- Email: [email protected]
Spectral analysis during sinus rhythm predicts an abnormal atrial substrate in patients with paroxysmal atrial fibrillation
Spectral analysis during sinus rhythm predicts an abnormal atrial substrate in patients with paroxysmal atrial fibrillation Yenn-Jiang Lin, MD,*† Tsair Kao, PhD,‡ Ching-Tai Tai, MD,* Shih-Lin Chang, MD,* Li-Wei Lo, MD,† Ta-Chuan Tuan, MD,† Ameya R. Udyavar, MD,† Yu-Feng Hu, MD,† Han-Wen Tso, MS,‡ Satoshi Higa, MD, PhD,§ Shih-Ann Chen, MD, FHRS*† From the *Division of Cardiology, Department of Medicine, Taipei Veterans General Hospital, †Institute of Clinical Medicine, and Cardiovascular Research Center, and ‡Institute of Biomedical Engineering, National Yang-Ming University, Taipei, Taiwan, and §Second Department of Internal Medicine, Faculty of Medicine, University of Ryukyus, Okinawa, Japan. BACKGROUND Regions of rapid and multiple deflections can be identified with high dominant frequency (DF) during sinus rhythm (SR). These areas may play a role in the perpetuation of atrial fibrillation (AF) and indicate an atrial substrate abnormality. OBJECTIVE The purpose of this study was to investigate the atrial substrate properties of the high-frequency sites in patients with paroxysmal AF. METHODS Forty patients (52 ⫾ 12 years of age) with paroxysmal AF were studied using a three-dimensional mapping system. Spectral analysis was performed on the bipolar electrograms in the left atrium (LA) during SR. Overall, 7708 electrograms were analyzed, and the DFs higher than 70 Hz were labeled as abnormal.
sites/patient). Type 2 includes high-DF sites in the LA or LA plus the PVs (n ⫽ 21, 11 ⫾ 5.6 sites/patient). In type 1, PV isolation (PVI) could eliminate the AF with negative AF inducibility testing after the PVI in 89% of patients. In type 2, additional LA substrate modification was needed in 81% of patients because sustained AF was induced after the PVI (P⬍.001, compared with type 1). Multivariate analysis showed that the lower mean voltage of the LA and high-frequency sites distribution both independently predicted a positive AF inducibility after the PVI (P⬍.05). CONCLUSIONS Spectral analysis during SR can detect an abnormal atrial substrate. A regional distribution of the high-DF sites predicts the efficacy of the PVI.
RESULTS The regional distribution of the high-DF sites in the LA could be divided into two types. Type 1 includes high-DF sites existing only in the pulmonary veins (PVs; n ⫽ 19, 6.6 ⫾ 3.4
KEYWORDS Ablation; Fibrillation; Electrogram
Introduction
Methods
Regions of abnormal atrial substrate can be identified by multiple rapid deflections, fractionated electrograms, and low electrogram voltage during sinus rhythm (SR).1–5 These areas with fibrillatory myocardium may play a role in the perpetuation of atrial fibrillation (AF) and indicate an atrial substrate abnormality. Recently, Pachon et al6 showed that the atrial substrate can be characterized by the frequency spectra obtained from single electrograms during SR. According to this study, clusters of fibrillatory myocardium presented with higher frequencies than the other surrounding normal atrial substrate.6 However, the substrate characteristics, anatomic distribution, and role of the high-frequency sites during the AF were not clear. The aims of this study were (1) to investigate the atrial substrate properties of the high-frequency sites identified during SR and (2) to assess the catheter ablation results based on the different types of distribution of the high-frequency sites.
Patient characteristics
Address reprint requests and correspondence: Shih-Ann Chen, M.D., Division of Cardiology, Taipei Veterans General Hospital, 201, Section 2, Shih-Pai Road, Taipei, Taiwan. E-mail address: [email protected]. (Received February 4, 2008; accepted March 26, 2008.)
(Heart Rhythm 2008;5:968 –974) © 2008 Heart Rhythm Society. All rights reserved.
This study enrolled 40 consecutive paroxysmal AF patients (mean age 52 ⫾ 12 years) who underwent radiofrequency (RF) catheter ablation under the guidance of a NavX mapping system (NavX 6.0, St. Jude Medical, Inc., St. Paul, MN). Patients with repeated procedures were not included in this study.
Electroanatomic mapping and electrophysiological study Each patient underwent an electrophysiological study and catheter ablation in the fasting state, after informed consent was obtained. The method of three-dimensional (3D) electroanatomic mapping has been described elsewhere.7,8 In brief, after acquiring the LA geometry, the 4-mm-tip catheter (EP Technologies, Boston Scientific, Inc., Boston) was selected as the roving catheter for sequential contact mapping. The roving catheter was used to collect local bipolar signals while it was swiped across the left atrium (LA) during SR. The signal from the mapping catheter was used to build a sequential map. After completion of the sequential map, all bipolar signals were exported for further analysis.
1547-5271/$ -see front matter © 2008 Heart Rhythm Society. All rights reserved.
doi:10.1016/j.hrthm.2008.03.039
Lin et al
Figure 1
Spectral Analysis during Sinus Rhythm
969
Schematic presentation of the electrogram morphology analysis and fast Fourier transform of single discrete electrograms during SR.
Provocation of AF was performed in all patients. We attempted to find the spontaneous onset of atrial ectopic beats or repetitive episodes of short runs or sustained AF before or after the infusion of isoproterenol or after an algorithm designed for facilitating the initiation of AF. The methods used to provoke spontaneous AF were attempted at least twice to ensure reproducibility.7–10
Signal recording and the analysis Bipolar atrial electrograms were collected in the LA with a point-by-point approach sequentially during SR under the guidance of a NavX system.7,8 The signals for the frequency analysis and electrogram morphology analysis were both exported from the NavX mapping system and analyzed by the Matlab computer program (MathWorks Inc. Natick, MA). The sampling rate of the NavX mapping system was 1200 Hz. Regarding the electrogram morphology analysis, the electrogram characteristics (activation duration, number of deflections, and peak-to-peak bipolar voltage) were measured before and after the pulmonary vein (PV) isolation (Figure 1).11 The duration of the atrial electrogram was measured from the earliest atrial activation site to the latest atrial activation site from a stable baseline. The number of deflections was assessed by counting the number of negative deflections crossing the baseline at least 10 ms apart. Spectral analysis was performed on the single discrete bipolar electrogram during SR (unrectified, Hanning window, 1 second in duration; Figure 1). Then the data were exported to an external computer program including single discrete electrograms during SR and the baseline on both sides of the discrete electrogram. The fast Fourier transform analysis was performed using a Hanning window function on each segment from all recording sites in the LA. The dominant frequency (DF) was defined as the frequency with the maximum power in the frequency range. Higher frequencies were mapped toward the purple end of the NavX color spectrum, as a real time built-in function. To ensure the reliability of the DF detection, the lowest noise signal was chosen for the analysis. Overall, 10.6 ⫾ 10 sites (equivalent to 4.8% ⫾ 3.6% of LA sites) per patient were excluded. Further, to avoid too high density in some regions, any mapping site with a distance of less than 0.5
mm to the nearest neighboring site was excluded. Spectral analysis was performed before the catheter ablation in all patients. In the preliminary study, the consistency of the DF value from two consecutive beats during SR was investigated in a total of 344 points in the initial five patients. The BlandAltman agreement test showed an acceptable result (22/344: 6.2% of the data pairs outside 2 standard deviations [SDs]), indicating the consistency of the DF during SR over time. In addition, the averaged density of the mapping sites was 2.1 ⫾ 0. 44 sites/cm2 in all patients. The maximal distance between any two mapping sites was 16.3 ⫾ 4.9 mm. The point density was 3.1 ⫾ 0.9 sites/cm2 in the anteroseptal wall of the LA, 2.2 ⫾ 0.38 sites/cm2 in the LA roof area, 2.3 ⫾ 0.43 sites/cm2 in the LA posterior wall, and 3.0 ⫾ 1.2 sites/cm2 in the PV area (P⬎.05). Therefore, the point density was nearly uniformly distributed and acceptable for the analysis.
Definitions Dominant frequency peak: The largest power of the frequency peak was identified as the DF peak. High frequency sites: The determination of the abnormal atrial substrate was based on the frequency spectrum of each site during SR. The frequency analysis was obtained from a total of 7708 electrograms in the LA during SR, with an average of 193 ⫾ 53 sites per patient. The mean DF value was 47 ⫾ 20 Hz. According to the normal distribution curve of the DF values, the top 5% were higher than 70 Hz, and those sites were defined as high frequency sites.
Catheter ablation of paroxysmal AF As described in the previous study, the location of the PV ostium was determined by both PV angiography and the local electrogram morphology.12 It was also confirmed by the 3D mapping system with the LA geometry reconstruction.8 Continuous circumferential lesions were created by encircling the right and left PV ostia guided by the NavX system using a 4-mm-tip ablation catheter. RF energy was applied continuously while repositioning the catheter tip every 40 seconds with a target temperature of 50°–55° and maximum power of 50 W in the temperature control mode. The temperature was reduced to 45°–50° when RF energy to
970 Table 1
Heart Rhythm, Vol 5, No 7, July 2008 Clinical and electrophysiological characteristics of the study patients.
Gender, male/female Age, years Duration of AF, years LV ejection fraction, % LA diameter, mm Underlying cardiovascular disease (coronary artery disease, heart failure, hypertrophic cardiomyopathy) (%) Mean no. of mapping sites in the LA Mean voltage of the LA, mV No. of high-frequency sites in the LA before the PV isolation
Type 1 (n ⫽ 19)
Type 2 (n ⫽ 21)
P
10/9 51 ⫾ 4.5 ⫾ 60 ⫾ 36 ⫾ 3 (16)
16/5 53 ⫾ 5.3 ⫾ 58 ⫾ 41 ⫾ 4 (19)
.11 .50 .80 .45 .06 .65
14 3.0 7.2 6.3
210 ⫾ 63 2.3 ⫾ 0.63 6.6 ⫾ 3.4
10 6.2 9.9 6.0
209 ⫾ 35 1.7 ⫾ 0.78 11 ⫾ 5.6
.98 .02 .01
Data are presented as mean value ⫾ SD unless otherwise indicated.
the posterior wall near the esophagus was necessary. The intention was to place the RF lesions at least 1–2 cm away from the angiographically defined ostia. After completion of the circumferential lesion set, the ipsilateral superior and inferior PVs were mapped carefully by circular catheter recording (Spiral, AF Division, St. Jude Medical, Inc., Minnetonka, MN) during SR or coronary sinus (CS) pacing. After successful isolation of all four PVs, which was confirmed by circumferential mapping of the PVs, high-current (3–5 times the pacing threshold) and wide (8 ms) pulse duration stimulation from the proximal and distal CS was performed (in 10-ms decrements from 250 to 150 ms, with a duration of each pacing cycle length of 5–10 seconds) and repeated 3–5 times. If an induced AF/LA flutter was sustained for ⬎1 minute, an additional linear ablation was performed randomly at either the anterior roof or mitral isthmus; the reentry circuit of the LA flutter was identified by isochrone mapping, entrainment maneuvers, and the postpacing interval analysis. If sustained LA AF/flutter was still induced, cardioversion was performed to restore SR.
Follow-up of AF recurrence After discharge, the patients underwent follow-up (2 weeks after the catheter ablation, then every 1–3 months) at our cardiology clinic or with the referring physicians, and antiarrhythmic drugs were prescribed for 8 weeks to prevent the early recurrence of paroxysmal AF (⬍1 month after the ablation). When the patients experienced symptoms suggestive of a tachycardia after the ablation, 24-hour Holter monitoring or cardiac event recordings were performed to define the cause of the clinical symptoms. AF recurrence was defined as an episode lasting more than 1 minute and confirmed by electrocardiograms 2 months after the ablation (blanking period).
Statistical analysis All continuous data were presented as the mean value ⫾ SD. Comparisons of the continuous data were performed with one-way analysis of variance. The univariate variables of various electrophysiologic variables were ascertained to determine the predictors of a positive inducibility test after the PV isolation. Various clinical and electrophysiological
variables including sex, age, structural heart disease, LA size, left ventricular ejection fraction, presence of highfrequency sites in the LA, and mean bipolar voltage of the LA were used to assess the predictors of a positive inducibility test after the PV isolation. The variables that were selected to be tested in the multivariate analysis were those with a P-value of less than 0.2 in the univariate models. A two-sided P⬍.05 was considered statistically significant.
Results Patient characteristics A total of 40 patients who underwent RF ablation of paroxysmal AF were included. The clinical and electrophysiological characteristics of the study patients are shown in Table 1.
Frequency analysis results Characteristics of the high-frequency sites identified during SR The electrogram morphology and spectral morphology differed between the high-frequency sites and non-high-frequency sites. As shown in Table 2, the high-frequency sites harbored a longer activation duration, more deflections, and lower peak-to-peak bipolar voltage compared with the nonhigh-frequency sites. Regarding the frequency spectra, 94% of the non-high-frequency sites presented a single DF peak, but only 31% of the high-frequency sites presented a single DF peak (P⬍.05). The other 69% of the high-frequency sites presented two or more than two DF peaks (Figure 1). The average DF value of the fundamental peaks in the high-frequency sites was similar to that of the DF peaks of the non-high-frequency sites (36.8 ⫾ 8.7 Hz vs. 40.3 ⫾ 11.7 Hz; P ⫽ .1). Distribution of the high-frequency sites in the LA Overall, 342 mapping sites had a DF value higher than 70 Hz in all 40 patients. These high-DF mapping sites were found in the specific regions and not uniformly distributed in the LA (Figure 2). The highest DF value of each patient was 104 ⫾ 24 Hz. There were 8.7 ⫾ 5.1 (range 1–19) high-frequency sites in the LA and PV region in each
Lin et al Table 2
Spectral Analysis during Sinus Rhythm
971
Characteristics of the high-frequency sites and non-high-frequency sites
Variables Electrogram morphology: Activation duration, ms No. of deflections Peak-to-peak bipolar voltage, mV Spectral morphology: Percentage of single DF peak morphologies (of overall electrograms) DF peak value, Hz
High-frequency sites
Non-high-frequency sites
49 ⫾ 7.1 3.10 ⫾ 0.76 1.57 ⫾ 0.56
32 ⫾ 9.2 1.86 ⫾ 0.63 1.99 ⫾ 0.74
31 82.1 ⫾ 11.6
94 40.3 ⫾ 11.7
P⬍.001 for all comparisons.
patient. The distribution of the high-frequency sites was as follows: 215 sites (63%) inside or around the PVs, 68 sites (20%) in the LA septum and anterior wall, 37 sites (11%) in the posterior wall, and 20 sites (6%) in the LA roof. Patients with a larger LA had more high-frequency sites in the LA (Pearson’s correlation, r ⫽ 0.37, P ⫽ .02). On the other hand, this did not correlate with the other clinical variables, such as the sex, age, left ventricular ejection fraction, or presence of structural heart disease (P⬎.05).
Characterization of the atrial substrate based on the high-frequency sites distribution The regional distribution of the high-frequency sites can be divided into two types. In type 1 (n ⫽ 19, 48%; Figure 2), the high-frequency sites existed only in the PVs. The highest DF sites were inside the PV isolation line. In these patients, the high-frequency sites were observed in the right superior PV in 15/19 patients (79%), in the right inferior PV in 8/19 patients (42%), in the left superior PV in 9/19
Figure 2 Electrogram morphology and frequency spectra and the regional distribution of the DF sites in a patient with a type 1 LA substrate before and after PV isolation. The electrogram at the high-frequency sites exhibited rapid and multiple deflections with high fragmented peaks of greater than 70 Hz. The purple areas in the reconstructed geometry of the LA indicate the high-DF sites that are higher than 70 Hz. Note that after the ablation, the high-DF peaks of 118 Hz were eliminated with the remaining DF around 32 Hz. The time-domain electrograms after the PV isolation were also less fractionated than those before the ablation. The high-frequency sites were eliminated by right PV isolation.
972
Heart Rhythm, Vol 5, No 7, July 2008
Figure 3 The regional distribution of the high-frequency sites in a patient with a type 2 LA substrate during SR is shown in the left panel. The high-frequency sites were in the PV and LA. The intracardiac bipolar electrograms and their corresponding frequency spectra of the high-frequency sites (sites 1 and 2) and the non-high-frequency sites (sites 3 and 4) are shown. The spectral morphology in the AF termination site in the LA anterior roof (site 1) showed a high-DF peak of 71 Hz during SR. Electrogram morphology showed multiple and rapid deflections. On the other hand, in the non-high-frequency sites in the low anterior septum (site 2), the bipolar electrograms were not complex, and the spectral morphology exhibited a single DF peak value of less than 70 Hz.
patients (47%), and in the left inferior PV in 3/19 patients (16%). In type 2 (n ⫽ 21, 52%; Figure 3), the high-frequency sites were observed in the LA or in the LA plus the PVs. The highest DF sites were outside the circumferential isolation line with DF gradient into the PVs. Among them, 13 patients (59%) had high-frequency sites in both the LA and PVs. Table 1 shows the clinical and substrate properties of the type 1 and type 2 distributions. Type 2 patients were associated with a larger LA size and a lower mean peak-topeak bipolar voltage compared with type 1 patients.
isolation line. Meanwhile, the high-frequency sites in the LA were more frequent in the patients with a positive AF inducibility compared with a negative AF inducibility (12/ 17, 70.6% vs. 6/23, 26.1%; P ⫽ .006). After the PV isolation plus linear lines, six patients in whom AF was induced after ablation had high-frequency sites in the LA (6/6, 100%). Those were located at the septum in four patients (67%), in the LA roof in four patients (67%), in the posterior wall in three patients (50%), and in the lateral mitral isthmus in one patient (17%).
Effects of the PV isolation and LA linear ablation As shown in Figure 4, in type 1 (n ⫽ 19, Figure 2) patients, with high-frequency sites in the PVs, circumferential PV isolation rendered the AF noninducible in 17 (93%) of 19 patients (Figure 3). After linear ablation (LA roof line in two patients), the AF became noninducible in all patients (19/19, 100%). In type 2 (n ⫽ 21; Figure 3) patients, with high-frequency sites in the LA or LA plus the PVs before the ablation, the AF remained inducible in 17 (81%) out of the total 21 patients after the circumferential PV isolation. After the linear ablation (a roof line in 17 patients and lateral mitral isthmus line in 11 patients), the AF inducibility remained positive in six of the 21 patients (29%; P⬍.05, when compared with type 1). In type 1, the number of high-frequency sites within the PVs decreased by 68% (from overall 215 to 69 mapping sites) after the PV isolation. In type 2, most of the highfrequency sites in the LA remained unchanged after the PV isolation, and the number of high-frequency sites decreased by 39% (from overall 127 to 76 mapping sites). These high-frequency sites eliminated by the PV isolation were all located in the LA septum and were adjacent to the right PV
Figure 4 Flow chart of the high-frequency sites mapping and ablation results in all 40 patients. Note that the inducibility was negative in all patients with a type 1 distribution after the circumferential PV isolation and linear ablation. On the other hand, the AF inducibility test was negative in 76% of the type 2 AF patients.
Lin et al Table 3
Spectral Analysis during Sinus Rhythm
973
Univariate and multivariate predictors of the AF inducibility after PV isolation Multivariate analysis
Factors
Positive AF inducibility (n ⫽ 19; 45%)
Negative AF inducibility (n ⫽ 21; 55%)
Univariate analysis (P-value)
Sex Age Structural heart disease (%) LA size, mm LV ejection fraction, % The presence of high-frequency sites in LA (%) Mean peak-to-peak LA bipolar voltage, mV
16/3 54 ⫾ 9.9 5 (26) 41 ⫾ 5.5 57 ⫾ 8.6 17 (89.5%) 1.47 ⫾ 0.63
10/11 50 ⫾ 14 2 (9.5) 36 ⫾ 5.9 60 ⫾ 8.7 4 (19.1%) 2.27 ⫾ 0.51
0.01 0.27 0.9 0.007 0.338 ⬍0.001 ⬍0.001
Hazard ratio (confidence interval 95%) 2.6 (0.01-6.98) 11.4 (0.98–1.45) 34.4 (2.0–506) 37.6 (2.1–665)
P .375 — — .08 — .01 .013
Data are presented as mean value ⫾ SD unless otherwise indicated.
Prediction of a positive inducibility test after PV isolation Univariate and multivariate predictors of the requirement for additional LA linear ablation lesions after circumferential PV isolation are detailed in Table 3. The clinical and electrophysiologic variables associated with a positive AF inducibility after PV isolation were a male gender, larger LA dimension, lower mean bipolar electrogram voltage of the LA, and the presence of high-frequency sites in the LA. The other factors did not differ in regard to the AF inducibility testing after the PV isolation. A multivariate analysis showed that the lower bipolar voltage of the LA (cutoff value of 1.9 mV) and the distribution of the high-frequency sites in the LA independently predicted a positive AF inducibility after the PV isolation, indicating abnormal LA substrate in these patients.
Long-term follow-up of the catheter ablation With a mean follow-up of 10.5 ⫾ 3.2 months, there were four patients who had recurrence of AF. The long-term success rate on or off previously ineffective antiarrhythmic drugs was 95% and 86% (in types 1 and 2, respectively; P⬎.05).
Discussion Main findings Spectral analysis during SR can identify an abnormal LA substrate with a DF peak above 70 Hz. These high-frequency sites were associated with multiple rapid deflections, a prolonged electrogram, and low bipolar voltage of the electrogram morphology. The LA substrate could be characterized by a distribution of the high-frequency sites in the LA and PVs. The presence of a high-frequency site in the LA independently predicted a positive AF inducibility after the circumferential PV isolation.
Compared with previous studies Pachon et al6 demonstrated that an abnormal atrial substrate could be differentiated from a normal or compact atrial substrate by using a fast Fourier analysis during SR. Those abnormal atrial sites were characterized by fractionated electrograms in the time-domain signals and high-frequency
peaks in the frequency spectra. Areas of fractionation provided potential targets for the AF ablation. RF ablation of those “AF nest sites” eliminated the high-frequency peaks, and the resultant frequency spectra were similar to those of the “non–AF nest sites.”6 This study demonstrated that the high-frequency sites were characterized by specific electrogram and spectral morphologies during SR. Spectral analysis revealed multiple and high-DF peaks above 70 Hz. The electrogram morphology can also provide the same information, and these can also be identified by the multiple rapid deflections and prolonged and fractionated bipolar electrograms, as in previous studies.2,3,6,7 Currently, circumferential ablation during PV isolation has become the standard procedure of catheter ablation of paroxysmal AF with an optimal long-term success.13,14 Pachon et al6 showed that selective LA catheter ablation based on the spectral analysis results can be effective without circumferential PV isolation. However, the effect of PV isolation on the high-frequency sites was not clear. This study showed that high-frequency sites within the PV ostia could be identified in 80% of the paroxysmal AF patients. PV isolation can eliminate 69% of the high-frequency sites within the PV isolation line. Although catheter ablation of the high-frequency sites based on the spectral analysis was not performed in this study, further studies are needed to show whether or not an ablation strategy targeting those high-frequency sites would be additive to the circumferential PV isolation.
Characteristics of an atrial substrate with highfrequency sites The majority of the high-frequency sites were located in the PVs and/or PV ostia. These results fit well with the findings of the high-frequency AF sources in the animal and human studies.15–17 After PV isolation, DF peaks of the highfrequency sites above 70 Hz were greatly reduced or eliminated (Figure 2). Concomitantly, the PV potentials were eliminated, suggesting that the high-DF peaks and conduction disturbances were related to the PV activation. In those patients, the PV isolation could treat the AF without any additional LA substrate modification. On the other hand, in
974 type 2 patients, LA high-frequency sites were frequently observed in the LA septum, posterior wall, and roof areas. In patients with a type 2 LA substrate, they were associated with a lower mean bipolar voltage and dilated LA before the ablation, which indicates a diseased atrial substrate. Previous studies have also indicated that the positive inducibility of AF after circumferential PV isolation required additional LA substrate modification for a better clinical outcome.8,18 The higher AF inducibility in the type 2 patients further confirmed the abnormal LA substrate in those patients. The underlying mechanism of the high-frequency sites is not known. The complex electrograms may be explained first by a localized fibrosis and anisotropy with discontinuous conduction.19 Second, the parasympathetic innervation sites of the LA also harbored similar fractionated and prolonged sinus electrograms; however, they were characterized by a high-voltage amplitude, and the characteristics may have differed from the high-frequency sites in the present study.5
Clinical implications Sustained AF could be induced by burst atrial pacing after PV isolation in around 50% of patients with paroxysmal AF. Overall, 30%–50% of patients with paroxysmal AF and more than 70% of patients with chronic AF needed additional atrial substrate modification.8,18 Previous studies showed that more extensive scarring of the LA or a lower mean bipolar voltage of the LA indicated a higher recurrence of AF after circumferential PV isolation.4,20 This study showed that the distribution of the high-frequency sites and presence of low-voltage areas both independently indicated an abnormal LA substrate. However, the optimal strategy for the LA substrate modification based on the voltage distribution in paroxysmal or chronic AF patients has not yet been determined. Identification of the distribution of the high-frequency sites may be beneficial for finding the AF ablation targets.6
Limitations First, mapping of the atrial substrate should be performed in SR; therefore, patients with a focal-type paroxysmal AF with rapid and frequent initiating AF were not included in this study. Second, the effect of the activation rhythm on the spectral analysis results was not assessed in this study. Third, patients without AF were not enrolled, and frequency analysis results during SR were not available in normal control patients. The high-frequency sites may be observed in the normal LA substrate.
Conclusion Spectral analysis during SR can detect an abnormal atrial substrate before the catheter ablation of AF. These highfrequency sites were associated with multiple rapid deflections, a prolonged electrogram, and low bipolar voltage of
Heart Rhythm, Vol 5, No 7, July 2008 the electrogram morphology. The regional distribution of the high-DF sites (higher than 70 Hz) in the LA predicted the efficacy of the PV isolation and the need for additional LA substrate modification.
References 1. Lin YJ, Tai CT, Huang JL, et al. Characterization of right atrial substrate in patients with supraventricular tachyarrhythmias. J Cardiovasc Electrophysiol 2005;16:173–180. 2. Nakao K, Seto S, Ueyama C, et al. Extended distribution of prolonged and fractionated right atrial electrograms predicts development of chronic atrial fibrillation in patients with idiopathic paroxysmal atrial fibrillation. J Cardiovasc Electrophysiol 2002;13:996 –1002. 3. Tanigawa M, Fukatani M, Konoe A, et al. Prolonged and fractionated right atrial electrograms during sinus rhythm in patients with paroxysmal atrial fibrillation and sick sinus node syndrome. J Am Coll Cardiol 1991;17:403– 408. 4. Verma A, Wazni OM, Marrouche NF, et al. Pre-existent left atrial scarring in patients undergoing pulmonary vein antrum isolation: an independent predictor of procedural failure. J Am Coll Cardiol 2005;45:285–292. 5. Lellouche N, Buch E, Celigoj A, et al. Functional characterization of atrial electrograms in sinus rhythm delineates sites of parasympathetic innervation in patients with paroxysmal atrial fibrillation. J Am Coll Cardiol 2007;50:1324 – 1331. 6. Pachon MJ, Pachon ME, Pachon MJ, et al. A new treatment for atrial fibrillation based on spectral analysis to guide the catheter RF-ablation. Europace 2004;6: 590 – 601. 7. Lo LW, Tai CT, Lin YJ, et al. Progressive remodeling of the atrial substrate—a novel finding from consecutive voltage mapping in patients with recurrence of atrial fibrillation after catheter ablation. J Cardiovasc Electrophysiol 2007;18: 258 –265. 8. Chang SL, Tai CT, Lin YJ, et al. The efficacy of inducibility and circumferential ablation with pulmonary vein isolation in patients with paroxysmal atrial fibrillation. J Cardiovasc Electrophysiol 2007;18:607– 611. 9. Chen SA, Hsieh MH, Tai CT, et al. Initiation of atrial fibrillation by ectopic beats originating from the pulmonary veins: electrophysiological characteristics, pharmacological responses, and effects of radiofrequency ablation. Circulation 1999;100:1879 –1886. 10. Higa S, Tai CT, Chen SA. Catheter ablation of atrial fibrillation originating from extrapulmonary vein areas: Taipei approach. Heart Rhythm 2006;3:1386 –1390. 11. Tai CT, Chen SA, Tzeng JW, et al. Prolonged fractionation of paced right atrial electrograms in patients with atrial flutter and fibrillation. J Am Coll Cardiol 2001;37:1651–1657. 12. Tsao HM, Wu MH, Huang BH, et al. Morphologic remodeling of pulmonary veins and left atrium after catheter ablation of atrial fibrillation: insight from long-term follow-up of three-dimensional magnetic resonance imaging. J Cardiovasc Electrophysiol 2005;16:7–12. 13. Cheema A, Dong J, Dalal D, et al. Long-term safety and efficacy of circumferential ablation with pulmonary vein isolation. J Cardiovasc Electrophysiol 2006; 17:1080 –1085. 14. Cheema A, Vasamreddy CR, Dalal D, et al. Long-term single procedure efficacy of catheter ablation of atrial fibrillation. J Interv Card Electrophysiol 2006;15: 145–155. 15. Lin YJ, Tai CT, Kao T, et al. Frequency analysis in different types of atrial fibrillation. J Am Coll Cardiol 2006;47:1401–1407. 16. Kalifa J, Jalife J, Zaitsev AV, et al. Intra-atrial pressure increases rate and organization of waves emanating from the superior pulmonary veins during atrial fibrillation. Circulation 2003;108:668 – 671. 17. Sanders P, Nalliah CJ, Dubois R, et al. Frequency mapping of the pulmonary veins in paroxysmal versus permanent atrial fibrillation. J Cardiovasc Electrophysiol 2006;17:965–972. 18. Oral H, Chugh A, Lemola K, et al. Noninducibility of atrial fibrillation as an end point of left atrial circumferential ablation for paroxysmal atrial fibrillation: a randomized study. Circulation 2004;110:2797–2801. 19. Spach MS, Dolber PC. Relating extracellular potentials and their derivatives to anisotropic propagation at a microscopic level in human cardiac muscle. Evidence for electrical uncoupling of side-to-side fiber connections with increasing age. Circ Res 1986;58:356 –371. 20. Chang SL, Tai CT, Lin YJ, et al. Biatrial substrate properties in patients with atrial fibrillation. J Cardiovasc Electrophysiol 2007;18:1134 –1139.
sites/patient). Type 2 includes high-DF sites in the LA or LA plus the PVs (n ⫽ 21, 11 ⫾ 5.6 sites/patient). In type 1, PV isolation (PVI) could eliminate the AF with negative AF inducibility testing after the PVI in 89% of patients. In type 2, additional LA substrate modification was needed in 81% of patients because sustained AF was induced after the PVI (P⬍.001, compared with type 1). Multivariate analysis showed that the lower mean voltage of the LA and high-frequency sites distribution both independently predicted a positive AF inducibility after the PVI (P⬍.05). CONCLUSIONS Spectral analysis during SR can detect an abnormal atrial substrate. A regional distribution of the high-DF sites predicts the efficacy of the PVI.
RESULTS The regional distribution of the high-DF sites in the LA could be divided into two types. Type 1 includes high-DF sites existing only in the pulmonary veins (PVs; n ⫽ 19, 6.6 ⫾ 3.4
KEYWORDS Ablation; Fibrillation; Electrogram
Introduction
Methods
Regions of abnormal atrial substrate can be identified by multiple rapid deflections, fractionated electrograms, and low electrogram voltage during sinus rhythm (SR).1–5 These areas with fibrillatory myocardium may play a role in the perpetuation of atrial fibrillation (AF) and indicate an atrial substrate abnormality. Recently, Pachon et al6 showed that the atrial substrate can be characterized by the frequency spectra obtained from single electrograms during SR. According to this study, clusters of fibrillatory myocardium presented with higher frequencies than the other surrounding normal atrial substrate.6 However, the substrate characteristics, anatomic distribution, and role of the high-frequency sites during the AF were not clear. The aims of this study were (1) to investigate the atrial substrate properties of the high-frequency sites identified during SR and (2) to assess the catheter ablation results based on the different types of distribution of the high-frequency sites.
Patient characteristics
Address reprint requests and correspondence: Shih-Ann Chen, M.D., Division of Cardiology, Taipei Veterans General Hospital, 201, Section 2, Shih-Pai Road, Taipei, Taiwan. E-mail address: [email protected]. (Received February 4, 2008; accepted March 26, 2008.)
(Heart Rhythm 2008;5:968 –974) © 2008 Heart Rhythm Society. All rights reserved.
This study enrolled 40 consecutive paroxysmal AF patients (mean age 52 ⫾ 12 years) who underwent radiofrequency (RF) catheter ablation under the guidance of a NavX mapping system (NavX 6.0, St. Jude Medical, Inc., St. Paul, MN). Patients with repeated procedures were not included in this study.
Electroanatomic mapping and electrophysiological study Each patient underwent an electrophysiological study and catheter ablation in the fasting state, after informed consent was obtained. The method of three-dimensional (3D) electroanatomic mapping has been described elsewhere.7,8 In brief, after acquiring the LA geometry, the 4-mm-tip catheter (EP Technologies, Boston Scientific, Inc., Boston) was selected as the roving catheter for sequential contact mapping. The roving catheter was used to collect local bipolar signals while it was swiped across the left atrium (LA) during SR. The signal from the mapping catheter was used to build a sequential map. After completion of the sequential map, all bipolar signals were exported for further analysis.
1547-5271/$ -see front matter © 2008 Heart Rhythm Society. All rights reserved.
doi:10.1016/j.hrthm.2008.03.039
Lin et al
Figure 1
Spectral Analysis during Sinus Rhythm
969
Schematic presentation of the electrogram morphology analysis and fast Fourier transform of single discrete electrograms during SR.
Provocation of AF was performed in all patients. We attempted to find the spontaneous onset of atrial ectopic beats or repetitive episodes of short runs or sustained AF before or after the infusion of isoproterenol or after an algorithm designed for facilitating the initiation of AF. The methods used to provoke spontaneous AF were attempted at least twice to ensure reproducibility.7–10
Signal recording and the analysis Bipolar atrial electrograms were collected in the LA with a point-by-point approach sequentially during SR under the guidance of a NavX system.7,8 The signals for the frequency analysis and electrogram morphology analysis were both exported from the NavX mapping system and analyzed by the Matlab computer program (MathWorks Inc. Natick, MA). The sampling rate of the NavX mapping system was 1200 Hz. Regarding the electrogram morphology analysis, the electrogram characteristics (activation duration, number of deflections, and peak-to-peak bipolar voltage) were measured before and after the pulmonary vein (PV) isolation (Figure 1).11 The duration of the atrial electrogram was measured from the earliest atrial activation site to the latest atrial activation site from a stable baseline. The number of deflections was assessed by counting the number of negative deflections crossing the baseline at least 10 ms apart. Spectral analysis was performed on the single discrete bipolar electrogram during SR (unrectified, Hanning window, 1 second in duration; Figure 1). Then the data were exported to an external computer program including single discrete electrograms during SR and the baseline on both sides of the discrete electrogram. The fast Fourier transform analysis was performed using a Hanning window function on each segment from all recording sites in the LA. The dominant frequency (DF) was defined as the frequency with the maximum power in the frequency range. Higher frequencies were mapped toward the purple end of the NavX color spectrum, as a real time built-in function. To ensure the reliability of the DF detection, the lowest noise signal was chosen for the analysis. Overall, 10.6 ⫾ 10 sites (equivalent to 4.8% ⫾ 3.6% of LA sites) per patient were excluded. Further, to avoid too high density in some regions, any mapping site with a distance of less than 0.5
mm to the nearest neighboring site was excluded. Spectral analysis was performed before the catheter ablation in all patients. In the preliminary study, the consistency of the DF value from two consecutive beats during SR was investigated in a total of 344 points in the initial five patients. The BlandAltman agreement test showed an acceptable result (22/344: 6.2% of the data pairs outside 2 standard deviations [SDs]), indicating the consistency of the DF during SR over time. In addition, the averaged density of the mapping sites was 2.1 ⫾ 0. 44 sites/cm2 in all patients. The maximal distance between any two mapping sites was 16.3 ⫾ 4.9 mm. The point density was 3.1 ⫾ 0.9 sites/cm2 in the anteroseptal wall of the LA, 2.2 ⫾ 0.38 sites/cm2 in the LA roof area, 2.3 ⫾ 0.43 sites/cm2 in the LA posterior wall, and 3.0 ⫾ 1.2 sites/cm2 in the PV area (P⬎.05). Therefore, the point density was nearly uniformly distributed and acceptable for the analysis.
Definitions Dominant frequency peak: The largest power of the frequency peak was identified as the DF peak. High frequency sites: The determination of the abnormal atrial substrate was based on the frequency spectrum of each site during SR. The frequency analysis was obtained from a total of 7708 electrograms in the LA during SR, with an average of 193 ⫾ 53 sites per patient. The mean DF value was 47 ⫾ 20 Hz. According to the normal distribution curve of the DF values, the top 5% were higher than 70 Hz, and those sites were defined as high frequency sites.
Catheter ablation of paroxysmal AF As described in the previous study, the location of the PV ostium was determined by both PV angiography and the local electrogram morphology.12 It was also confirmed by the 3D mapping system with the LA geometry reconstruction.8 Continuous circumferential lesions were created by encircling the right and left PV ostia guided by the NavX system using a 4-mm-tip ablation catheter. RF energy was applied continuously while repositioning the catheter tip every 40 seconds with a target temperature of 50°–55° and maximum power of 50 W in the temperature control mode. The temperature was reduced to 45°–50° when RF energy to
970 Table 1
Heart Rhythm, Vol 5, No 7, July 2008 Clinical and electrophysiological characteristics of the study patients.
Gender, male/female Age, years Duration of AF, years LV ejection fraction, % LA diameter, mm Underlying cardiovascular disease (coronary artery disease, heart failure, hypertrophic cardiomyopathy) (%) Mean no. of mapping sites in the LA Mean voltage of the LA, mV No. of high-frequency sites in the LA before the PV isolation
Type 1 (n ⫽ 19)
Type 2 (n ⫽ 21)
P
10/9 51 ⫾ 4.5 ⫾ 60 ⫾ 36 ⫾ 3 (16)
16/5 53 ⫾ 5.3 ⫾ 58 ⫾ 41 ⫾ 4 (19)
.11 .50 .80 .45 .06 .65
14 3.0 7.2 6.3
210 ⫾ 63 2.3 ⫾ 0.63 6.6 ⫾ 3.4
10 6.2 9.9 6.0
209 ⫾ 35 1.7 ⫾ 0.78 11 ⫾ 5.6
.98 .02 .01
Data are presented as mean value ⫾ SD unless otherwise indicated.
the posterior wall near the esophagus was necessary. The intention was to place the RF lesions at least 1–2 cm away from the angiographically defined ostia. After completion of the circumferential lesion set, the ipsilateral superior and inferior PVs were mapped carefully by circular catheter recording (Spiral, AF Division, St. Jude Medical, Inc., Minnetonka, MN) during SR or coronary sinus (CS) pacing. After successful isolation of all four PVs, which was confirmed by circumferential mapping of the PVs, high-current (3–5 times the pacing threshold) and wide (8 ms) pulse duration stimulation from the proximal and distal CS was performed (in 10-ms decrements from 250 to 150 ms, with a duration of each pacing cycle length of 5–10 seconds) and repeated 3–5 times. If an induced AF/LA flutter was sustained for ⬎1 minute, an additional linear ablation was performed randomly at either the anterior roof or mitral isthmus; the reentry circuit of the LA flutter was identified by isochrone mapping, entrainment maneuvers, and the postpacing interval analysis. If sustained LA AF/flutter was still induced, cardioversion was performed to restore SR.
Follow-up of AF recurrence After discharge, the patients underwent follow-up (2 weeks after the catheter ablation, then every 1–3 months) at our cardiology clinic or with the referring physicians, and antiarrhythmic drugs were prescribed for 8 weeks to prevent the early recurrence of paroxysmal AF (⬍1 month after the ablation). When the patients experienced symptoms suggestive of a tachycardia after the ablation, 24-hour Holter monitoring or cardiac event recordings were performed to define the cause of the clinical symptoms. AF recurrence was defined as an episode lasting more than 1 minute and confirmed by electrocardiograms 2 months after the ablation (blanking period).
Statistical analysis All continuous data were presented as the mean value ⫾ SD. Comparisons of the continuous data were performed with one-way analysis of variance. The univariate variables of various electrophysiologic variables were ascertained to determine the predictors of a positive inducibility test after the PV isolation. Various clinical and electrophysiological
variables including sex, age, structural heart disease, LA size, left ventricular ejection fraction, presence of highfrequency sites in the LA, and mean bipolar voltage of the LA were used to assess the predictors of a positive inducibility test after the PV isolation. The variables that were selected to be tested in the multivariate analysis were those with a P-value of less than 0.2 in the univariate models. A two-sided P⬍.05 was considered statistically significant.
Results Patient characteristics A total of 40 patients who underwent RF ablation of paroxysmal AF were included. The clinical and electrophysiological characteristics of the study patients are shown in Table 1.
Frequency analysis results Characteristics of the high-frequency sites identified during SR The electrogram morphology and spectral morphology differed between the high-frequency sites and non-high-frequency sites. As shown in Table 2, the high-frequency sites harbored a longer activation duration, more deflections, and lower peak-to-peak bipolar voltage compared with the nonhigh-frequency sites. Regarding the frequency spectra, 94% of the non-high-frequency sites presented a single DF peak, but only 31% of the high-frequency sites presented a single DF peak (P⬍.05). The other 69% of the high-frequency sites presented two or more than two DF peaks (Figure 1). The average DF value of the fundamental peaks in the high-frequency sites was similar to that of the DF peaks of the non-high-frequency sites (36.8 ⫾ 8.7 Hz vs. 40.3 ⫾ 11.7 Hz; P ⫽ .1). Distribution of the high-frequency sites in the LA Overall, 342 mapping sites had a DF value higher than 70 Hz in all 40 patients. These high-DF mapping sites were found in the specific regions and not uniformly distributed in the LA (Figure 2). The highest DF value of each patient was 104 ⫾ 24 Hz. There were 8.7 ⫾ 5.1 (range 1–19) high-frequency sites in the LA and PV region in each
Lin et al Table 2
Spectral Analysis during Sinus Rhythm
971
Characteristics of the high-frequency sites and non-high-frequency sites
Variables Electrogram morphology: Activation duration, ms No. of deflections Peak-to-peak bipolar voltage, mV Spectral morphology: Percentage of single DF peak morphologies (of overall electrograms) DF peak value, Hz
High-frequency sites
Non-high-frequency sites
49 ⫾ 7.1 3.10 ⫾ 0.76 1.57 ⫾ 0.56
32 ⫾ 9.2 1.86 ⫾ 0.63 1.99 ⫾ 0.74
31 82.1 ⫾ 11.6
94 40.3 ⫾ 11.7
P⬍.001 for all comparisons.
patient. The distribution of the high-frequency sites was as follows: 215 sites (63%) inside or around the PVs, 68 sites (20%) in the LA septum and anterior wall, 37 sites (11%) in the posterior wall, and 20 sites (6%) in the LA roof. Patients with a larger LA had more high-frequency sites in the LA (Pearson’s correlation, r ⫽ 0.37, P ⫽ .02). On the other hand, this did not correlate with the other clinical variables, such as the sex, age, left ventricular ejection fraction, or presence of structural heart disease (P⬎.05).
Characterization of the atrial substrate based on the high-frequency sites distribution The regional distribution of the high-frequency sites can be divided into two types. In type 1 (n ⫽ 19, 48%; Figure 2), the high-frequency sites existed only in the PVs. The highest DF sites were inside the PV isolation line. In these patients, the high-frequency sites were observed in the right superior PV in 15/19 patients (79%), in the right inferior PV in 8/19 patients (42%), in the left superior PV in 9/19
Figure 2 Electrogram morphology and frequency spectra and the regional distribution of the DF sites in a patient with a type 1 LA substrate before and after PV isolation. The electrogram at the high-frequency sites exhibited rapid and multiple deflections with high fragmented peaks of greater than 70 Hz. The purple areas in the reconstructed geometry of the LA indicate the high-DF sites that are higher than 70 Hz. Note that after the ablation, the high-DF peaks of 118 Hz were eliminated with the remaining DF around 32 Hz. The time-domain electrograms after the PV isolation were also less fractionated than those before the ablation. The high-frequency sites were eliminated by right PV isolation.
972
Heart Rhythm, Vol 5, No 7, July 2008
Figure 3 The regional distribution of the high-frequency sites in a patient with a type 2 LA substrate during SR is shown in the left panel. The high-frequency sites were in the PV and LA. The intracardiac bipolar electrograms and their corresponding frequency spectra of the high-frequency sites (sites 1 and 2) and the non-high-frequency sites (sites 3 and 4) are shown. The spectral morphology in the AF termination site in the LA anterior roof (site 1) showed a high-DF peak of 71 Hz during SR. Electrogram morphology showed multiple and rapid deflections. On the other hand, in the non-high-frequency sites in the low anterior septum (site 2), the bipolar electrograms were not complex, and the spectral morphology exhibited a single DF peak value of less than 70 Hz.
patients (47%), and in the left inferior PV in 3/19 patients (16%). In type 2 (n ⫽ 21, 52%; Figure 3), the high-frequency sites were observed in the LA or in the LA plus the PVs. The highest DF sites were outside the circumferential isolation line with DF gradient into the PVs. Among them, 13 patients (59%) had high-frequency sites in both the LA and PVs. Table 1 shows the clinical and substrate properties of the type 1 and type 2 distributions. Type 2 patients were associated with a larger LA size and a lower mean peak-topeak bipolar voltage compared with type 1 patients.
isolation line. Meanwhile, the high-frequency sites in the LA were more frequent in the patients with a positive AF inducibility compared with a negative AF inducibility (12/ 17, 70.6% vs. 6/23, 26.1%; P ⫽ .006). After the PV isolation plus linear lines, six patients in whom AF was induced after ablation had high-frequency sites in the LA (6/6, 100%). Those were located at the septum in four patients (67%), in the LA roof in four patients (67%), in the posterior wall in three patients (50%), and in the lateral mitral isthmus in one patient (17%).
Effects of the PV isolation and LA linear ablation As shown in Figure 4, in type 1 (n ⫽ 19, Figure 2) patients, with high-frequency sites in the PVs, circumferential PV isolation rendered the AF noninducible in 17 (93%) of 19 patients (Figure 3). After linear ablation (LA roof line in two patients), the AF became noninducible in all patients (19/19, 100%). In type 2 (n ⫽ 21; Figure 3) patients, with high-frequency sites in the LA or LA plus the PVs before the ablation, the AF remained inducible in 17 (81%) out of the total 21 patients after the circumferential PV isolation. After the linear ablation (a roof line in 17 patients and lateral mitral isthmus line in 11 patients), the AF inducibility remained positive in six of the 21 patients (29%; P⬍.05, when compared with type 1). In type 1, the number of high-frequency sites within the PVs decreased by 68% (from overall 215 to 69 mapping sites) after the PV isolation. In type 2, most of the highfrequency sites in the LA remained unchanged after the PV isolation, and the number of high-frequency sites decreased by 39% (from overall 127 to 76 mapping sites). These high-frequency sites eliminated by the PV isolation were all located in the LA septum and were adjacent to the right PV
Figure 4 Flow chart of the high-frequency sites mapping and ablation results in all 40 patients. Note that the inducibility was negative in all patients with a type 1 distribution after the circumferential PV isolation and linear ablation. On the other hand, the AF inducibility test was negative in 76% of the type 2 AF patients.
Lin et al Table 3
Spectral Analysis during Sinus Rhythm
973
Univariate and multivariate predictors of the AF inducibility after PV isolation Multivariate analysis
Factors
Positive AF inducibility (n ⫽ 19; 45%)
Negative AF inducibility (n ⫽ 21; 55%)
Univariate analysis (P-value)
Sex Age Structural heart disease (%) LA size, mm LV ejection fraction, % The presence of high-frequency sites in LA (%) Mean peak-to-peak LA bipolar voltage, mV
16/3 54 ⫾ 9.9 5 (26) 41 ⫾ 5.5 57 ⫾ 8.6 17 (89.5%) 1.47 ⫾ 0.63
10/11 50 ⫾ 14 2 (9.5) 36 ⫾ 5.9 60 ⫾ 8.7 4 (19.1%) 2.27 ⫾ 0.51
0.01 0.27 0.9 0.007 0.338 ⬍0.001 ⬍0.001
Hazard ratio (confidence interval 95%) 2.6 (0.01-6.98) 11.4 (0.98–1.45) 34.4 (2.0–506) 37.6 (2.1–665)
P .375 — — .08 — .01 .013
Data are presented as mean value ⫾ SD unless otherwise indicated.
Prediction of a positive inducibility test after PV isolation Univariate and multivariate predictors of the requirement for additional LA linear ablation lesions after circumferential PV isolation are detailed in Table 3. The clinical and electrophysiologic variables associated with a positive AF inducibility after PV isolation were a male gender, larger LA dimension, lower mean bipolar electrogram voltage of the LA, and the presence of high-frequency sites in the LA. The other factors did not differ in regard to the AF inducibility testing after the PV isolation. A multivariate analysis showed that the lower bipolar voltage of the LA (cutoff value of 1.9 mV) and the distribution of the high-frequency sites in the LA independently predicted a positive AF inducibility after the PV isolation, indicating abnormal LA substrate in these patients.
Long-term follow-up of the catheter ablation With a mean follow-up of 10.5 ⫾ 3.2 months, there were four patients who had recurrence of AF. The long-term success rate on or off previously ineffective antiarrhythmic drugs was 95% and 86% (in types 1 and 2, respectively; P⬎.05).
Discussion Main findings Spectral analysis during SR can identify an abnormal LA substrate with a DF peak above 70 Hz. These high-frequency sites were associated with multiple rapid deflections, a prolonged electrogram, and low bipolar voltage of the electrogram morphology. The LA substrate could be characterized by a distribution of the high-frequency sites in the LA and PVs. The presence of a high-frequency site in the LA independently predicted a positive AF inducibility after the circumferential PV isolation.
Compared with previous studies Pachon et al6 demonstrated that an abnormal atrial substrate could be differentiated from a normal or compact atrial substrate by using a fast Fourier analysis during SR. Those abnormal atrial sites were characterized by fractionated electrograms in the time-domain signals and high-frequency
peaks in the frequency spectra. Areas of fractionation provided potential targets for the AF ablation. RF ablation of those “AF nest sites” eliminated the high-frequency peaks, and the resultant frequency spectra were similar to those of the “non–AF nest sites.”6 This study demonstrated that the high-frequency sites were characterized by specific electrogram and spectral morphologies during SR. Spectral analysis revealed multiple and high-DF peaks above 70 Hz. The electrogram morphology can also provide the same information, and these can also be identified by the multiple rapid deflections and prolonged and fractionated bipolar electrograms, as in previous studies.2,3,6,7 Currently, circumferential ablation during PV isolation has become the standard procedure of catheter ablation of paroxysmal AF with an optimal long-term success.13,14 Pachon et al6 showed that selective LA catheter ablation based on the spectral analysis results can be effective without circumferential PV isolation. However, the effect of PV isolation on the high-frequency sites was not clear. This study showed that high-frequency sites within the PV ostia could be identified in 80% of the paroxysmal AF patients. PV isolation can eliminate 69% of the high-frequency sites within the PV isolation line. Although catheter ablation of the high-frequency sites based on the spectral analysis was not performed in this study, further studies are needed to show whether or not an ablation strategy targeting those high-frequency sites would be additive to the circumferential PV isolation.
Characteristics of an atrial substrate with highfrequency sites The majority of the high-frequency sites were located in the PVs and/or PV ostia. These results fit well with the findings of the high-frequency AF sources in the animal and human studies.15–17 After PV isolation, DF peaks of the highfrequency sites above 70 Hz were greatly reduced or eliminated (Figure 2). Concomitantly, the PV potentials were eliminated, suggesting that the high-DF peaks and conduction disturbances were related to the PV activation. In those patients, the PV isolation could treat the AF without any additional LA substrate modification. On the other hand, in
974 type 2 patients, LA high-frequency sites were frequently observed in the LA septum, posterior wall, and roof areas. In patients with a type 2 LA substrate, they were associated with a lower mean bipolar voltage and dilated LA before the ablation, which indicates a diseased atrial substrate. Previous studies have also indicated that the positive inducibility of AF after circumferential PV isolation required additional LA substrate modification for a better clinical outcome.8,18 The higher AF inducibility in the type 2 patients further confirmed the abnormal LA substrate in those patients. The underlying mechanism of the high-frequency sites is not known. The complex electrograms may be explained first by a localized fibrosis and anisotropy with discontinuous conduction.19 Second, the parasympathetic innervation sites of the LA also harbored similar fractionated and prolonged sinus electrograms; however, they were characterized by a high-voltage amplitude, and the characteristics may have differed from the high-frequency sites in the present study.5
Clinical implications Sustained AF could be induced by burst atrial pacing after PV isolation in around 50% of patients with paroxysmal AF. Overall, 30%–50% of patients with paroxysmal AF and more than 70% of patients with chronic AF needed additional atrial substrate modification.8,18 Previous studies showed that more extensive scarring of the LA or a lower mean bipolar voltage of the LA indicated a higher recurrence of AF after circumferential PV isolation.4,20 This study showed that the distribution of the high-frequency sites and presence of low-voltage areas both independently indicated an abnormal LA substrate. However, the optimal strategy for the LA substrate modification based on the voltage distribution in paroxysmal or chronic AF patients has not yet been determined. Identification of the distribution of the high-frequency sites may be beneficial for finding the AF ablation targets.6
Limitations First, mapping of the atrial substrate should be performed in SR; therefore, patients with a focal-type paroxysmal AF with rapid and frequent initiating AF were not included in this study. Second, the effect of the activation rhythm on the spectral analysis results was not assessed in this study. Third, patients without AF were not enrolled, and frequency analysis results during SR were not available in normal control patients. The high-frequency sites may be observed in the normal LA substrate.
Conclusion Spectral analysis during SR can detect an abnormal atrial substrate before the catheter ablation of AF. These highfrequency sites were associated with multiple rapid deflections, a prolonged electrogram, and low bipolar voltage of
Heart Rhythm, Vol 5, No 7, July 2008 the electrogram morphology. The regional distribution of the high-DF sites (higher than 70 Hz) in the LA predicted the efficacy of the PV isolation and the need for additional LA substrate modification.
References 1. Lin YJ, Tai CT, Huang JL, et al. Characterization of right atrial substrate in patients with supraventricular tachyarrhythmias. J Cardiovasc Electrophysiol 2005;16:173–180. 2. Nakao K, Seto S, Ueyama C, et al. Extended distribution of prolonged and fractionated right atrial electrograms predicts development of chronic atrial fibrillation in patients with idiopathic paroxysmal atrial fibrillation. J Cardiovasc Electrophysiol 2002;13:996 –1002. 3. Tanigawa M, Fukatani M, Konoe A, et al. Prolonged and fractionated right atrial electrograms during sinus rhythm in patients with paroxysmal atrial fibrillation and sick sinus node syndrome. J Am Coll Cardiol 1991;17:403– 408. 4. Verma A, Wazni OM, Marrouche NF, et al. Pre-existent left atrial scarring in patients undergoing pulmonary vein antrum isolation: an independent predictor of procedural failure. J Am Coll Cardiol 2005;45:285–292. 5. Lellouche N, Buch E, Celigoj A, et al. Functional characterization of atrial electrograms in sinus rhythm delineates sites of parasympathetic innervation in patients with paroxysmal atrial fibrillation. J Am Coll Cardiol 2007;50:1324 – 1331. 6. Pachon MJ, Pachon ME, Pachon MJ, et al. A new treatment for atrial fibrillation based on spectral analysis to guide the catheter RF-ablation. Europace 2004;6: 590 – 601. 7. Lo LW, Tai CT, Lin YJ, et al. Progressive remodeling of the atrial substrate—a novel finding from consecutive voltage mapping in patients with recurrence of atrial fibrillation after catheter ablation. J Cardiovasc Electrophysiol 2007;18: 258 –265. 8. Chang SL, Tai CT, Lin YJ, et al. The efficacy of inducibility and circumferential ablation with pulmonary vein isolation in patients with paroxysmal atrial fibrillation. J Cardiovasc Electrophysiol 2007;18:607– 611. 9. Chen SA, Hsieh MH, Tai CT, et al. Initiation of atrial fibrillation by ectopic beats originating from the pulmonary veins: electrophysiological characteristics, pharmacological responses, and effects of radiofrequency ablation. Circulation 1999;100:1879 –1886. 10. Higa S, Tai CT, Chen SA. Catheter ablation of atrial fibrillation originating from extrapulmonary vein areas: Taipei approach. Heart Rhythm 2006;3:1386 –1390. 11. Tai CT, Chen SA, Tzeng JW, et al. Prolonged fractionation of paced right atrial electrograms in patients with atrial flutter and fibrillation. J Am Coll Cardiol 2001;37:1651–1657. 12. Tsao HM, Wu MH, Huang BH, et al. Morphologic remodeling of pulmonary veins and left atrium after catheter ablation of atrial fibrillation: insight from long-term follow-up of three-dimensional magnetic resonance imaging. J Cardiovasc Electrophysiol 2005;16:7–12. 13. Cheema A, Dong J, Dalal D, et al. Long-term safety and efficacy of circumferential ablation with pulmonary vein isolation. J Cardiovasc Electrophysiol 2006; 17:1080 –1085. 14. Cheema A, Vasamreddy CR, Dalal D, et al. Long-term single procedure efficacy of catheter ablation of atrial fibrillation. J Interv Card Electrophysiol 2006;15: 145–155. 15. Lin YJ, Tai CT, Kao T, et al. Frequency analysis in different types of atrial fibrillation. J Am Coll Cardiol 2006;47:1401–1407. 16. Kalifa J, Jalife J, Zaitsev AV, et al. Intra-atrial pressure increases rate and organization of waves emanating from the superior pulmonary veins during atrial fibrillation. Circulation 2003;108:668 – 671. 17. Sanders P, Nalliah CJ, Dubois R, et al. Frequency mapping of the pulmonary veins in paroxysmal versus permanent atrial fibrillation. J Cardiovasc Electrophysiol 2006;17:965–972. 18. Oral H, Chugh A, Lemola K, et al. Noninducibility of atrial fibrillation as an end point of left atrial circumferential ablation for paroxysmal atrial fibrillation: a randomized study. Circulation 2004;110:2797–2801. 19. Spach MS, Dolber PC. Relating extracellular potentials and their derivatives to anisotropic propagation at a microscopic level in human cardiac muscle. Evidence for electrical uncoupling of side-to-side fiber connections with increasing age. Circ Res 1986;58:356 –371. 20. Chang SL, Tai CT, Lin YJ, et al. Biatrial substrate properties in patients with atrial fibrillation. J Cardiovasc Electrophysiol 2007;18:1134 –1139.