Author + information
- Received June 22, 2015
- Revision received August 12, 2015
- Accepted September 1, 2015
- Published online February 1, 2016.
- Jacob Laughner, PhDa,
- Shibaji Shome, PhDa,
- Nicholas Child, MB BSb,
- Allan Shuros, MSa,
- Petr Neuzil, MD, PhDc,
- Jaswinder Gill, MDb and
- Matthew Wright, PhD, MB BSb,∗ ()
- aBoston Scientific St. Paul, Minnesota
- bKings College London BHF Centre, Cardiovascular Division, NIHR Biomedical Research Centre at Guy’s and St. Thomas’ NHS Foundation Trust, London, United Kingdom
- cDepartment of Cardiology, Na Homolce Hospital, Prague, Czech Republic
- ↵∗Reprint requests and correspondence:
Dr. Matthew Wright, Divisions of Imaging Sciences and Biomedical Engineering and Cardiovascular Medicine, St. Thomas' Hospital, Westminster Bridge Road, London SE1 7EH, United Kingdom.
Objectives This study sought to evaluate basket catheter deployment, catheter-tissue contact, and time-space stability of unipolar atrial electrograms (aEGMs) recorded in persistent atrial fibrillation (AF) patients.
Background Panoramic mapping of human AF using multiple-electrode basket catheters may identify AF sources. Although clinical results using this technique are provocative, questions remain about its effectiveness.
Methods Data were collected from patients (N = 25) undergoing catheter ablation for AF during the multicenter STARLIGHT (Signal Transfer of Atrial Fibrillation Data to Guide Human Treatment) trial (NCT01765075). Left and right aEGM signals were recorded using basket catheters during baseline AF, following ablation and during sinus rhythm. Data were analyzed for basket deployment, peak-to-peak voltage, and electrogram stability and organization. Electrogram stability and organization were evaluated via time-frequency analysis (TFA).
Results Basket catheters displayed equatorial bunching when deployed in atria. Interspline spacing ranged from 1.7 to 64.0 mm in the right atrial and from 1.5 to 85.08 mm in the left atrial basket. Approximately one-third of mapping electrodes failed to demonstrate a median peak-to-peak voltage >2× the low-voltage threshold. Time-space stability and organization was observed in 13 of 22 (59.09%) right atrial and 10 of 22 (45.45%) left atrial baskets.
Conclusions Despite poor deployment and a large number of low-voltage electrodes, stability and organization was observed in about one-half of the mapped patients. Although this study suggests that basket catheters have limitations for patient-specific AF mapping, concordant activation occurs in some persistent AF patients, which may be amenable to high-density mapping techniques.
Theories on atrial fibrillation (AF) drivers are in constant development, supported by pre-clinical and clinical observations. The development of AF theory has been heavily influenced by interpretation of data obtained by available technology. In the early part of the 20th century, 2 competing theories, focal ectopy and circus re-entry, established the mechanistic debate for fibrillation sustenance (1,2). These theories were primarily derived from visual inspection of mechanical kymograph measurements (3–5) or electrical measurements from the body surface via string galvanometers (6). Using a computational model, Moe and Abildskov (7) hypothesized that neither multiple foci nor sustained, stable re-entry were sufficient mechanistic explanations for AF. Rather, they proposed that multiple re-entrant wavelets, sustained separately from an initiating mechanism drove AF. Experimental validation of the Moe and Abildskov computational model followed the development of a novel microelectrode apparatus designed by Allessie et al. (8–11). Modern research using intravenous catheters, multielectrode plaques, and optical mapping have produced intriguing insights on AF, but consensus on a specific mechanism is lacking.
Recently, the concept of “sustained, stable re-entry” received renewed interest as a potential persistent AF driver. Mapping AF using either contact endocardial electrograms from basket electrode catheters or reconstructed endocardial electrograms from body surface recordings (electrocardiographic imaging [ECGI]) have provided clinical evidence of this concept. In the CONFIRM (Conventional Ablation for Atrial Fibrillation With or Without Focal Impulse and Rotor Modulation) clinical trial, rotors or focal patterns were prospectively identified with a large basket mapping catheter in 96% of patients with persistent AF (12–14). Ablation of stable rotational or focal sources, in conjunction with conventional ablation, demonstrated a significantly higher freedom from AF after a median of 273 days compared to conventional ablation (82.4% vs. 44.9%, p < 0.001) (12). Stable rotors in the CONFIRM trial were defined as >50 rotations spanning several recording epochs (13). Although maps generated from endocardial basket electrode catheters suggest that rotors may be stable and stationary, rotors revealed via ECGI appear to be both fleeting and wandering over a considerably large area of the endocardium (15). Therefore strategies used to ablate rotors, assuming that rotors are AF sources, indirectly depend on the modality used to reconstruct cardiac activity. This apparent contradiction in the characteristics of a purportedly ubiquitous phenomenon in persistent AF highlights the role of the modality used to reconstruct cardiac activity.
Although the reemergence of “rotors” as a driver of AF is a provocative development, critical questions remain on technical limitations of whole chamber multipolar basket catheters (MBCs) for mapping rotors in persistent AF patients. Before rotor mapping with a MBC becomes a clinical standard, further clinical evaluations are necessary not just to evaluate rotor theory and procedural success, but also the limitations of the mapping tool. In this study, we evaluated 3 aspects of MBCs as a tool for mapping human persistent AF: 1) contact; 2) coverage; and 3) rhythm stability.
The STARLIGHT (Signal Transfer of Atrial Fibrillation Data to Guide Human Treatment) study (NCT01765075) was a nonrandomized, single-arm, multicenter observational study designed to collect intracardiac atrial electrogram (aEGM) data during a cardiac ablation procedure for treatment of persistent AF. The STARLIGHT study enrolled 25 subjects, was conducted at 2 European centers, and required 1 follow-up visit at 30 ± 7 days following the procedure. The study workflow is shown in Figure 1.
Unfiltered simultaneous left and right unipolar aEGMs were collected with 64-pole Constellation basket catheters (Boston Scientific, St. Paul, Minnesota; sizes 48, 60, and 75 mm) during baseline AF (presenting or induced), following ablation and during sinus rhythm using the Ensite Velocity cardiac mapping system (St. Jude Medical). Electrograms were recorded as continuous 4-min episodes at 2000 Hz sampling. The mapping system was also used to simultaneously collect electroanatomical maps, electrode positional data, chamber geometry, and surface electrocardiograms. Four pins in the right atrial basket pin block were reserved for the system reference, lowering the potential number of poles to 60 in the right atrial MBC.
Electroanatomic data (electrode locations, anatomical shells, aEGMs, 12-lead electrocardiogram) were analyzed post-hoc using Matlab (The MathWorks, Natick, Massachusetts). Briefly, aEGM ventricular artifact was minimized using an Average Beat Subtraction technique with 5 consecutive QRST complexes (16). The residual aEGMs were bidirectionally filtered with an infinite impulse response band-pass filter (Butterworth, n = 4, 3 to 30 Hz) to isolate atrial activity.
Analysis of basket deployment
Electrode locations and anatomical shells were reconstructed to qualitatively and quantitatively analyze spline coverage and bunching within atrial chambers (Figure 2). Quantitative analysis involved calculation of linear distances between neighboring equatorial electrodes. Equatorial interspline distance was used as a measure of basket deployment and was compared to the interspline spacing reported in the directions for use (DFU). For each patient, the coefficient of variation (CV) (standard deviation [STD]/mean) of the interspline distance was calculated and compared to hypothetical tolerance amounts of 20%, 40%, and 60% of STD on the spacing listed in the DFU.
Analysis of electrogram amplitudes
We used aEGM peak-to-peak voltage at each MBC electrode as an electrode-tissue contact surrogate (Figure 3). Due to inherent time variability present in aEGMs during tachycardias, median peak-to-peak voltage over a 7-s recording epoch was measured for patients with presenting or inducible tachycardia (n = 22 of 24, 91.67%) using a streaming Max-Min filter algorithm. In most mapping systems, a peak-to-peak voltage of 0.5 mV or less is set as the low voltage threshold (17). We used a threshold of 1.0 mV (2× low voltage threshold) to indicate that the signal-to-noise ratio may be inadequate for electrogram interpretation (i.e., noise, scar, noncontact).
Analysis of rhythm stability
Time-frequency analysis (TFA) with wavelets was used to analyze stability and organization within and between electrodes on the basket catheter over a randomly chosen 7-s window (Figure 4). TFA with wavelets produces similar information to the short-time Fourier transform method used by Schuessler et al. (22), but offers a multiresolution analysis of signal features and the ability for parallel implementation in computer languages. Although short-time Fourier analysis has a fixed time and frequency resolution, wavelet-based TFA approaches rely on scaling factors that enable variable time and frequency resolution. This approach favors aEGM analysis during AF where a combination of rapid temporal changes in dominant frequency (DF) and closely spaced subdominant frequency peaks occurs. In our analysis, a continuous-wavelet transform using the Fast Fourier Transform algorithm with a Morlet wavelet was applied to each filtered aEGM (Figure 4A) to create a time-frequency power spectrogram (Figure 4B) in order to understand interelectrode relationships. Spectrograms for each pole were then combined into a global median spectrogram (Figure 4C) to more easily identify patients lacking global stability and organization. Correlation between the median spectrogram and each individual electrode was performed by calculating the maximum cross correlation (Figure 4D). Electrodes possessing >80% correlation to the median TFA were considered to have a similar TFA signature. Using the global median spectrogram, DF was tracked over 7 s with a custom contour-based tracking algorithm for each recording. STD of the DF and the mean half-maximum power enveloped (HME) around the DF were analyzed in each TFA over the specified time epoch (Figure 5W).
Clinical characteristics for all patients enrolled in the STARLIGHT study are summarized in Table 1. The study population was predominately male (n = 21 of 24, 84%). Mean age for the study population was 59.5 ± 10.6 years. Right and left atrial baskets were sized according to long- and short-axis cardiac measurements acquired via echo and the Constellation DFU. The majority of patients (n = 20, 80%) did not have a prior ablation. One patient was excluded from analysis due to corrupted tracking and electrogram data.
An assessment of MBC deployment is presented in Figure 2. Considerable efforts (e.g., catheter torquing, prolapsing) were attempted to achieve uniform distribution of the MBC splines. Despite these efforts, right and left atrial MBCs displayed considerable spline spread and bunching (Figures 2B and 2D). Data for PT10 was excluded from statistical analysis due to data corruption but is displayed in Figure 2 for reporting completeness. Interspline spacing ranges were 1.7 to 64.0 mm in the right and 1.5 to 85.08 mm in the left atrial basket. The Constellation catheter has nominal equatorial interspline spacing of 18.8, 23.6, and 29.5 mm for 48, 60, and 75 mm catheter sizes, respectively, when deployed ex vivo. Allowing for some distortion due to nonspherical geometry of the right left atrium, the CV for each MBC was compared against hypothetical amounts of deviation of 20%, 40%, and 60% on the DFU-listed equatorial spacing. Assuming 20% deviation in the interspline spacing, 0 of 23 (0.00%) right atrial baskets or left atrial baskets displayed a CV ≤0.20. For a 40% deviation in the interspline spacing, 10 of 23 (43.48%) right atrial baskets had a CV less than or equal to the corresponding CV of 0.40 whereas 3 of 23 (13.04%) left atrial baskets had a CV ≤0.40. At the 60% deviation threshold, 13 of 23 (56.52%) right atrial baskets and 9 of 23 (34.78%) left atrial baskets had a CV ≤0.60.
Two representative examples of variable basket deployments from PT02 and PT04 are provided in Figures 2B and 2D. For PT02, considerable spline bunching was observed in the left atrium (Figure 2B); 4 splines gather in the left atrial anterior roof, 2 splines gather on the left atrial posterior wall, and 2 splines traverse the mitral valve. For PT04, spline bunching in the right and left atrium was not as severe as PT02 (Figure 2D). However, a bias of electrodes to the left atrial roof and 2 splines passing over the mitral valve plane was observed. In both PT02 and PT04, coverage of the pulmonary veins, left atrial septum, and left atrial lateral wall was limited. Similar spline deployments were observed among the other study subjects.
Figure 3 summarizes the results of the contact assessment. Median peak-to-peak voltage was higher in the right atrial MBC than in the left atrial MBC (15 of 22 patients, 68.18%). Further, median peak-to-peak voltages in the right atrial MBC were >1.0 mV in 15 of 22 (68.18%) patients, compared to 12 of 22 (54.54%) patients in the left atrial MBC. In 4 of 22 (18.18%) patients (PT01, PT09, PT20, PT23), 75% of aEGMs recorded by the basket in the left atrium (i.e., 48 of 64 electrodes in each case) had peak-to-peak amplitudes of <1.0 mV. Only 2 of 22 (9.09%) patients (PT08 and PT20) had right atrial baskets with 75% of the aEGMs displaying peak-to-peak amplitudes <1.0 mV. On average, right atrial baskets had 39.8 ± 14.4 of 60 (62.14 ± 24.95%) electrodes with peak-to-peak voltages >1.0 mV whereas left atrial baskets had 41.0 ± 15.4 of 64 (63.99 ± 24.09%) electrodes with peak-to-peak voltage >1.0 mV.
Electrogram stability and organization
Figure 5 summarizes the TFA analysis from all patients with an inducible/stable tachycardia (n = 22). The mean DF from all patients in this cohort was 5.05 ± 0.83 Hz in the right atrium and 5.01 ± 0.77 Hz in the left atrium. Patient PT19 met the study inclusion criteria, however periprocedural mapping revealed a left atrial flutter rhythm. Patients were further dichotomized by their DF STD and mean HME over the recording window. Empirical thresholds of 0.5 Hz DF STD and a mean HME of 3 Hz were established to classify spectrograms as stable or unstable. As shown in Figure 5W for the right atrial basket, 13 of 22 (59.09%) patients had a DF STD <0.5 Hz and a HME <3 Hz. Additionally, right atrial baskets falling below the STD and HME thresholds also had more electrodes with a TFA highly correlated (i.e., >80%) to the median TFA (43.08 ± 8.53 vs. 28.67 ± 9.45). For the left atrial basket, 10 of 22 (45.45%) patients had a DF STD <0.5 Hz and a HME <3 Hz. Similarly, left atrial baskets falling below the STD and HME thresholds had more electrodes with a TFA highly correlated (i.e., >80%) to the median TFA (41.50 ± 11.96 vs. 33.83 ± 10.09).
AF theory has historically been driven by the development of novel computation models, pre-clinical/clinical tools, and signal processing techniques. Recently, rotor theory has been suggested as a persistent AF driver. Although clinical results have aided proponents of rotor theory, several questions remain unanswered of mapping tools and methodologies. Here, we evaluated practical considerations of whole-chamber mapping with identical tools (i.e., Constellation catheters) used in most rotor mapping studies. Although previous work focused on activation and phase patterns produced from basket recordings, we attempted to evaluate the limitations of basket deployment, tissue-catheter contact, and rhythm stability.
Electrogram stability and organization
For activation time and phase maps to show stable sources of AF, aEGMs recorded during AF must display some degree of spatiotemporal stability and organization over many seconds. Without organization and stability, activation and/or phase maps may suggest spurious sources, making map-guided treatment difficult.
Several studies have attempted to map various aspects of AF stability and organization using both time-domain and frequency-domain approaches. Early research by Michelucci et al. (18) and Barbaro et al. (19) focused on time-domain analysis of the number of baseline deviations in bipolar electrograms. Michelucci et al. (18) used basket catheters to demonstrate well-defined patterns of organization over 10 min of mapping in the right atrium of persistent AF patients following an induced tachycardia. This observation was further explored in similar experiments by Ravelli et al. (20,21) with an alternative analytical approach by quantifying the mapped morphological similarity of bipolar atrial complexes. Ravelli et al. (20,21) identified a range of waveform similarity dependent on degree of AF: high spatiotemporal waveform similarity in paroxysmal AF patients and decreased spatiotemporal waveform similarity permanent AF patients.
Frequency-based analyses have also been used to investigate activation stability. Schuessler et al. (22) demonstrated discrete regions of stability and organization over 7-s recordings in DF maps identified from high-density epicardial measurements in AF using a time-frequency analysis. Although persistent AF patients demonstrated very consistent global DF magnitudes, Schuessler et al. (22) noted that the anatomic location of dominant frequency drivers moved in one-half of the examined patients. In another study of spectral content in AF patients (primarily persistent AF) with a Constellation catheter, Habel et al. (23) found no significant difference in mean DF when comparing the first and second 5-s time segments of 10-s aEGM recordings (5.55 ± 1.82 Hz vs. 5.55 ± 1.84 Hz). Although the spatial location of DF was observed to be time dependent over longer epochs, minimal spatial variability in the DF (26.9% or 1.48 Hz of deviation on a 5.55 Hz mean DF) was observed among electrodes.
As described previously, we measured stability and organization on the basket electrodes with a wavelet-based TFA approach. Specifically, we evaluated each median TFA plot for DF consistency over time (i.e., STD in DF) and the power spectrum distribution around the DF (HME). In theory, patients with less DF variation and less spread in the power spectrum distribution over the time window should display high temporal stability (i.e., stable tissue activation) and spatial coherence (i.e., strong coupling between electrodes). This behavior is illustrated well during flutter in Patient PT19 in Figure 5. Under these assumptions, we identified approximately 60% and 45% of the STARLIGHT study population to have stable right and left MBC recordings, respectively. In these patients, approximately 65% of the aEGMs displayed similar time-frequency behavior compared to the median TFA. This implies that a map generated with these electrodes is likely to display a concordant pattern.
The accuracy and utility of activation and phase maps of cardiac arrhythmias is closely linked with map resolution. One aspect that can affect map density and resolution is low voltage electrograms. Electrodes containing low-voltage information can result from poor contact with the myocardium or contact with scar. Due to consistent amplitude changes over the length of the recording, median peak-to-peak amplitude over a 7-s window was used to identify consistently low-voltage electrodes. Assuming that electrograms above the low-voltage voltage threshold are meaningful, this analysis provides a first approximation of electrode density in a typical panoramic map. We found that approximately 60% of the electrodes in the right and left atrial baskets are eligible for creation of phase or activation maps. We also observed a considerable amount of variability across all patients. In previous work, Habel et al. (23) found similar results: 21.6 ± 4.3 of 32 (67.50 ± 13.44%) bipolar electrodes on the MBC had “adequate signal-to-noise.” In a more recent study by Benharash et al. (24), 48% of electrograms were deemed decipherable for rotor mapping. On the basis of these results, it is likely that an interpolation strategy will be required to fill the missing activation and/or phase values in a panoramic map.
In addition to number of decipherable electrograms, basket coverage and deployment may limit map resolution. In a recent investigation of rotor mapping with an electroanatomic mapping system, Benharash et al. (24) found the MBCs provide poor coverage of the atrial anatomy (e.g., 55% of the left atrial surface area). Although previous studies have reported uniform deployment of MBC using fluoroscopy (13), we found 2D fluoroscopy images of basket deployment to be misleading compared to 3-dimensional electrode coordinates from the electroanatomic mapping system. Despite best attempts to deploy the basket panoramically (e.g., torqueing and prolapsing), significant spline bunching was observed in both right and left atria for a majority of the mapped patients. Anatomical geometry constraints in the left atrium appear to cause more basket distortion compared to the right atrium.
In previous computational studies by Rappel and Narayan (25), it was demonstrated that electrode spacing <1.1 cm is necessary to map spiral wave activation patterns. As shown in Figure 2, interspline distances for equatorial pairs of electrodes often exceed this limit. Moreover, the manufacturing-listed interspline distances for equatorial pairs are almost double this maximum threshold when deployed outside the body.
Representing basket catheters with 2-dimensional uniform 8 × 8 grids (or up-sampled grids of 32 × 32) may lead to map misinterpretation given nonuniform 3-dimensional spline bunching. The limitation of a 2-dimensional uniform grid display may be overcome with knowledge of substrate characteristics such as conduction velocities. However, accurate conduction velocity calculations are dependent on determining interelectrode distances. Further, electrode locations must be included to accurately account for distortion in basket deployment, rather than simply assuming electrodes are arranged in a flat grid, when using interpolation strategies to estimate missing data. In the absence of dense uniform data, Masé and Ravelli (26) demonstrated the use of radial basis functions to accurately interpolate sparse activation maps. This technique could be applied, provided electrode locations are known, to produce reliable activation and phase maps.
The Constellation catheter was the only MBC examined in this study. At the time of study enrollment and data collection, no other basket catheters were available for inclusion. The limitations of MBCs for panoramic mapping presented in this paper are therefore limited to the Constellation catheter design. Future whole-chamber basket catheters may be designed to overcome some of the presented limitations and enable high-density panoramic mapping.
Panoramic mapping with MBCs provides insufficient spatial resolution due to poor contact, demonstrates frequent bunching of the basket splines, and possesses inadequate electrode density to accurately detect rotors near the equatorial electrodes (Figure 6). Despite limitations of the current basket as a tool for mapping AF, organization and stability were identified in some patients in the STARLIGHT study. High density mapping strategies may be able to capture concordant activity in this subset of patients and provide a reliable tool for evaluating rotor theory.
COMPETENCY IN MEDICAL KNOWLEDGE: Panoramic mapping with multipolar whole-chamber basket catheters has become a technique for localizing sources of persistent AF. Ablation of AF sources identified by panoramic mapping may lead to a patient specific treatment of AF and an improvement over traditional anatomical-only ablation strategies.
TRANSLATIONAL OUTLOOK: Although panoramic imaging with whole-chamber basket catheters shows stability and organization within aEGMs during, present MBCs suffer from low signal quality and poor spatial resolution. Without a significant change in MBC design, panoramic mapping of AF will be limited by useable information. Alternative strategies based on sequential mapping may improve mapping results but need to be evaluated in controlled clinical trials.
The authors thank Dr. Jan Petru and Dr. Jan Skoda for procedural assistance and Dr. Pramod Thakur and Kevin Stalsberg for technical assistance.
Dr. Laughner, Dr. Shome, and Mr. Shuros are paid employees of Boston Scientific and own Boston Scientific stock. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- atrial electrogram
- atrial fibrillation
- electrocardiographic imaging
- coefficient of variation
- dominant frequency
- directions for use
- half-maximum envelop
- multipolar basket catheter
- standard deviation
- time-frequency analysis
- Received June 22, 2015.
- Revision received August 12, 2015.
- Accepted September 1, 2015.
- American College of Cardiology Foundation
- Jalife J.,
- Berenfeld O.,
- Mansour M.
- Mayer A.
- Garrey W.E.
- Lewis T.
- Allessie M.A.,
- Bonke F.I.,
- Schopman F.J.
- Allessie M.A.,
- Bonke F.I.,
- Schopman F.J.
- Allessie M.A.,
- Bonke F.I.,
- Schopman F.J.
- Allessie M.,
- Lammers W.,
- Bonke F.,
- Hollen J.
- Narayan S.M.,
- Krummen D.E.,
- Shivkumar K.,
- et al.
- Cuculich P.S.,
- Wang Y.,
- Lindsay B.D.,
- et al.
- Slocum J.,
- Byrom E.,
- McCarthy L.,
- et al.
- Ravelli F.,
- Masè M.
- Benharash P.,
- Buch E.,
- Frank P.,
- et al.