Author + information
- Received July 14, 2005
- Revision received October 25, 2005
- Accepted October 31, 2005
- Published online April 4, 2006.
- Yenn-Jiang Lin, MD⁎,†,
- Ching-Tai Tai, MD⁎,
- Tsair Kao, PhD‡,
- Han-Wen Tso, MS‡,
- Satoshi Higa, MD⁎,
- Hsuan-Ming Tsao, MD⁎,
- Shih-Lin Chang, MD⁎,†,
- Ming-Hsiung Hsieh, MD⁎,† and
- Shih-Ann Chen, MD⁎,†,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Shih-Ann Chen, Division of Cardiology, Taipei Veterans General Hospital, 201, Sec. 2, Shih-Pai Road, Taipei, Taiwan
Objectives This study sought to investigate the regional frequency distribution from multiple bi-atrial sites in different types of paroxysmal atrial fibrillation (AF).
Background A previous study showed a left atrium (LA) to right atrium (RA) frequency gradient in patients with paroxysmal AF.
Methods Forty-four patients (age = 60 ± 16, male patients = 27) with paroxysmal AF originating from the pulmonary veins (PVs) (n = 31) or superior vena cava (SVC) (n = 13) were included. Frequency analysis was performed on the intracardiac electrograms (7 s, 1 kHz/channel) recorded from PV, posterior LA, coronary sinus (CS), posterolateral RA, and SVC. The largest peak frequency was identified as the dominant frequency (DF).
Results In the PV-AF patients, there was a frequency gradient from the PV ostium to the LA, RA, and SVC (8.5 ± 3.3 Hz vs. 5.9 ± 1.1 Hz vs. 5.2 ± 0.85 Hz vs. 5.5 ± 0.48 Hz, respectively, p < 0.001). The highest DFs were mostly located at the arrhythmogenic PV ostium (58%). The DFs of the arrhythmogenic PV and PV ostium were significantly higher than those of the non-arrhythmogenic PVs and PV ostia (p < 0.05). In the SVC-AF patients, there was a frequency gradient from the SVC to the RA, LA, and PV (8.0 ± 2.4 Hz vs. 5.9 ± 1.1 Hz vs. 5.9 ± 0.7 Hz vs. 5.8 ± 0.7 Hz, respectively, p = 0.001). The highest DFs were mostly located inside the SVC (77%) instead of the SVC ostium (as compared with PV-AF patients, p = 0.035).
Conclusions The location of the highest DF depended on the arrhythmogenic PV or SVC. A frequency gradient was present between the arrhythmogenic thoracic vein and atrium in all patients.
Atrial fibrillation (AF) is initiated by pulmonary vein (PV) and non-PV foci, and the ablation of AF foci can prevent the re-initiation of AF (1–3). Once initiated, AF has originally been described as involving multiple wavelets in the atria that are important for the maintenance of the AF (4). However, animal and human studies have shown various degrees of spatiotemporal organization during sustained AF. Recent evidence from high-density mapping and spectral analysis also has shown that AF is associated with rotors and regular repetitive activation in part of the atrium. In previous animal studies, the AF waves from high-frequency sources in the left atrial (LA)/PV area showed an LA to right atrium (RA) frequency gradient across the Bachmann bundle, suggesting that the LA/PV area is important for maintaining sustained AF (5,6). In human studies, frequency analysis also has shown that there is a frequency gradient from the LA to the RA in AF patients (7,8). However, this regional distribution of the dominant frequency (DF) has not been described for different types of paroxysmal AF, originated from PVs and the superior vena cava (SVC). The purpose of this study was to investigate the distribution of the DF by using frequency domain analysis on multiple bi-atrial recordings, including from inside the PVs, PV ostium, LA (posterior wall near the ostium), coronary sinus (CS), SVC, and RA (posterolateral wall), during different types of paroxysmal AF.
The study consisted of 44 patients with frequent episodes of PAF (more than one episode per week, documented by electrocardiogram and 24-h Holter recording) referred to this institute for an electrophysiological study and catheter ablation. They were refractory to or intolerant of 1 to 3 (mean, 2 ± 2) antiarrhythmic drugs. All of the patients underwent echocardiography, and the size of the LA and left ventricle and the ejection fraction of the left ventricle were measured.
Each patient underwent an electrophysiological study and catheter ablation in the fasting, non-sedated state after informed consent was obtained. As described previously, all antiarrhythmic drugs except for amiodarone were discontinued for at least five half-lives before the procedure (3,9). A duodecapolar catheter (electrode width, 1 mm; interelectrode spacing, 2 mm; Daig Corp., Minnetonka, Minnesota) was placed along the posterolateral RA and advanced to the SVC with the proximal five bipolar pairs in the RA. The SVC to RA junction was determined fluoroscopically during SVC angiography (9). A 7-F deflectable decapolar catheter with a 2-mm interelectrode distance and 5-mm spacing between each electrode pair (Daig Corp.) was inserted into the CS via the internal jugular vein. After selective PV angiography or the venous phase of the pulmonary artery angiography to define the PV ostium, two decapolar catheters (same electrode spacing as the CS catheter) were inserted into the PVs via the transseptal approach with the second pair of bipolar electrodes straddling the PV ostium and the first bipolar pair (the most proximal) outside the PV. These two catheters were first placed at the superior PVs and then at the inferior PVs for recordings (Fig. 1).The location of the PV ostium was also confirmed by the local electrogram morphology, which had no PV potentials in the first bipolar recording.
We attempted to find the spontaneous onset of atrial ectopic beats or repetitive episodes of short runs or sustained AF and predict the location of the initiating foci (3,9). When the AF patients presented to the electrophysiology laboratory in sinus rhythm, we first attempted to observe the spontaneous activation during baseline or after an isoproterenol infusion. If spontaneous initiation of AF did not appear, burst atrial pacing was used to induce AF. After AF was sustained for more than 5 min, external cardioversion was attempted to convert the AF to sinus rhythm and observe for the spontaneous initiation of AF. The methods used to provoke spontaneous AF were attempted at least twice to ensure the reproducibility. If the AF patients presented to the electrophysiology laboratory in sustained AF, all the bipolar recordings for the frequency analysis were made immediately after the transseptal puncture. Patients with induced or spontaneous initiation of AF had at least a 2-min waiting period before the bipolar recordings were made. If the patients required isoproterenol to induce the AF, the drug was discontinued after the AF occurred and a 5-min waiting period was carried out before any intracardiac recordings were made for the frequency analysis.
Data acquisition and signal analysis
The bipolar atrial electrograms from the multiple recording sites in the LA and RA during AF were recorded using a Cardiolab system (Prucka Engineering Inc., Houston, Texas). Seven-second recordings were sampled at 1 kHz and stored on a removable hard disk for offline analysis. Each intra-atrial recording was filtered with a second-order, zero-phase Butterworth filter at 40 to 250 Hz. A second-order, zero-phase low-pass filter at 20 Hz was then applied to the absolute value of the resulting signal. An attenuation of the QRS-T complex of bipolar signals was performed by an adaptive cancellation technique before the frequency analysis, which has been described previously (10). The final step of the process involved frequency analysis. A fast Fourier transform with a Hamming window was performed for each 7-s segment from the multiple recording sites in the LA and RA. The largest peak frequency of the resulting spectrum was identified as the DF. To ensure the reliability of the DF detection, the clearest signal was chosen for analysis. The DF of the CS was measured from the proximal and distal bipoles of the CS catheter. The DF of the LA (posterior wall near the PV ostium), PV ostium, and inside of the PV was measured from the bipolar signals of the first bipole (most proximal), second bipole, and third to fifth bipoles of the decapolar catheter, respectively. The DF of the RA (posterolateral wall) was averaged from the proximal five bipoles of the duodecapolar catheter.
All continuous data were presented as the mean value ± standard deviation (SD). Comparisons between patients with PV-AF and SVC-AF were made with the Student ttest or chi-square test as appropriate. The regional differences in the DFs from multiple mapping sites were compared by one-way ANOVA. Paired DF comparisons were made after performing a Bonferroni correction after multiple comparisons. To detect the consistency of the DFs over time, two measurements of the DF, obtained from recordings 5 min apart at the same site, were evaluated using an intra-class correlation coefficient. In addition, a repeated-measure ANOVA was used to analyze the difference in the first and second DF measurements (time effect) and whether there was an interaction between the various mapping sites (CS, RA, LA, and SVC) over time. Statistical significance was considered when the two-sided p value was <0.05.
This study consisted of 44 patients (60 ± 16 years of age, male/female ratio, 27:17) enrolled retrospectively, including 13 consecutive patients with AF from the SVC and 31 age-matched control patients with AF from PVs selected randomly in the contemporary period. There was no difference between the age, underlying disease, LA dimension, left ventricular dimension, left ventricular ejection fraction, number of anti-arrhythmic drugs used, previous history of AF, and success rate of the ablation between the patients with paroxysmal AF initiated by pulmonary vein foci (n = 31) and those with SVC foci (n = 13) (Table 1).Only a predominance of female patients in the SVC-AF patients was noted (p = 0.01). In both groups of patients, no anti-arrhythmic drugs were used during the procedure.
During the electrophysiological study, an overall number of 40 arrhythmogenic foci originating from the PVs were responsible for the initiation of the sustained AF (1.3 PVs per patient), including 18 in the right superior PV (RSPV, 45%), 11 in the left superior PV (LSPV, 27.5%), 5 in the right inferior PV (RIPV, 12.5%), and 6 in the left inferior PV (LIPV, 15%). Regional differences in the DFs from the bi-atrial and thoracic vein recordings were observed in all patients. There was no significant difference among the overall DFs recorded from the four PVs (RSPV, 6.8 ± 1.9 Hz; LSPV, 6.2 ± 2.2 Hz; RIPV, 7.6 ± 2.5 Hz; LIPV, 6.7 ± 1.4 Hz, p = NS). In all patients the highest DF was located inside one PV or around the PV ostium. Among those, the highest DF was located within one of the arrhythmogenic PV ostia (n = 18, 58%) or inside the arrhythmogenic PV (n = 13, 42%).
In Figure 2,an example of the regional distribution of the fast Fourier transform and bipolar electrograms in a patient with PAF from the RSPV is shown. The recordings from the RSPV ostium showed rapid repetitive activity with the highest DF compared with the other recording sites. In patients with PV-AF, there was a frequency gradient from the arrhythmogenic PV ostium to the LA (posterior wall near the PV ostium), RA (posterolateral wall), SVC, proximal CS, and distal CS (8.5 ± 3.3 Hz, 5.9 ± 1.1 Hz, 5.2 ± 0.85 Hz, 5.5 ± 0.48 Hz, 5.2 ± 0.82 Hz, and 5.2 ± 0.94 Hz, respectively; p < 0.001) (Fig. 3).Furthermore, the DFs of the arrhythmogenic PVs and PV ostia were significantly higher than those of the non-arrhythmogenic PVs and PV ostia (7.3 ± 2.2 Hz vs. 6.3 ± 1.9 Hz, p = 0.032 for PV; 8.5 ± 3.3 Hz vs. 6.6 ± 1.8 Hz, p = 0.044 for PV ostium). The DFs of the other non-arrhythmogenic thoracic veins were lower than or equal to that of the adjacent atrium (5.11 ± 0.92 Hz vs. 5.59 ± 1.04 Hz, p = 0.184), suggesting passive activation in those veins.
In the patients with paroxysmal AF from the SVC, bipolar signals were obtained from the SVC, SVC/RA junction, RA posterolateral wall, CS, and two superior PVs in all patients. In these patients, the highest DF was located inside the SVC (n = 10, 77%) or SVC/RA junction (n = 3, 23%). In Figure 4,an example of regional bipolar electrograms and their corresponding frequency spectrum in a patient with SVC-AF are shown. The regional DF distribution showed a frequency gradient from inside the SVC to the SVC ostium, RA (posterolateral wall), LA (posterior wall near the PV ostium), PV, and CS (8.0 ± 2.4 Hz, 6.4 ± 1.3 Hz, 5.9 ± 1.1 Hz, 5.9 ± 0.7 Hz, 5.8 ± 0.7 Hz, and 5.6 ± 1.0 Hz, respectively; p = 0.001; Fig. 3). There was no significant difference in the DFs between the RA and LA (p = NS). In five patients (38%), after a successful SVC isolation, the DFs of the isolated AF inside the SVC were similar to those recorded before the ablation (7.5 ± 1.4 Hz vs. 7.4 ± 1.7 Hz; p = NS), whereas the intracardiac electrograms in the RA and the surface electrocardiogram showed sinus rhythm.
Differences in the frequency analysis results in the two groups of patients
The frequency analysis results showed different characteristics between those obtained in arrhythmogenic PVs and those obtained in the SVC. First, in the patients with PV-AF, the highest DFs within the arrhythmogenic PV or PV ostium were significantly higher than those in the arrhythmogenic SVC or SVC ostium (9.3 ± 2.9 Hz vs. 8.0 ± 2.4 Hz, p = 0.045). Second, the highest DFs were mostly located at the arrhythmogenic PV ostium, as compared with the patients with SVC-AF, in whom the highest DF was mostly located inside the SVC, instead of the ostium (p = 0.035). The most important observation of all was that there was a frequency gradient from the arrhythmogenic vein to the nearby atrium in both groups of patients.
The RA (posterolateral wall) frequency was higher in the SVC-AF patients (5.9 ± 1.1 Hz vs. 5.2 ± 0.85 Hz, for SVC-AF and PV-AF, respectively; p = 0.02), whereas the LA (posterior wall near the PV ostium) frequency was similar between the two groups (5.9 ± 0.69 Hz vs. 5.9 ± 1.1 Hz, for SVC-AF and PV-AF, respectively; p = NS). The frequency gradient between the atria was larger in the PV-AF patients than in the SVC patients (2.2 ± 2.2 Hz vs. 0.18 ± 0.78 Hz, p = 0.014).
The LA-RA-CS relationship in patients with paroxysmal atrial fibrillation
Do the CS recordings reflect the activity of the RA or the LA? In both groups of patients, we compared the DFs among those obtained from the CS, LA (posterior wall near the PV ostium), and RA (posterolateral wall) recordings. In the PV-AF patients, the DF of the CS was 44% and 15% lower than in the arrhythmogenic PVs and LA, respectively (r = 0.2 and 0.25, p = NS). On the other hand, the DF of the CS recording was only 2% lower than the recordings from the posterolateral RA (r = 0.9 and 0.92 for proximal CS and distal CS, respectively; p < 0.001). In the SVC-AF patients, the DFs of the CS correlated with the PV/LA recordings and were only 5% and 4% lower than the PV and LA recordings, respectively (r = 0.88 and 0.95, respectively; p < 0.001). Although the DF of the proximal CS correlated with the RA recording (r = 0.59, p = 0.01), there was a significant variance between the DFs of the distal CS and RA (r = 0.39, p = 0.1).
From the initial five patients with paroxysmal AF from the PVs, we recorded the frequency spectra 5 min apart to examine the consistency of the DFs over time. During sustained AF, the DFs calculated from the intracardiac electrocardiograms (in overall 42 mapping sites) showed excellent agreement (intraclass correlation coefficient, r = 0.862, 95% confidence interval 0.759 to 0.923), suggesting the temporal stability of the DF during these two different recording times. Furthermore, a repeated-measure ANOVA showed that the consistency of the DF measurement did not depend on the various mapping sites (RA, LA, CS, SVC, and PV, p > 0.05).
The present study confirms the finding that the regional distribution of the DFs differed for the different types of human paroxysmal AF. In both groups of patients with paroxysmal AF, originating from the PVs or SVC, there was a frequency gradient from the arrhythmogenic thoracic vein to the nearby atrium during sustained AF. The highest DFs were located inside or around the arrhythmogenic thoracic vein ostium. An LA-to-RA frequency gradient was only observed in the patients with paroxysmal AF originating from the PVs.
Comparison with previous studies and clinical implications
Previous frequency analysis showed a single dominant peak with rapid regular activity in part of the atrium during AF (5–8). Analysis of the spatial distribution of the DFs in animal models of AF has indicated that the area with the highest frequency of activation is likely to be located in the PV or posterior wall of the LA (5,6). In human studies, analysis of fibrillatory waves recorded from the epicardial electrodes or endocardial recordings showed a similar frequency distribution with the highest DF in the PVs (11,12). Detailed activation mapping also showed that re-entry in the distal PV and PV-LA junction was a necessary substrate for AF maintenance (13,14). On the other hand, Ndrepepa et al. (15) used basket catheters to show that the fastest AF activity was in the posterior wall of the LA, suggesting that the AF activity originated from the posterior wall instead of focal firing from the PV area during sustained AF. These findings support the hypothesis that AF is maintained by rapid repetitive activity from the PV/LA in the majority of AF, which originates from the PVs.
The present study showed that the highest DF was located within the arrhythmogenic PV or its ostium in the PV-AF patients. Non-arrhythmogenic PVs are mainly passively activated by the conduction during sustained AF. In a substantial number of patients with SVC-AF, the rapid repetitive activation was located in the SVC or its ostium. Therefore, an empirical anatomic PV isolation cannot be applied to all patients with paroxysmal AF. Some patients (e.g., AF from SVC) have a different location for the AF maintenance and need a different approach. This raises the possibility that selective isolation of the arrhythmogenic vein may be enough to treat some patients with paroxysmal AF.
Frequency distribution in the two groups of AF patients
The present study showed a regional distribution of the DFs in patients with paroxysmal AF initiated by PV or SVC ectopy. During sustained AF, a focal source with high-frequency activity may be the underlying cause of paroxysmal AF. The highest DF was located within the arrhythmogenic vein or at its ostium. More activation wave fronts are present at the area closer to the highest DF, and lesser activation wave fronts with conduction block are considered when propagating away from the driver because of the sink-to-source effect as fibrillatory conduction.
We observed the lack of RA-to-LA frequency gradient in SVC-AF patients. It could be attributable to the following reasons. First, the highest DFs within the arrhythmogenic SVC were lower than those in the arrhythmogenic PV. Therefore, the frequency gradient between the sink and the source would be less clear in SVC-AF patients. Second, this may be caused by the distance between the source of the DF and the mapping site. The LA mapping site was close to sites of the highest DF at PV ostium in PV-AF patients. On the other hand, the RA mapping site (posterolateral wall) was far away from the sites of highest DF inside the SVC in SVC-AF patients. Therefore, the RA-to-LA frequency gradient may be not evident in SVC-AF patients.
The present study also showed different frequency distributions in the two groups of AF patients. First, the location of the highest DFs in the SVC-AF patients was mostly located within the SVC, rather than in the PV ostium as in the PV-AF patients. These results suggested that the myocardial sleeves inside the distal SVC provided enough arrhythmogenic substrate for the high-frequency source of the AF. Therefore, local AF confined to the SVC after SVC isolation may be more prominent in SVC-AF patients (16). Second, the DFs of the arrhythmogenic SVC were lower than those of arrhythmogenic PVs. As the high-frequency waves from the arrhythmogenic vessels cross the veno-atrial junction to the atrium, they encounter an area of electrophysiological heterogeneity that may lead to wave break and result in the frequency gradients. The driver at the SVC or SVC ostium harboring the lower DF could still drive the AF with fibrillatory conduction. The frequency gradient between the RA and LA may not have been evident in these SVC-AF patients; however, the frequency gradient between the arrhythmogenic vessels and atrium still existed, providing the optimal sites to create electric disconnection.
Coronary sinus recording in patients with paroxysmal AF
The present study showed that the DF of the CS did not correlate with the PV/LA recordings during AF in the PV-AF patients. These results were compatible with those of a previous study that found that the CS activity did not reflect the LA activity (17). However, the present study showed that in the SVC-AF patients, the DF of the CS could reflect the posterior LA activity, because both the LA and CS are mainly activated only passively by the conduction during AF. On the other hand, the DF of the proximal CS, instead of the distal CS recording, resembled those of the posterolateral RA in both groups of AF patients.
First, the results of this study were based on a retrospective analysis. Future prospective research is needed to explore whether these areas with high DFs are critical for the maintenance of AF. Second, only limited mapping sites around the PVs, PV ostium, posterior LA near the PV ostium, CS, and posterolateral RA were obtained. Therefore, they cannot represent the activation of the other areas. Third, intracardiac signals from the superior PVs and inferior PVs were obtained sequentially. However, the accuracy of frequency analysis was based on the temporal stability of the AF activity in the previous study (8) and in this study. Simultaneous recordings from other sites may provide more information about the frequency distribution. Fourth, the different distance between each bipolar pair in decapolar (5 mm) and duodecapolar (2 mm) catheters may affect the results of the frequency analysis. Because the micro–re-entrant circuit of the AF rotor is approximately 1 cm in diameter (18), the duodecapolar and decapolar catheters used in this study may provide sufficient spatial resolution to detect the area with the highest DF during AF.
Frequency analysis during sustained AF accurately identified the arrhythmogenic thoracic veins with the highest DF (or shortest cycle length of the activation) inside the arrhythmogenic veins or their ostium. Non-arrhythmogenic thoracic veins mainly act as passive conduction during sustained AF. A frequency gradient was present between the arrhythmogenic veins and nearby atrium in all patients. A left-to-right atrium frequency gradient was only observed in patients with paroxysmal AF originating from PVs.
- Abbreviations and Acronyms
- atrial fibrillation
- coronary sinus
- dominant frequency
- left atrium/atrial
- left inferior pulmonary vein
- left superior pulmonary vein
- pulmonary vein
- right atrium/atrial
- right inferior pulmonary vein
- right superior pulmonary vein
- superior vena cava
- Received July 14, 2005.
- Revision received October 25, 2005.
- Accepted October 31, 2005.
- American College of Cardiology Foundation
- Chen S.A.,
- Hsieh M.H.,
- Tai C.T.,
- et al.
- Moe G.K.
- Mansour M.,
- Mandapati R.,
- Berenfeld O.,
- Chen J.,
- Samie F.H.,
- Jalife J.
- Kalifa J.,
- Jalife J.,
- Zaitsev A.V.,
- et al.
- Lazar S.,
- Dixit S.,
- Marchlinski F.E.,
- Callans D.J.,
- Gerstenfeld E.P.
- Lin W.S.,
- Tai C.T.,
- Hsieh M.H.,
- et al.
- Konings K.T.,
- Smeets J.L.,
- Penn O.C.,
- et al.
- Kumagai K.,
- Ogawa M.,
- Noguchi H.,
- Yasuda T.,
- Nakashima H.,
- Saku K.
- Arora R.,
- Verheule S.,
- Scott L.,
- et al.
- Tsai C.F.,
- Tai C.T.,
- Hsieh M.H.,
- et al.
- Mandapati R.,
- Skanes A.,
- Chen J.,
- et al.