Circumferential Radiofrequency Ablation of Pulmonary Vein Ostia
A New Anatomic Approach for Curing Atrial Fibrillation
Background—The pulmonary veins (PVs) and surrounding ostial areas frequently house focal triggers or reentrant circuits critical to the genesis of atrial fibrillation (AF). We developed an anatomic approach aimed at isolating each PV from the left atrium (LA) by circumferential radiofrequency (RF) lesions around their ostia.
Methods and Results—We selected 26 patients with resistant AF, either paroxysmal (n=14) or permanent (n=12). A nonfluoroscopic mapping system was used to generate 3D electroanatomic LA maps and deliver RF energy. Two maps were acquired during coronary sinus and right atrial pacing to validate the lateral and septal PV lesions, respectively. Patients were followed up closely for ≥6 months. Procedures lasted 290±58 minutes, including 80±22 minutes for acquisition of all maps, and 118±16 RF pulses were deployed. Among 14 patients in AF at the beginning of the procedure, 64% had sinus rhythm restoration during ablation. PV isolation was demonstrated in 76% of 104 PVs treated by low peak-to-peak electrogram amplitude (0.08±0.02 mV) inside the circular line and by disparity in activation times (58±11 ms) across the lesion. After 9±3 months, 22 patients (85%) were AF-free, including 62% not taking and 23% taking antiarrhythmic drugs, with no difference (P=NS) between paroxysmal and permanent AF. No thromboembolic events or PV stenoses were observed by transesophageal echocardiography.
Conclusions—Radiofrequency PV isolation with electroanatomic guidance is safe and effective in either paroxysmal or permanent AF.
Atrial fibrillation (AF) is a common arrhythmia associated with significant morbidity and mortality. For many years, the only curative treatment has been surgical, with extensive atrial incisions used to compartmentalize the atrial mass below that critical for perpetuating AF.1 Recently, transcatheter linear radiofrequency (RF) ablation in the right atrium (RA) and/or left atrium (LA) has been used to replicate the surgical procedures in patients with paroxysmal or chronic AF.2 3 However, uncertainty remains concerning the requisite number of lesions, their optimal location, and the need for continuous lines. Indeed, focal ablation has been proposed as an alternative approach on the basis of the demonstration that ectopic beats originating within or at the ostium of the pulmonary veins (PVs) may be the source of paroxysmal and even persistent AF.4 5 6 Despite high acute success rates, the feasibility of this technique is limited by the difficulty in mapping the focus if the patient is in AF or has no consistent firing, the frequent existence of multiple foci causing high recurrence rates, and an incidence of PV narrowing as high as 42%.5 To circumvent these limitations, we developed an anatomic approach in which circumferential RF lesions are created around the ostia of each PV, with the aim to isolate these veins from the LA while reducing the risk of PV stenosis. A nonfluoroscopic 3D electroanatomic navigation system was used for generating and validating the continuity of circular lines. We report the feasibility, safety, and clinical outcome of this technique in patients with resistant AF, either paroxysmal or permanent.
Patients were enrolled who had (1) paroxysmal AF, with documented daily sustained (>1 hour) episodes despite the use of antiarrhythmic drugs at therapeutic doses (number of drugs, 3±1), or (2) symptomatic permanent AF for ≥6 months, resistant to pharmacological and/or electrical cardioversion (≥2 unsuccessful attempts or AF recurrence within 1 month despite prophylaxis with 4±1 drugs). All patients had to be taking effective oral anticoagulation for ≥4 weeks. Diagnostic workup included Holter monitoring, transthoracic and transesophageal echocardiography, laboratory tests, and thyroid function evaluation.
Of 26 patients selected, 14 had paroxysmal and 12 had permanent AF, and 18 (69%) had no structural heart disease (Table 1⇓). Informed consent was obtained from each patient according to a protocol approved by the Institutional Human Research Committee.
Antiarrhythmic drugs (except amiodarone) and digoxin were discontinued for ≥5 half-lives. Quadripolar 6F catheters were placed in the coronary sinus (CS), RA, and right ventricular apex. The LA and PVs were explored through transseptal catheterization. Heparin was titrated to maintain a partial thromboplastin time of 60 to 90 seconds.
The nonfluoroscopic navigation system (CARTO; Biosense Webster) has been described elsewhere.2 7 With use of a special mapping and ablation catheter, 3D electroanatomic maps are reconstructed that display the spatial distribution of local endocardial activation times (LATs) relative to a reference electrogram.
The navigator catheter was placed 2 to 4 cm into each PV and slowly pulled back. Along pullback, multiple locations were recorded to tag the vein. The ostium was identified by fluoroscopic visualization of the catheter tip entering the cardiac silhouette with simultaneous impedance decrease and appearance of atrial potential. Three locations were recorded along the mitral annulus to tag valve orifice. LA maps were obtained by sequentially acquiring a minimum of 50 points.
Figure 1⇓ portrays the study protocol. In patients in sinus rhythm (SR) at the beginning of the procedure, maps were acquired during pacing from the CS or RA appendage at a cycle length (CL) of 600 ms (Figure 2⇓). Each endocardial location was recorded while a stable catheter position was maintained, as assessed by both end-diastolic stability (a distance <2 mm between 2 successive locations) and LAT stability (an interval <2 ms between 2 successive LATs). When split potentials were recorded, the LAT was derived from the steeper of the two. For patients in AF, maps were acquired to assess the distribution and types of electrograms by a previously reported method.7 Local CLs were automatically analyzed and displayed as histograms, which were classified as follows: type A, defined as fairly regular activation with a clear isoelectric baseline; type B, irregular activation with perturbations of baseline and/or highly fragmented electrograms; and type C, alternation between A and B (Figure 3⇓). According to an anatomic classification,8 the LA was divided into 9 regions, listed in Table 2⇓. The proportion of type A, B, or C signals in each region was computed.
RF pulses were delivered in unipolar mode to a cutaneous ground patch via the distal catheter electrode. Because all 4 PVs may serve as a source of AF,4 our end point was the creation of circumferential lines of conduction block around each PV. These lines consisted of contiguous focal lesions deployed at a distance ≥5 mm from the ostia. With a maximum temperature setting of 60°C, RF energy (up to 50 W) was applied for 60 to 120 seconds until local electrogram amplitude was reduced by 80%. During AF, the same power titration technique was used, but current was always delivered for 60 to 120 seconds. If there was an impedance rise, or the patient had cough, burning pain, or severe bradycardia, RF delivery was stopped.
Remap Process and Lesion Validation
In patients in SR, postablation remap was performed utilizing the preablation anatomic map for the acquisition of new points to allow accurate comparison of pre- and post-RF activation sequence. In patients in AF, after restoration of SR, the remap was done with the anatomic map acquired during AF, to maintain the same landmarks and lesion tags for reliable lesion verification. We found a small intrapatient difference between the anatomic map of a fibrillating, noncontracting atrium and the map during pacing, in which locations are recorded at end diastole. This was validated by measuring the distance between corresponding locations acquired during AF and pacing. We tested a set of 5 points per patient in a sample of 10 patients. No differences were noted between 3 paired measurements (mean difference 0.18±0.05 mm, t=0.74, P=0.86).
Lesion validation required acquisition of 2 maps during CS and RA pacing for the lateral and septal PVs, respectively. The rationale behind this setting was to pace from a site close to the lesions and shorten conduction time to the ablation site, thereby allowing detection of delayed activation inside the circular line.
The following criteria were used to define line continuity:
1. Low peak-to-peak bipolar potentials (≤0.1 mV) inside the lesion, as determined by local electrogram analysis and voltage maps (Figure 4⇓).
2. LAT delay >30 ms between contiguous points lying in the same axial plane on the external and internal sides of the line, as assessed by activation maps (Figure 2⇑). Changes in activation spread were also evaluated with propagation maps (Figures 5⇓ and 6⇓).
The presence of double potentials straddling the line was interpreted as a gap.
After ablation, patients underwent 48-hour telemetry monitoring. Nineteen patients (73%) were discharged without the need for antiarrhythmic drugs. Of the remaining 7 (27%) patients, amiodarone was maintained for 4 because of other arrhythmias, and 3 who had in-hospital AF episodes were given a previously ineffective antiarrhythmic drug. Other medications, including calcium-antagonists, β-blockers, and digoxin when appropriate, were prescribed to patients who underwent electrical cardioversion and those with cardiac disease. Oral anticoagulation was continued for 3 months.
Follow-up consisted of outpatient visits with serial echocardiograms and Holter monitoring performed on symptom recurrence or routinely at 1 week and every month for ≥6 months. The procedure was considered successful if no recurrences of AF lasting >30 seconds were present during postdischarge follow-up. Transesophageal echocardiography was performed within 3 days and 1 to 6 months after ablation to assess potential PV stenosis.
Dichotomous variables were compared by χ2 tests. Log-linear techniques for multiway contingency tables were used to compare the proportion of AF types at different LA regions, with Bonferroni correction for pairwise comparisons. Continuous measures are expressed as mean±SD and were compared by ANOVA. Statistical significance was inferred at P<0.05.
Mapping and Ablation Procedure
Preablation maps comprised 94±21 points. For AF maps, 787 sites were sampled, and 258 (33%) showed type A electrograms, 102 (13%) type B, and 427 (54%) type C. Table 2⇑ reports spatial electrogram distribution. More than 50% of signals around the PVs (except the inferior septal vein) were of type A.
Overall procedure duration was 290±58 minutes, and 118±16 RF pulses were deployed (Table 3⇓). For patients in SR, procedures lasted 247±41 minutes (range, 198 to 298 minutes) versus 327±43 minutes (range, 282 to 390 minutes) for those in AF (P<0.05), with shorter mapping time in SR patients (60±5 versus 96±17 minutes, P<0.05) and similar fluoroscopy time (25±3 versus 27±3 minutes, respectively; P=NS).
In 9 (64%) of 14 patients in AF at the beginning of the procedure, SR was acutely restored during ablation; RF was being delivered around the superior lateral PV in 6 (67%) of these 9 patients, although this was not the initial site of ablation in all but 1 patient. In 5 patients who were still in AF after completion of ablation, the arrhythmia was terminated by direct current shocks.
After ablation, among 104 PVs treated, complete isolation was demonstrated by absence of discrete electrical activity (voltage ≤0.1 mV) at all sites inside the lesion in 79 (76%) (Table 4⇓). Such lesions were associated with marked LAT delay (58±11 ms). Incomplete lesions were dichotomized into those with LAT delay >30 ms (11/104 [11%], including 2 superior septal, 6 inferior septal, and 3 inferior lateral PVs) and those without delay (14/104 [13%], including 8 inferior septal and 6 inferior lateral PVs). Interestingly, incomplete lesions without delay were associated with significant LAT prolongation and amplitude decrease versus pre-RF signals.
No patient developed intolerable pain or severe cough during RF delivery. One patient (4%) who had hemopericardium recovered well after pericardiocentesis. There were no strokes or other thromboembolic events. Two patients had mild pericardial effusion managed medically. During the first 48 hours from ablation, 3 patients (12%) developed spontaneously terminating AF episodes lasting from 7 minutes to 2 hours. All patients were discharged in SR. In patients with paroxysmal AF, predischarge echocardiography demonstrated unchanged LA transport function (peak A-wave velocity 0.54±0.06 m/s before RF versus 0.51±0.07 m/s after RF; P=NS). Of the 12 patients with permanent AF, mitral A waves were detectable in all patients who returned to SR during ablation and in only 1 of the 5 who were defibrillated (peak A velocity 0.37±0.12 m/s). During follow-up, all patients without AF recurrence showed preserved LA contraction, with mitral inflow tracings demonstrating progressive improvement (peak A velocity 0.60±0.09 m/s, P<0.05 versus early post-RF). Transesophageal echocardiography showed no high-velocity turbulence near the ostia that suggested PV stenosis in any of the patients (peak flow velocities [m/s]: pre-RF, 0.59±0.10; post-RF, 0.65±0.13; follow-up, 0.62±0.09; P=NS for all comparisons).
After 9±3 months of follow-up, 22 patients (85%) had stable SR, including 16 (62%) who were no longer taking antiarrhythmic drugs and 6 (23%) who were still taking drugs (Table 3⇑). Overall freedom from AF was not dissimilar (P=NS) for paroxysmal AF (12/14 patients [86%], including 7 not taking antiarrhythmic drugs) and permanent AF (10/12 patients [83%], including 9 not taking drugs). AF recurred in 4 patients (15%): 2 ablated for paroxysmal AF developed brief (<1 hour) and rare (<3 per month) episodes, and 2 who had permanent AF developed sustained episodes responsive to drugs. There was no difference between patients with and without recurrence in the number of PVs with incomplete lines (6/16 veins [38%] in patients with recurrence versus 19/88 veins [22%] in those without recurrence, P=NS).
This study describes a new ablative technique for curing AF by creating circumferential lesions around the PV ostia based on an anatomic map. This approach appears more feasible than CARTO-guided LA compartmentalization through a single long line encircling the PVs altogether and connected to the mitral annulus by a curvilinear line along the lateral LA wall. Difficulties in deploying such complex lesions result in long procedural times (up to 10 hours) and extensive use of fluoroscopy.2 9 In the present study, procedures for patients in SR lasted ≈250 minutes, significantly less than for patients in AF, with very short fluoroscopic times in both groups. Thus, in a routine setting, when one may not acquire detailed AF maps and patients may even be cardioverted before ablation, procedural times are quite satisfactory. Development of new catheter designs allowing generation of ring lesions at PV ostia with a single application will further improve the feasibility of our approach.
Electrophysiological Evidence of Lesion Completeness
Unlike linear lesions, a circular line of block around an isolated tube such as a PV should result in no discrete local electrical activity beyond the line. With this criterion, PV isolation was demonstrated by voltage maps in 76% of the veins treated. For such lesions, however, visualization of entry, albeit delayed, into the ablated area (activation and propagation maps) is in apparent contrast with the existence of complete conduction block. In this case, either the lesion is not truly complete or the CARTO system, which is able to sense amplitudes as low as 10-3mV, is recording far-field electrical activity.
Interestingly, 11% of the lesions did not meet the voltage criterion but had a significant conduction delay. This suggests that the empiric rules regarding what constitutes an adequate amount of delay to denote a complete line are particularly difficult to define for a circumferential lesion. In addition, even lesions that did not satisfy both the activation delay and voltage criteria were associated with significant changes in the activation sequence and amplitude compared with preablation. This is noteworthy because such lesions may have produced enough atrial injury to achieve a therapeutic effect, as suggested by the lack of relationship between lesion completeness and AF recurrence in the present study.
RF applications around the PV orifices were well tolerated. The rate of important pericardial effusion (4%) was similar to that in previous transseptal studies.3 5 9 We had no cases of PV stenosis, probably because lesions were deployed ≥5 mm apart from the ostia, thereby avoiding scarring and contraction of the venous wall resulting from thermal injury.
The rate of in-hospital AF episodes was 12% in the present study, lower than that occurring after surgical maze procedure (47%) or linear ablation (63%), probably because of lesser lesion extent.1 2 After a mean follow-up of 9 months, 85% of the patients were free of AF, with 62% no longer taking antiarrhythmic drugs (true success rate) and 23% taking drugs that had been ineffective before ablation. These results are better than those obtained with CARTO-guided linear LA ablation alone and comparable to those of biatrial ablation in paroxysmal AF,2 and they appear satisfactory especially considering that RA triggers were not weighed and that “complete” block was obtained only in two thirds of veins.
Our finding of similar outcomes in paroxysmal and permanent AF raises important issues as to which is the mechanism underlying the efficacy of PV isolation. During AF in our patients, the areas around the PVs predominantly showed organized atrial electrograms, in keeping with previous studies using multielectrode mapping systems.3 10 Such regular electrogram patterns have been shown to correlate closely with the shortest atrial effective refractory periods, which, in turn, represent an important determinant of AF persistence.11 Thus, our anatomy-based RF lesions may have altered AF arrhythmogenic substrate, as also suggested by the acute SR restoration during ablation in 64% of patients. This interpretation is further supported by a study showing termination of chronic AF with ablation that targeted sites of organized activity adjacent to PV openings.3
Alternatively, PV isolation might have interrupted pathways crucial in the genesis of AF located at the PV-LA junction. This hypothesis is consistent with experimental studies showing that ablation of the ligament of Marshall (which is adjacent to the superior lateral PV) can prevent isoproterenol-induced AF in dogs12 and that PV isolation alone without maze procedure can eliminate chronic AF in a sheep model.13 At the same time, electrical disconnection of all 4 PVs at their ostia may block the egress of potential AF triggers arising from within the veins.4 5 6 The lack of correlation between AF recurrence and achievement of complete lines of block may be due to the fact that the discontinuous lines involved mainly the inferior PVs, which contain less-developed myocardial sleeves and have less-frequent ectopic potentials than the superior veins.4 5 8 Finally, atrial debulking and/or denervation may have contributed to suppression of AF.14
The complexity of this technique should not be underestimated. Our results have been obtained at a center with extensive experience in performing CARTO-guided atrial ablation (>500 procedures in the last 2 years). For this approach to be used safely, detailed institutional protocols are needed, including special staff training, expert monitoring of the patients, and careful outcome assessment.
Insight into the mechanism of AF is limited by the lack of RA mapping and the fact that only qualitative assessment of electrogram types was performed.
Although 89% to 94% of AF triggers have been shown in the PVs, the arrhythmia can be initiated by ectopy from the crista terminalis, CS ostium, and atrial free wall.4 5 Therefore, a limitation to a purely anatomic map is that a complex PV isolation procedure can be performed, yet the source of AF may not be the PVs. However, ablation may still be effective through mechanisms other than isolation of PV foci.
Our population was predominantly composed of young patients who had no structural heart disease, normal LA size, and good cardiac function. This clinical profile is commonly associated with resistant AF and a high likelihood of a focal source.4 5 6 Whether our results can be extrapolated to a broader AF population is unknown.
This study lends further evidence to the concept of the dominance of the LA in the region of the PVs in the initiation and/or maintenance of AF. The efficacy of PV isolation in both paroxysmal and permanent AF supports the current notion of a common pathogenesis (involving various combinations of focal activity and reentry) with a spectrum of clinical presentations. On the basis of our findings, one can envision that PV isolation may be proposed as a valuable alternative to either focal ablation or biatrial compartmentalization.
- Received April 5, 2000.
- Revision received June 12, 2000.
- Accepted July 7, 2000.
- Copyright © 2000 by American Heart Association
Pappone C, Oreto G, Lamberti F, et al. Catheter ablation of paroxysmal atrial fibrillation using a 3D mapping system. Circulation. 1999;100:1203–1208.
Chen SA, Hsieh MH, Tai TC, et al. Initiation of atrial fibrillation by ectopic beats originating from the pulmonary veins. Circulation. 1999;100:1879–1866.
Ernst S, Schluter M, Ouyang F, et al. Modification of the substrate for maintenance of idiopathic human atrial fibrillation. Circulation. 1999;100:2085–2092.
Doshi RN, Wu TJ, Yashima M, et al. Relation between ligament of Marshall and adrenergic atrial tachycardia. Circulation. 1999;100:876–883.
Fieguth HG, Wahlers T, Borst HG. Inhibition of atrial fibrillation by pulmonary vein isolation and auricular resection: experimental study in a sheep model. Eur J Cardiothorac Surg. 1997;11:714–721.
Elvan A, Pride HP, Eble JN, et al. Radiofrequency catheter ablation of the atria reduces inducibility and duration of atrial fibrillation in dogs. Circulation. 1995;91:2235–2244.