All animal protocols were reviewed and approved by the University of California San Francisco Laboratory Animal Resource Center’s Institutional Animal Care and Use Committee, conformed to the regulations for humane care and treatment of animals established by the National Institute of Health, and were conducted with the assistance/supervision of the Animal Resource Department veterinary staff.
Twenty-five mongrel dogs weighing 25–30 Kg were divided into five different animal models: control (n=7), mitral regurgitation (n=6), congestive heart failure (n=6), rapid atrial pacing (n=6), and methylcholine (n=6, given to controls).
(MR) was induced in 6 dogs through catheter avulsion of the mitral chordae as previously described.2
A 7Fr steerable catheter with a stiff 2-mm wire hook at its terminus was placed in the left ventricle via a transseptal sheath. This catheter was manipulated until mitral chordae were ensnared and avulsed. After severe MR was achieved and acute left atrial dilatation was observed on TEE, the animals were recovered and they underwent the mapping protocol (see below) after 4 weeks. This time frame of severe MR has previously been shown to increase AF vulnerability, but prior to the development of depressed LV ejection fraction2
Rapid Atrial Pacing
(RAP) was performed in 6 dogs for at least 6 weeks, as previously described7
. AV conduction was eliminated by ablation of the AV junction, and endocardial pacing leads were placed into the right atrial appendage (RAA) and the right ventricle (RV). The pacemakers were programmed at 4 times capture threshold, with an atrial rate of 600 bpm and a ventricular rate of 100 bpm. Six weeks follow-up was chosen to allow for enough time for self-sustained atrial fibrillation, but prior to the development of significant atrial fibrosis.8, 9
(CHF) was induced in 6 dogs via four weeks of rapid ventricular pacing via a lead placed in the RV and pulse generator set to pace at 240 bpm1
followed by ablation of the AV node to create complete heart block. Ventricular function was monitored weekly with transthoracic echocardiography, and the mapping protocol was performed after 4 weeks of rapid ventricular pacing. Four weeks was chosen based on previous data demonstrating significant atrial fibrosis and AF vulnerability in that time.1, 10
(METH) infusion was used as a model of parasympathetic AF. In 6 of the Control animals, methylcholine at 1 g/250 ml saline IV11
was infused until a 20-mmHg decrease in the blood pressure and a 20% decrease in heart rate or spontaneous AF was observed. Once the blood pressure stabilized, the mapping protocol was performed.
Non-Contact Mapping of Atrial Fibrillation
Animals were intubated, mechanically ventilated and anesthetized with isoflurane (2%). Surface ECG leads I, II, III, aVL, aVR, and aVF were recorded throughout the procedure. Two transseptal catheterizations were performed using a Brockenbrough needle, and two sheaths were then placed in the LA. A non-contact balloon mapping catheter (Ensite 3000, ESI) was positioned in the LA along with a standard EP catheter (EP Technologies, San Jose, CA).
Non-contact mapping was performed using the Ensite 3000 (Ensite 3000 TM, ESI) mapping system, which is described in detail elsewhere12–14
. The EP catheter that was also inserted into the LA, was used to create a geometry of the LA in the ESI software with a methodology that has been described elsewhere.15, 16
Detailed geometries (> 500 points) of the LA were obtained prior to mapping and initiation of AF. Anatomic structures were marked on the reconstructed image. The non-contact mapping system samples all 64 cavitary potentials on the balloon at 1200 Hz and inversely applies them through Laplace’s equation in real time. This method generates >3,000 unipolar electrograms, projected onto the geometry of the LA. The methodology has been extensively described and validated previously17–19
Plaque Mapping Protocol
While non-contact mapping was performed on the endocardial surface, contact plaque mapping was simultaneously performed as previously described6
on the epicardial surface . Briefly, custom-built plaques with 240 unipoles were placed on the epicardial surface of the atria. Electrograms displaying mostly ventricular activity or 60 Hz noise were excluded from the analysis (<15%). Using the location algorithm in the ESI, each electrode of the LA plaques that was in contact with the epicardial surface was located and labeled on the LA geometry that was created.
Atrial Fibrillation Mapping
Once catheters and plaque electrodes were in place, and detailed LA geometries obtained, episodes of AF were mapped. At the time of follow-up, AF was only present in the RAP group, which was induced by chronic rapid atrial burst pacing (via the implanted pacemaker). In the other models, AF was initiated with rapid atrial burst pacing from the left atrium with a cycle length of 50 ms, a pulse width of 9.9 ms, and an output of 9.9 mA. AF was defined as a rapid, atrial rhythm that produced an irregular ventricular response and no identifiable p-wave on the surface ECG. For the CHF and RAP groups, in which the ventricles were paced during the follow-up mapping study at 50 bpm, AF was present if no identifiable p-wave was on the surface ECG during episodes of rapid atrial rhythms. Multiple 30-second epochs of AF were recorded per animal. In the control group, for which it was more difficult to induce sustained AF, recordings lasting longer than 1 minute were analyzed. At the onset of AF, two pacing spikes were delivered at 300 ms to synchronize the ESI and plaque recordings. In order to determine the activation of the AF, isopotential movies of the endocardial activation were analyzed using the ESI software, and the epicardial activation was analyzed using customized Matlab (Mathworks, Natick, MA) programs. Isochronal maps were constructed with local activation determined by the maximum dV/dt of either the virtual electrogram (ESI) or the contact electrogram (plaques). Episodes of AF were categorized as focal, reentrant, or multiple wavelet based on the characteristics of the activation in the isopotential movies and isochronal maps. Focal activation was defined as repetitive activation occurring in a radial fashion away from a source while reentrant activation was defined as repetitive activation over the entire cycle where the latest activation neighbored the area of earliest activation. Multiple wavelets were defined as multiple sources of early activation with no stability or reentry.
Validation of Non – Contact Electrograms
To validate the non-contact electrograms, the position of the EP catheter was labeled on the geometry of the LA that was created with the EnGuide location signal. The signals recorded from the EP catheter at that particular site were then cross-correlated with its corresponding non-contact signal that was calculated for that same site (Data Supplement fig. 1 A–C
) at 3 – 5 uniformly distributed sites per dog. This same procedure was done for the AF signals in each animal and in sinus rhythm in the control, MR, and CHF dogs. The cross-correlation (XC) function was calculated at zero lag for each electrogram combination, and the peak value was considered the correlation coefficient, representing the degree of correlation between the two signals.
Signal Processing and Frequency Domain Analysis
Unipolar electrograms were obtained from the epicardial plaque electrodes filtered at 0.2 – 300 Hz and sampled at 1000 Hz (Cardiomapp, Prucka GE, Marquette, FL). All signals recorded by the ESI system were filtered at 2 – 300 Hz and sampled at 1200Hz. Two thousand forty eight virtual electrograms were exported for analysis along with the location (x, y, z coordinates) of each signal on the LA geometry. Also included in this file is the location (x, y, z coordinates) of each plaque electrode that was mapped to the LA geometry. A virtual endocardial electrogram was then matched to its corresponding labeled point from the plaque epicardial electrograms by calculating the distance between the plaque electrode and each of the 2048 virtual electrogram points. The two with the shortest distance was used as a matching pair. This was performed for each plaque labeled electrode. Data were filtered and analyzed using Matlab (The Math Works Inc, Natick, MA) as previously described.20, 21
Briefly, all signals were band-pass filtered using a 40–250 Hz second-order digital Butterworth filter. The absolute value of the filtered waveform was low pass filtered using a 20 Hz second-order digital Butterworth filter.
To perform the frequency domain analysis, the resulting signal was detrended and multiplied by a Hamming window. A FFT was calculated on the final digitally filtered waveform over a sliding two-second window every 1.0 seconds. The largest peak of the resulting magnitude spectrum was identified and defined to represent the dominant frequency (DF). The position of the harmonic peaks was determined based on the position of the DF. The areas under the largest peak and 3 of its harmonic peaks were each calculated over a 1 Hz window. This produced an area under 4 peaks. The total area of the spectrum was calculated from 2 Hz up to but not including the 5th
harmonic peak. Higher frequencies were excluded because they were assumed to exceed the physiological range of frequencies for AF wavelets. The ratio of the power under the harmonic peaks to the total power in this range was calculated, and the resulting number was defined as the organization index (OI). The OI was theorized to represent the organization of AF at that period in time20, 21
. Frequency domain analysis was performed on all of the downloaded virtual electrograms and all of the electrograms recorded by the plaque electrodes. To calculate the variance of both the OI and the DFs, spatial coefficient of variance (SCoV; SD/mean) of the DFs and OI during each episode of AF among all recording sites and temporal coefficient of variance (TCoV) of average DFs between 2-s windows were calculated. These measures were used to measure the stability of the DFs spatially and temporally.
The plaque electrograms that were recorded on the epicardial surface were then compared to its corresponding virtual electrogram from the endocardial surface. This comparison was performed with both frequency domain analysis and XC. During the FFT analysis, each 2 second window was compared between the two signals. The DFs of the signals during that window were considered similar if they fell within the frequency resolution of the FFT which was 0.48 Hz. If the signals had similar frequencies for more than 90% of the windows, it was considered a match. For the XC analysis, the plaque electrograms were upsampled to 1200Hz to match the sampling rate of the ESI signals.
Data were expressed as the mean±SD. Comparisons among all mapping analysis variables (DF, maximum DF, DF spatial CoV, DF temporal CoV, OI, maximum OI, OI spatial CoV, OI temporal CoV, and XC) were performed with an ANOVA between AF models. Individual comparisons were performed with a Fisher’s exact test. Paired comparisons between endocardial and epicardial electrograms were performed with 2-tailed Student t tests. Statistical significance was defined as p<0.05.