613

Prospective Study of Atrial Fibrillation TerminationDuring Ablation Guided by Automated Detection

of Fractionated ElectrogramsMICHAEL PORTER, M.D., WILLIAM SPEAR, M.D., JOSEPH G. AKAR, M.D., PH.D.,

RAY HELMS, M.D., NEIL BRYSIEWICZ, B.S.E., PETER SANTUCCI, M.D.,and DAVID J. WILBER, M.D.

From the Cardiovascular Institute, Loyola University Medical Center, Maywood, Illinois, USA

AF Ablation Guided by Automated CFAE Detection. Introduction: Complex fractionated atrialelectrograms (CFAE) may identify critical sites for perpetuation of atrial fibrillation (AF) and provideuseful targets for ablation. Current assessment of CFAE is subjective; automated detection algorithms mayimprove reproducibility, but their utility in guiding ablation has not been tested.

Methods and Results: In 67 patients presenting for initial AF ablation (42 paroxysmal, 25 persistent), LAand CS mapping were performed during induced or spontaneous AF. CFAE were identified by an onlineautomated computer algorithm and displayed on electroanatomical maps. A mean of 28 ± 18 sites/patientwere identified (20 ± 13% of mapped sites), and were more frequent during persistent AF. CFAE occurredmost commonly within the CS, on the atrial septum, and around the pulmonary veins. Ablation initiallytargeting CFAE terminated AF in 88% of paroxysmal AF, but only 20% of persistent AF (P < 0.001).Subsequently, additional ablation was performed in all patients (PV isolation for paroxysmal AF, PV iso-lation + mitral and roof lines for persistent AF). Minimum follow-up was 1 year. One-year freedom fromrecurrent atrial arrhythmias without antiarrhythmic drug therapy after a single procedure was 90% forparoxysmal AF, and 68% for persistent AF.

Conclusions: Ablation guided by automated detection of CFAE proved feasible, and was associated with ahigh AF termination rate in paroxysmal, but not persistent AF. As an adjunct to conventional techniques, itwas associated with excellent long-term single procedure outcomes in both groups. Criteria for identifyingoptimal CFAE sites for ablation, and selection of patients most likely to benefit, require additional study.(J Cardiovasc Electrophysiol, Vol. 19, pp. 613-620, June 2008.)

atrial fibrillation, ablation, mapping, electrograms, atrium

Introduction

The identification of rapid repetitive discharges from thepulmonary veins (PV) as initiating and perpetuating triggersfor AF by Haissaguerre et al. in 1998 focused attention the im-portance of the musculature of the PVs and peri-ostial atriumin the genesis of AF.1 However, PV/antral isolation aloneis associated with subsequent AF recurrence in 20–40% ofpatients with paroxysmal AF, and 40–70% with AF of long-standing (>1 year) duration after a single procedure.5-8 Whilea substantial portion of recurrences appear related to electri-cal reconnection through previously isolated areas,9,10 focalsources for initiating and maintaining AF remote from the

This work was supported in part by a grant from the George M. EisenbergFoundation.

Dr. Wilber serves as a consultant on an AF advisory board and has receivedhonoraria relevant to this topic.

Address for correspondence: David J. Wilber, M.D., Bldg. 110, Room 6232,Loyola University Medical Center, 2160 S. First Ave., Maywood, IL 60153,USA. Fax: 708-327-2377; E-mail: dwilber@lumc.edu

Manuscript received 10 March 2008; Revised manuscript received 12 March2008; Accepted for publication 14 March 2008.

doi: 10.1111/j.1540-8167.2008.01189.x

pulmonary veins are also identified.11,12 Linear lesions con-necting the PVs to other anatomic barriers provide incremen-tal improvements in outcome over PV isolation alone,13,14

although conduction block across these lines is often diffi-cult to attain and regions of incomplete block may be proar-rhythmic. These data have prompted a continued search foradjunctive/alternative strategies to improve outcomes.

Nademanee and colleagues were the first to propose thattargeting sites with complex fractionated atrial electrograms(CFAE) could be an effective primary strategy for AF abla-tion.15 They defined CFAE as low amplitude multicomponentpotentials (generally 0.05–0.25 mv) that are either continuousor separated by short isoelectric intervals (<120 ms) aver-aged over a 10-second recording interval. They hypothesizedthat CFAE identified critical anatomic regions that perpet-uate reentry. Subsequent experience with CFAE ablation asa primary16 or adjunctive12,17-20 strategy by other investi-gators has yielded conflicting results. An important limita-tion of most clinical studies to date is reliance on subjectiveand qualitative visual assessment of local electrograms to de-termine the presence of CFAE during AF. This approach ishighly dependent on operator experience and judgment. Inter-and intraobserver variability and the reproducibility of sub-jective classification are unknown. An objective quantitativeapproach would facilitate comparison of outcomes betweeninvestigators, refinement of “high yield” criteria, and a short-ened learning curve for new operators.

614 Journal of Cardiovascular Electrophysiology Vol. 19, No. 6, June 2008

Recently, computerized algorithms for automated recog-nition and quantitation of CFAE have been introduced as in-tegrated modules in 3D mapping systems.21,22 However, theoutcomes of CFAE ablation at sites characterized by auto-mated algorithm have not been reported. The purpose of thisstudy was to prospectively evaluate the role of automatedCFAE detection in identifying ablation sites associated withtermination of ongoing AF, and in reducing long-term re-currence as an adjunct strategy in patients undergoing AFablation.

Methods

Patients

The study included 67 patients referred for initial catheterablation of paroxysmal (N = 42) or persistent (N = 25) AFunresponsive to at least one antiarrhythmic drug. The char-acteristics of the study population are displayed in Table 1.Patients with persistent AF were younger, had a greater fre-quency of left ventricular dysfunction, and larger left atrial(LA) size. In patients with persistent AF, the median durationof continuous AF preceding the procedure was 13 months(range 3–60 months), and was >1 year in 17 patients (68%).All patients underwent preprocedure evaluation, including64 slice computed tomography (CT) for the evaluation ofLA and PV anatomy and transesophageal echocardiographyto exclude LA thrombus. All patients gave written informedconsent, and the study protocol was approved by the institu-tional review board.

Electrophysiological Study

Patients were studied in the absence of antiarrhythmicdrugs. In patients previously receiving amiodarone, the drugwas stopped at least 1 month prior to the procedure. A 14-pole catheter was positioned via the right internal jugular veinsuch that the distal group of seven electrodes was in the coro-nary sinus (CS), and the proximal group in the lateral rightatrium (TZ Medical, Portland, OR, USA). This catheter wasused for reference and internal cardioversion. The LA wasinstrumented with a 3.5 mm irrigated tip catheter (Thermo-cool Navistar, Biosense Webster, Diamond Bar CA, USA)and a circular mapping catheter (Lasso, Biosense Webster),delivered through a single transseptal puncture. A heparinbolus (100 units/kg) was given at the time of transseptalpuncture, and the activated clotting time kept at 300–350 sec-onds by additional heparin bolus throughout the remainder of

TABLE 1

Baseline Clinical Characteristics

All Patients Paroxysmal Persistent(N = 67) (N = 42) (N = 25)

Age (years)∗ 59 ± 10 61 ± 9 56 ± 10Men (n,%) 51 (76%) 29 (69%) 22 (88%)AF history (years) 6.3 ± 6.1 6.7 ± 6.1 5.4 ± 5.9Structural heart disease (n,%) 34 (51%) 20 (48%) 14 (56%)Hypertension (n,%) 36 (54%) 23 (55%) 13 (52%)LVEF < 50%∗ 17 (25%) 5 (12%) 12 (48%)LA volume (mL)∗ 96 ± 31 85 ± 23 116 ± 34

∗P < 0.05 paroxysmal vs persistent.AF = atrial fibrillation; LA = left atrium; LVEF = left ventricular ejectionfraction.

the procedure. Following transseptal catheterization, LA andCS electroanatomical mapping (Carto XP, Biosense Webster)was performed during spontaneous or induced AF (>5 minduration prior to initiation of mapping). Mapping of the rightatrium to detect CFAE was not performed. A mean of 143 ±38 sites/patient were sampled, with the fill threshold set to10 mm to facilitate homogenous mapping density. Bipolarelectrograms were filtered at 30–400 kHz. CT registrationand fusion of LA reconstructions with the electroanatomicalmap was subsequently performed (<2.5 mm mean surfaceregistration error). Intracardiac echocardiography (Acunav,Siemens Medical, Malvern, PA, USA) was used to guidetransseptal catheterization, confirm the accuracy of PV ostiallocation registered on CT, guide subsequent positioning ofmapping and ablation catheters, and identify collateral struc-tures. All mapping points were acquired outside the PV ostia,as confirmed by echo and CT.

Identification of CFAE

At each mapping site, a 2.5-second window of bipo-lar electrograms was analyzed online by programmablesoftware (CFAE Software Module, Biosense Webster) thatprovides online automated identification and electroanatom-ical display of CFAE. The extent and repetitiveness ofelectrogram fractionation were determined by algorithms de-scribed below and illustrated in Figure 1. Initially, low ampli-tude high-frequency electrograms were identified by taggingthe peak voltage of bipolar deflections (positive or negative)that exceeded a programmable lower threshold (±0.05 mvfor this study) to exclude noise. Voltage peaks greater thanthis threshold but less than an upper threshold (±0.15 mvfor this study) were subsequently identified. Next, the inter-vals between successive peaks falling within the low voltagewindow of 0.05 and 0.15 mv were calculated. Finally, inter-vals that fell within a programmable duration (60–120 msfor this study) were identified, and the number of such inter-vals during the entire 2.5 second sampling window summed(designated the interval confidence level [ICL]). Sites withgreater number of short intervals between low amplitude mul-tideflection complexes (higher ICL) reflected more frequentand repetitive CFAE, and thus potential targets for ablation.21

Since no standardized criteria were available to identify theoptimal extent of fractionation to guide ablation, we chosea relatively conservative cutoff for the ICL (>7 intervals),which represented approximately 15% of all sampled sitesin a preliminary series of 10 patients with paroxysmal AF.These sites were considered to represent primary CFAE. Pri-mary CFAE sites were subsequently represented as discretepoints on electroanatomical maps, which also depicted ICLlevel as a color-coded continuous gradient (Figs. 2 and 3).Primary CFAE sites were subsequently categorized into oneof nine anatomic regions: coronary sinus, peri left PV, periright PV, septum, inferior, lateral, anterior (including LA ap-pendage), roof, and posterior.

Study Protocol

The study was designed primarily to determine the impactof CFAE ablation based on automated detection algorithmson the termination of AF. For all patients, the initial step ofablation was to target all primary CFAE sites, beginning inthe posterior wall and peri-PV region, followed by the LAroof, septum, inferior wall, anterior wall, and, finally, the CS.

Porter et al. AF Ablation Guided by Automated CFAE Detection 615

Figure 1. Application of the automated detection algorithm. The example is a 2.5 second recording displaying a bipolar left atrial electrogram (top) andbody surface ECG lead V3. The horizontal brown line represents the lower voltage boundary (±0.05) and the blue line the upper voltage boundary (±0.15)for detection of complex fractionated atrial electrograms (CFAE). Electrogram peaks within this voltage range are tagged (white dots). The intervals betweentagged peaks are calculated, some of which are displayed in red below the electrogram. Intervals falling within a specified range (60–120 ms) are counted,with the number representing the interval confidence level (ICL). In this example, the ICL is 9, representing a primary CFAE site. See text for details.

Radiofrequency (RF) energy was delivered with a peak pow-ers of 25 W (CS), 35 W (posterior wall), and 45 W (otherLA sites), and a maximum temperature of 45◦C. Energy wasapplied for 30 seconds or until local electrograms becameorganized. During ablation, unipolar electrograms were dis-played along with bipolar electrograms, as the former facil-itated monitoring of voltage reduction during ablation as anadditional endpoint for energy delivery. Termination of AFwas defined as a direct transition to sinus rhythm or to an or-

Figure 2. Representative electrograms and electroanatomical maps of the left atrium in the posterior-anterior (PA) and right anterior oblique (RAO)projections from a patient with paroxysmal atrial fibrillation (AF). The primary CFAE sites (ICL >7) are depicted as red dots on the electroanatomical maps.The map colors reflect the ICL gradient from lowest values (purple) to highest (red), with a range of 1–14 in this example. The proximal pulmonary veins(PV) are shown in dark blue and were not mapped. Panels A-C depict electrograms from closely adjacent primary CFAE sites near the left superior PV.Ablation at the site illustrated by panel A terminated AF. Panel E depicts an electrogram from the inferior LA, where ablation also terminated AF. Panel D isan example of an electrogram where few peaks or intervals fall within the specified ranges, and does not demonstrate CFAE. See text for additional details.Other abbreviations as in previous figures.

ganized atrial tachycardia or flutter. In patients with paroxys-mal AF, if AF terminated during CFAE ablation, a minimumof five attempts were made to reinduce AF by burst pacingfrom the CS at the shortest cycle length permitting 1:1 cap-ture for 10 seconds. If AF was reinduced, additional primaryCFAE sites were ablated during AF. No attempt was made toreinduce AF in patients with persistent AF. For all patients,by protocol, CFAE ablation continued until all primary siteswere ablated.

616 Journal of Cardiovascular Electrophysiology Vol. 19, No. 6, June 2008

Figure 3. Illustration of a representativeelectrogram and electroanatomical maps ofthe left atrium in a patient with persistent AF.Note that there are multiple high ICL regionsand primary CFAE sites in the posterior, in-ferior, septal, and anterior regions of the LA.Ablation at a site in the inferior LA adjacent tothe coronary sinus terminated AF; the electro-gram from this site is shown in the top panel.Abbreviations as in previous figures.

Since the long-term efficacy of ablation of CFAE detectedby automated algorithm was unknown at the onset of thisstudy, by design all patients underwent additional ablationirrespective of AF termination or noninducibility followingCFAE ablation. Patients with paroxysmal AF underwent widearea isolation of right and left PVs in ipsilateral pairs. Theendpoint of PV isolation was the elimination of all PV poten-tials detectable by a circular mapping catheter positioned atmultiple sites within and around the PV ostia. In patients withpersistent AF, in addition to PV isolation, linear lesions wereproduced between the upper pulmonary veins (roof line) andbetween the left inferior PV and the mitral annulus (mitralline). Endpoints for the latter ablation were conduction blockassessed as previously described.13,14 If AF did not terminateby CFAE or additional ablation, patients underwent internalDC cardioversion. At the end of the ablation procedure, all pa-tients underwent infusion of high dose isoproterenol (10–20µg/min). PV isolation and lines of block were reconfirmed,and additional ablation of isoproterenol-induced spontaneousfocal activity in the right or left atria, if present, was under-taken. Ablation of the cavotricuspid isthmus was performedin patients with induced atrial flutter or a prior history of thatarrhythmia.

Follow-Up

Patients were seen in the arrhythmia clinic at 1 and3 months and at 3 monthly intervals thereafter. Symp-toms suggestive of arrhythmias were evaluated by ECG ortranstelephonic monitoring at any time during follow-up. Inaddition, a 24-hour Holter was performed at 6 months, fol-lowed by a transtelephonic monitor for 1 month with dailytransmissions to ascertain the presence of asymptomatic ar-rhythmias. AAD (except amiodarone) were continued for1 month in patients with paroxysmal AF and 2–3 monthsin patients with persistent AF. A CT scan of the LA andPV was performed at 3 months to identify the presence ofsignificant (>50%) PV stenosis. AAD were withheld in all

patients after 3 months. Recurrent atrial arrhythmias duringthe first 3 months of follow-up were not considered treatmentfailures. In patients with persistent AF, early recurrences thatlasted more than a few days were treated by cardioversion.After 3 months, any documented atrial arrhythmia lasting≥30 seconds was considered a recurrence.5 All patients werefollowed for a minimum of 1 year.

Statistics

Categorical data were presented as frequencies and con-tinuous data as mean ± standard deviation. Comparisons be-tween continuous variables were made by unpaired t-test andbetween categorical variables by chi-square or Fisher’s exacttest, as appropriate. A two-tailed P-value <0.05 was con-sidered significant. All analyses were performed with SYS-TAT 12.0 statistical software (Systat Software Inc., San Jose,CA, USA).

Results

Frequency and Distribution of CFAE

Electrograms from a total of 9,581 sites were analyzed. In83% of these electrograms, at least one CFAE complex (ICL≥1) was present, underscoring the importance of more selec-tive criteria for repetitive CAFE to identify ablation targets.Primary CFAE (ICL >7) were identified at a mean of 28 ±18 sites per patient (20 ± 13% of mapped sites), and tendedto be more frequent in patients with persistent AF (Table2). Sites classified as primary CFAE by automated analy-sis demonstrated repetitive low amplitude multicomponentelectrograms (Fig. 2A,E), less complex deflections with veryshort coupling intervals (Fig. 2B), or intermediate patterns(Fig. 2C). There were significant variations in probabilityof finding primary CFAE sites among different LA regionsacross all patients (P < 0.001, Fig. 4). While CFAE werewidely distributed throughout the LA, they were most com-

Porter et al. AF Ablation Guided by Automated CFAE Detection 617

TABLE 2

Procedural Data

All Patients Paroxysmal Persistent(N = 67) (N = 42) (N = 25)

Primary CFAE sites (n)∗ 28 ± 18 21 ± 15 35 ± 22Proportion primary 20 ± 13% 18 ± 13% 24 ± 14%

CFAE sites (n)CFAE ablation (min)∗ 12 ± 6 10 ± 5 15 ± 7Total ablation (min)∗ 50 ± 13 42 ± 6 64 ± 9CTI ablation (n,%) 33 (49%) 24 (57%) 9 (36%)Additional RA ablation (n,%) 6 (9%) 3 (7%) 3 (12%)Procedure time (min)∗ 197 ± 39 180 ± 23 225 ± 44Fluoroscopy time (min)∗ 33 ± 9 30 ± 7 37 ± 10

∗P < 0.05 paroxysmal vs persistent.CFAE = complex fractionated atrial electrograms; CTI = cavotricuspidisthmus ablation; RA = right atrium.

monly found in the CS (79% of patients), LA septum (69% ofpatients), and adjacent to the left PVs (64%). There were nosignificant differences in the regional distribution of CFAEbetween patients with persistent compared with paroxysmalAF. The mean time for completion of CFAE mapping anddisplay was 21 ± 8 min.

CFAE Ablation and Termination of AF

During ablation at CFAE sites, at least one terminationof AF occurred in 37 of 42 patients (88%) presenting withparoxysmal AF (Figs. 2 and 5, Table 2). In these 37 patients,there were a total of 54 termination sites (2 sites in 9 patients,3 sites in 4 pts). These sites were located adjacent to the leftPVs in 18 (33%), adjacent to the right PVs in 8 (15%), theLA roof in 8 (15%), the CS in 6 (11%), LA septum in 5 (9%),inferior LA in 5 (9%), posterior LA in 2 (4%), and anterior LAin 2 (4%). Following completion of primary CFAE ablationalone, AF could no longer be induced in 33 patients (79%). Ofthe five patients without AF termination after completion ofprimary CFAE ablation, subsequent AF termination occurredin four during PV isolation. Of the four patients with at leastone AF termination during completion of CFAE ablation,but persistently inducible AF, two had termination duringPV isolation. Four of these six patients had no inducible AF

Figure 4. Proportion of patients with at least one primary CFAE site withineach of the predefined regions of the left atrium. CFAE sites were mostcommon in the coronary sinus (CS), septum, and around the left pulmonaryveins (PV). There was significant variation in the frequency of CFAE acrossregions (P < 0.001).

Figure 5. Proportion of atrial fibrillation (AF) termination during ablationstratified by clinical presentation with paroxysmal or persistent AF. At leastone termination occurred during CFAE ablation in 88% of paroxysmal, butonly 20% of persistent AF (P < 0.001).

at the end of the procedure. The remaining three patients hadcontinued AF and required cardioversion at the end of theprocedure.

In contrast, termination during CFAE ablation occurred inonly 5 of 25 patients (20%) with persistent AF (P < 0.001vs paroxysmal AF). In four of these five patients, the mostrecent episode of AF was >1 year in duration. Of the fivetermination sites, three were located in the coronary sinus,and two in the inferior LA (Figs. 3 and 5, Table 2). Of theremaining 20 patients with continued AF following CFAEablation, AF termination occurred during PV isolation in onepatient, and during deployment of roof or mitral lines in ninepatients. The remaining eight patients had continued AF andrequired cardioversion at the end of the procedure.

Following CFAE ablation, all PV ostia were examinedby the circular mapping catheter prior to additional ablation.No patient had evidence of isolation of any vein, irrespec-tive of prior AF termination. This observation suggests thatPV isolation did not play a role in AF termination associatedwith CFAE ablation. The duration of radiofrequency energyapplication at CFAE sites was greater for patients with per-sistent AF, reflecting the greater number of such sites in thesepatients (Table 2).

Follow-Up

In patients with paroxysmal AF, late recurrences of atrialarrhythmias after 3 months occurred in four patients (AF in3, atrial flutter in 1). The remaining 38 patients (90%) havebeen free of recurrent atrial arrhythmias without AAD 1 yearafter a single ablation procedure. In patients with persistentAF, late recurrences of atrial arrhythmias after 3 months oc-curred in eight patients (AF in 3, atrial flutter in 4, both in 1).The remaining 17 patients (68%) have been free of recurrentatrial arrhythmias without AAD 1 year after a single ablationprocedure.

Additional procedural details are provided in Table 2.Both total RF duration and procedural times were signifi-cantly longer for patients with persistent AF. One patient hada moderate pericardial effusion at the end of the procedurethat was treated by pericardiocentesis without long-term se-quela. There were no other complications and no patient hadevidence of significant PV stenosis.

618 Journal of Cardiovascular Electrophysiology Vol. 19, No. 6, June 2008

Discussion

Major Findings

This study demonstrates for the first time the feasibilityand outcomes of using CFAE detected by automated algo-rithm to guide ablation of AF. Paroxysmal AF terminatedin 88% of patients during CFAE ablation at sites identi-fied by automated detection, similar or better than the 55–89% termination rates reported in comparable groups of pa-tients by investigators using subjective visual assessment ofCFAE.15,18,19 Furthermore, in 79% of patients presentingwith paroxysmal AF, AF was no longer inducible by fol-lowing ablation of CFAE alone. In this study, terminationduring CFAE ablation in persistent AF was observed in only20% of patients. In patients with persistent AF undergoingablation guided by visually identified CFAE, termination ororganization of AF has been reported in 16–80%, althoughpersistent AF recurred in 40–70% of patients during the firstyear of follow-up irrespective of acute termination.15,16,19

Differences in the area and duration of CFAE ablation be-tween studies may be partially responsible for the variableoutcome in this population.

This study was not designed to assess the impact of CFAEas the sole strategy for AF ablation, and all patients receivedadditional ablation determined both by clinical presentationand provoked arrhythmias during the procedure. It is pos-sible that some patients, particularly those with paroxysmalAF who were noninducible after CFAE ablation alone, mayhave done well without additional ablation. However, the sin-gle procedure freedom from recurrent atrial arrhythmias inthe absence of antiarrhythmic therapy at one year follow-upassociated with adjunct automated detection and ablation ofCFAE (90% in paroxysmal, and 68% in persistent AF) com-pares favorably with other strategies.5-7,23

We found that repetitive frequent computer detected CFAEhad a nonhomogenous regional distribution and were mostfrequently located in the coronary sinus, atrial septum, andadjacent to the pulmonary veins, consistent with the reportof Nademanee and coworkers for visually detected CFAE.15

There were no significant differences in the regional distribu-tion of CFAE between paroxysmal and persistent AF. How-ever, AF termination sites were adjacent to the PV in ∼50%of patients with paroxysmal AF, but in none of the patientswith persistent AF, similar to the findings of Schmitt et al.for visually detected CFAE.19 These findings potentially canbe explained by both differing mechanisms for the genera-tion of CFAE between the two groups of patients, as well asdifferences in the critical structures required for maintenanceof AF. It is also possible that CFAE ablation is less effectiveas an isolated strategy in the termination and elimination ofpersistent AF.16

Significance of CFAE

Experimental and clinical data provide a number of poten-tial mechanisms by which CFAE are generated during AF, in-cluding (1) sites of reentrant circuit pivot points, wave turningand slow conduction,24,25 (2) sites of anisotropic conductionor focal reentry,26,27 (3) sites of wavebreak at the peripheryof high-frequency rotors,28,29 and (4) sites of local autonomicinnervation and activation.30-33 Some of these mechanismsindicate that CFAE sites represent or are closely adjacent tostructures critical to the maintenance of AF, while others in-

dicate a more passive role. Multiple mechanisms (and bothactive and passive) may coexist in the same patient at differ-ent sites or times.34 These data suggest that all CFAE maynot be the same, and that stability over time, relationship toactivity at adjacent sites, and specific electrogram character-istics may all play a role in differentiating sites critical to AFmaintenance.

Automated Detection Algorithms

The identification of CFAE by automated algorithms pro-vides objective reproducible criteria for evaluating potentialablation sites, and in comparing outcomes between investi-gators. Acquisition of mapping data and 3D display of CFAEsites both as points and regional gradients is relatively rapid,at most 15–30 minutes. The impact of varying criteria for theextent of electrogram fractionation can be rapidly assessed,and may facilitate a more nuanced evaluation of “high-yield”sites.

Alternative criteria to those employed in this study for se-lection of sites with sufficient electrogram fractionation towarrant ablation may have been more effective, particularlyin patients with persistent AF. However, our goal in this ini-tial study was to test the effectiveness of ablation at siteswith the highest degree of electrogram fractionation usinguniform and objective standards, while limiting unnecessarytissue destruction. Refinements in these criteria may comefrom correlations with the subjective ratings of experiencedphysicians, though it is unclear that such ratings represent anideal “gold standard.” Further investigation into the patternsand mechanisms of electrogram fractionation in various clini-cal settings, and in experimental models, are likely to provideadditional clarification.

The algorithm employed in this study analyzed a 2.5-second window of electrograms. While Nademanee reportedlong-term temporal stability of visually assessed CFAE,15

the impact of more extended sampling durations on the auto-mated characterization of CFAE is unclear. Preliminary dataindicate that CFAE detected by automated criteria remain rel-atively stable and reproducible over short-term intervals.35,36

Role of CFAE Elimination During AF Ablation

By study design, CFAE ablation was performed as the ini-tial procedure in all patients. Other ablation strategies mayincorporate potential CFAE sites without specifically exam-ining electrogram characteristics. If wide area or antral PVisolation had been performed first, at least 50% of all CFAEtermination sites in patients with paroxysmal AF would havebeen incorporated within the ablation area. However, a sig-nificant minority of termination sites were remote from thePVs. Such sites could potentially contribute to the recurrenceof AF following initial PV isolation procedures. CFAE abla-tion as an adjunct strategy in patients with paroxysmal AF athigh risk for recurrence following PV isolation alone (such aspersistent inducibility of AF,17,36-38 or very dilated LA) mayreduce recurrences and the need for additional procedures.The results of this study and others17 suggest that such anapproach may be useful. However, the superiority of adjunctCFAE ablation over PV isolation alone requires additionalprospective validation.

In patients with persistent AF, PV isolation alone un-commonly terminates AF and is associated with a highrecurrence rate. Haissaguerre and colleagues incorporated

Porter et al. AF Ablation Guided by Automated CFAE Detection 619

ablation at sites of visually identified CFAE as part of a step-wise strategy, including PV isolation and linear lesions.20,23

This strategy resulted in an 87% acute termination rate of per-sistent AF, with 57% of patients free of AF during a follow-up of 11 ± 6 months after a single procedure, and 95% afteran additional procedure. Ablation of CFAE identified by au-tomated detection could play a similar adjunct role in thispopulation.

Conclusions

Automated detection algorithms provide a rapid, objec-tive, and systematic approach to the identification of siteswith highly fractionated electrograms. CFAE characterizedby these algorithms have a regional distribution and responseto ablation similar to those identified by subjective visual as-sessment. As an adjunct to conventional techniques, CFAEablation guided by automated detection was associated witha single procedure 1-year freedom from recurrent atrial ar-rhythmias without antiarrhythmic drug therapy in 90% ofpatients with paroxysmal AF, and 68% of patients with per-sistent AF. Optimal criteria for identifying the most effec-tive CFAE sites for ablation, and the selection of patientsmost likely to benefit from this approach, require additionalinvestigation.

References

1. Haissaguerre M, Jais P, Shah DC, Takahashi A, Hocini M, QuiniouG, Garrigue S, Le Mouroux A, Le Metayer P, Clementy J.: Sponta-neous initiation of atrial fibrillation by ectopic beats originating in thepulmonary veins. N Engl J Med 1998;339:659-666.

2. Hocini M, Ho SY, Kawara T, Linnenbank AC, Potse M, Shah D, JaisP, Janse MJ, Haissaguerre M, De Bakker JM: Electrical conduction incanine pulmonary veins: Electrophysiological and anatomic correlation.Circulation 2002;105:2442-2448.

3. Arora R, Verheule S, Scott L, Navarrete A, Katari V, Wilson E, Vaz D,Olgin JE: Arrhythmogenic substrate of the pulmonary veins assessedby high-resolution optical mapping. Circulation 2003;107:1816-1821.

4. Chen YJ, Chen SA: Electrophysiology of pulmonary veins. J CardiovascElectrophysiol 2006;17:220-224.

5. Calkins H, Brugada J, Packer DL, Cappato R, Chen SA, Crijns HJ,Damiano RJ Jr, Davies DW, Haines DE, Haissaguerre M, Iesaka Y,Jackman W, Jais P, Kottkamp H, Kuck KH, Lindsay BD, March-linski FE, McCarthy PM, Mont JL, Morady F, Nademanee K, Na-tale A, Pappone C, Prystowsky E, Raviele A, Ruskin JN, Shemin RJ:HRS/EHRA/ECAS expert consensus statement on catheter and surgicalablation of atrial fibrillation: Recommendations for personnel, policy,procedures and follow-up. Heart Rhythm 2007;4:816-861.

6. Natale A, Raviele A, Arentz T, Calkins H, Chen SA, HaissaguerreM, Hindricks G, Ho Y, Kuck KH, Marchlinski F, Napolitano C,Packer D, Pappone C, Prystowsky EN, Schilling R, Shah D, Themis-toclakis S, Verma A: Venice Chart international consensus documenton atrial fibrillation ablation. J Cardiovasc Electrophysiol 2007;18:560-580.

7. Willems S, Klemm H, Rostock T, Brandstrup B, Ventura R, Steven D,Risius T, Lutomsky B, Meinertz T: Substrate modification combinedwith pulmonary vein isolation improves outcome of catheter ablationin patients with persistent atrial fibrillation: A prospective randomizedcomparison. Eur Heart J 2006;27:2871-2878.

8. Kanj M, Wazni O, Natale A: Pulmonary vein antrum isolation. HeartRhythm 2007;S73–S79.

9. Ouyang F, Antz M, Ernst S, Hachiya H, Mavrakis H, Deger FT, Schau-mann A, Chun J, Falk P, Hennig D, et al: Recovered pulmonary veinconduction as a dominant factor for recurrent atrial tachyarrhythmiasafter complete circular isolation of the pulmonary veins: Lessons fromdouble Lasso technique. Circulation 2005;111:127-135.

10. Verma A, Kilicaslan F, Pisano E, Marrouche NF, Fanelli R, BrachmannJ, Geunther J, Potenza D, Martin DO, Cummings J, Burkhardt JD,Saliba W, Schweikert RA, Natale A: Response of atrial fibrillation topulmonary vein antrum isolation is directly related to resumption anddelay of pulmonary vein conduction. Circulation 2005;112:627-635.

11. Callans DJ, Gerstenfeld EP, Dixit S, Zado E, Vanderhoff M, Ren JF,Marchlinski FE: Efficacy of repeat pulmonary vein isolation proceduresin patients with recurrent atrial fibrillation. J Cardiovasc Electrophysiol2004;15:1050-1055.

12. Arentz T, von RJ, Blum T, Stockinger J, Burkle G, Haissaguerre M,Hocini M, Sanders P, Takahashi Y, Rotter M, Sacher F, Rostock T,Hsu LF, Jonsson A, O’Neill MD, Bordachar P, Reuter S, Roudaut R,Clementy J, Jais P: Localized sources maintaining atrial fibrillationorganized by prior ablation. Circulation 2006;113:616-625.

13. Hocini M, Jais P, Sanders P, Takahashi Y, Rotter M, Rostock T, Hsu LF,Sacher F, Reuter S, Clementy J, et al.: Techniques, evaluation, and con-sequences of linear block at the left atrial roof in paroxysmal atrial fib-rillation: A prospective randomized study. Circulation 2005;112:3688-3696.

14. Jais P, Hocini M, Hsu LF, Sanders P, Scavee C, Weerasooriya R, MacleL, Raybaud F, Garrigue S, Shah DC, Le Metayer P, Clementy J, Haissa-guerre M: Technique and results of linear ablation at the mitral isthmus.Circulation 2004;110:2996-3002.

15. Nademanee K, McKenzie J, Kosar E, Schwab M, SunsaneewitayakulB, Vasavakul T, Khunnawat C, Ngarmukos T: A new approach forcatheter ablation of atrial fibrillation: Mapping of the electrophysiologicsubstrate. J Am Coll Cardiol 2004;43:2044-2053.

16. Oral H, Chugh A, Good E, Wimmer A, Dey S, Gadeela N, Sankaran S,Crawford T, Sarrazin JF, Kuhne M, Chalfoun N, Wells D, Frederick M,Fortino J, Benloucif-Moore S, Jongnarangsin K, Pelosi F Jr, Bogun F,Morady F: Radiofrequency catheter ablation of chronic atrial fibrillationguided by complex electrograms. Circulation 2007;115:2606-2612.

17. Oral H, Chugh A, Lemola K, Cheung P, Hall B, Good E, Han J, TamirisaK, Bogun F, Pelosi F Jr, Morady F: Noninducibility of atrial fibrilla-tion as an end point of left atrial circumferential ablation for paroxys-mal atrial fibrillation: A randomized study. Circulation 2004;110:2797-2801.

18. Oral H, Chugh A, Good E, Sankaran S, Reich SS, Igic P, Elmouchi D,Tschopp D, Crawford T, Dey S, Wimmer A, Lemola K, JongnarangsinK, Bogun F, Pelosi F Jr, Morady F: A tailored approach to catheterablation of paroxysmal atrial fibrillation. Circulation 2006;113:1824-1831.

19. Schmitt C, Estner H, Hecher B, Luik A, Kolb C, Karch M, NdrepepaG, Zrenner B, Hessling G, Deisenhofer I: Radiofrequency ablation ofcomplex fractionated atrial electrograms (CFAE): Preferential sites ofacute termination and regularization in paroxysmal and persistent atrialfibrillation. J Cardiovasc Electrophysiol 2007;18:1039-1046.

20. Haissaguerre M, Sanders P, Hocini M, Takahashi Y, Rotter M, Sacher F,Rostock T, Hsu LF, Bordachar P, Reuter S, Roudaut R, Clementy J, JaisP: Catheter ablation of long lasting atrial fibrillation. Critical structuresfor termination. J Cardiovasc Electrophysiol 2005;15:1125-1137.

21. Nademanee K, Schwab M, Porath J, Abbo A: How to per-form electrogram-guided atrial fibrillation ablation. Heart Rhythm2006;3:981-984.

22. Scherr D, Dalal D, Cheema A, Cheng A, Henrikson CA, Spragg D,Marine JE, Berger RD, Calkins H, Dong J: Automated detection andcharacterization of complex fractionated atrial electrograms in hu-man left atrium during atrial fibrillation. Heart Rhythm 2007;4:1013-1020.

23. Haissaguerre M, Sanders P, Hocini M, Takahashi Y, Rotter M, Sacher F,Rostock T, Hsu LF, Bordachar P, Reuter S, Roudaut R, Clementy J, JaisP: Catheter ablation of long lasting persistent atrial fibrillation: Clini-cal outcome and mechanisms of subsequent arrhythmias. J CardiovascElectrophysiol 2005;16:1138-1147.

24. Konings KT, Kirchhof CJ, Smeets JR, Wellens HJ, Penn OC, AllessieMA: High-density mapping of electrically induced atrial fibrillation inhumans. Circulation 1994;89:1665-1680.

25. Konings KTS, Smeets JLRM, Penn OC, et al.: Configuration of unipo-lar atrial electrograms during electrically induced atrial fibrillation inhumans. Circulation 1997;95:1231-1241.

26. Spach MS, Dolber PC, Heidlage JF: Influence of the passive anisotropicproperties on directional differences in propagation following modifi-cation of the sodium conductance in human atrial muscle. A modelof reentry based on anisotropic discontinuous conduction. Circ Res1988;62:811-832.

27. Spach MS, Heidlage JF, Dolber PC, Barr RC: Mechanism of origin ofconduction disturbances in aging human atrial bundles: Experimentaland model study. Heart Rhythm 2007;4:175-185.

28. Kalifa J, Tanaka K, Zaitsev AV, Warren M, Vaidyanathan R, AuerbachD, Pandit S, Vikstrom KL, Ploutz-Snyder R, Talkachou A, AtienzaF, Guiraudon G, Jalife J, Berenfeld O: Mechanisms of wave fraction-ation at boundaries of high-frequency excitation in the posterior left

620 Journal of Cardiovascular Electrophysiology Vol. 19, No. 6, June 2008

atrium of the isolated sheep heart during atrial fibrillation. Circulation2006;113:626-633.

29. Tanaka K, Zlochiver S, Vikstrom KL, Yamazaki M, Moreno J, KlosM, Zaitsev AV, Vaidyanathan R, Auerbach DS, Landas S, GuiraudonG, Jalife J, Berenfeld O, Kalifa J: Spatial distribution of fibrosis gov-erns fibrillation wave dynamics in the posterior left atrium during heartfailure. Circ Res 2007;101:839-847.

30. Armour JA, Murphy DA, Yuan BX, Macdonald S, Hopkins DA: Grossand microscopic anatomy of the human intrinsic cardiac nervous sys-tem. Anat Rec 1997;247:289-298.

31. Lin J, Scherlag BJ, Zhou J, Lu Z, Patterson E, Jackman WM, LazzaraR, Po SS: Autonomic mechanism to explain complex fractionated atrialelectrograms (CFAE). J Cardiovasc Electrophysiol 2007;18:1197-1205.

32. Lemery R, Birnie D, Tang AS, Green M, Gollob M: Feasibility study ofendocardial mapping of ganglionated plexuses during catheter ablationof atrial fibrillation. Heart Rhythm 2006;3:387-396.

33. Mcclelland JH, Duke D, Reddy R: Preliminary results of a limitedthoracotomy: New approach to treat atrial fibrillation. J CardiovascElectrophysiol 2007;18:1289-1295.

34. Rostock T, Rotter M, Sanders P, Takahashi Y, Jais P, Hocini M, Hsu LF,Sacher F, Clementy J, Hassaguerre M: High-density activation mapping

of fractionated electrograms in the atria of patients with paroxysmalatrial fibrillation. Heart Rhythm 2006;3:27-34.

35. Scherr D, Dalal D, Cheema A, Nazarian S, Bilchick K, Cheng A, Henrik-son C, Spragg D, Marine JE, Berger RD, Calkins H: Temporal stabilityof complex fractionated atrial electrograms in human left atrium duringatrial fibrillation (abstract). Circulation 2007;116:II-392.

36. Stiles MK, Brooks A, John B, Dhar S, Wilson L, Kuklik P, DimitriH, Lau H, Roberts-Thomson RL, Mackenzie L, Willoughby S, YoungGD, Sanders P: Effect of the electrogram duration on quantificationof complex fractionated atrial electrograms and dominant frequency. JCardiovasc Electrophysiol 2008;19:252-258.

37. Chang SL, Tai CT, Lin YJ, Wongcharoen W, Lo LW, Tuan TC, UdyavarAR, Chang SH, Tsao HM, Hsieh MH, Hu YF, Chen YJ, Chen SA: Theefficacy of inducibility and circumferential ablation with pulmonaryvein isolation in patients with paroxysmal atrial fibrillation. J CardiovascElectrophysiol 2007;18:607-611.

38. Haissaguerre M, Sanders P, Hocini M, Hsu LF, Shah DC, ScaveeC, Takahashi Y, Rotter M, Pasquie JL, Garrigue S, Clementy J,Jais P: Changes in atrial fibrillation cycle length and inducibilityduring catheter ablation and their relation to outcome. Circulation2004;109:3007-3013.