In paroxysmal atrial fibrillation (AF), success rates of catheter ablation (CA) exceed 80%, and recurrence is associated mostly with pulmonary vein (PV) reconnection (1- 2). By combining PV isolation, complex and fractionated atrial electrographic ablation, and linear lesions in patients with persistent AF, success rates without antiarrhythmic drugs (AADs) surpass 70% (3- 4). However, multiple procedures are often necessary, and creating linear lesions is sometimes challenging.

Surgical AF ablation has evolved from the original Cox maze procedure toward a minimally invasive, video-assisted procedure with new ablation tools to isolate the PVs and create linear lesions without opening the heart (5). Nevertheless, even bipolar radiofrequency (RF) energy cannot guarantee transmural lesions (6). Linear lesions such as the mitral isthmus cannot be created solely from the epicardium, and proving bidirectional block epicardially can be challenging (7).

Combining a transvenous endocardial and thoracoscopic epicardial approach in a single procedure overcomes these shortcomings.

We report our initial experience with long-term follow-up of minimally invasive epicardial bilateral PV isolation and linear lesions in combination with endocardial proof of conduction block and endocardial touch-up if indicated.

Twenty-six consecutive patients with symptomatic AF underwent hybrid thoracoscopic surgical and transvenous CA with follow-up of 1 year. Definitions of paroxysmal, persistent, and longstanding persistent AF, success and failure of ablation, and follow-up monitoring were based on the Heart Rhythm Society, European Heart Rhythm Association, and European Cardiac Arrhythmia Society consensus statement (3). Therapy with at least 1 AAD had failed. Other selection criteria were previously failed CA, left atrial volume ≥29 ml/m², persistent or longstanding persistent AF, or patient preference for a hybrid procedure instead of a percutaneous approach. Eleven patients (44%) had prior CA for AF or atrial flutter (AFL). All patients underwent transthoracic echocardiography, cardiac computed tomography, and pulmonary function testing preoperatively.

The procedure was performed under general anesthesia with double-lumen endotracheal tube placement for selective lung ventilation. Transesophageal echocardiography was used to exclude a thrombus in the left atrium (LA). On the right side, a 12-mm camera port was placed in the fifth intercostal space midaxillary line and in the sixth or seventh intercostal space anterior axillary line. A 5-mm working port was placed in the third intercostal space anterior axillary line. The pericardium was opened anterior to the phrenic nerve. Blunt dissection was used to open the transverse and oblique sinuses.

Via the femoral venous approach, a His bundle (St. Jude Medical, St. Paul, Minnesota) and coronary sinus catheter (Medtronic, Minneapolis, Minnesota) were placed under fluoroscopy, and transseptal puncture was performed with a long 8-F sheath (SL0, St. Jude Medical) into the LA. The patient then underwent heparinization (1,000 U heparin per 10 kg body weight and a heparin infusion), with activated clotting time >300 s. During rapid ventricular pacing, we injected contrast through the long sheath to visualize left atrial anatomy. The PVs were mapped with a circular mapping catheter (Lasso, Biosense Webster, Diamond Bar, California). Antral isolation of the right PVs as a pair was performed with 4 to 6 applications using a bipolar RF clamp (Atricure, West Chester, Ohio) (Figure 1). Each application had a duration of about 15 s, with a median output of 10 to 15 W. The same port incisions were made on the left side but placed more posteriorly (Figure 2). The pericardium was divided posterior to the phrenic nerve. Left PV isolation was conducted as described earlier. We did not attempt ablation of the ganglionated plexi.

In patients with severe chronic obstructive pulmonary disease , we performed thoracoscopic epicardial isolation of the PVs only on the right, and the left PVs were isolated using a cryothermal energy balloon catheter (Arctic Front, Cryocath, Montreal, Quebec City, Canada) endocardially to avoid bilateral sequential lung deflation.

The end point for PV ablation was entrance and exit block. We defined exit block as local capture in the PV during pacing from the Lasso catheter (output 10 mA, pulse width 2 ms) without conduction to the LA. In the case of sinus rhythm after PV isolation, reinduction of AF was attempted 5 times by pacing in the coronary sinus for 10 seconds at the shortest cycle length resulting in 1:1 atrial capture. AF was considered inducible if it lasted more than 1 min. If AF became noninducible, isoproterenol was infused at rates of 10 to 30 μg/min. If AF had not terminated or still was inducible, linear lesions were deployed.

A roof line (connecting both superior PVs) and an inferior line (connecting both inferior PVs) were made epicardially using a bipolar RF pen or linear pen device (Isolater Pen and Coolrail, Atricure). If the right atrium was dilated, 2 additional ablation lines were placed: 1 encircling the superior caval vein using the clamp, the other connecting both caval veins using the pen.

By making a roof and an inferior line, we isolated the posterior LA (box lesion). If entrance and exit block were not reached, we identified the conduction gaps endocardially and ablated those with a 3.5-mm-tip catheter (ThermoCool, Biosense Webster). The location of the linear lesions was visualized with the linear pen device in situ and using fluoroscopy.

A left isthmus line was made in 3 patients using the bipolar RF pen device. The line was started from the ablation line on the antrum of the left inferior PV toward and crossing the coronary sinus. All patients needed endocardial touch-up ablation to reach bidirectional block, starting from the mitral annulus toward or inside the coronary sinus. If the patient was known to have typical AFL or if this arrhythmia occurred during the procedure, the cavotricuspid isthmus (CTI) was ablated endocardially. The endpoint was bidirectional block. In 7 patients, the left atrial appendage was removed using a stapling device.

The pericardium was approximated with a stitch, and a chest tube was placed in both pleural cavities. There was no drain left in the pericardial space.

Low-molecular-weight heparin was started 6 h after the procedure, and on the second post-operative day, acenocoumarol was reinitiated.

Patients restarted as soon as possible their pre-operative AAD regimens. Acenocoumarol and AADs were discontinued after the 6-month monitoring visit confirmed the absence of atrial arrhythmia.

Any symptomatic patient not in sinus rhythm was cardioverted before the 3-month follow-up visit. One patient had a pacemaker, which was used for monitoring. The remaining patients underwent 7-day continuous Holter monitoring at 3, 6, 9, and 12 months. If 7-day Holter monitoring was not available, patients underwent at least 24-h Holter monitoring. According to current guidelines, success was defined as no episode of AF, AFL, or any atrial tachycardia (AT) lasting more than 30 s off AAD after the 3-month blanking period.

Data were retrospectively entered into a database. Statistical analysis was performed using SPSS version 16.0 (SPSS, Inc., Chicago, Illinois). Continuous variables are summarized with means and standard deviations. Outcomes are displayed using Kaplan-Meier plots and were compared using log-rank tests. Any episode of AF, AFL, or AT lasting more than 30 seconds detected after the 3-month post-procedural period by electrocardiography, pacemaker interrogation, or 7-day, 48-h, or 24-h continuous Holter monitoring performed at 6, 9, and 12 months was considered failure.

Twenty-six patients underwent hybrid procedures between May 29, 2008, and February 25, 2010. Patients' baseline characteristics are shown in (Table 1). Ten patients had persistent AF, and 1 had longstanding persistent AF. Eleven patients had 1 or 2 (Patient #25) previous CAs. None of the PVs were isolated in those patients, and there were no epicardial scars indicative of transmural lesions. The mean follow-up period was 470 ± 154 days (range: 221 to 858 days). The mean length of hospital stay was 7 ± 2 days (range: 5 to 13 days), and mean intensive care unit stay for post-procedural recovery was 1.0 ± 0.2 days (range: 1 to 2 days). The median length of the hybrid procedure (from initial skin incision to skin closure) was 280 ± 84 min (range: 195 to 505 min).

Eleven patients were in AF at the start of the procedure. In all patients, we achieved bidirectional block of all the PVs (Figure 3). In 2 patients, we did not perform any other lesion, because AF was not inducible. In 1 patient with inducible AF after PV isolation, we performed an additional roof line only. In 1 patient with inducible right AFL after PV isolation, we performed an additional CTI line only. Thereafter, no arrhythmia could be induced. In 22 patients, we created box lesions epicardially. In 17 patients (77%), we were able to demonstrate endocardial entrance and exit block in the box during sinus rhythm. After endocardial touch-up in 5 patients (23%), we completed the box lesions. In 3 patients, there was a gap at the junction of the right superior PV with the roof of the LA. One patient had a gap in the lateral portion of the roof line, and 1 patient in the middle of the inferior line. Because of ongoing AF or organization into mitral isthmus–dependent AFL, we performed in 3 patients mitral isthmus lines epicardially. In all cases, the lines had to be completed from the endocardium or coronary sinus, with a mean of 1.6 ± 0.5 applications of RF energy, resulting in bidirectional block (Figure 4). In 2 patients, CTI ablation was performed. One patient needed electrical cardioversion at the end of the procedure because of ongoing AF. All other patients were in sinus rhythm. Twenty-four patients (92%) were discharged on their pre-operative AAD regimens (amiodarone in 6 of 24 [25%]). All patients were discharged on acenocoumarol.

Twenty-four patients (92%) reached 1-year follow-up, and 96% underwent 7-day Holter monitoring 1 year after the procedure. The remaining 2 patients (8%) reached 6-month follow-up, and there was 100% compliance with 7-day Holter monitoring (Table 2).

At 1 year, 22 of 24 patients (92%) were in sinus rhythm, with no episodes of AF, AFL, or AT lasting longer than 30 s on office follow-up, Holter monitoring, or pacemaker interrogation (Figure 5). None of those patients were on AADs, and 5 (21%) were on acenocoumarol. One-year success, defined according to the Heart Rhythm Society, European Heart Rhythm Association, and European Cardiac Arrhythmia Society consensus statement (freedom from AF, AFL, and AT off AADs), was 93% for patients with paroxysmal AF and 90% for patients with persistent AF. Two of those 22 patients (9%) underwent CA for recurrent AF or left AFL after the hybrid procedure (Table 3). There was no statistically significant difference in patients with or without previous CA regarding the presence of sinus rhythm after 1 year without AADs (p = 1.00).

This means a single-procedure success rate (sinus rhythm without AAD and/or redo procedure) of 79% at 1 year for paroxysmal AF (11 of 14 patients) and 90% for persistent AF (9 of 10 patients) and an overall single-procedure success rate of 83% at 1 year.

At 6 months, 2 of 2 patients (100%) were in sinus rhythm, with no episodes of AF, AFL, or AT lasting longer than 30 s. One of these patients was taking flecainide, and they were both on acenocoumarol. None of those patients had an additional CA procedure.

The management of patients with recurrent supraventricular tachycardia (SVT) after the 3-month blanking period is summarized in (Table 3). Four of 26 patients (15%) had SVT recurrence, with a mean of 99 ± 94 (range: 29 to 239) SVT-free days after the blanking period. Two of these 4 patients underwent electrophysiologic studies and CA. One patient was found to have AF due to a gap at the junction of the right superior PV with the roof of the LA. The other patient had left AFL. These 2 patients were followed for at least 5 months and did not experience further SVT off AADs. The remaining 2 patients with SVT recurrence were treated with AADs for paroxysmal AF.

No deaths or conversion to cardiopulmonary bypass were encountered. No patient demonstrated paralysis of the phrenic nerve. One patient had a pleural effusion drained 3 weeks after surgery. One patient stayed hospitalized for 13 days because of difficulty controlling chest pain at the insertion sites of the working ports, without signs of infection.

This report describes for the first time combined simultaneous thoracoscopic surgical and transvenous CA of paroxysmal and persistent AF. These 2 complementary techniques performed in combination bear the potential of treating AF with a single ablation procedure. According to accepted definitions, our overall 1-year freedom from arrhythmia without AADs was 92%, with only 2 patients needing second CA procedures (3). Of note, almost half of the patients with persistent AF and over a third of all patients had 1 or more previous CA. The robustness of the hybrid approach lies in its complementary nature, as represented by the fact that in almost a quarter of patients, CA was needed to finish incomplete epicardial surgical lesions by endocardial touch-up, and additional lesions such as the mitral isthmus line can be performed rather quickly.

Jaïs et al. (8) found that if AF was inducible after PV isolation in patients with paroxysmal AF, the addition of linear lesions, with the end point of noninducibility, resulted in 91% of patients free from arrhythmia without AADs during a follow-up period of 18 ± 4 months. To reach that number, however, 31% of patients needed second procedures. Using the same end point, our success rate off AADs was comparable, but only 9% of patients needed second procedures.

The single-procedure, drug-free success rate of the stepwise ablation approach for persistent AF as described by Haïssaguerre et al. (9) was 62% at 11 ± 6 months. Allowing for repeat procedures and the use of AADs, the 1-year success rate increased to 95%. In our series, we followed a comparable ablation strategy, and the single-procedure, drug-free success rate at 1 year was 83%. Allowing for redo procedures, this number increased to 92%.

The hybrid approach appears to result in better end-points regarding rhythm control, for both paroxysmal and persistent AF, using the same end points and ablation strategy as described in other trials.

Approximately 80% of patients have at least partial recovery of PV conduction at 4 months after PV isolation with unipolar RF energy (10). Bipolar RF energy, as used in our series, overcomes the heat sink by clamping the tissue and excluding the effect of the circulating blood on ablation, which seems to result in more persistent lesions (11). But even for such devices, epicardial fat can prevent transmurality. In our series, box lesions were incomplete after epicardial ablation alone in 23% of the patients, and endocardial touch-up with unipolar RF energy was necessary to achieve bidirectional block. The possibility to perform such endocardial touch-up is 1 of the major advantages of this procedure.

By replacing the incisions of the traditional Cox maze III procedure with less invasive linear lesions of ablation using bipolar RF, Shen et al. (12) introduced the Cox maze IV procedure. This approach still requires cardiopulmonary bypass and at least 1 small right thoracotomy. Freedom from AF recurrence was 91% at 12 months, with 67% of patients off AADs. These figures are comparable with our results, although in the case of the hybrid procedure, neither cardiopulmonary bypass nor a thoracotomy is needed. None of the existing surgical ablation technologies (even bipolar RF energy) can guarantee complete transmurality. We solved this issue by applying RF energy endocardially in the case of incomplete lesions. Another shortcoming of the surgical approach is the inability to precisely locate AF triggers and to map AT or re-entry arrhythmias known to occur during AF ablation (13). In the setting of a hybrid procedure, however, it is possible to perform extensive mapping to tailor the lesion set. Finally, during an epicardial surgical AF ablation on the beating heart, it is technically not possible to create a linear lesion across the CTI. In our series, this was safely performed endocardially in 3 patients.

In 2009, Lockwood et al. (7) described techniques for assessing conduction block across surgical lesions on the basis of epicardial mapping. Using combinations of a focal and bipolar RF device, as in our series, they achieved complete block across linear lesions by the first set of RF applications in only 21%. Several factors, such as epicardial fat and local myocardial thickness, limit the depth of penetration of RF energy and thus the creation of transmural lesions (14). After localization of the gaps, epicardial ablation was repeated until complete bidirectional block across all the linear lesions was achieved.

In our series, pacing maneuvers and mapping techniques were performed from the endocardial side. In 23% of our patients, we were not able to create complete box lesions, even after repeating epicardial ablation. To create completely transmural lesions, we applied unipolar RF energy endocardially at the remaining gaps.

Krul et al. (15) described a series of 31 patients with AF treated with thoracoscopic PV isolation and ganglionated plexi ablation. In patients with nonparoxysmal AF, left atrial ablation lines were created, and conduction block was verified epicardially with custom-made catheters. After 1 year, they reported comparable results to ours (86% of patients had no recurrence of SVT off AADs). Although the autonomic nervous system seems to play an important role in AF, there is until now too little evidence to advocate a systematic ganglionated plexi ablation strategy (16- 17). In animal models, return of ganglionated plexi activity was observed 4 weeks after selective ablation, and little is known about the electrophysiological impact reinnervation will have on long-term outcomes (18). An important conceptual difference between studies is that Krul et al. (15) could only perform epicardial lesions, without the possibility of adding endocardial lesions including endocardial touch-ups to improve transmurality, as well as performing CTI and left-sided isthmus ablation. In addition, they could check the completeness of ablation lesions only from the epicardium, which may be insufficient to show complete electrical block.

The small number of patients in this single-center retrospective study prevents definitive conclusions, also in view of the limited power of the log-rank test we used to compare groups. However, future larger studies may corroborate our results.

Although patients with previous CA had at most a PV isolation procedure, and none of the PVs appeared to be isolated in any of these patients, it cannot be excluded that previous CA favorably influenced the results.

The single-procedure success rate was similar, rather than lower, in patients with persistent compared with those with paroxysmal AF. Although this may have been due to the small sample size, it may also be considered an expression of the robustness of the hybrid procedure.

In the present series, the overall clinical success may have not recognized some episodes of asymptomatic arrhythmia by the relatively short periods of ambulatory monitoring performed.

The safety of this procedure may be a concern because of the extent of ablation and the full heparinization during the procedure.

The longer-term impact of this ablation strategy on the atrial systolic function remains unknown.

An ablation strategy based on noninducibility could lead to overtreatment of some patients. As demonstrated by Jaïs et al. (8), despite being still inducible after PV isolation and deployment of linear lesions, some patients remained arrhythmia free without AADs.

A combined transvenous endocardial and thoracoscopic epicardial ablation procedure for paroxysmal and recent persistent AF resistant to AADs has a single-procedure success rate of 83% at 1 year. Recurrent arrhythmias can be handled with AADs or CA.