Chapter 58. Surgery for Atrial Fibrillation

Atrial fibrillation (AF) is the most common arrhythmia in the world. It is associated with significant morbidity and mortality secondary to its detrimental sequelae: (1) palpitations resulting in patient discomfort and anxiety; (2) loss of atrioventricular (AV) synchrony, which can compromise cardiac hemodynamics, resulting in various degrees of ventricular dysfunction or congestive heart failure; (3) stasis of blood flow in the left atrium, increasing the risk of thromboembolism and stroke.1–10

Medical treatment of AF has many shortcomings. Because of this, interest in nonpharmacologic treatment approaches led to the development of catheter-based and surgical techniques beginning in the 1980s. Initial attempts aimed at providing rate control failed to address the detrimental hemodynamic and thromboembolic sequelae of atrial fibrillation. The early attempts at finding a surgical treatment culminated in the introduction of the Maze procedure in 1987, which became the gold standard for many years.

The following sections describe the historical aspects of surgery for AF, and the current state of surgical ablation for the treatment of AF, including recent minimally invasive techniques.

The Left Atrial Isolation Procedure

The first surgery designed specifically to eliminate AF, the left atrial isolation procedure, was described in 1980 in the laboratory of Dr. James Cox at Duke University. This approach confined AF to the left atrium, and restored the remainder of the heart to sinus rhythm (Fig. 58-1).11 This procedure reestablished a regular ventricular rate without requiring a permanent pacemaker. Isolating the left atrium allowed the right atrium and the right ventricle to contract in synchrony, providing a normal right-sided cardiac output. This effectively restored normal cardiac hemodynamics.

By confining AF to the left atrium, the left atrial isolation procedure only eliminated two of the three detrimental sequelae of AF: an irregular heartbeat and compromised cardiac hemodynamics. It did not eliminate the thromboembolic risk because the left atrium usually remained in fibrillation. This procedure never achieved clinical acceptance, although it was performed by Dr. Cox in a single patient.

Catheter Ablation of the Atrioventricular Node–His Bundle Complex

In 1982, Scheinman and coworkers introduced catheter fulguration of the His bundle, a procedure that controlled the irregular cardiac rhythm associated with AF and other refractory supraventricular arrhythmias.12 This procedure electrically isolated the fibrillation to the atria. Unfortunately, ablating the bundle of His required permanent ventricular pacemaker implantation to restore a normal ventricular rate.

The shortcoming of this intervention was that it only eliminated the irregular heartbeat. Both atria remained in fibrillation, and the risk of thromboembolism persisted. AV contraction remained desynchronized, compromising cardiac hemodynamics. In addition, patients become pacemaker dependant for the remainder of their lives. Nevertheless, AV node ablation has remained a common treatment for medically refractory AF.

The Corridor Procedure

In 1985, Guiraudon and associates developed the corridor procedure for the treatment of AF.13 This operation isolated a strip of atrial septum harboring both the sinoatrial (SA) node and the AV node, allowing the SA node to drive both the ventricles. This procedure effectively eliminated the irregular heartbeat associated with AF, but both atria either remained in fibrillation or developed their own asynchronous intrinsic rhythm because they were isolated from the septal "corridor." Furthermore, the atria were isolated from their respective ventricles, thereby precluding the possibility of AV synchrony. The corridor procedure was abandoned because it had no effect on the hemodynamic compromise or the risk of thromboembolism associated with AF.

The Atrial Transection Procedure

In 1985, Dr. James Cox and associates described the first procedure that attempted to terminate AF.14 This was different from the prior surgical procedures described that only isolated or confined AF to a certain region of the atria. Using a canine model, Cox's group found that a single long incision around both atria and down into the septum could terminate AF. This atrial transection procedure prevented the induction and maintenance of AF or atrial flutter in every canine that underwent the operation.15 Although this procedure was not effective clinically and was soon abandoned, it laid the ground work of the development of the Cox-Maze procedure.

The Maze procedure was clinically introduced in 1987 by Dr. Cox after extensive animal investigation at Washington University in St. Louis.15–17 The Cox-Maze procedure was originally developed to interrupt any and all macro-reentrant circuits that were felt to develop in the atria, thereby precluding the ability of the atrium to flutter or fibrillate (Fig. 58-2). Unlike the previous procedures, the Cox-Maze procedure successfully restored both AV synchrony and sinus rhythm, thus potentially reducing the risk of thromboembolism and stroke.18 The operation consisted of creating an array of surgical incisions across both the right and left atria. These incisions were placed so that the sinoatrial node could still direct the propagation of the sinus impulse throughout both atria. It allowed for most of the atrial myocardium to be activated, resulting in preservation of atrial transport function in most patients.19

The first iteration of the Maze procedure was modified because of problems with late chronotropic incompetence and a high incidence of pacemaker implantation. The resulting Maze II procedure, however, was extremely difficult to perform technically, and it was soon replaced by the Maze III procedure (Fig. 58-3).20,21

The Cox-Maze III procedure—often referred to as the "cut-and-sew" Maze—became the gold standard for the surgical treatment of AF. In a long-term study of patients who underwent the Cox-Maze III procedure at our institution, 97% of the patients at late follow-up were free of symptomatic AF.22 These excellent results have been reproduced by other groups worldwide.23–25

Although the Cox-Maze III procedure was effective in eliminating AF, it was technically difficult and invasive, and only a handful of cardiac surgeons still perform the cut-and-sew operation today. At the present time, a variety of ablation devices are available that can replicate the surgical incisions on the atrium and create conduction block. This has made AF ablation much simpler to perform. These ablation-assisted procedures have greatly expanded the field of AF surgery in the past decade.26 With present ablation technology, surgery can be performed with low morbidity and mortality and often with less invasive incisions.

The development of surgical ablation technology has transformed a difficult and time-consuming operation that few surgeons were willing to perform into a procedure that is technically easier, shorter, and less invasive. Several ablation technologies exist, each with its relative advantages and disadvantages.

For an ablation technology to successfully replace surgical incisions, it must meet several criteria. First, it must reliably produce bidirectional conduction block across the line of ablation. This is the mechanism by which incisions prevent AF, by either blocking macro-reentrant or micro-reentrant circuits or isolating focal triggers. Our laboratory and others have shown that this requires a transmural lesion, as even small gaps in ablation lines can conduct both sinus and fibrillatory impulses.27–29 Second, the ablation device must be safe. This requires a precise definition of dose–response curves to limit excessive or inadequate ablation, and potential hazards to surrounding vital cardiac structures, such as the coronary sinus, coronary arteries, and valvular structures. Third, the ablation device should make AF surgery simpler and require less time to perform. This would require the device to create lesions rapidly, be simple to use, and have adequate length and flexibility. Finally, the device should be adaptable to a minimally invasive approach. This would include the ability to insert the device through minimal access incisions or ports. As of the present time, no device has met all of these criteria. The following sections will briefly summarize the current ablation technologies.

Cryoablation

Cryoablation technology is unique in that it destroys myocardial tissue by freezing rather than heating. It has the benefit of preserving the myocardial fibrous skeleton and collagen structure and is thus one of the safest energy sources available. These devices work by pumping a refrigerant to the electrode tip where it undergoes transformation from a liquid to a gas phase, absorbing heat energy from the tissue in contact with the tip. The formation of intracellular and extracellular ice crystals disrupts the cell membrane and causes cell death. There is also evidence that the induction of apoptosis plays a role in late lesion expansion. Lesion size depends on the temperature of the probe and thermal conductivity and temperature of the tissue.30

There are currently two commercially available sources of cryothermal energy that are being used in cardiac surgery. The older technology, based on nitrous oxide, is manufactured by AtriCure (Cincinnati, OH). More recently, ATS Medical (Minneapolis, MN) has developed a device based on argon. At 1 atmosphere of pressure, nitrous oxide is capable of cooling tissue to −89.5°C, whereas argon has a minimum temperature of −185.7°C. The nitrous oxide technology has a well-defined efficacy and safety profile and is generally safe except around the coronary arteries.31,32 The potential disadvantage of cryoablation, however, is its relatively long time required to create lesions (1 to 3 minutes). There is also difficulty in creating lesions on the beating heart because of the "heat sink" of the circulating blood volume.33 Furthermore, if blood is frozen during epicardial ablation on the beating heart, it may coagulate, creating a potential thromboembolic risk.

Radiofrequency Energy

Radiofrequency (RF) energy has been used for cardiac ablation for many years in the electrophysiology laboratory, and was one of the first energy sources to be applied in the operating room.34 Resistive RF energy can be delivered by either unipolar or bipolar electrodes, and the electrodes can be either dry or irrigated. With unipolar RF devices, the energy is dispersed between the electrode tip and an indifferent electrode, usually the grounding pad applied to the patient. In bipolar RF devices, alternating current is passed between two closely approximated electrodes. The lesion size depends on electrode–tissue contact area, the interface temperature, the current and voltage (power), and the duration of delivery. The depth of the lesion can be limited by char formation, epicardial fat, myocardial and endocavity blood flow, and tissue thickness.

There have been numerous unipolar RF devices developed for ablation. These include both dry and irrigated devices and devices that incorporate suction. Although dry unipolar RF has been shown to create transmural lesions on the arrested heart in animals with sufficiently long ablation times, it has not been consistently successful in humans. After 2-minute endocardial ablations during mitral valve surgery, only 20% of the in vivo lesions were transmural.35 Epicardial ablation on the beating heart has been even more problematic. Animal studies have consistently shown that unipolar RF is incapable of creating epicardial transmural lesions on the beating heart.36,37 Epicardial RF ablation in humans resulted in only 10% of the lesions being transmural.38 This deficiency of unipolar RF has been felt to be caused by the heat sink of the circulating blood.39 This has led industry to examine adding both irrigation and suction to improve lesion formation. Although these additions have improved depth of penetration, there has been no unipolar device that has been shown by independent laboratories to be capable of creating reliable transmural lesions on the beating heart.

To overcome this problem, bipolar RF clamps were developed. With bipolar RF, the electrodes are embedded in the jaws of a clamp to focus the delivery of energy. By shielding the electrodes from the circulating blood pool, this improves and shortens lesion formation and limits collateral injury. Bipolar ablation has been shown to be capable of creating transmural lesions on the beating heart both in animals and humans with short ablation times.40–42 Three companies (AtriCure, West Chester, OH; Medtronic, Minneapolis, MN; and Estech, San Ramon, CA) currently market bipolar RF devices.

Another advantage of bipolar RF energy over unipolar RF is its safety profile. A number of clinical complications of unipolar RF devices have been reported, including coronary artery injuries, cerebrovascular accidents, and esophageal perforation leading to atrioesophageal fistula.43–46 Bipolar RF technology has virtually eliminated this collateral damage; there have been no injuries described with these devices despite extensive clinical use.

High-Intensity Focused Ultrasound

High-intensity focused ultrasound, or HIFU, is another modality being applied clinically for surgical ablation (St. Jude Medical, St. Paul, MN). Ultrasound ablates tissue via mechanical hyperthermia. Ultrasound waves travel through the tissue, causing compression, refraction, and particle movement, which is translated into kinetic energy and ultimately creates thermal coagulative tissue necrosis. HIFU produces high-concentration energy in a focused area, and is reportedly able to create transmural epicardial lesions through epicardial fat in a short time.47

HIFU is unique in that it is able to create noninvasive, noncontact focal ablation in three-dimensional volume without affecting intervening and surrounding tissue. It uses ultrasound beams in the frequency range of 1 to 5 MHz or higher. The focused beams increase the temperature of the targeted tissue to above 80°C, effectively killing the cells. By focusing ultrasound waves, HIFU is able to create targeted thermal coagulation of tissue at a defined distance from the probe without heating intervening tissue. Its ability to focus the target of ablation at specific depths is a potential advantage over other energy modalities.

Another advantage of HIFU technology is its mechanism of thermal ablation. Unlike all other energy sources that heat or cool tissue by thermal conduction, HIFU ablates tissue by directly heating the tissue in the acoustic focal volume, and is therefore less affected by the "heat sink" of the circulating endocardial blood pool. A few clinical studies using HIFU have shown some encouraging results.47–50 However, there has been no independent experimental verification of the efficacy of HIFU devices to reliably create transmural lesions, and some limited clinical experience has had poor results.51 The fixed depth of penetration of these devices may be a problem in pathologically thickened atrial tissue. Moreover, these devices are somewhat bulky and expensive to manufacture.

Both microwave and laser energy have been used clinically, but limitations of each technology led to limited use and/or withdrawal of those devices from the market.39

In summary, each ablation technology has its own advantages and disadvantages. Continued research investigating the effects of each surgical ablation technology on atrial hemodynamics, function, and electrophysiology will allow for more appropriate use in the operating room. The inability to create reliable linear lesions on the beating heart remains a shortcoming of most devices and has impeded the development of minimally invasive procedures.

There are essentially three categories of procedures that presently are performed to surgically treat AF: the Cox-Maze procedure, left atrial lesion sets, and pulmonary vein isolation (PVI). Each of these approaches is described in the following.

The Cox-Maze IV Procedure

The original "cut-and-sew" Cox-Maze III procedure is rarely performed today. At most centers, the surgical incisions have been replaced with lines of ablation using a variety of energy sources. At our institution, bipolar RF energy has been used successfully to replace most of the surgical incisions of the Cox-Maze III procedure. Our current RF ablation-assisted procedure, termed the Cox-Maze IV, incorporates the lesions of the Cox-Maze III (Fig. 58-4).52 Our clinical results have shown that this modified procedure has significantly shortened operative time while maintaining the high success rate of the original Cox-Maze III procedure.53,54

The Cox-Maze IV procedure is performed on cardiopulmonary bypass. The operation can be done either through a median sternotomy or a less invasive right minithoracotomy. The right and left pulmonary veins (PVs) are bluntly dissected. If the patient is in AF, amiodarone is administered and the patient is electrically cardioverted. Pacing thresholds are obtained from each pulmonary vein. Using a bipolar RF ablation device, the PVs are individually isolated by ablating a cuff of atrial tissue surrounding the right and left PVs. Proof of electrical isolation is confirmed after ablation by demonstrating exit block from each PV.

The right atrial lesion set is performed on the beating heart (Fig. 58-5). A bipolar RF clamp is used to create most of the lesions. A unipolar device, either cryoablation or radiofrequency energy, is used to complete the ablation lines endocardially down to the tricuspid annulus because of the difficulty of clamping in this area.

The left-sided lesion set (Fig. 58-6) is performed via a standard left atriotomy on the arrested heart. The incision is extended inferiorly around the right inferior PV and superiorly onto the dome of the left atrium. Connecting lesions are made with the bipolar RF device into the left superior and inferior pulmonary veins and a final ablation is performed down toward the mitral annulus. Our group has shown that isolating the entire posterior left atrium with ablation lines into both left pulmonary veins resulted in a better drug-free freedom from AF at 6 and 12 months than a single connection lesion.55 A unipolar device, usually cryoablation, is used to connect the lesion to the mitral annulus and complete the left atrial isthmus line. The left atrial appendage is amputated, and a final ablation is performed through the amputated left atrial appendage into the one of the left pulmonary veins.

Left Atrial Procedures

Over the past decade, there have been a number of new surgical procedures introduced in attempt to cure AF. The results have been variable and have had a wide range of success rates.43,56–63 All of the ablation technologies have been used to create a large number of different lesion patterns. All of these procedures have generally involved some subset of the left atrial lesion set of the Cox-Maze procedure. Results have been dependent on the technology used, the lesion set, and the patient population. From a technical standpoint, all of the approaches have attempted to isolate the pulmonary veins. The importance of the rest of the left atrial Cox-Maze lesion set remains controversial. However, Gillinov et al published a large series demonstrating that the omission of the left atrial isthmus lesion resulted in a significantly higher incidence of recurrent AF in patients with permanent AF.64 To complete this lesion, it is mandatory to also ablate the coronary sinus in line with the endocardial lesion. In addition, our clinical results have shown that it is important to isolate the entire posterior left atrium.55

Pulmonary Vein Isolation

Pulmonary vein isolation is an attractive therapeutic strategy because the procedure can be done without cardiopulmonary bypass and with minimally invasive techniques, using either a thoracoscopic approach or small incisions. It can also be easily added to another cardiac surgery intervention (eg, coronary bypass graft or a valve procedure) with minimum increase of time to the case. Based on the original report of Hassaiguerre, it has been well documented that the triggers for paroxysmal AF originate from the pulmonary veins in the majority of cases.65 However, it is important to remember that up to 30% of triggers may originate outside the pulmonary veins.66 To increase efficacy, some investigators have added ablation of the ganglionic plexi (GP).67–69

The pulmonary veins can be isolated separately or as a box (Fig. 58-7). The most common approach for treatment of lone AF uses an endoscopic, port-based approach to minimize incision size and pain for the patient. At our center, bipolar RF clamps are favored to isolate the pulmonary veins, but unipolar radiofrequency, cryoablation, and HIFU devices have also been used.49,70,71

Patient preparation begins with double-lumen endotracheal intubation. A transesophageal echocardiogram is performed to confirm the absence of thrombus in the left atrial appendage. If a thrombus is found, the procedure is either aborted or converted to an open procedure, in which the risk of systemic thromboembolism from left atrial clot can be minimized. External defibrillator pads are placed on the patient and the patient is positioned with the right side turned upward 45 to 60 degrees and the right arm positioned above the head to expose the right axilla.

An initial port for the thoracoscopic camera is placed at the sixth intercostal space. Under thoracoscopic vision, a small working port can then be placed in either the third or fourth intercostal space at the midaxillary line depending on surgeon preference and patient anatomy. The right phrenic nerve is identified to avoid injury to this structure. An incision is made in the pericardium, anterior and parallel to the phrenic nerve, to expose the heart from the superior vena cava to the diaphragm. Through this pericardiotomy, the space above and below the right pulmonary veins is dissected to allow enough room for insertion of a specialized thoracoscopic dissector. This includes opening into the oblique sinus and dissecting the space between the right superior PV and the right pulmonary artery. The dissector and a guiding sheath is introduced through a second port, either lateral or medial to the scope port, and guided into the space between the right superior pulmonary vein and right pulmonary artery. After the dissector is carefully removed from the chest, the sheath remains in place as a guide for the insertion of the bipolar RF clamp. At this point, the patient is cardioverted into sinus rhythm so that pacing thresholds can be obtained. As with a Cox-Maze IV procedure, it is critical to document pacing thresholds from the pulmonary veins before isolation. Some surgeons also use the opportunity provided by surgical exposure to test and ablate ganglionated plexi.

The sheath is attached to the lower jaw of a bipolar RF clamp. The clamp is introduced into the chest, and the left atrium surrounding the pulmonary veins is clamped and ablated. After confirmation of exit block by pacing, the instruments are removed from the right chest and the right chest ports are closed.

The approach to the left chest is similar to the right. The patient is repositioned such that the left chest is elevated 45 to 60 degrees and the left arm is held up to expose the left axilla. A port for the thoracoscopic camera is placed in the sixth intercostal space, slightly more posterior than on the right side. With thoracoscopic visualization, a left-sided working port is created in the third or fourth intercostal space (Fig. 58-8). The left phrenic nerve is identified, and the pericardium is opened posterior to the course of the nerve. The ligament of Marshall is identified and divided. The dissector is then introduced through a second port, also in the sixth intercostal space. This port site is placed to allow for a straight line introduction of the dissector around the pulmonary veins. The guiding sheath is used to position the RF clamp around the left pulmonary veins, and they are isolated. Again, conduction block is confirmed with pacing.

The procedure is not complete until the left atrial appendage has been addressed. Traditionally, this has been done by stapling across the base of the left atrial appendage with an endoscopic stapler. This requires careful surgical technique and attention because it can result in tears and bleeding.72 There are clip devices in development designed to address this difficulty.73,74 Further investigation is needed to determine their efficacy and safety for this purpose. After ablation of the left pulmonary veins and exclusion of the left atrial appendage, the left side of the pericardium is closed.

The Cox-Maze Procedure

The Cox-Maze III procedure has had excellent long-term results. In our series at Washington University, 97% of 198 consecutive patients who underwent the procedure with a mean follow-up of 5.4 years were free from symptomatic AF. There was no difference in the cure rates between patients undergoing a stand-alone Cox-Maze procedure and those undergoing concomitant procedures.22 Similar results have been obtained from other institutions around the world with the traditional "cut-and-sew" method.23,25,75

Our results from patients who underwent the Cox-Maze IV procedure have been encouraging as well. In a prospective, single-center trial from our institution, 91% of patients at the 6-month follow-up were free from AF.42,54 The Cox-Maze IV procedure has significantly shortened the mean cross-clamp times for a lone Cox-Maze from 93 ± 34 minutes for the Cox-Maze III to 47 ± 26 minutes for the Cox-Maze IV (p < .001), and from 122 ± 37 minutes for a concomitant Cox-Maze III procedure to 92 ± 37 minutes (p < .005) in those undergoing the Cox-Maze IV procedure concomitantly with another cardiac operation.42 A propensity analysis performed by our group has shown that there was no significant difference in the freedom from AF at 3, 6, or 12 months between the Cox-Maze III and IV groups.54

The Cox-Maze IV procedure also worked as well in patients with paroxysmal as for longstanding persistent AF. The 6-month freedom for AF was 91% in the paroxysmal group compared with 88% in the persistent group (p = .53). The drug-free success rate was also similar.76 Our present series consists of 263 consecutive patients undergoing a Cox-Maze IV between January 2002 and June 2009. All patients have been followed with ECGs or prolonged monitoring. Our 12-month freedom from AF was 93%, with 78% of patients also off all antiarrhythmic drugs. We have had no intraoperative mortalities in 90 consecutive patients with lone atrial fibrillation undergoing this procedure.

Left Atrial Lesion Sets

A number of centers around the world have suggested performing ablation confined to the left atrium only to cure AF. This concept is supported by the fact that the majority of paroxysmal AF appears to originate around the PVs and the posterior left atrium. A left atrial lesion set typically involves pulmonary vein isolation with a lesion to the mitral annulus as well as removal of the left atrial appendage. Many ablation technologies have been used to create these lesion sets with varied degrees of success.43,56–63

There have been no randomized trials of biatrial versus left atrial ablation only in the surgical population. Because of this, the importance of the right atrial lesions of the traditional Cox-Maze procedure is difficult to determine. A meta-analysis of the published literature by Barnett and Ad revealed that a biatrial lesion set resulted in a significantly higher late freedom from AF when compared with a left atrial lesion set alone (87% versus 73%, p = .05).77 This is not surprising considering the results of intraoperative mapping of patients with atrial fibrillation. Both our group and others have shown that AF can originate from the right atrium in 10 to 50% of cases.78–80

Of the specific left atrial lesions of the Cox-Maze procedure, it is difficult to determine the precise importance of each particular ablation. All surgeons agree on the importance of isolating the pulmonary veins. As stated previously, work from Gillinov et al has shown the importance of the left atrial isthmus in a retrospective study.64 In a rare randomized trial, Gaita and coauthors examined pulmonary vein isolation alone versus two alternate lesion sets that both included ablation of the left atrial isthmus. In this study, normal sinus rhythm at 2-year follow-up was only seen in 20% in the PVI group versus 57% in the other groups (p < .006).62 Finally, our group has shown that isolating the entire posterior left atrium significantly improved our drug-free freedom from AF at 6 months (54% versus 79%, p = .011) in a retrospective analysis of our results. Thus, most of the left atrial Cox-Maze lesion set is likely needed to ensure a high success rate. However, there has been no randomized trial to conclusively demonstrate the correct left atrial lesion set.

Pulmonary Vein Isolation

The results of PVI alone have been variable and have been dependent on patient selection. In the first report of surgical PVI, Wolf and colleagues reported that 91% of patients undergoing a video-assisted bilateral PVI and left atrial appendage exclusion were free from AF at three months follow-up.81 Edgerton et al reported on 57 patients undergoing PVI with GP ablation with more thorough follow-up and found 82% of their patients with paroxysmal AF to be free from AF at 6 months, with 74% off antiarrhythmic drugs.82 Subsequent studies have shown encouraging results in patients with paroxysmal AF. In a study involving 21 patients with paroxysmal AF undergoing PVI with GP ablation, McClelland et al reported 88% freedom from AF at 1 year without antiarrhythmic drugs.68 A larger, single-center trial recently reported a 65% single procedure success at 1 year in a series of 45 patients undergoing PVI with GP ablation, including patients with persistent and paroxysmal AF.83 A multicenter trial reported 87% normal sinus rhythm rate in a more diverse patient population, including some patients with longstanding persistent AF; however, those patients with longstanding persistent AF only had a 71% incidence of normal sinus rhythm.84

In patients with longstanding or persistent AF, the results have been worse. In a study from Edgerton and his group, only 56% of patients were free from AF at 6 months follow-up (35% off antiarrhythmics).85 With concomitant procedures, the success rate of PVI is even lower. Of 23 patients undergoing mitral valve surgery or coronary revascularization with concomitant PVI, only half of patients were free from AF at a follow-up of 57 ± 37 months.86 In the setting of mitral valve disease, Tada and colleagues report 61% freedom from AF and an only 17% freedom from antiarrhythmic drugs in their series of 66 patients undergoing PVI.87

Ganglionated Plexus Ablation

There have been convincing experimental data demonstrating that autonomic ganglia in ganglionated plexi (GP) play a role in the initiation and maintenance of AF.88,89 As a result, some surgeons have added GP ablation to pulmonary vein isolation in hopes to increase procedural efficacy. Some of the initial surgical results have been encouraging. However, there have been no randomized trials demonstrating the efficacy of adding GP to PVI. In 2005, Scherlag and colleagues reported a study of GP ablation combined with catheter PVI in 74 patients with lone AF. After a relatively short median follow-up of 5 months, 91% of patients were free from AF.89

However, the effects of vagal denervation are not clearly defined. Experimental evidence in our laboratory and others have demonstrated recovery of autonomic function as early as 4 weeks after GP ablation.90–92 It is worrisome that the reinnervation may not be homogenous and could result in a more arrhythmogenic substrate. In a more recent report, Katritsis and colleagues used left atrial GP ablation alone to treat 19 patients with paroxysmal AF. Fourteen of these patients had recurrent AF during 1-year follow-up.93 Because of these suboptimal results and the lack of any long-term follow-up regarding the effects of GP ablation, our practice is not to perform GP ablation to treat AF. GP ablation should be reserved for centers participating in clinical trials.

The indications for surgery have been defined in a recent consensus statement.94 These presently include: (1) All symptomatic patients with documented atrial fibrillation undergoing other cardiac surgical procedures; (2) selected asymptomatic patients with atrial fibrillation undergoing cardiac surgery in which the ablation can be performed with minimal risk in experienced centers; and (3) stand-alone atrial fibrillation ablation should be considered for symptomatic patients with atrial fibrillation who either prefer a surgical approach, have failed one or more attempts at catheter ablation, or are not candidates for catheter ablation.

There is still controversy regarding the referral of patients for surgery with medically refractory, symptomatic atrial fibrillation in lieu of catheter ablation. No randomized controlled trials exist that compare outcomes with catheter versus surgical ablation. In these instances, clinical decisions should be made based on the individual institution's experience with catheter and surgical ablation, the relative outcomes and risks of each in the individual patient, and patient preference. Programs involved in the surgical treatment of atrial fibrillation should develop a team approach to these patients, including both electrophysiologists and surgeons, to ensure appropriate selection of patients.

There are relative indications for surgery that were not included in the consensus statement. The first is a contraindication to long-term anticoagulation in patients with persistent AF and a high risk for stroke (CHADS score ≥2). Up to one-third of patients with AF screened for participation in clinical trials of warfarin were deemed ineligible for chronic anticoagulation, mainly because of a high perceived risk for bleeding complications.95–97 In one study, the annual rate of intracranial hemorrhage in anticoagulated patients with AF was 0.9% per year, and the overall major bleeding complication rate was 2.3% per year.96 The stroke rate following the Cox-Maze procedure off anticoagulation has been remarkably low, even in high-risk patients. At our institution, only 5 of 450 patients had a stroke after a mean follow-up of 6.9 ± 5.1 years. There was no difference in stroke rate in those patients with CHADS scores above or below 2.98 This low risk of stroke after the Cox-Maze procedure has been noted in other series.18,99,100 The decrease in stroke risk is likely due to the high cure rate, as well as the resection of the left atrial appendage.

Surgical treatment for AF with amputation of the left atrial appendage also should be considered in patients with chronic AF who have suffered a cerebrovascular accident despite adequate anticoagulation, as these patients are at high risk for repeat neurologic events. Anticoagulation with warfarin reduces the risk of ischemic and hemorrhagic strokes by more than 60% in patients with AF, but does not completely eliminate this serious complication.9,101 At our institution, 20% of patients who underwent the Cox-Maze III procedure had experienced at least one episode of cerebral thromboembolism that resulted in a significant temporary or permanent neurologic deficit before undergoing the operation.22 There were no late strokes in this population, even with 90% of patients off anticoagulation at last follow-up.22

In patients undergoing concomitant valve surgery, studies have shown that adding the Cox-Maze procedure can decrease the late risk of cardiac- and stroke-related deaths.102,103 However, there have been no prospective, randomized studies demonstrating the survival or other benefits of adding a Cox-Maze procedure in patients with AF. A contraindication to a Cox-Maze procedure in the concomitant surgery group would be in asymptomatic patients who have tolerated their AF well and have not had problems with anticoagulation, in whom adding the ablation would increase their surgical risk.

Although the Cox-Maze procedure achieves high rates of success and recent technological advances have made it easier to perform, it still represents an invasive procedure that requires cardiopulmonary bypass. A simple, minimally invasive operation that does not require cardiopulmonary bypass would be preferable. Such a procedure should preserve normal atrial physiology, and have minimal morbidity and a high success rate. Achieving this goal will require progress in understanding the mechanism of AF in individual patients. The surgical approach may need to be redesigned based on a better understanding of the mechanisms and the effects of surgical ablation on atrial electrophysiology.

It is now known that there are multiple different possible mechanisms of AF and that multipoint mapping is necessary to describe this complex arrhythmia.79-81,104,105 Epicardial activation sequence mapping has been the traditional gold standard for mapping of AF, but is both invasive and time consuming, limiting its clinical use.17 A newer noninvasive technique, electrocardiographic imaging (ECGI), offers a potentially useful way to describe the atrial activation sequence and determine mechanistic information from conscious patients.106 The technique uses body surface potentials, measured by surface electrodes, and anatomical data obtained through computed tomographic scanning to indirectly calculate the surface potentials on the surface of the heart. This technique has been shown to work well for normal sinus rhythm and atrial flutter as well as ventricular arrhythmias.107–109 Currently our group is testing the technique in patients with AF, in collaboration with Yoram Rudy at Washington University, the developer of ECGI. The initial results are promising.107,110

The information acquired from ECGI can be analyzed to determine activation sequence and frequency maps for individual patients noninvasively. Using these data, a strategy for designing patient-specific optimal lesion sets is being developed, taking into account the patient's atrial geometry, conduction velocity, and refractory period.111 Previous work has demonstrated that a critical mass is necessary to maintain AF, and initial lesions could be determined by a calculation of the critical area needed to maintain AF in the individual patient.112 This could be done using mechanistic information derived from activation data and anatomical data from a CT scan. It will also allow identification of focal sources that can then be either isolated or ablated in the electrophysiologic laboratory or operating room.

In instances in which a specific mechanism of AF cannot be defined, the goal will be to create a lesion pattern that will make the atria unable to fibrillate. Failures of the Cox-Maze III and IV procedures have been seen in patients with increasing left atrial size or longstanding AF.99,112,113 A recent study performed by our laboratory on a canine model found that the probability of maintaining AF is correlated with increasing atrial tissue areas, widths, and weights, as well as the length of the effective refractory period and the conduction velocity of the tissue.111 All of these data can be acquired noninvasively through ECGI. Using these data may allow surgeons to design custom operations for each patient based on the mechanism of their arrhythmia and their specific atrial anatomy or electrophysiology.

Finally, the limitations of present ablation devices have impeded the development of a truly minimally invasive procedure. Unfortunately, the creation of reliable transmural lines of ablation on the beating heart has been difficult. This has been because of the heat sink of the circulating endocardial blood pool.39 Future advances may be anticipated with devices that overcome this limitation or by the introduction of hybrid procedures in which surgeons and electrophysiologists work together to complete lines of block.

In conclusion, the development of ablation technologies has dramatically changed the field of AF surgery. The replacement of the surgical incisions with linear lines of ablation has transformed a complex, technically demanding procedure into one accessible to the majority of surgeons. More importantly, these new ablation technologies have introduced the possibility of minimally invasive surgery for AF, prompting numerous efforts to develop simpler procedures that can be preformed epicardially on the beating heart. There is already strong evidence that PVI may be effective in a subset of patients with paroxysmal AF. With extended lesions sets, it may be possible to extend the efficacy of minimally invasive procedures to patients with persistent and longstanding AF. However, surgeons must remember that the Cox-Maze procedure has good efficacy in these patients and can be performed using a small thoracotomy with acceptable success and low morbidity.100 It is imperative for surgeons trying new procedures to carefully follow their results and publish them in peer-reviewed journals. For surgeons performing AF ablation, it is mandatory to adhere to the recently published guidelines for follow-up of patients and for determining success or failure after these procedures. As we learn more about the mechanisms of AF and develop improved preoperative diagnostic technologies capable of precisely locating the areas responsible for AF, it will become possible to tailor specific lesion sets and ablation modalities to individual patients, making the surgical treatment of AF more effective and available to a larger population of patients.