Noninvasive M apping to G u i d e At r i a l F i b r i l l a t i o n Ablation Han S. Lim, MBBS, PhD, Stephan Zellerhoff, MD, Nicolas Derval, MD, Arnaud Denis, MD, Seigo Yamashita, MD, Benjamin Berte, MD, Saagar Mahida, MBChB, Darren Hooks, MBChB, Nora Aljefairi, MD, Ashok J. Shah, MD, Frédéric Sacher, MD, Meleze Hocini, MD, Pierre Jais, MD, Michel Haissaguerre, MD* KEYWORDS Noninvasive mapping Atrial fibrillation Catheter ablation Localized drivers
KEY POINTS Noninvasive mapping overcomes previous obstacles to panoramic atrial fibrillation (AF) mapping and enables biatrial mapping of the underlying mechanisms in AF. Localized reentrant and focal drivers are identified with the use of noninvasive mapping and phase mapping algorithms in patients with AF. A strategy of driver-based ablation guided by noninvasive mapping yields promising results and minimizes the extent of ablation. Future studies will enhance the role of noninvasive mapping to direct personalized therapy to patients with AF.
Atrial fibrillation (AF) is traditionally thought to be maintained by multiple wave fronts that propagate randomly throughout the atria.1 In paroxysmal AF, the realization of the role of triggers, mainly from the pulmonary veins (PVs), initiating AF has led to the elimination of PV triggers as the cornerstone of AF ablation therapy.2,3 However, in persistent AF, the elimination of PV triggers alone does not yield similar results, largely because of wider substrate involvement contributing to the maintenance of the arrhythmia. Various mechanisms, such as multiple wavelets, reentry, and focal sources, have been proposed to maintain persistent AF.1,2,4,5
Accumulating evidence now suggests that persistent AF may be driven by localized reentrant and focal sources in the atria.6–10 However, these localized AF drivers prove difficult to detect, because of AF being a dynamic rhythm and the potential spatiotemporal dynamicity of the underlying sources. Previous attempts to map these localized drivers using endocardial single-point or multielectrode catheters and epicardial surgical plaques have been restricted by the inability to simultaneously map different points in the left and right atria, contact issues, and limited surgical access.7,10,11 Noninvasive mapping overcomes these difficulties by enabling panoramic beatto-beat mapping of this dynamic rhythm.9,12 Initial
Disclosure: Drs Haissaguerre, Jais and Hocini are stockholders in and Dr Shah is a paid consultant to CardioInsight Inc., Cleveland, OH, USA. This work was supported through the Investment of the Future grant, ANR-10-IAHU-04, from the government of France through the Agence National de la Recherche. IHU LIRYC, Electrophysiology and Heart Modeling Institute, Fondation Bordeaux Universite´, Bordeaux, France * Corresponding author. E-mail address: [email protected]
Card Electrophysiol Clin 7 (2015) 89–98 http://dx.doi.org/10.1016/j.ccep.2014.11.004 1877-9182/15/$ – see front matter Ó 2015 Elsevier Inc. All rights reserved.
Lim et al studies with ablation guided by noninvasive mapping to identify and target these localized sources have yielded promising results.8,9,12 This article discusses the role of noninvasive mapping to guide AF ablation therapy, including the specific methodology of noninvasive mapping to map AF drivers, current experience and clinical results, and future areas of investigation.
METHODOLOGY OF NONINVASIVE MAPPING TO DETECT ATRIAL FIBRILLATION DRIVERS The use of noninvasive mapping in the atria has been validated in numerous studies. This technique includes noninvasive mapping of focal atrial tachycardias (ATs), macroreentrant ATs, WolffParkinson-White syndrome, and AF.13–17 Several advances in the approach to body surface mapping have enabled improved precision mapping of the atria. First, noninvasive body surface mapping is based on the principle that the electric cardiac potential generated and its relation to the surrounding torso volume and the body surface follows the Laplace equation.14 An inverse solution to the forward problem of electrocardiography is therefore applied. This transformation is refined by applying several computational algorithms during this process to suppress potential errors.14 Second, the patient-specific threedimensional (3D) atrial geometry is obtained from noncontrast thoracic computed tomography (CT) scan to improve mapping precision. During the CT scan, the exact locations of all the body surface electrodes on the patient’s torso are simultaneously acquired. To record body surface potentials, we use a commercially available noninvasive mapping system (ECVue, Cardioinsight Technologies Inc, Cleveland, OH). The body vest consists of 252 electrodes, and is applied to the patient’s torso. Data from the recorded body surface potentials and segmented patient-specific atrial geometry are subsequently combined. Cardiac potentials derived from the body surface potentials are reconstructed during each beat and projected onto the biatrial shell of the patient.9 To avoid interference from the QRST signal, we select the T-Q segments for analysis. Windows with R-R pauses greater than or equal to 1000 milliseconds during AF are consecutively recorded. In select patients with rapid ventricular rates, diltiazem is administered to slow the ventricular rate in order to provide adequate recording windows. Noninvasive mapping can be performed at the bedside for patients presenting in AF, before entering the electrophysiology laboratory. For patients presenting in sinus rhythm, AF can be
induced in the electrophysiology laboratory by rapid atrial pacing. AF drivers are subsequently analyzed after a predetermined waiting period (z30 minutes in initial studies). During this time, other procedural steps, such as left atrial access and creation of the atrial geometry using an electroanatomic mapping system, may be undertaken. From the recorded potentials, activation maps and phase maps can both be created for analysis.11,12 Activation maps are created using the conventional unipolar electrogram intrinsic deflection-based method ( dV/dTmax [function of voltage over time]). Filtering processes are applied to remove signal artifacts and to optimize phase transformation.9,11 Phase maps during AF are created using specific phase mapping algorithms, whereby a representation of the depolarization and repolarization wave fronts are computed from the isophase values corresponding respectively with p/2 and p/2.12 Videos of the wave propagation patterns during AF are subsequently displayed on the 3D biatrial geometry of each patient. With the panoramic phase mapping approach described earlier, localized drivers in AF may be identified, which are largely in one of 2 groups: reentrant and focal. Drivers are classified as reentrant (rotor) when a wave is recorded to rotate fully around a functional core on phase progression. The reentrant drivers are confirmed by sequential activation of local unipolar electrograms covering the local cycle length around a pivot point. Drivers are classified as focal when a wave front originates from a focal site with centrifugal activation, which is verified by a QS pattern on unipolar electrograms. To avoid nonphysiologic errors, in some studies drivers are only classified when 2 or more repeated events are recorded. In our experience, the localized drivers in each patient display certain spatiotemporal characteristics, but tend to recur around the same regions. Hence, a driver-density map is created in each patient by summating all the drivers recorded in each window onto an aggregated map that is projected on the patient’s biatrial geometry. This aggregated driver-density map serves as the roadmap for ablation. To classify the distribution of the observed reentrant and focal drivers, the biatrial geometry is divided into several anatomic domains, for example (1) left PVs and left atrial appendage, (2) right PV/posterior interatrial septum, (3) inferoposterior left atrium (LA) and coronary sinus, (4) superior right atrium (RA), (5) inferior RA, (6) anterior LA/roof, and (7) anteroseptal region.
Noninvasive Mapping for AF Ablation PAROXYSMAL ATRIAL FIBRILLATION In patients with paroxysmal AF, ectopic beats originating from the PVs are frequently observed as the trigger initiating AF, thus leading to the concept of PV isolation to eliminate these triggers.2,3 Zellerhoff and colleagues18 recently investigated 17 consecutive patients with paroxysmal AF using the described noninvasive mapping system (ECVue, Cardioinsight Technologies Inc, Cleveland, OH). These findings were correlated with invasive multielectrode mapping and radiofrequency (RF) ablation was performed according to the roadmap provided by noninvasive mapping with the end point of AF termination.
Demonstration of Pulmonary Vein Triggers and Extra–Pulmonary Vein Sources The PVs and their respective junctions with the posterior LA harbored localized drivers in nearly 90% of the patients. Apart from displaying AF drivers in these anticipated PV regions, noninvasive panoramic imaging provided further insight into potential AF drivers localized remote from the PVs. In approximately 50% of the patients, non-PV foci and reentry could be recorded outside the PV regions and the neighboring posterior LA. The right atrial appendage (RAA), the inferior LA, the coronary sinus ostium, and the anterior LA could feature such foci and reentrant drivers. Of the observed localized drivers, 74% were reentrant and 26% were focal drivers. An example of a focal and reentrant drivers in paroxysmal AF is shown in Fig. 1.
Results of Invasive Mapping and Impact of Radiofrequency Ablation on Atrial Fibrillation Termination The results of noninvasive panoramic imaging correlated well with invasive mapping: arrhythmogenic regions frequently harbored rapid electrical activity during AF, or were shown after return to sinus rhythm to spontaneously induce AF (12 of 17 patients). RF ablation of these noninvasively identified regions terminated AF reliably, further supporting the results of the mapping: ablation of the PVs terminated AF in 15 of 17 (88%) patients. Only 2 of 17 patients required ablation of non-PV regions identified by noninvasive mapping to terminate AF. This phenomenon may be explained by triggered activity during AF and potentially indicated additional sources that may manifest after PV isolation.19 In conclusion, noninvasive AF mapping shows that paroxysmal AF is related to a dynamic interplay of PV foci and ostial reentrant drivers in
humans, matching well with experimental results using optical mapping.20 Moreover, identification of non-PV drivers is feasible and reliable, guiding further RF ablation in selected patients.
PERSISTENT ATRIAL FIBRILLATION Persistent Atrial Fibrillation Drivers Identified by Noninvasive Mapping The feasibility of noninvasive mapping to identify underlying AF patterns was shown by Cuculich and colleagues11 in a cohort of patients with paroxysmal and persistent AF. With the use of activation mapping, diverse AF patterns were observed, which included multiple wavelets, rotors, and focal sources.11 Locations that were critical to AF maintenance were identified by noninvasive mapping, which responded to catheter ablation. In a recent study, Haissaguerre and colleagues12 described the use of noninvasive mapping to guide ablation in patients with persistent AF. In 103 consecutive patients with persistent AF, noninvasive phase maps during AF were acquired, either at the bedside before entering the electrophysiology laboratory or just before ablation. These maps revealed constantly changing dynamic beat-to-beat wave fronts during AF. Approximately 80% of the observed drivers were reentrant and approximately 20% were focal. Driver activity displayed spatiotemporal periodicity and reentrant drivers were observed to meander. However, these reentrant drivers recurred at the same region. Reentrant activity was periodic and not sustained, akin to that described in optical mapping and animal studies.4,21 For reentrant activity, the median number of continuous rotations was 2.6 (2.3–3.3), whereas an average of 6 events was observed from a focal site.12 Overall, a median of 4 driver regions has to be targeted per patient. In the entire study cohort, reentrant drivers were commonly located in the right PV/septal region, left PV/left atrial appendage region, left inferior wall/coronary sinus region, and upper RA. However, this distribution varied among individual patients. Fig. 2 shows a reentrant driver identified by noninvasive mapping in the inferior LA.
Persistent Atrial Fibrillation Drivers, Electrogram Characteristics, and Atrial Fibrosis Because of the spatiotemporal variability of the AF drivers, identifying them using conventional pointby-point mapping or even with the use of regional multielectrode catheters confers inherent difficulties.11 Bipolar electrogram characteristics of driver versus nondriver regions detected by noninvasive mapping were compared in a subset of
Lim et al
Fig. 1. Phase mapping shows the posterior view of the LA during paroxysmal AF. (A) Serial snapshots of a single wave emerging out of the left inferior PV (white star) and reaching the right veins in 30 milliseconds while expanding radially to the roof and inferior walls. (B) Serial snapshots of 2 successive rotations (white arrows) of a rotor located near the ostia of the right veins. The core of the rotor (the white star at the center of the rainbow-colored phases of the rotor) is seen meandering in a small region in this example. The blue wave indicates the depolarizing front, which makes 1 full rotation in 160 milliseconds. The phases of wave propagation are color coded using a rainbow scale. Blue represents the depolarizing wave; green represents the end of repolarization. The wave front can be read by following the blue color. The time (milliseconds) at the bottom of each snapshot represents the moment in the time window when the snapshot was taken. LIPV, left inferior PV; LSPV, left superior PV; RIPV, right inferior PV; RSPV, right superior PV; SVC, superior vena cava. (From Haissaguerre M, Hocini M, Shah AJ, et al. Noninvasive panoramic mapping of human atrial fibrillation mechanisms: a feasibility report. J Cardiovasc Electrophysiol 2013;24:713; with permission.)
patients in the study by Haissaguerre and colleagues.12 Prolonged fractionated electrograms were more commonly found in regions that harbored reentrant AF drivers, conceivably because of local tissue heterogeneity anchoring reentry. In addition, electrograms recorded on a multielectrode catheter covered a large part of the AF cycle length, which is consistent with recordings of localized reentry. At present, studies are underway to further correlate the underlying
diverse AF patterns and drivers revealed by noninvasive mapping with conventional invasive mapping. There are recent data to indicate that reentrant drivers tend to reside in the patchy zones and border zones of dense fibrosis detected by atrial MRI.22 In addition, complex fractionated atrial electrograms were found mostly in areas with patchy or no fibrosis, rather than in the densely fibrotic regions.22 However, although fractionation
Noninvasive Mapping for AF Ablation
Fig. 2. Intracardiac recordings (left), noninvasive phase maps (middle), and reconstructed electrograms (right) of the posterior LA during persistent AF. The 4 snapshots in the middle show a rotor meandering in the inferior LA at different times. The blue wave indicating the depolarizing front makes a full rotation in 170 to 180 milliseconds. The core of the rotor (the white star at the center of the rainbow-colored phases of the rotor) can be seen meandering in a small region. The phases of wave propagation are color coded using a rainbow scale. Blue represents the depolarizing wave; green represents the end of repolarization. The wave front can be read by following the blue color. The time (milliseconds) at the bottom of each snapshot represents the moment in the time window when the snapshot was taken. Right: unipolar electrograms along the path of 1 rotor rotation, before any specific signal processing. Note the varying morphology of electrograms recorded within a small area surrounding the core, and the presence of potentials covering the cycle length (172 milliseconds). (From Haissaguerre M, Hocini M, Shah AJ, et al. Noninvasive panoramic mapping of human atrial fibrillation mechanisms: a feasibility report. J Cardiovasc Electrophysiol 2013;24:715; with permission.)
may signal proximity to a driver site, it is not a specific pointer.23–25 The degree of fractionation may also be influenced by other factors, such as the collision of multiple wave fronts, anisotropic/slow conduction, and adjacent anatomic structures.26
Driver-based Catheter Ablation for Persistent Atrial Fibrillation The optimal ablation approach for these localized drivers is still under investigation. In a recent study, an aggregated driver-density map was created in each individual patient and used as a roadmap to guide catheter ablation.12 Over a cumulative recording period, specific regions of dense driver activity emerge on the aggregated AF map. The driver regions are then ranked according to overall prevalence of driver activity. Catheter ablation is started at the region of highest driver density, where point-by-point RF lesions are applied at the area covering the reentrant or focal drivers. The local ablation end point at each driver region is local cycle length slowing (Fig. 3). Complete abolition of regional electrograms is not required and not pursued, to avoid excessive ablation and potentially arrhythmogenic scar. Each driver region is progressively ablated in a decreasing
order, aiming for the overall procedural end point of AF termination. RF power delivery was set at 30 to 40 W using an irrigated-tip catheter, with a lower setting at the posterior LA wall, and temperature was limited to 45 C. In the aforementioned study, PV isolation was completed at the end of the procedure during sinus rhythm if required. Should patients remain in AF at the end of the procedure, direct-current cardioversion was performed. In patients in whom AF terminated into AT, these intervening ATs were subsequently ablated.12 Figs. 4 and 5 show 2 case examples of patients with persistent AF who underwent driver-guided ablation.
Clinical Outcomes The median number of targeted driver regions varied with the duration of continuous AF, from 3 regions in AF lasting less than or equal to 3 months to 6 regions in AF lasting greater than or equal to 6 months. Ablation of driver regions alone terminated AF in 75% of patients with early persistent AF, which declined with longer duration of AF (