Poor techniques to classify AF and the absence of a widely accepted classification system are two problems. The terms paroxysmal, persistent and chronic provide little information about the pathology of AF. Further, there is no clear delineation between the focal and nonfocal mechanisms.

None of the present descriptors of AF are clear or distinct. It is unclear whether the left or right atrium is primarily involved, despite talk about associated heart disease. Whether the AF that results from heart disease is the same or different is unclear. The underlying electrophysiologic (EP) properties, such as zones of slow conduction, dispersion of refractoriness, and anisotropy, are difficult to classify. It is unclear how to describe the findings from intracardiac mapping.

A new mapping technique is needed to identify the focal drivers of AF. Intracardiac electrocardiograms are difficult to analyze during AF, due to the lack of a surface or intracardiac signal to serve as a marker of when to begin the evaluation. Activation mapping requires evaluation of multiple sites, which may be particularly difficult to accomplish in humans, as is characterizing the maps clearly and succinctly. The recording duration required to insure reproducibility is unclear.

Attempts to treat right atrial AF with catheters introduced into the right atrium have had limited success. Notably, right atrial ablation is successful in preventing AF recurrence in some patients, and seems most likely to be effective when the right atrium is the "driving force".

AF occurring in the setting of severe tricuspid regurgitation (right atrial predominant disease) was studied, as most patients with AF have underlying heart disease and thus anatomic atrial remodeling. AF occurring in the setting of combined severe triscuspid regurgitation and rapid atrial pacing was studied, as electrical remodeling may occur due to repeated bouts of paroxysmal atrial fibrillation or tachycardia. The hypothesis was that right atrial pathology results in AF with a higher activation frequency in the right atrium versus the left atrium.

The utility of multisite mapping using a basket catheter with 32 bipoles was evaluated using the technique of frequency analysis (analyzing the component frequencies of a signal) of AF electrograms. The parameters of interest were 1) the average frequency of activation in each atrium, 2) peak frequency of activation in each atrium, and 3) the energy in each 5-9.5 Hz bandwidth. This corresponds to a heart rate of 300-570 beats per minute, typical ranges for AF activation. They began with 32-second recordings and used this duration to assess the reproducibility of the results.

Twelve goats were studied. Three models were studied. One, 3 animals in whom AF was inducible in the baseline state. Two, 6 animals that underwent only anatomic remodeling [creation of severe tricuspid regurgitation (TR)]. Three, 3 animals that underwent anatomic and electrical remodeling, including severe tricuspid regurgitation and rapid atrial pacing. All acute and chronic studies were performed with the animals intubated, ventilated, and anesthetized with halothane.

Anatomic remodeling was performed via a lateral incision, with the renal vein identified and cannulated. Under fluoroscopic guidance, a transvenous hook catheter was used to create TR until an 8-12 mmHg increase in right atrial pressure was noted. The vein was repaired when possible or a nephrectomy was performed. Anatomic and electrical remodeling was created by creating TR in the same manner. A transvenous pacing electrode was positioned via the jugular vein in the right atrium and attached to a high rate pacemaker programmed to a rate of 400 bpm.

Baseline testing was performed before TR was created in the three animals. The other animals returned to the lab for EP testing after 2-3 months. Echocardiography documented moderate to severe TR with right atrial enlargement. The renal vein was again cannulated, and blood pressure monitored via a catheter positioned in the renal artery.

A multielectrode (32 bipoles) basket catheter was positioned under fluoroscopic guidance in the right atrium via the renal vein. AF was induced by either burst or programmed stimulation if needed. One-minute recordings of sustained AF (lasting > 3 min) were acquired on a digital recorder with a bandpass of 3-250 Hz. Transeptal catheterization of the left atrium was attempted after recording right atrial AF. The multielectrode basket catheter was positioned under fluoroscopic guidance. AF was induced by burst or programmed stimulation if needed. One-minute recordings of AF were acquired.

Digitally stored data was downloaded from the EP recording system to a NEXT computer system and analyzed using Mathematica. The first and last 32 seconds were analyzed individually to evaluate the reproducibility of the results. Each 32-second sample was divided into 8-second sequential 4-second segments of AF for analysis.

Each 4-second sample underwent mean removal. A Hanning window was applied and the data was zero-padded to 8192 points; standard techniques in fast Fourier transform (FFT) analysis. An FFT was performed on each 4-second sample from which the power spectrum was calculated. The 8 power spectra were averaged and smoothed using a 5-point moving average fitter.

The analysis focused on the 5-9.5 Hz range (300-570 bpm). At each site, the average frequency of activation was calculated. In each atrium, the average and peak frequency of activation was calculated. The energy in the 5-9.5 Hz bandwidth was calculated.

A typical recording in a goat with atrial flutter demonstrated the utility of this technique, as shown in Figure 1. Intracardiac electrograms recorded from the basket catheter showed very uniform atrial activation at approximately a cycle length of 200 ms; typical atrial flutter. The power spectra based on the FFT analysis showed a peak at about 5 Hz, corresponding to the activation of 300 bpm or the atrial cycle length of 200 ms. Recordings in multiple sites in the atrium showed that in the same atrium there can be very simple and more complex atrial electrograms. Figure 2 shows a typical recording made from an animal with combined anatomic and electric remodeling in the right atrium. Fairly discrete atrial activation can be seen, which is difficult to characterize. The peak seen in the FFT in Figure 2 is much broader than that noted in atrial flutter because of more dispersion in the cycle length. As illustrated in Figure 3, recordings in multiple sites in the atrium result in more complex AF electrograms. The peak at about 5 Hz in the FFT identifies that to be the site of activation frequency. The electrograms shown in Figures 4 and 5 from the left atrium again illustrate that the electrograms can be more complex during AF with a broader range in frequency (Fig 5).

Figure 3. Recordings in multiple sites in the same atrium result in more complex atrial fibrillation electrograms. The peak at about 5 Hz in the FFT identifies that to be the site of activation frequency. (Goldberger 2000)

Figure 5. Left atrium electrogram illustrating the complexity of electrograms during atrial fibrillation with a broader rage in frequency. (Goldberger 2000)

The average frequency in each atrium in the 5-9.5 Hz bandwidth was highly correlated between the first and second 32 seconds (R²=0.97); the data are very reproducible (Fig. 6). No difference in the average frequency in the right atrium between baseline and the two models was seen. No statistically significant difference between the average frequency in the right and the left atriums was seen.

The peak frequency in each atrium was above 8 Hz (Fig. 7). The correlation between the first and second 32 seconds was quite high (R²=0.88). No differences between any of the models in the right atrium were seen, or any statistically significant differences between left and right atriums in the peak frequency of activation. The correlation for the energy under the curve in the 5-9.5 Hz range was slightly less (R²=0.76) (Fig. 8). A fair amount of dispersion in the data seemed to be present between the baseline and the right atrium and the left atrium with large standard deviations (Fig. 8).

Figure 8. The correlation for the energy under the curve in the 5-9.5 Hz range was slightly less, compared to the correlations for the average and peak frequencies. A fair amount of dispersion in the data seems to be present between the baseline and the right atrium and the left atrium with large standard deviations. (Goldberger 2000)

Study limitations include the small numbers in each group. Although the basket catheter records from multiple atrial sites, it is possible that the site responsible for driving AF (highest activation frequency) may be remote from any of the recording sites. Differences in spontaneous versus induced episodes may exist.

In conclusion, despite the presence of right atrial structure abnormalities, high frequency AF was not confined to the right atrium. This suggests that even in this setting, the right atrium may not necessarily be the driving force of the AF. Frequency analysis of AF electrograms is useful to identify the local activation frequency of AF. The frequency data are highly reproducible. Frequency analysis is an efficient way to map AF during episodes of AF.