Relationship between the spectral characteristics of atrial brillation and atrial tachycardias that occur after catheter ablation

of atrial brillation
Kentaro Yoshida, MD, Aman Chugh, MD, Magnus Ulfarsson, PhD,* Eric Good, DO, Michael Kuhne, MD, Thomas Crawford, MD, Jean F. Sarrazin, MD, Nagib Chalfoun, MD, Darryl Wells, MD, Warangkna Boonyapisit, MD, Srikar Veerareddy, MD, Sreedhar Billakanty, MD, Wai S. Wong, MD, Krit Jongnarangsin, MD, Frank Pelosi, Jr., MD, Frank Bogun, MD, Fred Morady, MD, Hakan Oral, MD
From the Division of Cardiovascular Medicine, University of Michigan, Ann Arbor, Michigan, and *Department of Electrical and Computer Engineering, University of Iceland, Reykjavik, Iceland.
BACKGROUND During catheter ablation of complex fractionated atrial electrograms, persistent atrial brillation (AF) may convert to an atrial tachycardia (AT). OBJECTIVE The purpose of this study was to investigate the possible mechanisms of AT by examining the spectral and electrophysiologic characteristics of AF and ATs that occur after catheter ablation of AF. METHODS The subjects of this study were 33 consecutive patients with persistent AF who had conversion of AF to AT during ablation of AF (group I) and 20 consecutive patients who underwent ablation of persistent AT that developed more than 1 month after AF ablation (group II). Spectral analysis of the coronary sinus (CS) electrograms and lead V1 was performed during AF at baseline, before conversion, and during AT. The spatial relationship between the AT mechanism and ablation sites was examined. RESULTS A spectral component with a frequency that matched the frequency of AT was present in the baseline periodogram of AF more often in group I (52%) than in group II (20%, P .02). Ablation resulted in a decrease in the dominant frequency of AF but not in the frequency of the spectral component that matched the AT. There was a signicant direct relationship between the baseline dominant frequency of AF and the frequency of AT in the CS in group I (r 0.76, P .0001) but not in group II (r 0.38, P .09). ATs were macroreentrant in 64% and 60% of patients in groups I and II, respectively (P .8). The AT site was more likely to be distant ( 1 cm) from AF ablation sites in group I (70%) than in group II (35%, P .007). CONCLUSION The ndings of this study suggest that ATs observed during ablation of AF often may be drivers of AF that become manifest after elimination of higher-frequency sources and brillatory conduction. KEYWORDS Atrial brillation; Atrial tachycardia; Catheter ablation; Spectral analysis (Heart Rhythm 2009;6:1117) 2009 Heart Rhythm Society. Published by Elsevier Inc. All rights reserved.

Persistent atrial brillation (AF) often involves drivers outside the pulmonary veins (PVs).1 4 Complex fractionated atrial electrograms are targeted in an attempt to ablate these drivers.37 When AF converts during ablation of complex fractionated atrial electrograms, it converts much more often to an atrial tachycardia (AT) than to sinus rhythm.1,4,8 The mechanistic implication of ATs that become manifest during ablation of complex fractionated atrial electrograms is unclear. The hypothesis underlying this study was that ATs to which AF converts during ablation represent drivers of persistent AF that were obscured by brillatory conduction
Supported in part by a grant from St. Jude Medical, Inc. Drs. Oral and Morady are founders and equity owners of Ablation Frontiers, Inc. Address reprint requests and correspondence: Dr. Hakan Oral, Cardiovascular Center, SPC 5853, 1500 East Medical Center Drive, Ann Arbor, Michigan 48109-5853. E-mail address: oralh@umich.edu. (Received June 17, 2008; accepted September 25, 2008.)

and by higher-frequency drivers. To test this hypothesis, we examined the spectral and electrophysiologic characteristics of AF and the ATs that the AF converted to during an ablation procedure.

Methods
Study subjects
The subjects of this study consisted of two groups of patients who underwent radiofrequency catheter ablation of persistent or long-lasting persistent AF. Group I consisted of 33 consecutive patients in whom AF converted to AT at the time of the catheter ablation procedure. Group II consisted of a control group of 20 consecutive patients who underwent a repeat ablation procedure for persistent AT at a mean of 7 5 months after catheter ablation of AF. In each of these 20 patients, AF had been converted to sinus rhythm by direct-current countershock during the index procedure. The
doi:10.1016/j.hrthm.2008.09.031

1547-5271/$ -see front matter 2009 Heart Rhythm Society. Published by Elsevier Inc. All rights reserved.

12
Table 1 Clinical characteristics of the study patients Group I (N 33) Age (years) Gender (male/female) Duration of atrial brillation (months) Left atrial size (mm) Left ventricular ejection fraction% Structural heart disease Ischemic heart disease Nonischemic cardiomyopathy Hypertensive heart disease
Values are given as mean SD.

Heart Rhythm, Vol 6, No 1, January 2009 grams recorded in the left atrium and CS with 20 ms variability in cycle length; and (3) stable activation sequence. As described previously, the mechanism of AT was considered to be macroreentry if activation mapping accounted for 90% of the tachycardia cycle length and if the diameter of the reentrant circuit was 3 cm.9 The mechanism of the tachycardia was considered to be due to microreentry if the diameter of the circuit was 3 cm. Sites where the postpacing interval was within 20 ms of the tachycardia cycle length were considered to be within the reentrant circuit. A focal mechanism was conrmed if mapping showed centrifugal activation from a point source. Among the 20 patients with AT in group II, 7 received an antiarrhythmic drug after the index ablation procedure for recurrent atrial arrhythmia: amiodarone in 4, sotalol in 1, and propafenone in 2. Therapy with propafenone and sotalol was discontinued four half-lives before the ablation procedure for AT, and therapy with amiodarone was discontinued 4 weeks before the procedure.

Group II (N 20) 63 9 12/8 57 52 46 56 9 5 1 7 4 8

P value .3 .1 .02 .8 .3 .7

61 8 27/6 32 24 45 53 13 5 4 8 6 8

clinical characteristics of the study subjects are given in Table 1.

Electrophysiologic study and catheter ablation

The study protocol was approved by the Institutional Review Board, and all patients provided informed written consent. Electrophysiologic studies were performed in the fasting state. Antiarrhythmic drug therapy was discontinued four to ve half-lives before the procedure, except for amiodarone, which was discontinued 8 to 12 weeks before the procedure. Vascular access was obtained through a femoral vein. A steerable quadripolar catheter with 2.5-mm interelectrode spacing (EP Technologies, Sunnyvale, CA, USA) was positioned in the coronary sinus (CS). The CS catheter was positioned in the posterolateral CS throughout the procedure. After the transseptal puncture, systemic anticoagulation was achieved with intravenous heparin to maintain an activated clotting time of 300 to 350 seconds. The PVs were mapped with a decapolar ring catheter (Lasso, Biosense Webster, Diamond Bar, CA, USA). An open-irrigation, 3.5-mm-tip deectable catheter (ThermoCool, Biosense Webster) was used for mapping and ablation. Bipolar electrograms were displayed and recorded at lter settings of 30 to 500 Hz during the procedure (EPMed Systems, West Berlin, NJ); they also were recorded at 0.5 to 200 Hz for ofine spectral analysis. Left atrial and PV geometry was reconstructed with an electroanatomic mapping system (CARTO, Biosense Webster). Conscious sedation was achieved with fentanyl and midazolam after barium swallow for visualization of the esophagus. Radiofrequency energy was delivered at a power of 25 to 35 W, maximum ow rate of 30 mL/min, and maximum temperature of 45C. The ablation strategy consisted of antral ablation to isolate all of the PVs, followed by ablation of complex fractionated atrial electrograms in the left atrium, CS, and right atrium aimed at conversion of AF to AT or sinus rhythm.

Digital signal processing and data analysis

All patients in group I presented to the laboratory in AF. CS electrograms and 12-lead ECG were acquired for 60 seconds at baseline, at 3 minutes before conversion of AF to AT, and during sustained AT 2 minutes after conversion of AF to AT. Spectral analysis of only the rst AT to which AF converted was performed in this study. In group II, 12 of 20 patients were in AT upon presentation to the electrophysiology laboratory, and sustained AT was induced by rapid atrial pacing in the remaining 8 patients. CS electrograms were acquired during sustained AT for 60 seconds before radiofrequency ablation. In addition, electrograms of baseline AF were acquired from the index procedure. The sites at which ATs were ablated were annotated on the electroanatomic maps. Electrograms recorded for 60 seconds in the CS and lead V1 were processed ofine in the MatLab environment (MathWorks, Inc., Natick, MA, USA) using custom software. As described previously, the QRS or QRST complexes (in lead V1) were subtracted.10 First, digitized bipolar electrograms, sampled for 60 seconds at 1,000 Hz (60,000 points), underwent the following preprocessing steps of band-pass ltering at 40 to 250 Hz, rectication, and lowpass ltering at 20 Hz. Then the discrete Fourier transform of the preprocessed signal was computed using the fast Fourier transformation algorithm to analyze the 0.5- to 80-Hz spectral band. An estimate of the signal spectrum was obtained by computing the periodogram, which is the modulus squared of the discrete Fourier transform. The frequency resolution was 0.017 Hz. The dominant frequency (DF) was dened as the frequency of the highest peak of the smoothed periodogram in the interval from 0.5 to 20 Hz.11 AF has signicant periodic elements with varying degrees of regularity. The Fourier transform takes advantage of the fact that continuous signals can be decomposed to a sum of weighted sinusoidal functions.12 Although the DF is

Mapping and ablation of AT

AT was dened by three criteria: (1) discrete and monomorphic P waves on the ECG; (2) regularity of the electro-

Yoshida et al

Atrial Tachycardia and Atrial Fibrillation

13 cycle length of AT was 225 20 ms (4.48 0.40 Hz) in group I and 229 33 ms (4.46 0.65 Hz) in group II (P .62). Among patients in group II, there was no signicant difference in the mean cycle length of induced and spontaneous ATs (218 27 ms vs 236 36 ms, P .25). The baseline DF of AF in lead V1 also was similar between groups I and II (5.69 0.72 Hz and 5.87 0.68 Hz, respectively, P .38).

A
(202 ms)

6.60 6 60 Hz 4.96 Hz

B
4.85 Hz
(206 ms)

5.76 Hz

AT frequency and spectral components of AF in the CS

A spectral component of AF with a frequency similar to the AT frequency was identied in the baseline AF periodogram in 17 (52%) of 33 patients in group I (Figure 1) and in 4 (20%) of 20 patients in group II (P .02). A similar spectral component was identied in the periodogram of the AF shortly before conversion to AT in 26 (79%) of 33 patients in group I and before transthoracic cardioversion of AF to sinus rhythm in 10 (50%) of 20 patients in group II (P .03). A matching spectral component was identied in 7 (78%) of 9 patients in group I with an AT involving the mitral isthmus or the CS and in none of the 8 patients with a microreentry localized to the roof or the anterior wall (P .001). A matching spectral component was identied in 10 (63%) of 16 of the remaining ATs (Table 2).

(207 ms)

4.84 Hz

Figure 1 Periodogram of atrial brillation (AF) recorded in the coronary sinus at baseline (A), 3 minutes prior to conversion (B), and the frequency of atrial tachycardia (AT; C). At baseline, the dominant frequency (DF) of AF is 6.60 Hz (A). There is a spectral component with a frequency of 4.96 Hz (arrow). Ablation of complex fractionated atrial electrograms results in a decrease in DF of AF; however, there is no change in the frequency of the spectral component (B, arrow). After termination of AF to AT, the frequency of AT (4.84 Hz) is similar to the frequency of the spectral component identied in the periodogram of AF (C). The mechanism of AT was mitral isthmus dependent utter in this example. Cycle length is given in parentheses.

AT frequency and spectral components of AF in lead V1

A spectral component with a frequency similar to the AT frequency was identied in the baseline periodogram of AF in lead V1 in 10 (30%) of 33 patients in group I and in 3 (15%) of 20 patients in group II (P .21). A similar spectral component was identied in the periodogram of AF in lead V1 shortly before conversion to AT in 14 (42%) of 33 patients in group I and before cardioversion of AF to sinus rhythm in 3 (15%) of 20 patients in group II (P .04). A matching spectral component was identied in 7 (78%) of 9 patients in group I who had a cavotricuspid
Table 2 Mechanisms of atrial tachycardia Group I (N 33) Macroreentry Cavotricuspid isthmus utter Mitral isthmus utter Interatrial septum Roof Right-sided PV antrum Microreentry Anterior wall Coronary sinus Left-sided PV antrum Focal Left atrial septum Right-sided PV antrum
PV pulmonary vein.

the frequency of the sinusoidal waveform with the highest power, the Fourier transform also shows neighboring periodic and stable waves with less power as other peaks in the periodogram. The periodogram during AF was systematically analyzed to identify spectral components other than the DF (and its harmonics) that matched the frequency of the AT (within 5%). All spectral components with a peak frequency power 20% of the DF were compared with the frequency of the AT.

Group II (N 20) 12 0 10 1 0 1 7 1 3 3 1 0 1

P value .8 .008 .003 .4 .1 .3 .7 .06 .8 .1 .9 .4 .4

Statistical analysis
Continuous variables are expressed as mean 1 SD and were compared by Students t-test or paired t-test, as appropriate. Categorical variables were compared by Chi-square analysis or with Fishers exact test, as appropriate. P .05 was considered signicant.

Results
DF of AF and frequency of AT
The baseline DF of AF in the CS was similar in groups I and II (5.64 0.69 Hz vs 5.73 0.58 Hz, P .66). The mean

21 9 6 3 2 1 10 6 3 1 2 1 1

14

Heart Rhythm, Vol 6, No 1, January 2009

Figure 2 Conversion of atrial brillation (AF) to atrial tachycardia (AT). Shown are ECG leads I, II, III, aVF, and V1 intracardiac electrograms recorded from the distal bipole of an ablation catheter positioned in the left atrium (Abl) and distal bipole of a quadripolar catheter positioned in the coronary sinus (CS). Electrograms were recorded at baseline (A), 3 minutes before conversion to AT (B), during conversion of AF to AT (C) and during AT (D). During transition from AF to AT, the AF became more organized before converting to AT. At times during the ablation procedure, the degree of organization varied, suggesting a gradual effect of ablation on brillatory conduction.

isthmus-dependent atrial utter. In the other two patients, the mechanism of AT was macroreentry around the rightsided PVs in one and a focal tachycardia originating close to the right-sided antrum in the other patient. In group I, the prevalence of a matching spectral component in lead V1 was higher when AT was cavotricuspid isthmus dependent than when it was not (78% vs 13%, P .001).

Figure 5). A similar signicant correlation also existed between the baseline DF of AF in lead V1 and the frequency of AT in group I (r 0.58, P .0005) but not in group II (r 0.03, P .92).

Mechanisms of AT in group I
In group I, all ATs were targeted for ablation during the index procedure. AT converted to sinus rhythm during ablation in 22 (67%) of 33 patients. The remaining 11 patients underwent transthoracic cardioversion to terminate AT because of a long procedure duration ( 5 hours). The mechanism of the 33 ATs was macroreentry in 21 (64%), microreentry in 10 (30%), and focal AT in 2 (6%) patients (Table 2). The macroreentrant circuit involved the cavotricuspid isthmus in 9 patients, the

Duration and effect of ablation and DF of AF and AT

The mean duration of RF for conversion of AF to AT was 75 18 minutes in group I (Figure 2). The mean duration of radiofrequency energy application was 62 13 minutes in group II. There was a signicant correlation between the DF of AF in the CS at baseline and the duration of RF required for conversion to AT (r 0.54, P .002). The DF of AF was signicantly lower shortly before conversion to AT (5.10 0.51 Hz) than at baseline (5.64 0.69 Hz) in group I (P .0001, Figure 3). However, there was no signicant change in the frequency of the spectral component of AF that matched the frequency of the AT (4.52 0.46 Hz and 4.54 0.46 Hz, P .8, Figure 3). In group II, the DF of AF shortly before cardioversion to sinus rhythm was signicantly lower than at baseline (5.13 0.62 Hz vs 5.73 0.58 Hz, P .0004). The analysis of DF of AF in lead V1 yielded similar results to those of DF of AF in CS as described earlier (Figure 3).

Relationship between the DF of AF and frequency of AT

There was a direct correlation between the baseline DF of AF in the CS and the frequency of AT in group I (r 0.76, P .0001, Figure 4) but not in group II (r 0.38, P .09, Figure 5). There also was a signicant correlation between the DF of AF shortly before conversion to AT and the frequency of AT in group I (r 0.66, P .0001, Figure 4) but not in group II (r 0.29, P .21,
Figure 3 Effect of ablation on dominant frequency (DF) of atrial brillation (AF). Shown are the DF of AF recorded in the coronary sinus (CS; hatched bars) and lead V1 (gray bars) at baseline and before conversion of AF to atrial tachycardia (AT) among patients in group I. The DF recorded from the CS and lead V1 were similar in all groups. Ablation resulted in a signicant decrease in the DF of AF. However, the frequency of the spectral component in the periodogram of AF that matched AT and frequency of AT after conversion was similar.

Yoshida et al

Atrial Tachycardia and Atrial Fibrillation

15

Figure 4 Relationship between atrial brillation (AF) and atrial tachycardia (AT) recorded in the coronary sinus among patients in group I. There was a signicant direct correlation between the dominant frequency (DF) of AF and the frequency of AT at baseline (A) and shortly before conversion to AT (B).

mitral isthmus in 6, interatrial septum in 3, and left atrial roof in 2, and there was macroreentry around the rightsided PVs in 1 patient. The 10 microreentrant circuits were localized to the left atrial anterior wall in 6 patients, the CS in 3, and the left-sided PV antrum in 1. In another 2 patients, the sites of AT were the left atrial septum and the right-sided PV antrum.

Discussion
Main ndings
The main ndings of this study were as follows. (1) There often was a spectral component in the periodogram of AF that matched the frequency of the AT to which AF converted. (2) Although ablation of AF resulted in a decrease in the DF of AF, it did not eliminate the frequency in the periodogram that matched the frequency of the subsequent AT. (3) There was a direct relationship between the duration of ablation necessary for conversion of AF to AT and the baseline DF of AF. (4) The baseline DF of AF strongly correlated with the frequency of AT to which the AF converted during ablation of AF but not with the frequency of ATs that developed late after ablation. (5) ATs to which AF converted during ablation usually were not adjacent to AF ablation sites, whereas ATs that developed late after AF ablation usually were adjacent to prior ablation sites. These ndings suggest that the majority of ATs that occur during ablation of AF coexist with higher-frequency sources during AF and may represent drivers of AF that become manifest only after elimination of higher-frequency drivers and brillatory conduction. On the other hand, as previously reported, ATs that develop late after AF ablation usually are a manifestation of proarrhythmia caused by ablation-induced conduction slowing or a gap in an ablation line.13

Mechanisms of AT in group II
The mechanisms of AT in group II were macroreentry in 12 (60%) patients, microreentry in 7 (35%), and a focal AT in 1 (5%, Table 2). The macroreentrant circuits involved the mitral isthmus in 10 patients, the left atrial septum in 1, and right-sided PV antrum in 1. The sites of 7 microreentrant ATs were the left-sided PV antrum in 3 patients, the CS in 3, and the left atrial anterior wall in 1 (Table 2).

Spatial relationship between AF ablation sites and AT

ATs that used a reentrant circuit or had a site of origin within 1 cm of a prior ablation site were considered to be secondary to radiofrequency ablation. In group I, 23 (70%) of 33 ATs were not adjacent to an AF ablation site, and 10 (30%) were considered secondary to ablation. In group II, 7 (35%) of 20 ATs were not adjacent to an AF ablation site, and 13 (65%) were secondary to ablation (P .01).

Figure 5 Relationship between atrial brillation (AF) and atrial tachycardia (AT) recorded in the coronary sinus among patients in group II. There was no signicant relationship at baseline (A) or shortly before cardioversion to sinus rhythm (B). DF dominant frequency.

16

Heart Rhythm, Vol 6, No 1, January 2009 dependent atrial utter is one of the lower-frequency drivers in some patients with AF, possibly explaining the longrecognized association between the two arrhythmias.18 Of note, microreentrant and focal ATs arising in the left atrium, because of their smaller size, would be more likely than macroreentrant ATs to be undetected in a CS electrogram. Therefore, the proportion of ATs caused by macroreentry may have been overestimated in this study.

Conversion of AF during catheter ablation

ATs that occur acutely during an AF ablation procedure could be due to either proarrhythmia or emergence of an underlying AF driver. ATs that occur late after AF ablation have been demonstrated to usually be related to a gap in a prior ablation line, consistent with proarrhythmia.13 The match between a frequency in the AF periodogram and the AT frequency strongly suggests that acutely occurring ATs are a manifestation of preexisting drivers that are uncovered by the ablation process. It seems likely that ablation eliminated higher-frequency drivers that created brillatory conduction in the atria. This presumably allowed an AT with a lower frequency to become manifest.

DF of AF and duration of ablation necessary to convert AF to AT

There was a signicant direct relationship between the DF of AF at baseline and the duration of ablation necessary to convert AF to AT in this study. This observation suggests that drivers with a higher frequency may be more likely to result in brillatory conduction and a higher prevalence of complex fractionated atrial electrograms. It also is possible that, in a remodeled atrium where the DF of AF is often higher, multiple drivers may develop and subsequently necessitate more extensive ablation.

Frequency of AF and AT
In a stable and electroanatomically homogeneous substrate, tachycardias with different frequencies would not coexist. However, during AF, anatomic and functional heterogeneities in conduction, refractoriness, and propagation exist in the atria. Fixed and functional barriers14,15 may explain how tachycardias with a lower frequency than the DF were identied in the AF periodograms in this study. In this study, there was a direct correlation between the frequency of the AF and the AT to which AF converted during ablation. The DF of AF is a function of the refractory period, and the atrial effective refractory period also is one of the determinants of AT cycle length. Different degrees of remodeling during persistent AF will inuence atrial refractoriness to different degrees. Therefore, the direct relationship between the DF of AF and the frequency of the ATs to which AF converted during ablation may be explained by the extent to which atrial remodeling affected atrial refractoriness among the patients in this study.

Periodogram of AF in the CS versus lead V1

CS electrograms reect both local CS and adjacent left atrial depolarizations, but they do not reect global spatiotemporal dynamics in the left and right atria. As would be expected, ATs closer to the CS (e.g., mitral isthmus utters) were more readily identied in the periodogram of the AF than the ATs remote from the CS (e.g., left atrial roof utters). Therefore, it is likely that spectral analysis of global atrial activation would have enabled identication of a higher proportion of ATs in the periodogram of AF than identied in this study. To better represent global atrial activation, lead V1 was also analyzed in this study. Although lead V1 often has a more favorable signal-to-noise ratio compared with the other leads, it may be more representative of right than left atrial depolarizations.19 Right atrial ATs, in fact, were easily identied in lead V1 in this study. Precise assessment of global left atrial depolarization may require simultaneous high-density multisite mapping, which often is not feasible during the course of an ablation procedure in human subjects.

Role of macroreentry in AF
The ATs in group I were mostly due to macroreentry, not localized microreentry or focal sources. This does not necessarily imply that the underlying drivers of AF are more often macroreentrant as opposed to microreentrant or focal. As evident from the baseline AF periodogram, the AT frequency always was lower than the DF of AF. It is possible that the higher-frequency components of the periodogram were caused by high-frequency rotors or focal discharges. In this study, cavotricuspid isthmus-dependent atrial utter accounted for approximately 40% of the macroreentrant ATs in group I. In none of the 9 cases ablation was performed in the right atrium before emergence of these right atrial utters. Therefore, proarrhythmia was an unlikely cause of these cavotricuspid isthmus-dependent atrial utters. In a prior study, the AF recurrence rate after PV isolation tended to be lower when cavotricuspid isthmus ablation was also performed than when it was not.16 In a more recent study, a history of atrial utter was associated with a higher AF recurrence rate after PV isolation, suggesting that non-PV drivers may be more likely in patients with atrial utter.17 It is possible that cavotricuspid isthmus-

Study limitations
The majority of ATs in group II occurred in proximity to prior ablation sites and therefore appeared to be gap related. However, the possibility that some of these ATs in group II were primary ATs that coexisted with AF during the index procedure and persisted after ablation cannot be excluded because AF did not terminate during ablation in any of the patients in group II. These residual ATs may have resurfaced during follow-up with a different frequency because of changes in the electrophysiologic properties of the atrium as a result of radiofrequency ablation.

Conclusion
The genesis of AF is multifactorial. The PVs and their antra play a critical role in the initiation and perpetuation of AF.

Yoshida et al

Atrial Tachycardia and Atrial Fibrillation

17
9. Chae S, Oral H, Good E, et al. Atrial tachycardia after circumferential pulmonary vein ablation of atrial brillation: mechanistic insights, results of catheter ablation, and risk factors for recurrence. J Am Coll Cardiol 2007;50:17811787. 10. Lemola K, Ting M, Gupta P, et al. Effects of two different catheter ablation techniques on spectral characteristics of atrial brillation. J Am Coll Cardiol 2006;48:340 348. 11. Yoshida K, Ulfarsson M, Tada H, et al. Complex electrograms within the coronary sinus: time- and frequency-domain characteristics, effects of antral pulmonary vein isolation and relationship to clinical outcome in patients with paroxysmal and persistent atrial brillation. J Cardiovasc Electrophysiol 2008; 19:10171023. 12. Ng J, Goldberger JJ. Understanding and interpreting dominant frequency analysis of AF electrograms. J Cardiovasc Electrophysiol 2007;18:680 685. 13. Chugh A, Oral H, Lemola K, et al. Prevalence, mechanisms, and clinical signicance of macroreentrant atrial tachycardia during and following left atrial ablation for atrial brillation. Heart Rhythm 2005;2:464 471. 14. Berenfeld O, Zaitsev AV, Mironov SF, et al. Frequency-dependent breakdown of wave propagation into brillatory conduction across the pectinate muscle network in the isolated sheep right atrium. Circ Res 2002;90:11731180. 15. Klos M, Calvo D, Yamazaki M, et al. The atrial septopulmonary bundle of the posterior left atrium provides a substrate for AF initiation in a model of vagally mediated pulmonary vein tachycardia of the structurally normal heart. Circ Arrhythmia Electrophysiol 2008;1:175183. 16. Scharf C, Veerareddy S, Ozaydin M, et al. Clinical signicance of inducible atrial utter during pulmonary vein isolation in patients with atrial brillation. J Am Coll Cardiol 2004;43:20572062. 17. Moreira W, Timmermans C, Wellens HJ, et al. Can common-type atrial utter be a sign of an arrhythmogenic substrate in paroxysmal atrial brillation? Clinical and ablative consequences in patients with coexistent paroxysmal atrial brillation/atrial utter. Circulation 2007;116:2786 2792. 18. Waldo AL, Feld GK. Inter-relationships of atrial brillation and atrial utter mechanisms and clinical implications. J Am Coll Cardiol 2008;51: 779 786. 19. Dibs SR, Ng J, Arora R, et al. Spatiotemporal characterization of atrial activation in persistent human atrial brillation: multisite electrogram analysis and surface electrocardiographic correlationsa pilot study. Heart Rhythm 2008;5:686 693.

However, drivers beyond the PVs also are important in persistent AF. The ndings of this study suggest that macroreentrant ATs often coexist with persistent AF, becoming manifest only after ablation of higher-frequency drivers and/or brillatory conduction. It is possible that these macroreentrant ATs also serve as drivers of AF. The most efcient way to identify and ablate AF drivers without having to perform extensive ablation at sites of passive brillatory conduction remains to be determined in future studies.

References
1. Haissaguerre M, Hocini M, Sanders P, et al. Catheter ablation of long-lasting persistent atrial brillation: clinical outcome and mechanisms of subsequent arrhythmias. J Cardiovasc Electrophysiol 2005;16:1138 1147. 2. Haissaguerre M, Hocini M, Sanders P, et al. Localized sources maintaining atrial brillation organized by prior ablation. Circulation 2006;113:616 625. 3. Nademanee K, McKenzie J, Kosar E, et al. A new approach for catheter ablation of atrial brillation: mapping of the electrophysiologic substrate. J Am Coll Cardiol 2004;43:2044 2053. 4. Oral H, Chugh A, Good E, et al. Radiofrequency catheter ablation of chronic atrial brillation guided by complex electrograms. Circulation 2007;115:2606 2612. 5. Jalife J, Berenfeld O, Mansour M. Mother rotors and brillatory conduction: a mechanism of atrial brillation. Cardiovasc Res 2002;54:204 216. 6. Kalifa J, Tanaka K, Zaitsev AV, et al. Mechanisms of wave fractionation at boundaries of high-frequency excitation in the posterior left atrium of the isolated sheep heart during atrial brillation. Circulation 2006;113:626 633. 7. Konings KT, Smeets JL, Penn OC, et al. Conguration of unipolar atrial electrograms during electrically induced atrial brillation in humans. Circulation 1997;95:12311241. 8. Haissaguerre M, Sanders P, Hocini M, et al. Catheter ablation of long-lasting persistent atrial brillation: critical structures for termination. J Cardiovasc Electrophysiol 2005;16:11251137.