing to analyze wavefront direction. This mathematical
filter calculates the gradient of image intensity at each point. These
data are then used to detect areas where the intensity changes rapidly,
which helps to identify edges of an image. The activation edge direction
was derived from these edges and was used to calculate the overall
wavefront direction. The researchers were interested in (1) whether an
advancing wavefront maintained its direction and (2) whether the plane
wavefront would be predictive of the subsequent wavefront direction.
In eight patients with three types of AF (paroxysmal, persistent without
amiodarone and persistent under amiodarone treatment), more than 34.000
activation edges of plane wavefronts (i.e., linear wavefront activations
of the mapping field) were detected. With increasing intervals between
new wavefront activations, the difference of wavefront directions also
increased in a linear relationship. Smaller direction changes were
observed in patients with persistent AF with amiodarone as compared to
those with persistent AF without amiodarone. However, the greatest
variability in activation changes was observed in paroxysmal AF.
Activation direction consistency, indicated by small variations between
wavefront directions, was highest in persistent AF with amiodarone and
lowest in paroxysmal AF. More than half of all wavefront activation
patterns predicted the pattern of the subsequent wavefront. In this
context, the highest predictive value of activation patterns for
subsequent wavefronts was found in persistent AF without amiodarone (the
AF type with the shortest cycle length).
With the presented study, Smith and colleagues shed some light on the
complexities surrounding AF activation patterns. Gleaned from human
high-density contact mapping of electrophysiological features during AF,
data were processed with a dedicated software in order to transform AF
activation patterns into a color-coded two-dimensional model. This novel
software package may provide real-time calculation of conduction
velocity and cycle length during a mapping/ablation procedure and could
potentially be incorporated into existing 3-dimensional mapping systems.
The authors are to be commended for providing a potential solution for
processing electrophysiological activation data of AF with the potential
to include this information into the ablation procedure. Furthermore,
the data of this study reinforce the paradigm that AF perpetuation
follows an organized and repetitive activation pattern (10).
While the study has a clear scientific basis, there are some practical
issues to be considered. First, the current technology is limited by the
apparent inapplicability to fibrotic atrial tissue which may represent
an anatomical substrate in patients with more advanced types of
persistent AF. Second, the algorithm characterized paroxysmal AF by
features indicating more “disorganized” patterns as compared to the
persistent AF types. The reasons for this observation currently remain
unclear since paroxysmal AF with focal triggering (and maintaining)
sources predominantly located within the pulmonary veins (and not the
atria) and longer AF cycle lengths appears to represent a more organized
type of AF. Third, with the current type of visualization of activation
patterns, focally activated wavefronts cannot be distinguished from
rotor activity or even passively activated areas. Fourth, only a small
area covered by the double-looped mapping catheter can be mapped at a
given time. However, even with larger high-density contact mapping
catheters (e.g., the basket catheter), it still remains impossible to
simultaneously cover the entire surface of the left atrium. In this
context, the data of the presented study also may help to overcome this
limitation since they revealed that a preceding wavefront predicted the
subsequent activation pattern, potentially indicating similar wavefront
configurations in neighboring areas. Sixth, calculation models used in
this study are evaluated with a limited number of electrodes. Thus, it
needs to be determined if larger areas represented by considerably more
electrodes (and data) can undergo computational analysis in the same way
(and in real-time). Finally, further and much larger studies are needed
to determine how best to incorporate information on AF mechanism and
visualization of atrial activation during AF in the clinical setting.
Thus, next steps of research may include characterization of appropriate
targets for the interventional treatment.
At the end of the day, the clinical electrophysiological community
should welcome efforts aimed at improving our understanding of the
underlying mechanisms of particularly non-pulmonary vein dependent AF.
This seems critical in order to define a tailored approach to the
individual patient. The alternative is a continued focus on
“empirical” ablation without fully appreciating the underlying
pathophysiology. For now, durable pulmonary vein isolation is the only
pathophysiologically established treatment in AF ablation, but hopefully
will not remain the only in the future.