1024

Diagnostic Accuracy of a New Software for ComplexFractionated Electrograms Identification in Patients

with Persistent and Permanent Atrial FibrillationLEONARDO CALO, M.D., F.E.S.C.,∗ ERMENEGILDO DE RUVO, M.D.,∗ LUIGI SCIARRA, M.D.,∗

ROBERTO GRICIA, B.S.,† GIOVANNA NAVONE, M.D.,∗ LUCIA DE LUCA, M.D.,∗

FRANCESCA NUCCIO, M.D.,∗ ANTONELLA SETTE, M.D.,∗ CRISTIAN PRISTIPINO, M.D.,‡ALESSANDRO DULIO, B.S.,† FIORENZO GAITA, M.D.,§ and ERNESTO LIOY, M.D.∗

From the ∗Division of Cardiology, Policlinico Casilino, ASL Roma B, Rome, Italy; †Biosense Webster Italy, Johnson & Johnson Medical,Milan, Italy; ‡Department of Cardiovascular Diseases, San Filippo Neri Hospital, Rome, Italy; and §Division of Cardiology,

Ospedale Civile di Asti, Asti, Italy

CARTO Mapping of CFAEs. Introduction: The elimination of complex fractionated atrial electro-grams (CFAEs) has been proposed as a potential target for guiding successful AF substrate ablation. Thepossibility to efficiently map the atria and rapidly identify CFAEs sites is necessary, before the CFAEsablation becomes a routine approach. The aims of this study, conducted in patients with persistent andpermanent atrial fibrillation (AF), were to analyze by CARTO mapping in the right (RA) and in the leftatrium (LA) during AF: (1) the diagnostic accuracy of a new software for CFAEs analysis, (2) the spatialdistribution of CFAEs, (3) the regional beat to beat AF intervals (FF).

Methods and Results: Twenty-five consecutive patients (four women, 58.8 ± 11.4 years) undergoing ra-diofrequency catheter ablation for persistent and permanent AF were enrolled in the study. The CFAEsoftware showed a high sensitivity (90%) and specificity (91%) in the identification of CFAEs, using a spe-cific setting of parameters. The LA had a significantly higher prevalence of CFAEs as compared with theRA (30.5% vs 20.3%, P = 0.016). The CFAEs were mostly present in the septum and in the area of coronarysinus ostium (CS os). The FF intervals were significantly shorter in the LA than in the RA (P < 0.01).

Conclusion: CARTO system has a high diagnostic accuracy in the identification of CFAEs. Atrial electricalactivity (CFAEs, mean FF intervals) during AF showed a significant spatial inhomogeneity. (J CardiovascElectrophysiol, Vol. 19, pp. 1024-1030, October 2008)

atrial fibrillation, mapping, catheter ablation, complex fractionated atrial electrograms

Introduction

In the last years, several clinical investigations1-4 haveobserved that atrial areas of slow conduction, functional con-duction of block, and pivot points can be implicated in thepersistence of atrial fibrillation (AF). Such areas are gener-ally characterized by the presence of complex fractionatedatrial electrograms (CFAEs).1 The elimination of CFAEs hasshown to be an effective target for guiding successful AFsubstrate ablation.5-7 However, little information exists onthe spatial distribution of CFAEs in both the left and the rightatrium (RA).

The specific purposes of this study, conducted in a groupof patients with persistent and permanent AF, were: (1) to de-termine the diagnostic accuracy of a new software for CFAEsanalysis of the CARTO� system; (2) to analyze the spatialdistribution of CFAEs; (3) to evaluate the regional FF intervalin both atria during AF.

Drs. Gricia and Dulio are employees of Biosense-Webster Inc.

Address for correspondence: Leonardo Calo, M.D., F.E.S.C., Division ofCardiology, Policlinico Casilino, Via Buonarroti, 16-00047 Marino, Rome,Italy. Fax: +39-06-93668162; E-mail: leonardo.calo@tin.it

Manuscript received 20 January 2008; Revised manuscript received 24February 2008; Accepted for publication 14 March 2008.

doi: 10.1111/j.1540-8167.2008.01219.x

Methods

Study Subjects

Twenty-five consecutive patients (four women, 58.8 ±11.4 years) undergoing radiofrequency catheter ablation forsymptomatic, persistent (15 patients) or permanent AF (10patients), were enrolled in the study. Inclusion criteria were(1) AF resistant to more than three attempts of pharmaco-logical and/or electrical cardioversion or (2) recurrent, per-sistent AF despite prophylaxis with at least three differentantiarrhythmic drugs (class I and/or III). AF was defined aspersistent when lasting more than 7 days and as permanentAF when resistant to cardioversion or relapsing within 24hour.8 The clinical characteristics of the patients are shown inTable 1. The study protocol was approved by the institutionalreview board.

Electrophysiological Study

All patients provided informed, written consent beforethe study. All antiarrhythmic drugs except amiodarone hadbeen discontinued for at least five half-lives. All patientsunderwent transesophageal echocardiography before theprocedure to exclude the presence of thrombus. Electrophys-iologic study was performed with the patient in the fastingstate using mild sedation. A quadripolar standard catheterwas placed in the coronary sinus (CS). After transseptalcatheterization, heparin was titrated to maintain an activated

Calo et al. CARTO Mapping of CFAEs 1025

TABLE 1

Baseline Characteristics

Studied Population(n = 25)

Age 58.8 ± 11.4Sex, M/F, n 21/4Duration of AF, y 7 ± 4Persistent AF, n 15Permanent AF, n 10Structural heart disease, n 16

Hypertension, n 8Coronary artery disease, n 3Idiopathic dilated cardiomyopathy, n 3Valvular disease, n 1Hypertrophic cardiomyopathy, n 1

Left atrial diameter, mm 47.5 ± 6Left ventricular ejection fraction,% 51.4 ± 8.6

clotting time of 250 to 300 seconds. An irrigated-tip quadripo-lar ablation catheter with a distal 3.5-mm tip and then three1-mm electrodes separated by 2, 5, and 2 mm (Navistar Ther-mocool, Biosense-Webster, Diamond Bar, CA, USA) wasused for mapping and ablation. Real-time 3D left atrial andright atrial maps were reconstructed using a nonfluoroscopicmapping system (CARTO�, Biosense-Webster). Poor con-tact with the endocardium was prevented by fluoroscopic vi-sualization of catheter motion in relation to cardiac shape andand by Carto XP system (a distance of 2 mm between twosuccessive beats is usually noticed).

Study Protocol

In all patients, the mapping procedure was performed dur-ing AF. The electroanatomic mapping was performed beforethe ablation procedure in the RA and in the left atrium (LA)by the CARTO� System, provided with a new software forCFAEs analysis. The software associates to each acquiredpoint the corresponding electrogram that lasts 2.5 seconds.On this electrogram, a measurement of the distance betweentwo consecutive deflections of the isoelectric is taken, ac-cording to a specific set of parameters. Parameters regardthe voltage amplitude of the signal and the distance betweenconsecutive peaks. The software annotates only spikes thathave amplitude included in a predefined window, and then itmeasures all intervals between two consecutive spikes (posi-tive or negative). All intervals with value included between aminimum duration and the maximum duration threshold (ex-pressed in ms) are assigned to each point of the map. The totalamount of assigned intervals leads the level of fragmentationof the potential. The CARTO� System with CFAE Softwareassociates to each point all intervals that satisfy this set ofparameters. The software for CFAE’s analysis, can display,in addition to the local activation time, unipolar and bipo-lar voltage, propagation and isochronal map, three additionalkinds of maps, according to the analysis of the intracardiacsignal of each acquired point. These three new maps are:(1) shortest complex interval (SCI) that displays the valueof the shortest interval between two consecutive deflections,according to an arbitrary predefined setting of the voltage andthe distance between consecutive peaks; (2) average complexinterval (ACI) that displays the average value for all intervalsconsidered for each point; (3) interval confidence level (ICL)that displays the total amount of intervals counted for each

point, with the possibility to show two different color tags forpoints with high (red) or medium (blue) level of fragmenta-tion. The threshold of ICL that distinguishes points with highICL (high fragmentation = HF) and medium ICL (mediumfragmentation = MF) can be determined for each patient. Forthis study, we have considered as high ICL the points with ≥20 intervals and medium ICL the points with 10 to 20 inter-vals. These cut-offs are arbitrary and have been determinedon the basis of a preliminary visual analysis of a biatrial map-ping in 10 patients, in order to avoid the inclusion of pointswith few or transient CFAEs.

Different types of analysis were performed: (1) compar-ison of various settings of CFAE software to determine thebest one in the identification of fragmented electrograms; (2)prevalence of CFAE in the different regions of the right andthe left atrium; (3) measurement of FF intervals. The level offragmentation and the analysis of FF interval were evaluatedin 12 different regions of both atria. In the LA: (1) poste-rior wall, including the PV ostial regions; (2) anterior wall;(3) lateral wall; (4) roof; (5) inferior wall; (6) septum, in theRA; (7) posterior wall; (8) lateral wall; (9) anterior wall; (10)cavotricuspid isthmus; (11) CS os; (12) septum. Four differ-ent parameter settings of CFAE software were compared toevaluate sensitivity, specificity, positive predictive value, andnegative predictive value in the identification of CFAE, withrespect to manual determination. The CARTO map used todetermine the presence of CFAEs was the ICL map. Both thepoints with HF and those with MF were considered togetherfor our analysis as fragmented electrograms. Given the factthat each CFAE per se is an electrogram composed of multipledeflections, the following settings were considered: (a) Min-imum duration of 15 ms with a maximum duration of 30 ms(setting A); (b) Minimum duration of 30 ms with a maximumduration of 50 ms (setting B); (c) Minimum duration of 50 mswith a maximum duration of 70 ms (setting C); (d) Minimumduration of 70 ms with a maximum duration of 120 ms (set-ting D). The last one is the default setting of CFAE software.For all settings, the minimum threshold voltage was 0.05 mV;in case of relevant noise, the minimum voltage was increasedto 0.07 mV. The maximum voltage was set to 1 mV in orderto consider all signals for the settings A, B, and C; for thesetting D, the maximum voltage was 0.15 mV (default).

We have chosen to analyze these settings considering thatthe CFAEs are multicomponent potentials with very short in-tervals between the different peaks of the electrograms andthat longer intervals could identify double potentials, that is,areas of collisions or blocks. The electrograms acquired werecarefully reviewed and classified as fragmented or not frag-mented by two expert electrophysiologists blinded to the soft-ware analysis. Differences were solved by consensus. CFAEswere defined as fractionated electrograms exhibiting multipledeflections from the isoelectric line (≥ 3 deflections) and/orpotentials with continuous electrical activity without an iso-electric line.2

The prevalence of CFAE was determined by using theICL map. The setting used for this analysis was the bestobtained in terms of sensitivity and specificity in the iden-tification of CFAE. Each point was automatically classifiedby the software into three categories: (1) HF; (2) MF; (3) notfragmented.

The FF interval was measured by using the ACI map. Thesetting used was: low voltage threshold 0.05 mV; high volt-age threshold 1 mV; minute duration 110 ms; max duration

1026 Journal of Cardiovascular Electrophysiology Vol. 19, No. 10, October

TABLE 2

Comparison of the Four Parameters Settings of Software for Identifying CFAEs

Setting A Setting B Setting C Setting D(15–30 ms) (30–50 ms) (50–70 ms) (70–120 ms)

Sensitivity 90 (87–93) 74 (70–78) 37 (32–42) 58 (53–62)Specificity 91 (94–97) 87 (84–90) 93 (91–95) 64 (60–67)Negative predictive value 93 (92–96) 85 (82–87) 71 (68–74) 71 (67–75)Positive predictive value 94 (90–94) 78 (74–82) 76 (70–82) 49 (45–54)ROC area 0.93 0.81 0.65 0.61

Values are expressed as percentage. In parentheses, 95% of upper and lower confidence intervals. ROC = receiver operating characteristic.

250 ms. Only points with ICL>10 were considered in orderto avoid the inclusion of points with few intervals measured.The mean value of the FF intervals obtained from the pointstaken from each region of both atria was used for the analysis.

This setting for the FF interval analysis was decided tobetter identify “real” sites of rapid atrial activity (rotors),considering that FF intervals shorter than 110 ms could be inmost cases expression of sites of block or collision, particu-larly using a bipolar mapping, and not the real atrial activityin the mapped points.

Statistical Analysis

Proportions were compared with Fisher’s exact test or Chisquared test, as appropriate. Shapiro-Wilks or D’Agostinotests were used to verify normal distribution of values, as ap-propriate. Normally distributed values are expressed as mean± standard error, while nonparametric values are expressed as(median, min-max). Comparisons of means were performedwith Mann-Whitney U test or Student’s t-test for unpairedgroups, as appropriate. Multiple comparisons among prede-fined groups were performed by two-way ANOVA test orby Friedman analysis of variance test, as appropriate; and,if this showed a statistically significant difference amonggroups, Tukey-Kramer test or Newman-Keuls test was ap-plied, as appropriate. Bonferroni adjustment was applied toposthoc analysis on subgroups, as appropriate. Diagnostic ac-curacy was assessed with 2 × 2 tables to calculate sensitivity,specificity, positive predictive value, and negative predictivevalue of each test setting. A value of P < 0.05 was con-sidered significant. A receiver operating characteristic curvewas constructed for each of the mapping criteria, and the areaunder the curve was calculated for comparisons based on anonparametric method.

Results

A mean of 132 ± 25 and 99 ± 29 atrial points was respec-tively taken for the LA and the RA. The left atrial volumewas 149 ± 38 mL and the right one was 99 ± 39 mL.

Comparison of the Different Settings of CFAE Software

A total of 5,775 points acquired in the LA and RA were an-alyzed. Thirty percent of electrograms were considered frag-mented by the electrophysiologist A and 29% of the pointsby the electrophysiologist B. The concordance was 97%.Table 2 shows the sensitivity, specificity, negative predictivevalue, and positive predictive value of the four different set-tings compared. The best one was the setting A (15–30 ms)

with a sensitivity of 90%, a specificity of 91%, and a receivingoperating characteristic area of 0.93 (Fig. 4).

Spatial Distribution of CFAEs

Overall, the total CFAEs were present in 30.5% of thepoints acquired in the LA and in 20.3% of the points in theRA (P = 0.016). The HF CFAEs were significantly prevalentin the LA as compared with the RA (7.3% vs 2.8%, P =0.009). Also, the MF CFAEs were more frequently found inthe LA than in the RA (23.2% vs 17.5%, P = 0.04).

Figure 1 shows the prevalence of CFAEs in the differentregions analyzed in the LA and the RA. The left septumshowed the highest presence of total CFAEs (46 ± 29%). Inthe RA, the area of coronary sinus ostium (CS os) (37 ± 29%)and the septum (32 ± 21%) showed the higher presence oftotal CFAEs. A statistically significant difference was reachedfor the left septum when compared with the left lateral (P <0.01) and the left anterior wall (P < 0.01), and for the CSos in comparison with the right posterior wall (P < 0.05).Moreover, the left septum was the region with a significantlyhigher prevalence of HF CFAEs (15 ± 15%), as comparedwith all regions (P < 0.01).

The HF CFAEs were observed in 24 of the 25 patients. TheHF CFAEs were localized in only one area in three patients(one patient with permanent AF), in two areas in five patients(two patients with permanent AF), and in three or more areasin 16 patients (seven patients with permanent AF).

The HF CFAEs were mainly found in the septum, posteriorwall and roof in the LA, and septum and CS os in the RA.Table 3 shows the sites where more than one HF CFAEs

Figure 1. Receiver operating characteristic curves comparing the four pa-rameters settings of software for identifying CFAEs.

Calo et al. CARTO Mapping of CFAEs 1027

TABLE 3

Sites Where ≥ 2 HF CFAEs Were Found

Patients,%(n = 25)

LASeptum 64Posterior wall 52Roof 40Inferior wall 28Anterior wall 16Lateral wall 12

RASeptum 40CS os 36Posterior wall 24Lateral wall 16Anterior wall 16Isthmus 12

were found. Figure 2 shows examples of CFAE maps fromfour patients.

No significant difference was found between patients withpersistent and permanent AF. In the group of patients withoutstructural disease, the prevalence of total CFAE was signifi-cantly higher in the roof of the LA (P < 0.05) and lower inthe anterior wall of the RA (P< 0.05) in comparison with thepatients with structural heart disease.

Regional FF Interval Durations

The FF intervals were significantly shorter in the LA (me-dian 152, range 120–240 ms) than in the RA (median 162,range 120–250, P = 0.002). Figure 3 shows the values of FFintervals in the different regions of both atria. In the LA, theshortest FF intervals were found in the septum and posteriorwall, but without reaching a statistically significant differ-ence as compared with the other LA locations. In the RA,the shortest FF intervals were observed in the CS os and inthe septum, but only the CS os had significantly shorter FFintervals than the other RA regions (Fig. 4).

The areas with the shortest FF intervals were distributedin both atria, even if they were mainly localized in the leftseptum (74%), close to the PV (59%), in the CS os (55%),and in the left atrial appendage (50%). In 77% of the cases theareas with the shortest FF intervals included or were adjacentto sites with HF electrograms (Fig. 5).

Figure 2. Prevalence of CFAEs in the different regions analyzed in the RA and the LA. Data are expressed as the mean value. White and black bars in thefigure indicate HF and MF prevalence, respectively. Y-error bars denote standard deviations. Ant = anterior wall; cs os = coronary sinus ostium; lat =lateral wall; post = posterior wall; sept = septum.

Discussion

Main Findings

In this mapping study, conducted in 25 patients with per-sistent and permanent AF, most with structural heart disease,the following were observed.1 The CFAE software demon-strated to have an high sensitivity (90%) and specificity (91%)in the identification of CFAEs, using a specific setting of pa-rameters.2 The LA showed a significantly higher prevalenceof CFAEs as compared with the RA. The CFAEs were mostlypresent in the septal region, mainly the left one, and in the areaof CS os. Also, the posterior wall and the roof in the LA hada significant presence of CFAEs, particularly the HF CFAEs.Eight of the 25 patients showed the HF CFAEs localized inlimited regions (< 3).3 The FF intervals were significantlyshorter in the LA than in the RA. The shortest FF intervalswere found in the septum and posterior wall in the LA and inthe CS os in the RA. The areas with the shortest FF intervalsgenerally included or were adjacent to CFAEs.4

CFAE Definition and Significance

CFAEs have been proposed as a new target of AF ab-lation, but its definition is almost unclear to the most. Infact, in the CFAE “pot” different types of electrograms aremelted together. The term CFAE was used for the first timeby Nademanee et al.5 They defined CFAE as follows: (1)atrial electrograms that have fractionated electrograms com-posed of two deflections or more, and/or perturbation of thebaseline with continuous deflection of a prolonged activa-tion complex over a 10-second recording period; (2) atrialelectrograms with a very short cycle length < 120 ms av-eraged over a 10-second recording period. Therefore, theyincluded in CFAE true continuous fragmented atrial electro-grams and rapid atrial activity. More recently, Haissaguerre etal.2 considered CFAE only fractionated potentials exhibitingmultiple deflections from the isoelectric line (≥ 3 deflections)and/or potentials with continuous electrical activity withoutan isoelectric line. Konings et al.1 defined fragmented po-tentials as electrograms exhibiting more than two negativedeflections within 50 ms. The limit of 50 ms was based onthe assumption that the atrial refractory period during AFand the interval between two successive fibrillation waveswere ≥50 ms. Rapid atrial activity (long double potentialsof Konings’ definition) is generally characterized by sharpatrial electrograms with a very short cycle length (<120 msin the Nademanee definition). It may be the expression of

1028 Journal of Cardiovascular Electrophysiology Vol. 19, No. 10, October

Figure 3. ICL maps of the RA and LA fromfour patients. Panel A and C, left anterioroblique view; panel B and D posteroanteriorview. This map displays the total amount of in-tervals counted for each point, with the pos-sibility to show two different color tags forpoint with HF (red) or MF (blue). These tagscan be displayed also superimposed on othermaps. According to the color bar of CARTOSystem, points with high number of intervalsare displayed in red (highest fragmentation);points with few or none interval are displayedin purple (absence of fragmentation).

collision/overlap of different wavelets entering the same areaor may be determined by focal sources. It is important to un-derline that Konings et al.1 conducted a unipolar mapping,while most of endocardial studies utilized bipolar recordings.CFAEs are thought to indicate areas of nonuniform waveletpropagation and slowed conduction, pivot points around areasof functional and anatomical blocks that may be implicatedin arrhythmia perpetuation.3 The fragmentation of atrial elec-trograms can be determined by: (1) conduction velocity (areaof fibrosis, tissue anisotropy, etc).1,9 (2) Spatial dispersion ofrefractory periods: in the areas with longer refractory periodsthe action potentials may not be able to be achieved whenthe activation rate encroaches on the refractory periods, inother words some areas are able and other areas are not ableto maintain 1:1 capture.10 (3) Presence of specific anatomi-cal characteristics (areas of wide muscle connections favor-ing wavefront collision or presence of anatomical barriers

Figure 4. FF intervals in the different regionsanalyzed in the RA and the LA. Data are ex-pressed as the mean value. X-error bars de-note standard deviations. Ant = anterior wall;cs os = coronary sinus ostium; lat = lateralwall; post = posterior wall; sept = septum.∗ P < 0.01 vs posterior and anterior wall andP < 0.05 vs septum, isthmus and lateral wall.

between atrial muscle bundles).3,11-14 Furthermore, CFAEscan be determined by high-frequency sources (mother ro-tors).15,16 Finally, it has been proposed that CFAEs are areasclose to ganglionated plexi, where the local vagal stimulationshortens the local effective refractory periods and causes theformation of CFAEs.17,18

Regional Distribution of CFAEs

Little information exists on the spatial distribution ofCFAEs in the LA and RA. It is generally accepted that re-gional differences in the atrial electrical activity exist in AFand that fragmentation of atrial electrograms are generallypresent in same areas. Nademanee et al.5 mainly localizedCFAEs in the interatrial septum, proximal CS, PVs, mitralannulus, and left atrial roof. Previous human mapping stud-ies4,19,20 have found the septum as the area where CFAEs

Calo et al. CARTO Mapping of CFAEs 1029

Figure 5. ACI and ICL maps of the RA and LA from two patients (pos-teroanterior view). The FF interval was measured by using the ACI map.ACI displays the average value for all intervals considered for each point.The setting of duration used was: min duration 110 ms and max duration250 ms. According to the color bar of CARTO System, points with shorter FFintervals are displayed in red; points with longer FF intervals are displayedin purple. For the ICL maps see Figure 3.

were mostly present. Oral et al.,21 using the Nademanee def-inition of CFAE, have observed the highest prevalence ofCFAEs in the anterior wall of the left atrium, in the PV antrumor ostium and in the left septum. Scherr et al.,22 using the de-fault setting (70–120 ms) of CARTO software for CFAEsanalysis, without validation by an electrophysiologist, foundthat CFAE complexes were predominately located at the in-teratrial septum, LA posterior wall, and PV ostia. No datawere referred about the RA. They reported that the distribu-tion of CAFEs was quite different from results by Nademaneeet al. Noteworthy, CFAE sites were identified by the customsoftware in 86% of all LA endocardial locations and 34%were categorized as highly repetitive.

In this study CFAEs were significantly prevalent in theseptum, particularly the left septum. Moreover, the area ofCS os showed a high presence of CFAEs. Also, the pos-terior wall and the roof of the LA (principally in patientswithout heart disease) had a relative high prevalence of HFCFAEs. The septal line performed at the level of the inter-atrial connections showed to be highly effective in exper-imental studies12,23,24 and in humans.5,7,20 In humans, theCS was found to be critical in the maintenance of AF,13,25,26

and, in some cases, the source of rapid repetitive electricalactivity. The electrical disconnection of the CS from the LAreduced the inducibility of sustained AF.13 Several studies ob-served that the septal wall presents conduction deterioration,increased nonuniform anisotropy, and disorganized electricalactivity.3,4,11,20 Moreover, this finding could be determinedby a disproportionate amount of fibrosis in the septum, asobserved by Corradi et al.27 in AF patients with mitral valvedisease. Furthermore, an experimental study in dogs found

that autonomic ganglia stimulation favors the initiation of AFby the appearance of CFAEs.17 When the ganglia stimulationstopped, the electrograms invariably showed the restorationof discrete potentials with isoelectric intervals before termi-nation. In this study, the CFAEs were found along the sulcusterminalis (interatrial septum) and adjacent to the right supe-rior PV, close to the anterior right ganglionated plexi. Morerecently, Hou et al.18 confirmed these data and observed theright vasosympathetic stimulation (anterior right ganglion-ated plexi, located between the base of right superior PV andthe sulcus terminalis separating the left and right atria) wasmore arrhythmogenic than left vagosympathetic stimulation(superior left ganglionated plexi, located near the junction ofleft superior PV and left pulmonary artery), even if the degreeof effective refractory period shortening was similar. The au-thors postulated that anterior right ganglionated plexi is largerin size and closer to atrial myocardium than the superior leftganglionated plexi.

FF Intervals

According to previous studies,2,4,12,28 our data confirmthat the FF intervals are shorter in the LA as compared withthe RA, and that FF intervals recorded at the CS os resem-bled LA in terms of duration. In the LA, the septum andthe posterior wall showed the shortest FF intervals duration,even if the difference between the various LA regions wasnot statistically significant. Probably, as observed in recentreports,15,29 the maintenance of persistent or chronic AF maybe less dependent on posterior LA. Moreover, we found thatthe areas with the shortest FF intervals generally includedor were adjacent to CFAEs. Probably, the fast centrifugalpropagation of the wavefront from the core of a rotor maydetermine wave break and fibrillatory conduction. In fact,mapping of AF in sheep models has demonstrated the pres-ence of CFAEs around areas of high-frequency excitation.30

Therefore, it could be hypothesized that CFAEs are adjacentto the AF driver and can favor the maintenance of AF bymultiple reentry. This hypothesis is also supported by the ob-servation that the occurrence of fragmented electrograms issignificantly associated with a preceding shortening of FF in-terval and the duration of complex atrial fragmented electro-grams is inversely correlated with the preceding FF interval(shorter FF interval greater duration of CFAE).2

Conclusions

In conclusion, our study demonstrated that this new CFAEsoftware of CARTO system has a high diagnostic accuracyin the identification of fragmented electrograms, only by tai-loring it in a specific way. Furthermore, the same softwaregives us the possibility to obtain the average FF intervalsmaps. Electrical activity during AF showed a significant spa-tial inhomogeneity. In fact, the septal region, particularly theleft one, as well as the area of CS os, presented a significanthigher prevalence of CFAEs.

CFAE ablation approach could be a suitable and repro-ducible “step” in ablation of AF substrate, particularly inlong-lasting AF. One of the difficulties is the relative subjec-tivity inherent in defining whether a particular electrogramis “complex” enough.1,2,5,6,21 In an effort to standardize thedefinition of a CFAE site, there has been a significant amountof work in developing signal processing software to analyze

1030 Journal of Cardiovascular Electrophysiology Vol. 19, No. 10, October

atrial electrograms during AF. CFAE mapping requires a rel-evant effort to evaluate all regions of the atria. Thus, thepossibility to use an advanced mapping system not only torepresent the atrial anatomy, but also to efficiently map thechambers during AF and rapidly identify CFAE sites, it isnecessary to become a routinely approach.

Further studies are needed to increase our knowledgeabout the CFAEs distribution, the locations sites of rapid atrialactivity (rotors), and their relationship in different subgroupof patients.

Study Limitations

A major limitation of the work is that we did not performselective ablation of CFAEs. While such confirmation neednot be a part of this study, this should be an objective of futureclinical validation. Another limitation is due to the difficultyto state a cutoff value for FF interval count in the settingof extensive electrogram fragmentation, so that any methodwill be to some extent arbitrary and may be not comparableto measures of local activation frequency by other methods.

References

1. Konings KT, Smeets JL, Penn OC, Wellens HJ, Allessie MA: Config-uration of unipolar atrial electrograms during electrical induced atrialfibrillation in humans.Circulation 1997;95:1231-1241.

2. Rostock T, Rotter M, Sanders P, Takahashi Y, Jaıs P, Hocini M, Hsu LF,Sacher F, Clementy J, Haıssaguerre M: High-density activation map-ping of fractionated electrograms in the atria of patients with paroxysmalatrial fibrillation. Heart Rhythm 2006;3:27-34.

3. Papageorgiou P, Monahan K, Boyle NG, Seifert MJ, Beswick P, ZebedeJ, Epstein LM, Josephson ME: Site-dependent intra-atrial conduc-tion delay: Relationship to initiation of atrial fibrillation. Circulation1996;94:384-389.

4. Gaita F, Calo L, Riccardi R, Garberoglio L, Scaglione M, LicciardelloG, Coda L, Di Donna P, Bocchiardo M, Caponi D, Antolini R, Orzan F,Trevi GP: Different patterns of atrial activation in idiopathic atrial fibril-lation: Simultaneous multisite atrial mapping in patients with paroxys-mal and chronic atrial fibrillation. J Am Coll Cardiol 2001;37:534-541.

5. 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.

6. Haıssaguerre M, Sanders P, Hocini M, Takahashi Y, Rotter M, Sacher F,Rostock T, Hsu LF, Bordachar P, Reuter S, Roudaut R, Clementy J, JaısP: Catheter ablation of long-lasting persistent atrial fibrillation: Criticalstructures for termination. J Cardiovasc Electrophysiol 2005;16:1125-1137.

7. Calo L, Lamberti F, Loricchio ML, De Ruvo E, Colivicchi F, BianconiL, Pandozi C, Santini M: Left atrial ablation versus biatrial ablation forpersistent and permanent atrial fibrillation: A prospective and random-ized study. J Am Coll Cardiol 2006;47:2504-2512.

8. Levy S, Camm AJ, Saksena S, Aliot E, Breithardt G, Crijns H, DaviesW, Kay N, Prystowsky E, Sutton R, Waldo A, Wyse DG: Working groupon arrhythmias, working group on cardiac pacing of the European so-ciety of cardiology, North American society of pacing and electrophys-iology. International consensus on nomenclature and classification ofatrial fibrillation. Europace 2003;5:119-122.

9. Saffitz JE, Kanter HL, Green KG, Tolley TK, Beyer EC: Tissue-specificdeterminants of anisotropic conduction velocity in canine atrial andventricular myocardium. Circ Res 1994;74:1065-1070.

10. Li H, Hare J, Mughal K, Krum D, Biehl M, Deshpande S, Dhala A,Blanck Z, Sra J, Jazayeri M, Akhtar M: Distribution of atrial electrogramtypes during atrial fibrillation: Effect of rapid atrial pacing and intercavaljunction ablation. J Am Coll Cardiol 1996;27:1713-1721.

11. Platonov PG, Yuan S, Hertervig E, Kongstad O, Roijer A, VygovskyAB, Chireikin LV, Olsson SB: Further evidence of localized posteriorinteratrial conduction delay in lone paroxysmal atrial fibrillation. Eu-ropace 2001;3:100-107.

12. Kumagai K, Uno K, Khrestian C, Waldo AL: Single site radiofrequencycatheter ablation of atrial fibrillation: Studies guided by simultaneous

multisite mapping in the canine sterile pericarditis model. J Am CollCardiol 2000;36:917-933.

13. Oral H, Ozaydin M, Chugh A, Scharf C, Tada H, Hall B, Cheung P,Pelosi F, Knight BP, Morady F: Role of the coronary sinus in mainte-nance of atrial fibrillation. J Cardiovasc Electrophysiol 2003;14:1329-1336.

14. Roithingher FX, SippensGroenewegen A, Karch MR, Steiner PR, EllisWS, Lesh MD: Organized activation during atrial fibrillation in man: En-docardial and electrocardiographic manifestations. J Cardiovasc Elec-trophysiol 1998;9:451-461.

15. Sanders P, Berenfeld O, Hocini M, Jaıs P, Vaidyanathan R, Hsu LF,Garrigue S, Takahashi Y, Rotter M, Sacher F, Scavee C, Ploutz-SnyderR, Jalife J, Haıssaguerre M: Spectral analysis identifies sites of high-frequency activity maintaining atrial fibrillation in humans. Circulation2005;112:789-797.

16. Jalife J, Berenfeld O, Mansour M. Mother rotors and fibrillatory conduc-tion: A mechanism of atrial fibrillation. Cardiovasc Res 2002;54:204-216.

17. Scherlag BJ, Yamanashi W, Patel U, Lazzara R, Jackman WM: Auto-nomically induced conversion of pulmonary vein focal into atrial fib-rillation. J Am Coll Cardiol 2005;45:1878-1886.

18. Hou Y, Scherlag BJ, Lin J, Zhang Y, Lu Z, Truong K, Patterson E,Lazzara R, Jackman WM, Po SS: Ganglionated plexi modulate extrinsiccardiac autonomic nerve input: Effect on sinus rate, atrioventricularconduction, refractoriness, and inducibility of atrial fibrillation. J AmColl Cardiol 2007;50:61-68.

19. Jaıs P, Haıssaguerre M, Shah DC, Chouairi S, Clementy J: Regional dis-parities of endocardial atrial activation in paroxysmal atrial fibrillation.Pacing Clin Electrophysiol 1996;19(Pt II):1998-2003.

20. Gaita F, Riccardi R, Calo L, Scaglione M, Garberoglio L, AntoliniR, Kirchner M, Lamberti F, Richiardi E: Atrial mapping and ra-diofrequency catheter ablation in patients with idiopathic atrial fibril-lation. Electrophysiological findings and ablation results. Circulation1998;97:2136-2145.

21. 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.

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

23. Tondo C, Scherlag BJ, Otomo K, Antz M, Patterson E, Arruda M, Jack-man WM, Lazzara R: Critical atrial site for ablation of pacing-inducedatrial fibrillation in the normal dog heart. J Cardiovasc Electrophysiol1997;8:1255-1265.

24. Betts TR, Roberts PR, Morgan JM: Feasibility of a left atrial electri-cal disconnection procedure for atrial fibrillation using transcatheterradiofrequency ablation. J Cardiovasc Electrophysiol 2001;12:1278-1283.

25. Cox JL, Schuessler RB, Lappas DG, Boineau JP: An 8-1/2-year clinicalexperience with surgery for atrial fibrillation. Ann Surg 1996;224:267-273.

26. Nair M, Shah P, Batra R, Kumar M, Mohan J, Kaul U, Arora R: Chronicatrial fibrillation in patients with rheumatic heart disease. Mapping andradiofrequency ablation of flutter circuits seen at initiation after car-dioversion. Circulation 2001;104:802-809.

27. Corradi D, Callegari S, Benussi S, Maestri R, Pastori P, NascimbeneS, Bosio S, Dorigo E, Grassani C, Rusconi R, Vettori MV, Alinovi R,Astorri E, Pappone C, Alfieri O: Myocyte changes and their left atrialdistribution in patients with chronic atrial fibrillation related to mitralvalve disease. Hum Pathol 2005;36:1080-1089.

28. Ndrepepa G, Karch MR, Schneider MA, Weyerbrock S, Schreieck J,Deisenhofer I, Zrenner B, Schomig A, Schmitt C: Characterization ofparoxysmal and persistent atrial fibrillation in the human left atriumduring initiation and sustained episodes. J Cardiovasc Electophysiol2002;13:525-532.

29. Lazar S, Dixit S, Marchlinski FE, Callans DJ, Gerstenfeld EP: Presenceof left-to-right atrial frequency gradient in paroxysmal but not persistentatrial fibrillation in humans. Circulation 2004;110:3181-3186.

30. Kalifa J, Tanaka K, Zaitsev AV, Warren M, Vaidyanathan R, AuerbachD, Pandit S, Vikstrom KL, Ploutz-Snyder R, Talkachou A, Atienza F,Guiraudon G, Jalife J, Berenfeld O: Mechanisms of wave fractionationat boundaries of high-frequency excitation in the posterior left atrium ofthe isolated sheep heart during atrial fibrillation. Circulation 2006;113:626-633.