Initial experience with the maze procedure for atrial fibrillation

Surgery for Acquired Heart Disease

Initial experience with the maze procedure for atrial fibrilla tion From January 1991 until May 1992, a total of 14 patients (mean age 48 years) underwent the maze procedure for refractory atrial fibrillation (mean duration, 7 years; mean number of antiarrhythmic medications, six). Three patients had had embolic events, one patient had had a cardiac arrest from flecainide, one had pulmonary fibrosis from amiodarone, and six of ten who were employed were temporarily disabled. Two patients underwent successful mitral valve repair in which the maze procedure was added as a secondary goal of the operation. Postoperative fluid retention was a problem in five patients (36 %). Six patients (43 %) were temporarily treated with an antiarrhythmic medication. Two patients (14 %) with preoperative sick sinus syndrome required pacemakers. One patient was discharged from the hospital but died suddenly less than 1 month after the operation (7 % operative mortality) of hyperkalemia caused by acute renal failure. All patients beyond 3 postoperative months (100 % "cure") are receiving no antiarrhythmic medications, have sinus rhythm, or have p-wave tracking with ventricular pacing. Atrial contraction has been documented by cinegraphic magnetic resonance imaging studies and by Doppler echocardiography performed when sinus rhythm had resumed. The maze procedure is an extensive operation but is indicated for selected patients who have the severe sequelae of atrial fibrillation. (J THORAC CARDIOVASC SURG 1993;105:1077-87)

Patrick M. McCarthy, MD, Lon W. Castle, MD, James D. Maloney, MD, Richard G. Trohman, MD, Tony W. Simmons, MD, Richard D. White, MD, Allan L. Klein, MD, and Delos M. Cosgrove III, MD, Cleveland, Ohio

Atrial fibrillation is a major health care problem. An estimated I million U.S. citizens are in atrial fibrillation and at risk from the effects of the arrhythmia.' Cox and colleagues? have identified three direct consequences of atrial fibrillation: symptomatic tachycardia, impaired hemodynamic function, and the risk of thromboemboli. To these three direct consequences should be added an From The Cleveland Clinic Foundation, Departments of Thoracic and Cardiovascular Surgery, Cardiology, and Diagnostic Radiology, Cleveland, Ohio. Read at the Eighteenth Annual Meeting of The Western Thoracic Surgical Association, Kauai, Hawaii, June 24-27, 1992. Address for reprints: Patrick M. McCarthy, MD, Department of Thoracie and Cardiovascular Surgery, The Cleveland Clinic Foundation, 9500 Euclid Ave., Cleveland, Ohio 44195. Copyright

1993 by Mosby-Year Book, Inc.

0022-5223/93 $1.00

+ .10 12/6/44669

indirect consequence, the risk of treatment of atrial fibrillation with anticoagulant and antiarrhythmic medications. The maze procedure was conceived, tested in the laboratory, and then introduced into clinical practice by Cox and associates.e' The procedure was designed to cure atrial fibrillation.? In theory, the operation is effective through three critical steps. First, the macroreentry circuits responsible for atrial fibrillation- 4 are surgically divided by full-thickness atrial incisions, which eliminate atrial fibrillation. Second, conduction of the wave front from the sinoatrial (SA) node to the atrioventricular (A V) node is preserved.i which restores sinus rhythm and alleviates the tachycardia symptoms. Third, conduction to both atria is also preserved and atrial systole is restored, which improves hemodynamic function and decreases atrial stasis and the risk of thromboemboli. Theoretical-

1077

1078

The Journal of Thoracic and Cardiovascular Surgery June 1993

McCarthy et al.

Table I. Preoperative clinical characteristics of 14 patients undergoing the maze procedure for AF Patient No.

Age (yr)

Sex

66 41 33 37 43 41 42 45 53 62 40 60 53 52

M M M

1* 2 3 4* 5* 6* 7 8 9

lOt 11 *.t 12 13* 14

F M

F M M M M M

F F F

Duration of AF (yr)

Predominant rhythm

10 2 2 3 12 5 3

Chronicatrial flutter Paroxysmal AF /SSS Paroxysmal AF Chronicatrial flutter Paroxysmal AF/SSS Chronic AF Chronic AF Chronic AF Chronic AF Chronic AF Chronic AF Paroxysmal AF Paroxysmal AF Chronic AF

8 I

30 3 2 18

2

No. of unsuccessful antiarrhythmic drugs

10 6 4 9

10 8 I

5 6 4 5 6 8

5

AF. Atrial fibrillation; SSS. sick sinus syndrome.

'Patient receiving amiodarone before the operation. tPatient with previous AV node ablation.

ly, patients return to sinus rhythm and can discontinue antiarrhythmic and anticoagulant medications. The fundamental work of Cox and his collaborators/? is best described in a series of four papers. Cox's maze procedure was first used at the Cleveland Clinic in January 1991. This report describes our initial experience with the maze procedure, our minor modifications of the original operative technique, our results, and preliminary risk-benefit analysis of the operation. Patients and methods Patient population. Fourteen patients were operated on between January 1991 and May 1992. Patient characteristics are given in Table I. The mean age was 48 years and the mean duration of atrial fibrillation was 7 years. By the time of the operation, the rhythm had converted to chronic a trial fibrillation (or atrial flutter) from paroxysmal atrial fibrillation in nine patients. Five patients were still predominantly in paroxysmal atrial fibrillation. The patients had had no relief from a mean of six antiarrhythmic medications. One patient with severe mitral regurgitation was treated with only one drug, digoxin. Six patients were receiving (or had discontinued) arniodarone, and six were receiving long-term anticoagulation with warfarin. All patients had had electrical cardioversion with, at best, temporary return of sinus rhythm. All patients reported severe symptoms related to tachycardia, impaired hemodynamic function, or both. All patients complained of fatigue and dyspnea on exertion, nine had exerciseinduced syncope or near-syncope, and eight complained of severe palpitations. One patient had amaurosis fugax and two others had had strokes. Patient 6 had had two previous strokes and had residual hemiparesis. Patient 10 had had two strokes with a subsequent seizure disorder and had had a renal embolus. The severe symptoms related to atrial fibrillation, or the drugs used to treat it, led to time missed from work for the ten

patients who were employed. Six of these patients became disabled. Drug intolerance or severe side effects were present in all patients except patient 7 (who underwent mitral valve repair, with the maze procedure added as a secondary goal of the operation"). Hyperthyroidism developed in two patients receiving amiodarone and pulmonary fibrosis developed in another. One patient had ventricular tachycardia and cardiac arrest after 48 hours of flecainide therapy. This patient had normal ventricular function and no structural heart disease. Two patients had pacemakers before the operation. Patient II had not responded to five drugs and, at another institution, underwent radiofrequency ablation of the A V node, which was unsuccessful. Further attempts at ethanol injection into the AV node artery were also unsuccessful. Finally, direct-current shock of the AV node led to heart block that was treated with a VVI-R pacemaker, but conduction returned 2 months later and the tachycardia symptoms recurred because of atrial fibrillation with a rapid ventricular response. During these manipulations the patient's right coronary artery became occluded. The patient had a right ventricular infarction and had residual right ventricular dysfunction. Patient 10 had had a successful AV node ablation and a permanent pacemaker implanted in 1989, but he had had three embolic events while receiving warfarin and had symptoms of impaired hemodynamics (fatigue and dyspnea). Most patients had lone atrial fibrillation, but studies revealed preexisting cardiac conditions in three patients. Patient 6 had Yamaguchi's disease;' with concentric left ventricular hypertrophy and apical obstructive cardiomyopathy. She also had a family history of cardiomyopathy. Patient 7 had degenerative mitral valve disease and previous endocarditis. He had a flail segment in the middle scallop of the posterior leaflet with dilated atria. Patient 14 had rheumatic mitral valve disease with mixed mitral stenosis and regurgitation. One patient had an incidental 50% obstruction in the left anterior descending artery, detected during preoperative cardiac catheterization. As noted

The Journal of Thoracic and Cardiovascular Surgery Volume 105, Number 6

McCarthy et al.

1079

Base of left atrial appendage

I

»:

IVC

Fig. 1. Pulmonary vein encircling incision as visualized through an incision across dome of left atrium and atrial septum. Separate incision to mitral valve anulus has been closed, and closure of encircling incision is completed from right side of heart (lower right). IVC, Inferior vena cava. earlier, patient II had a right ventricular infarction during AV node ablation. Informed consent was obtained from al1 patients including discussion of the foreseeable, and possible unforeseen, complications of the procedure. The maze procedure. Cox's maze procedure was performed according to the techniques he 5 has described. The technique is briefly reiterated here. Both venae cavae were completely mobilized. The interatrial groove was dissected, and the pericardial attachments to the dome of the left atrium were divided. Once cardiopulmonary bypass was established, the caval tapes were ensnared, and the heart was fibrillated with alternating current. This was done to avoid potential air embolus in a patient with previously undetected patent foramen ovale (present in patient 12). The right atrial appendage was excised, and incisions were made medially down to the atrial septum and lateral1y across the free wall of the right atrium for approximately 2 em. The aortic crossclamp was then applied, and cardioplegic solution was infused. The dome of the left atrium was opened to the baseofthe left atrial appendage, and the incision was extended through the limbus of the fossa ovalis, across the fossa ovalis, and to the tendon of Todaro. The pulmonary vein encircling incision was commenced by a left atriotomy in the interatrial groove extended cephalad belowthe superior vena cava and caudal1ybelow the inferior vena cava. This incision was extended to the left side of the pulmonary veins and superiorly up to the dome of the left atrium. A separate incision was made from the encircling incision to the mitral valve anulus (Fig. I). A 3 mm cryolesion was placed on the mitral valve anulus for 2 minutes at -60 0 C. The coronary sinus was dissected free and

was treated with a 3-minute cryolesion, The pulmonary vein encircling incision was completed by extending the left atriotomy up to the dome of the left atrium, along the medial aspect of the superior vena cava, The left atrial appendage was completely excised and the dome of the left atrium and the atrial septum were closed. Incisions were closed with 3-0 polypropylene sutures as the operation progressed, with all knots being left outside the atria. The heart was evacuated of air and the aortic crossclamp was removed. The remaining right atrial incisions were completed while the heart was beating, usual1y in a junctional rhythm. An incision was made from the atriotomy up the medial aspect of the superior vena cava. A "posterior counterincision" was made in the far lateral right atrium, and this incision was extended up to the superior vena cava. Along the superior extent of this incision, care was taken to rotate the incision from the lateral right atrium to the posterior right atrium and, subsequently, to the posterior superior vena cava. This rotation maximized the amount of undisturbed tissue around the SA node complex (Fig. 2). This posterior counterincision was extended down to the inferior vena cava and then across to the AV groove. Within the right atrium, the incision was completed to the tricuspid valve anulus. A 3 mm cryolesion as applied at the tricuspid valve anulus for 2 minutes. All right atrial incisions were then closed. A short patch of pericardium was used to close the medial superior vena cava incision (Fig. 3). In two patients the opening of the superior vena cava was thought to be sufficient so that a pericardial patch was not needed. In all patients, two sets of temporary atrial pacing wires were placed-one in the SA node area and the other along the medial right atrium. The maze incisions were modified in patient 12 after discussion with Dr. Cox. In this patient, the right atrial incisions were

1080

McCarthy et al.

The Journal of Thoracic and Cardiovascular Surgery June 1993

Fig. 2. Extensive posterior and lateral mobilization of right atrium-superior vena cava junction allows right atrial counterincision to be rotated far lateral and posterior to SA node.

Fig. 3. Completed incisions for Cox's maze procedure. A short piece of pericardium was usually used to patch the medial superior vena cava incision.

performed before the left atrial incisions; also, exposure into the left atrium was across the dome of the left atrium, after division of the superior vena cava 2 em above the sinus node. In most patients, the new incisions would preserve the artery to the SA node. However, in this patient the artery to the SA node arose from the circumflex coronary artery. It was divided by the usual incision in the dome of the left atrium. The standard maze procedure (used in the other 13 patients) divides the artery to the SA node with the medial right atrial incision that extends down to the atrial septum. The two patients with mitral valve disease posed new problems. For these patients success of the mitral valve repair was paramount; the maze procedure was only to be added if the mitral valve repair was successful and performed rapidly. The initial right atrial incisions were not made until the mitral valve repair was completed. The operation started through a standard left atriotomy in the interatrial groove. This incision was extended far underneath the inferior vena cava. An extension from this incision was made to the mitral valve anulus, where the cryolesion was placed. This incision was closed for approximately I em so that the coronary sinus was still visible. We made the initial incision to the mitral valve anulus, placed the cryolesion, and partially closed this incision because with later mitral valve annuloplasty this area would not be accessible. Patient 7 required a quadrangular resection of the middle scallop of the posterior leaflet. In addition. a piece of bovine pericardium was modified and used to plicate the posterior anulus according to a recent surgical technique of Cosgrove (unpublished data). Patient 14 had a rheumatic mitral valve, and fibrotic material was removed from both leaflets. The com-

missures were opened. The subvalvular apparatus was stripped of fibrotic deposits, and a posterior annuloplasty was performed. In both patients, cold saline solution infused into the left ventricle demonstrated excellent valve competence. The remainder of the maze incisions were then completed. Further modifications of the maze procedure were necessary in these two patients (and in patient 6) because of dilated atria. Theoretically, the standard incisions in dilated atria could leave enough atrial tissue between suture lines to sustain a macroreentry circuit. Therefore strips of atrium were removed to decrease the size of the left atrium and the area of atrial tissue between incisions. A large strip of atrium was removed from the posterior left atrium. The pulmonary vein encircling incisions were placed farther away from the orifice of the pulmonary veins than usual. When the atrial appendages were excised, a large amount of tissue at the base of the appendage was also excised. A strip of left atrium across the dome was removed, and a strip of right atrium in the lateral wall was also removed. Intraoperative transesophageal echocardiograms of patient 7 showed a substantial reduction in the size of the left atrium." A more moderate reduction was evident in patient 14 (Fig. 4). A I cm thrombus was found in patient 10 at the base of the left atrial appendage during the operation. This patient was on a program of warfarin therapy and had had three previous embolic events. No other atrial thrombi were identified in this series. This patient also had an abnormal origin of the left atrial appendage. It was located in the inferior left atrium, midway between the mitral valve anulus and the left inferior pulmonary vein. The defect resulting from excision of the left atrial appendage was closed without a patch.

The Journal of Thoracic and Cardiovascular Surgery Volume 105. Number 6

McCarthyel al.

108 I

The operations (Table II) were performed with oxygenated crystalloid cardioplegic solution with a mean aortic crossclamp time of 69 minutes (range 47 to 98 minutes). The mean cardiopulmonary bypass time was 143 minutes (range 112 to 172 minutes). Results Intraoperative and postoperative characteristics are given in Table II. All patients were extubated within 48 hours of the operation. There were no reoperations for bleeding. All patients were transferred to a regular nursing service within 36 hours except for patient I, who had pulmonary fibrosis as a result of amiodarone (3 days), patient 6, who had cardiomyopathy and marginal cardiac output (4 days), and patient 14, who awoke with right hemiparesis that gradually resolved (5 days). (Occlusion of the left internal carotid artery was subsequently detected in patient 14.) The mean length of hospitalization was 13 days (range 7 to 25 days). Postoperative cardiac rhythms. All patients had junctional rhythm (confirmed by atrial electrogram) soon after the operation. Three patients on a regimen of lowdose dobutamine initially had sinus rhythm that changed to junctional rhythm when dobutamine was discontinued. Atrial pacing usually was not effective from the SA node area during the initial 72 hours after the operation. Patients were paced artificially through the medial wires until sinus node function returned or until adequate junctional rhythm (stable blood pressure, positive chronotropic response to exercise) was established. Sinus rhythm (with intermittent periods of junctional rhythm) returned as early as 4 days after the operation in some patients but not until after hospital discharge in others. Patient 2 had a very slow junctional rhythm without chronotropic response to exercise. This patient had preoperative sick sinus syndrome; a DDD-R pacing system was placed 14 days after the operation. One-year followup studies showed the return of sinus rhythm. Patient 5 also had preoperative sick sinus syndrome. He had long sinus pauses after the operation and received a DDD-R pacing system on the fifteenth postoperative day. One month after the operation his predominant rhythm was normal sinus rhythm. Patients 12, 13, and 14 were started on theophylline therapy (100 to 200 mg orally twice a day) within 72 hours of the operation in an attempt to speed the recovery of sinus node function or a more rapid junctional rhythm. These patients were all in a stable, nonpaced rhythm within 5 days of the operation. These patients were discharged earlier than most (at 8, 8, and II days, respectively). Sustained episodes of atrial fibrillation necessitating

Fig. 4. Intraoperative transesophageal echocardiogram, using acoustic quantification techniques, documented reduced atrial sizeafter mitral valve repairand maze procedurewith resection of atrial tissue.

maintenance antiarrhythmic drug treatment developed during the early postoperative period (within 14 days) in six patients (43%). Other patients had short bursts of atrial fibrillation or flutter. The six patients with sustained atrial fibrillation were discharged receiving a variety of antiarrhythmic medications, depending on their preoperative profile of effectiveness and the side effects of the various drugs. Three patients were treated with tlecainide, two with digoxin and procainamide, and one with propafenone. In all patients the antiarrhythmic medications were stopped within 3 months after the operation. None of these patients has had any further episodes of atrial

I 082

The Journal of Thoracic and Cardiovascular Surgery June 1993

McCarthy et af.

Table II. Intraoperative and postoperative characteristics of 14 patients undergoing the maze procedure for chronic AF Patient No.

Crosse/amp time (min)

CPBtime (min)

59

145 150 171 151 139 114 148 117 130 149 172 143 112 160

1 2 3 4 5 6 7 8 9 10 11 12 13 14

77

98 76 66 61 80 57 62 77

69 47 47 90

Operative modifications

Hospital stay (days)

22 18 10 9

MY repair

21 14 11

L-LAD

8 9

25 S-RCA Modified

7 8 8

MY repair

11

Follow-up rhythm NSR NSR (DDD-R)* NSR NSR NSR (DDD-R)* JR (died)

NSR NSR NSR DDD* NSRt NSR NSR NSR

AF, Atrial fibrillation; CPR, cardiopulmonary bypass; L-LAD, Left internal mammary artery-left anterior descending artery bypass; S-RCA. Saphenous veinright coronary artery bypass; NSR, Normal sinus rhythm; JR, Junctional rhythm. *p-Wave tracking, with or without ventricular pacing.

tPreoperative pacemaker placed.

fibrillation. Twenty-four-hour Holter monitor studies in patients from 6 to 12 weeks after the operation did not show any atrial fibrillation. Late e1ectrophysiologic data through invasive testing have not yet been obtained. Postoperative complications. Unusually severe postoperative fluid retention was a problem in five patients (36%) (patients 4, 6, 9, II, and 14). Three of these patients had to be readmitted for the treatment of pleural effusions that caused dyspnea, severe peripheral edema, or both. Two patients had normal ventricular function; three others had impaired ventricular function (patient 6 from Yamaguchi's disease, patient II from previous right ventricular infarction, and patient 14 from mitral valve disease with ventricular dilatation). These patients were treated with a variety of diuretics and by thoracentesis when necessary. Fluid retention resolved in two patients. One patient's symptoms have improved substantially only 6 weeks after the operation. One patient with right ventricular infarction continued to have peripheral edema at 4 months' follow-up, and one patient (patient 6) died. Patient 6 was discharged from the hospital after 14 days. She was readmitted because of dyspnea and pleural effusions. After thoracentesis, there was a residual infiltrate in the right lower lobe and antibiotics (including gentamicin) were begun for possible pneumonia. Her cardiac rhythm during this time alternated between junctional rhythm and normal sinus rhythm. Three days after starting the antibiotics the patient suddenly died of renal failure and hyperkalemia. Autopsy showed severe concentric left ventricular hypertrophy with apical obstruction and a small left ventricular cavity consistent

with Yamaguchi's disease. The suture lines of the maze procedure were intact with no thrombus. A small pulmonary infarction had involved the right lower lobe. Although this patient was initially discharged from the hospital, her death is classified as an operative mortality (7% mortality) because she died less than 30 days after the operation. Cinegraphic magnetic resonance imaging studies were performed on five patients (Fig. 5) from 2 months to I year after the operation. In all patients there was clear evidence of atrial systole and this appeared to be near normal. Four patients could not be studied with magnetic resonance imaging because of the presence of permanent pacemaker systems. Transesophageal Doppler echocardiographic criteria were also used to assess atrial systole, including an A wave across the mitral or tricuspid valve inflow, and reversal of flow in the pulmonary veins. Patients in junctional rhythm did not clearly exhibit atrial systole. All other patients did have evidence of mechanical atrial contraction. All patients beyond the 6-week postoperative phase returned to New York Heart Association functional class I or II. One patient is permanently disabled and has not returned to work. Patient II had not yet returned to work in part because of peripheral edema and fatigue, but he anticipated returning to work. Three patients who had taken early retirement, or who were temporarily disabled, returned to work. Patient 9 returned to work but then began having edema and fatigue. He underwent pericardial stripping for constrictive pericarditis 8 months after the maze procedure and is recovering from that operation.

The Journal of Thoracic and Cardiovascular Surgery Volume 105. Number 6

McCarthy et al.

I083

Fig. 5. Impro vement in atria l size a nd contrac tility status afte r maze procedure documented on dynam ic "bright blood" cinegraphic magnetic resonance imaging (fo ur-cha mber orie nta tion). Before maze procedure (A) dilat ation of the fibrillating, minimally contracting left atrium (large asterisk ) a nd right atrium (small asterisk ) is noted a t two adjacent levels. Because of significant mitr al valvular insufficiency, the left ventricle (large circle) is a lso dilated whereas the right ventricle (small circle) is of norm al size. Afte r the ma ze procedure, includ ing m itral valve reconstruction, atrial size is mark edly improved. T his is related to resolution of mitral insufficiency a nd restorat ion of a trial cont raction; from ventricular end-systole (8) to late diastole (e), passive a nd active size reduction in the size of both a tria is noted . The la tter aids in opening of the mitral valve leaflets (arrowheads).

Discussion Need for the maze procedure. Cardiac surgeons, heretofore, have rarely been exposed to the exceptional group of patients with severe disabling atrial fibrillation.

Alth ough most symptoms of atrial fibrillation are controlled with medications, a sma ll percentage of patients have drug intolerance or severe side effects; especially notable are the myriad effects of long-term amiodarone

1 084

The Journal of Thoracic and Cardiovascular Surgery June 1993

McCarthy et al.

Table III. The annual risk of stroke in patients with nonrheumatic atrial fibrillation treated with placebo, aspirin, or warfarin Study

Patients (n)

Placebo (%)

Aspirin (%)

AFASAK22 SPAF23 BAATAF24 CAFA 25

1007 1244 420 378

5.5 8.3 3.0t 5.2

5.5 *[

Warfarin (%) 1.6*

2.0 ] 0.4 3.5

*Patients receiving aspirin or warfarin. tForty-six percent of patients in the placebo group were receiving aspirin.

use.8 , 9 More worrisome are the proarrhythmic potential of most antiarrhythmic drugs.'? The CAST study!' surprised many by revealing serious safety issues with the use of fiecainide and encainide after myocardial infarction. Reports are now appearing of cardiac arrest (such as in our patient 4) and death in patients treated with these drugs for supraventricular tachyarrhythrnias.l- 13 One of the first-line drugs, quinidine (which may induce torsades de pointes), recently has been implicated in a metaanalysis showing higher-than-expected mortality for patients having atrial fibrillation.!" These reports can be faulted, and they include only small numbers of patients, but no prospective study (such as CAST) has investigated the safety of antiarrhythmic medications used to treat atrial fibrillation.f For patients with severe symptoms whose drug therapy has failed, radiofrequency ablation of the AV node to produce heart block prevents the rapid ventricular response to the supraventricular tachycardia. Insertion of a VVI rate-responsive pacemaker converts most patients to a less symptomatic condition.P: 16 However, the atria continue to fibrillate, with the ongoing risk of thromboembolism, and hemodynamic function is still impaired by the loss of atrial systole (both of these problems were seen in patient 10). Radiofrequency ablation of the AV node appears to be safer than direct-current AV node ablation (which is associated with a 5.1% hospital mortality), 17,18 but the patients usually have permanent heart block and are pacemaker dependent. Late sudden death with both treatments, most likely caused by pacemaker failure, has been reported. I 5, 17 The procedure is well tolerated, however, and is effective in the majority of patients. From January 1991 until May 1992 (the same time period as the maze procedure experience), 61 patients underwent radiofrequency ablation of the AV node with insertion of a pacemaker for drug-resistant atrial fibrillation at the Cleveland Clinic. Although the symptoms of atrial fibrillation may be severe and even disabling in unusual patients, the most

feared complication is thromboembolism. In patient 10, with three previous emboli, the operation revealed a left atrial thrombus, despite long-term warfarin use. The patient with Yamaguchi's disease {a known risk factor for systemic embolismj!? also had had two previous strokes. Patients who have had emboli have a high risk for further emboli. 2o , 2 1 Recent prospective studies22-2S indicate an embolic risk of 5% to 8% per year for patients with nonrheumatic atrial fibrillation treated only with placebo (patients were treated with aspirin in one study"), This risk of stroke can be decreased to 0.4% to 3.5% per year by warfarin (Table III). However, the long-term administration of warfarin was associated with a significant risk of minor and major bleeding episodes. Patients less than 60 years old with lone atrial fibrillation, however, are at low risk for emboli.f 20, 21 Further evidence of the risks of chronic atrial fibrillation was obtained in a study of 3099 life insurance applicants with atrial fibrillation.i" Patients with paroxysmal atrial fibrillation and no other cardiovascular abnormality had a normal mortality rate according to standard insurance mortality tables. However, patients with chronic atrial fibrillation had a significantly increased mortality, especially if fibrillation was accompanied by mitral stenosis. A surgical attempt to "cure" atrial fibrillation can be justified from the previous studies with regard to stroke risk, 19,22-25 decreased life expectancy." and avoidance of toxic antiarrhythmics.vl" Patients less than 60 years of age with paroxysmal "lone" atrial fibrillation are at low risk for thromboemboli and can anticipate a normal life expectancy. For this group of patients the indication for operation would be multiple drug failures and disabling symptoms. The potential benefits of the maze procedure over AV node ablation for these patients is the maintenance of sinus rhythm and atrial systole and the avoidance of life-long pacemaker dependency. Patients with previous emboli are typically maintained on warfarin therapy but still may be at significant risk for emboli. These patients will not benefit from AV node ablation and may be best served by the maze procedure. Patients with mitral valve disease, chronic atrial fibrillation for more than 1 year in duration, and dilated atria are unlikely to regain sinus rhythm after a mitral valve operarion." These patients are at risk for emboli's 20 and face a decreased life expectancy.i" The maze procedure may be added to the mitral valve operation (especially repair, or, theoretically, replacement with a tissue valve) to avoid long-term anticoagulation and, ideally, to decrease the risk of emboli and perhaps death. Effectiveness of the maze procedure. The disabling symptoms and potential risks to patients with atrial

The Journal of Thoracic and Cardiovascular Surgery Volume 105, Number 6

fibrillation must be weighed against the effectiveness of the maze procedure in attaining its stated goals: the return of sinus rhythm and atrial systole, All but one of our patients returned to either normal sinus rhythm or paced atrial rhythm that also effectively restored atrial systole. Patient 12 was in junctional rhythm for approximately 2 months before the return of sinus rhythm. We had modified the maze procedure hoping to preserve the artery to the sinus node and perhaps to speed the recovery of sinus rhythm. In this patient, however, the sinus node artery arose from the circumflex coronary artery and no such benefit was derived. In addition, the modified operative approach is somewhat more complicated than the standard maze procedure and it is more difficult to combine with mitral valve operations; we therefore returned to the standard maze procedure. It would appear that the maze procedure effectively restores sinus rhythm, and subsequently atrial systole, in the majority of patients. However, in individual patients restoration may take weeks-up to 52 weeks in one of our patients-before sinus node function returns. Additionally, 3 months after the operation, all patients but one have been maintained without antiarrhythmic and anticoagulant medications. A follow-up echocardiogram of patient I I identified a thrombus on his remaining permanent pacemaker wire, and he is currently receiving a low doseof warfarin. Therefore, beyond 3 months the maze procedure was effective for rhythm control, return of atrial systole, and the avoidance of antiarrhythmic medications. The morbidity and mortality of the procedure must alsobe judged against the risks of atrial fibrillation. The most troublesome morbidity was perioperative fluid retention with edema and symptomatic pleural effusions. Fluid retention may be more likely to occur in patients withimpaired ventricular function, but it was also seen in two patients with normal ventricular function. Cox and colleaguesf have postulated that this may be the result of a relativeabsence of atrial natriuretic factor. We are currently planning a prospective study of our patients to determine the source of this fluid retention. Perioperative atrial fibrillation necessitating drug therapywas not uncommon. Our experience, and the reported experience of Cox and associates.f indicates that fibrillation may be an isolated perioperative event. The perioperative catecholamine release, inflammatory pericardial response, and trauma to the atria may favor a shorter atrial refractory period so that reentry circuits can be sustained within the suture lines.28 Additionally, experimental studies have shown that operative trauma to theatria can lead to lower atrial fibrillation thresholds.?" One patient had a perioperative stroke (transient right

McCarthy et al.

1085

hemiparesis), but this was most likely related to occlusion of the left internal carotid artery that was not recognized before the operation. The sole operative death occurred in a very disabled young patient with preoperative strokes (and persistent hemiparesis) and was related to the underlying apical obstructive cardiomyopathy. The perioperative cardiac function had been adequate but borderline and probably predisposed the patient to the rapid onset of hyperkalemia and acute renal failure with the introduction of aminoglycoside antibiotics. In summary, the morbidity and mortality of the operation have been acceptable, but they should not be underestimated. We found a relatively low need for permanent pacemaker insertion (14%) after the operation. It may be more than just coincidence that the need for pacemakers was most common in the earlier patients. Currently, we are extremely careful to avoid excessive traction on the "strip" of right atrium containing the SA node. In addition, the posterior counterincision is brought as close to the left atrial suture line as possible and rotated posteriorly to give the SA node complex as wide a berth as possible (Fig. I). The two patients who had preoperative pacemakers would not otherwise have required permanent pacing systems because sinus node function returned early in both. The early junctional rhythm with relatively slow ventricular response has kept patients in the hospital longer than desirable. To minimize hospital stay, and potentially to speed the rate of recovery of the sinus node, we administered oral theophylline soon after the operation. This decision to administer theophylline assumes that the clinical situation is similar to that of patients in whom sinus node dysfunction develops after heart transplantatiorr'" and is based on reports of early return of sinus node function and avoidance of pacemakers in transplant recipients treated with theophylline'! Further experience is necessary to determine if theophylline is effective for patients after the maze operation, but early results are encouraging. Conclusions

In summary, Cox's maze procedure is a complicated operation for the treatment of patients with severely symptomatic atrial fibrillation refractory to drug therapy. We have also applied it in patients undergoing mitral valve repair, in whom the return of sinus rhythm was highly unlikely. Patients chosen for this operation were at the greatest long-term risk from emboli or severe disability. The maze procedure may have a higher morbidity and mortality for patients with impaired ventricular function than for patients with lone atrial fibrillation and normal ventricular function. Many questions are as yet unan-

I 086

McCarthy et al.

swered regarding the long-term efficacy of the operation, the risk of permanent sick sinus syndrome and the postoperative need for pacemaker systems, and the cause of perioperative fluid retention. The operation is cautiously being offered to a select group of patients who are most likely to benefit from the return of normal sinus rhythm and atrial systole and who have a low operative risk. Currently we consider patients as candidates for the operation if they (I) are less than 70 years old, (2) have no, or only mild, ventricular dysfunction, (3) have had embolic events related to atrial fibrillation, (4) have severe symptoms, and disability, on drug therapy (or from drug therapy), or (5) are undergoing other cardiac operations, have had chronic atrial fibrillation for more than I year, and have a left atrial dimension more than 60 mm (ideally with an easily repairable mitral valve lesion). However, because this operation is extensive and new, many patients are treated medically or are referred for A V node ablation with insertion of a permanent pacemaker.

Addendum As of February 1993, a total of 23 patients had undergone Cox's maze procedure, including mitral valve repair in six. There have been no further postoperative pacemakers and no antiarrhythmic drugs after the perioperative period. REFERENCES I. Cox JL, Schuessler RB, Cain ME, et al. Surgery for atrial fibrillation. Semin Thorac Cardiovasc Surg 1989;1:67-73. 2. Cox JL, Schuessler RB, D'Agostino JH, et al. The surgical treatment of atrial fibrillation. III. Development of a definitive surgical procedure. J THORAC CARDIOVASC SURG 1991;101:569-83. 3. Cox JL, Schuessler RB, Boineau JP. The surgical treatment of atrial fibrillation. I. Summary of the current concepts of the mechanisms of atrial flutter and atrial fibrillation. J THORAC CARDIOVASC SURG 1991;101:402-5. 4. Cox JL, Canavan TE, Schuessler RB, et al. The surgical treatment of atrial fibrillation. II. Intraoperative electrophysiologic mapping and description of the electrophysiologic basis of atrial flutter and atrial fibrillation. J THORAC CARDIOVASC SURG 1991;101:406-26. 5. Cox JL. The surgical treatment of atrial fibrillation. IV. Surgical technique. J THORAC CARDIOVASC SURG 1991; 101:584-92. 6. McCarthy PM, Cosgrove DM, Castle LW, White RD, Klein AL. Combined treatment of mitral regurgitation and atrial fibrillation with valvuloplasty and the Maze procedure. Am J CardioI1993;71:483-6. 7. Yamaguchi H, Ishimura T, Nishiyama S, et al. Hypertrophic nonobstructive cardiomyopathy with giant negative T waves (apical hypertrophy): ventriculographic and echocardiographic features in 30 patients. Am J Cardiol 1979;44:401-12.

The Journal of Thoracic and Cardiovascular Surgery June 1993

8. Pritchett EL. Management of atrial fibrillation. N Engl J Med 1992;326:1264-71. 9. Horowitz LN, Spielman SR, Greenspan AM, et al. Use of amiodarone in the treatment of persistent and paroxysmal atrial fibrillation resistant to quinidine therapy. J Am Coli Cardiol 1985;6:1402-7. 10. Josephson ME. Antiarrhythmic agents and the danger of proarrhythmic events. Ann Intern Med 1989;111:101-3. II. Cardiac Arrhythmia Suppression Trial (CAST) Investigators. Preliminary report: effect of encainide and flecainide on mortality in a randomized trial of arrhythmia suppression after myocardial infarction. N Engl J Med 1989; 321:406-12. 12. Fish FA, Gillette PC, Benson DW. Proarrhythmia, cardiac arrest and death in young patients receiving encainide and flecainide. J Am Coli Cardiol 1991;18:356-65. 13. Falk RH. Flecainide-induced ventricular tachycardia and fibrillation in patients treated for atrial fibrillation. Ann Intern Med 1989;111:107-11. 14. Coplen SE, Antmann EM, Berlin JA, Hewitt P, Chalmers TC. Efficacy and safety of quinidine therapy for maintenance of sinus rhythm after cardioversion: a metaanalysis of randomized control trials. Circulation 1990;82:1106-16. 15. Rosenqvist M, Lee MA, Moulinier L, et al. Long-term follow-up of patients after transcatheter direct current ablation of the atrioventricular junction. J Am Coli Cardiol 1990;16:1467-74. 16. Baermann JM, Olshansky B, Wilber DJ. Advances in the non-pharmacologic treatment of atrial fibrillation. J Invasi Cardiol 1992;4:160-6. 17. Morady F, Calkins J, Langber JL, EI-Atassi R. A prospective, randomized comparison of direct-current shocks vs. radiofrequency energy for catheter ablation of the atrioventricular junction [Abstract]. J Am Coli Cardiol 1992; 19:183A. 18. Evans GT, Scheinman MM, Bardy G, et al. Predictors of in-hospital mortality after DC catheter ablation of atrioventricular junction: results of a prospective, international, multicenter study. Circulation 1991;84:1924-37. 19. KogureS, Yamamoto Y, TomonoS, Hasegawa A, Suzuki T, Murata K. High risk of systemic embolism in hypertrophic cardiomyopathy. Jpn Heart J 1986;27:475-80. 20. Repique LJ, Shah SN, Marais GE. Atrial fibrillation 1992: management strategies in flux. Chest 1992;101:1095-103. 21. Cairns JA, Connolly SJ. Nonrheumatic atrial fibrillation: risk of stroke and role of antithrombotic therapy. Circulation 1991;84:469-81. 22. Petersen P, Boysen G, Godtfredsen J, Andersen ED, Anderson B. Placebo-controlled, randomised trial of warfarin and aspirin for prevention of thromboembolic complications in chronic atrial fibrillation: The Copenhagen AFASAK Study. Lancet 1989;1:175-9. 23. Stroke Prevention in Atrial Fibrillation Study Group Investigators. Preliminary report of the stroke preventionin atrial fibrillation study. N Engl J Med 1990;322:863-8. 24. Boston Area Anticoagulation Trial for Atrial Fibrillation Investigators. The effect of low-dosewarfarin on the risk of

The Journal of Thoracic and Cardiovascular Surgery Volume 105, Number 6

25,

26. 27.

28.

29.

30.

31.

stroke in patients with nonrheumatic atrial fibrillation. N Engl J Med 1990;323:1505-11. Connolly SJ, Laupacis A, Gent M, Roberts RS, Cairns JA, Joyner C. Canadian atrial fibrillation anticoagulation CCAFA) study. J Am Coli Cardiol 1991;18:349-55. Gajewski J, Singer RB. Mortality in an insured population with atrial fibrillation. JAMA 1981;245:1540-4. Betriu A, Chaitman BR, Almazan A, Guiteras Val P, Pelletier C. Preoperative determinants of return to sinus rhythm after valve replacement. In: Cohn LH, Gallucci V, eds. Cardiac bioprosthesis. Proceedings of the Second International Symposium. New York: Yorke Medical Books,1982:184-91. Cox JL, Boineau JP, Schuessler RB, et al. Successful surgical treatment of atrial fibrillation: review and clinical update. JAMA 1991;266:1976-80. Torres EJ, McCarthy PM, Schaff HV, Trastek VF, Pairolero Pc. Atrial trauma during venous cannulation or pneumonectomy decreases atrial fibrillation threshold. Surg Forum 1987;38:223-5. Heinz G, Laufer G, Hirschi M, et al. Postoperative sinus node dysfunction in the transplanted heart: impaired automaticity but normal refractoriness. Chest 1992;101:603-6. Redmond JM, Zehr K, Gillinov MA, et al. Use of theophylline for treatment of prolonged sinus node dysfunction in human orthotopic heart transplantation [Abstract]. J Heart Lung Transplant 1992;11:203.

Discussion Dr. Walter P. Dembitsky (San Diego, Calif). These cases represent the first reported series of the maze procedure (outside of the St. Louis experience, which is now in excess of 30 cases) and as such is important to us all. Whom do you consider to be candidates for this procedure now? Dr. McCarthy. There are a million patients with atrial fibrillation, and we must decide who should have this operation. We need to be cautious in trying to apply this only to patients who are going to derive the most benefit and who are at the lowestsurgical risk-young patients in whom all other medical therapy has failed and patients who have had previous emboli,

McCarthy et al.

1087

because they are at risk for further emboli. Also, I think we will see more maze procedures combined with other operations. Dr. Dembitsky. Do you think that preserving the SA node artery will prevent the dela yed chronotropic response to exercise seen late in some of the St. Louis patients, and should the course of this artery be identified before the operation? Dr. McCarthy. Yes. We have been working on modifying the maze incisions. Our patient 12 had a sinus node artery that arose from the circumflex artery. No matter which operation I had done, the standard or modified maze, I would have cut that artery. I think, though, that the potential exists to try to modify the maze incisions to try to preserve the SA node artery. Dr. Dembitsky. Some have seen elevated postoperative pulmonary artery pressures and think the elevation might be due to diminished left atrial compliance. Have you made this observation? Dr. McCarthy. Patients who had a thermodilution catheter were those who had impaired ventricular function and those who had a combined maze operation and mitral valve repair. They have other reasons for elevated pulmonary artery pressure. Patients with lone atrial fibrillation and normal ventricular function have only a central venous pressure line during the operation, so I do not have enough patients to allow me to make a clear judgement. However, we have also suspected that atrial compliance is decreased in patients who have had late studies. Dr. Dembitsky. Dr. Cox believes that postoperative fluid retention is due to attenuated atrial natriuretic factor production, and he has successfully managed that by using aldactone. What are your thoughts in that regard? Dr. McCarthy. Five of our patients had substantial fluid retention. We treated three of those patients with aldactone and it did not seem to be effective. We have treated them with a variety of diuretics such as bumetanide and furosemide. I would like to make some closing comments about Dr. Cox. First, he perceived the clinical problem with atrial fibrillation. Second, he studied the basic mechanisms in the laboratory and then confirmed these mechanisms in human beings. Third, he designed and then tested the operation to treat the problem in the laboratory. Finally, he successfully applied that operation in the clinical setting. Because of these accomplishments, I think that it is appropriate that the maze operation be referred to as Cox's maze procedure.