Instantaneous Amplitude and Frequency Modulations Detect the Footprint of Rotational Activity and Reveal Stable Driver Regions as Targets for Persistent Atrial Fibrillation Ablation

Abstract

Rationale: Costly proprietary panoramic multielectrode (64-256) acquisition systems are being increasingly used together with conventional electroanatomical mapping systems for persistent atrial fibrillation (PersAF) ablation. However, such approaches target alleged drivers (rotational/focal) regardless of their activation frequency dynamics.

Objectives: To test the hypothesis that stable regions of higher than surrounding instantaneous frequency modulation (iFM) drive PersAF and determine whether rotational activity is specific for such regions.

Methods and results: First, novel single-signal algorithms based on instantaneous amplitude modulation (iAM) and iFM to detect rotational-footprints without panoramic multielectrode acquisition systems were tested in 125 optical movies from 5 ex vivo Langendorff-perfused PersAF sheep hearts (sensitivity/specificity, 92.6/97.5%; accuracy, 2.5-mm) and in computer simulations. Then, 16 pigs underwent high-rate atrial pacing to develop PersAF. After a median (interquartile range [IQR]) of 4.4 (IQR, 2.5-9.9) months of high-rate atrial pacing followed by 4.1 (IQR, 2.7-5.4) months of self-sustained PersAF, pigs underwent in vivo high-density electroanatomical atrial mapping (4920 [IQR, 4435-5855] 8-second unipolar signals per map). The first 4 out of 16 pigs were used to adapt ex vivo optical proccessing of iFM/iAM to in vivo electrical signals. In the remaining 12 out of 16 pigs, regions of higher than surrounding average iFM were considered leading-drivers. Two leading-driver + rotational-footprint maps were generated 2.6 (IQR, 2.4-2.9) hours apart to test leading-driver spatiotemporal stability and guide ablation. Leading-driver regions (2.5 [IQR, 2.0-4.0] regions/map) exactly colocalized (95.7%) in the 2 maps, and their ablation terminated PersAF in 92.3% of procedures (radiofrequency until termination, 16.9 [IQR, 9.2-35.8] minutes; until nonsustainability, 20.4 [IQR, 12.8-44.0] minutes). Rotational-footprints were found at every leading-driver region, albeit most (76.8% [IQR, 70.5%-83.6%]) were located outside. Finally, the translational ability of this approach was tested in 3 PersAF redo patients.

Conclusions: Both rotational-footprints and spatiotemporally stable leading-driver regions can be located using iFM/iAM algorithms without panoramic multielectrode acquisition systems. In pigs, ablation of leading-driver regions usually terminates PersAF and prevents its sustainability. Rotational activations are sensitive but not specific to such regions. Single-signal iFM/iAM algorithms could be integrated into conventional electroanatomical mapping systems to improve driver detection accuracy and reduce the cost of patient-tailored/mechanistic approaches.

Keywords: ablation; algorithms; atrial fibrillation; driver; mapping; rotor.

Figures

Figure 1.

Amplitude modulation (AM)/frequency modulation (FM) concept and study design.A, AM and FM in radio broadcasting. B, Left, during cardiac fibrillation, scroll wave/rotor drifting gives rise to AM and FM. When a drifting scroll wave filament/rotor core approaches the square, the amplitude of the action potential decreases increasing instantaneous AM (iAM; in red). Simultaneously, as the wave-emitting source (scroll wave filament/rotor core) approaches, the perceived instantaneous FM (iFM; (Continued )Figure 1 Continued. in blue) at the spot increases (Doppler Effect). Therefore, simultaneous iAM/iFM increase indicates drifting scroll waves/rotors in the surroundings. Additionally, the areas with the highest average iFM are those potentially driving fibrillation. Right, average iFM is estimated from its median and mean values (8 Hz both) and with the conventional dominant frequency (DF) spectral approach (5.6 Hz). Interestingly, the time intervals with the highest iFM usually show the lowest amplitudes and vice versa, which conditions the height of their corresponding power spectral peaks and limits the value of DF-based hierarchical approaches. C, Translational approach for the study. D, Porcine experimental model of persistent atrial fibrillation (PersAF). AF indicates atrial fibrillation; HRAP, high-rate atrial pacing; and PVI, pulmonary vein isolation.

Figure 2.

Examples of instantaneous frequency modulation (iFM)/instantaneous amplitude modulation (iAM) from an optical movie in a persistent atrial fibrillation (PersAF) sheep heart (Online Movie I).A, Top, signal from a pixel (gray square in bottom maps, enlarged to ease visualization) crossed by a phase singularity (PS) in a figure-of-eight reentry. Spirals mark the times at which a PS (white circles) passes by the pixel. Blue points, activation times. Red points, start and end of phase-0. Activation times and phase-0 amplitude excursions are used to generate the iFM (second row, in blue) and the iAM (second row, in red) signals, respectively. Time intervals with sustained simultaneously increasing iFM (second row, thick blue tracings) and iAM (thick red tracings) reaching a prespecified iAM threshold (horizontal red dashed line) are detected. Afterward, the rotor is still considered to be around while iAM keeps over the threshold regardless of iFM (see rotor no. 2 and Online Figure/Movie II). Time intervals with simultaneously high, but not necessarily increasing, iFM and iAM above prespecified thresholds (horizontal blue and red dashed lines, respectively) are detected as quasistationary rotors (see rotor no. 3). The third row displays a synthetic FM | AM signal in which the rotational-footprint positive intervals are highlighted. (Continued )Figure 2 Continued. B, Signal from a pixel close to areas swept by drifting rotors but not actually crossed by their associated PS. Note that there are still intervals with simultaneously increasing iFM and iAM. However, iAM does not reach the prespecified threshold. Therefore, the algorithm properly classifies the pixel as rotational-footprint negative. C, Signal from a pixel far from areas swept by drifting rotors. Increasing iAM, although still present, is not as noticeable as in pixels very close to or actually crossed by drifting rotors. Therefore, the algorithm also classifies this pixel as rotational-footprint negative. D, Comparison between the pixels actually crossed by a PS (PS map, “gold standard”) and the pixels detected by the single-signal algorithm as rotational-footprint positive (iFM/iAM map). The size of a conventional ablation catheter is shown for reference. The signals shown in A–C were taken from pixels a–c. Additional examples are shown in Online Figure IV.

Figure 3.

In vivoinstantaneous frequency modulation (iFM)/ instantaneous amplitude modulation (iAM) calculation and single-signal rotational-footprint detection from unipolar electrical signals.A, Activation times (first row, cyan points) and unipolar amplitude excursions between the starts and ends of negative deflections (first row, red points) were used to generate the iFM (second row, cyan tracing) and the iAM (second row, red tracing) signals, respectively. Amplitude excursions simultaneous with ventricular activations were interpolated from the surrounding excursions because ventricular far-field residues may affect them, even after applying QRS minimization strategies. Time intervals with sustained simultaneously increasing iFM (thick cyan) and iAM (thick red) reaching a prespecified iAM threshold (dotted horizontal red line) are detected. Afterward, the rotor is still considered to be around while iAM keeps over the threshold and regardless of iFM. The third row displays a synthetic FM | AM signal (morphologically similar to an optical signal) that incorporated both iFM and iAM dynamic changes. Rotational-footprint positive intervals are highlighted. B, Snapshots from the phase movie obtained by interpolating data from the 20 electrodes of a PentaRay catheter fully deployed in the right atrial appendage (RAA; Online Movie III). Electrode locations with rotational-footprint positive signals are highlighted with cyan squares. Note the high correlation between highlighted electrodes and the center of rotation in the phase movie. See also Online Movie IV from left atrial appendage (LAA). C, Unipolar electrograms confirming the reentrant activation displayed in B. Red arrows mark depolarizations that may be explained by precession of the rotational core (Online Figure/Movie II). D, Combined leading-driver (iFMmedian) + rotational-footprint map. Rotational-footprint positive locations are marked with black/white squares within light/dark areas, respectively. Interestingly, many regions displayed repetitive rotational activations, including the RAA and LAA. However, most of them did not seem hierarchically relevant to drive AF since it acutely terminated and was rendered nonsustainable after ablating only the purple area located in the coronary sinus. Importantly, rotational-footprints were also found in that region. *The location from which the signal in A was retrieved. ICV indicates inferior cava vein; and SCV, superior cava vein.

Figure 4.

Performance of the single-signal instantaneous frequency modulation (iFM)/ instantaneous amplitude modulation (iAM) algorithm to detect rotational-footprints in optical movies from sheep with persistent atrial fibrillation.A, Snapshots from sheep no. 1 (Online Movie V) with a drifting rotor (top row) that eventually leaves the field of view. Then, planar wavefronts are observed. The single-signal algorithm yielded positive results (white +) in the pixels near the pivoting point of the drifting rotor (phase singularity [PS]), but no pixel was tagged positive during planar wavefront intervals. Considering a 1.25 mm tolerance (light blue areas, width equal to the radius of an ablation electrode), both PS and iFM/iAM maps were extremely similar. B, Snapshots from sheep no. 2 (Online Movie VI) displaying an interval with centrifugal activation. C, Snapshots from sheep no. 3 (Online Movie VII) displaying an interval (1690–2595 ms) with breakthrough activations. The signal from the pixel marked with a gray square is shown during the same time interval, when a simultaneous increase in iFM and iAM is present. Therefore, the initial breakthroughs might be the result of a scroll wave with a changing filament approaching the mapped surface. Indeed, this breakthrough activation eventually turned into a drifting figure-of-eight reentry. D, Summary of sensitivity (sens) and specificity (spec) using the optimal parameter combinations for the iFM/iAM algorithm (n=117 optical mapping movies). Data displayed normal distribution (Shapiro-Wilk test) and are shown as mean and SD.

Figure 5.

In vivoquantitative results of the porcine model of self-sustained persistent atrial fibrillation (PersAF).A, Protocol times since high-rate atrial pacing initiation. B, Atrial diameters and left ventricular (LV) ejection fractions measured with echocardiography. The Wilcoxon signed-rank test was used to compare paired measurements from the same pigs at baseline and at the time of the electrophysiological study (EPS). The Mann-Whitney test was used for comparisons between sham-operated and self-sustained PersAF animals. C, Increased atrial fibrosis was found in self-sustained PersAF hearts (Mann-Whitney test). D, Anatomic distribution and size of leading-driver regions. The cumulative size (%) of the leading-driver regions found in the left atrium (LA) + coronary sinus (CS) was higher than the cumulative size of those in the right atrium (RA; Wilcoxon signed-rank test). E, The number of leading-driver regions was significantly higher in the LA+CS than in the RA (Wilcoxon signed-rank test). F, Rotational-footprint quantification and spatial correlation with leading-driver regions. The percentage of signals with rotational-footprints and the percentage of rotational-footprints outside leading-driver regions did not show significant differences between the first and second maps (Wilcoxon signed-rank test). Importantly, around 3-quarters of all rotational-footprints were found outside leading-driver regions. PV indicates pulmonary vein.

Figure 6.

Leading-driver regions are highly stable intraprocedure. A, Example of persistent atrial fibrillation (AF) termination after ablating the coronary sinus region delineated with a fuchsia dashed line for 11.4 min. Then, AF was reinduced and lasted >10 min. After 11 additional minutes of radiofrequency delivery (orange dashed line), AF terminated again and was no longer sustainable. B, Case with extensive atrial remodeling due to severe tricuspid regurgitation. Much higher median values of instantaneous frequency modulation (iFMmedian) than in the rest of animals were documented over large areas of the left atrium. AF did not terminate despite 97 min of radiofrequency delivery. However, ablation resulted in an important reduction in the overall atrial activation rate measured by dominant frequency (DF) from the 12-lead surface ECG (bottom panel), which indicated that ablation effectively modified the AF substrate. *The locations where the displayed signals were retrieved from. iAM indicates instantaneous amplitude modulation; and iFM, instantaneous frequency modulation.

Figure 7.

Leading-driver regions seem stable in the long-term.A, One pig underwent a successful first ablation procedure after which atrial fibrillation (AF) burden was temporally reduced to 0%. Driver areas not ablated in the first procedure remained as key regions in a second procedure >4 mo after resuming high-rate atrial pacing, and >3 mo after reaching self-sustained persistent AF again. Also, a new region with consistent high median values of instantaneous frequency modulation (iFMmedian) outliers was found in the posterior left atrium. After a total radiofrequency delivery of 19.8 min, AF was rendered nonsustainable. B, One pig underwent 2 mapping procedures 78 d apart. Note that leading-driver regions, although less electrically remodeled in the first mapping procedure (lower iFMmedian values), were roughly similar. Online Figures XXIII and XXIV show signals from the marked locations (*).

Figure 8.

Translation ability of the instantaneous frequency modulation (iFM)/instantaneous amplitude modulation (iAM) approach to patients with persistent atrial fibrillation (PersAF) despite ≥1 previous pulmonary vein isolation procedures.A, Map from a patient undergoing a third AF ablation procedure. Fast and large leading-driver regions covered a considerable portion of the atria and precluded a limited ablation strategy (fuchsia dashed line) from acutely terminating PersAF. Moreover, radiofrequency delivery did not modify atrial activation rates on the 12-lead ECG. B, Map from a patient with a well-delimited leading-driver region. Targeting that region with radiofrequency delivery for ≈10 min successfully terminated PersAF. Upon reinduction, common atrial flutter was the only inducible arrhythmia, which was eventually terminated by creating a linear lesion at the cavotricuspid isthmus. Abl indicates ablation; DF, dominant frequency; RF, radiofrequency; Rot, rotational; and SR, sinus rhythm.