Study patients .

We modified the original maze procedure [REF]1 and began to combine it with other open heart operations in March 1992 ; the initial 14 patients were selected mainly on the basis of simplicity of the combined procedure for safety .
After a further modification in atriotomy ( Fig .
1 ) [REF]7 , contraindications for the combined approach were abandoned in the next 37 patients with AF undergoing cardiac operation , with the exception of two patients during the same period who were judged unable to tolerate the combined operations .
Up to August 1993 , 51 patients with chronic AF ( Table 1 ) underwent the maze operation simultaneously with mitral valve operation ( n = 41 ) , isolated aortic valve surgery ( n = 2 ) and closure of atrial septal defect ( [ ASD ] ostium secundum , n = 7 ; ostium primum , n = 1 ) .
Concomitant tricuspid annuloplasty was carried out in 31 patients ( 60.8 % ) and left atrial plication in 3 .
Twelve patients ( 23.5 % ) had a previous valvular operation , with replacement in 9 and repair in 3 .

Control patients .

For each patient undergoing the maze procedure , a control patient was selected from patients undergoing the same procedures for the same underlying diseases , with the same history of previous operation , but without the maze procedure .
Control patients were selected retrospectively according to best match with regard to age and preoperative New York Heart Association functional class .

Surgical modifications .

Our initial modification of the original maze procedure [REF]1 included use of cryoablation and changes in atriotomy ( Fig .
1 ) to shorten the atrial suture line .
This procedure was further modified [REF]7 to avoid transecting the sinus node artery [REF]8 ( Fig .
1 ) .
Other modifications included transection of the superior vena cava for better exposure and easier manipulation of the mitral valve in all except one patient who had ASD closure and tricuspid annuloplasty .
Circumferential incision around the pulmonary veins mobilized the left ventricle and improved exposure of the mitral valve .

Postoperative management .

Atrial fibrillation occurring early after operation associated with hemodynamic compromise was first treated with overdrive pacing using temporary wires and intravenous medication and then by electrical cardioversion in both groups of patients .
Atrial fibrillation occurring later without obvious compromise was treated in the same way except for electrical cardioversion , which required an informed consent .
Direct current cardioversion was carried out under intravenous anesthesia , delivering 100 to 300 J with intravenous antiarrhythmic agents in case of failure .
Afterward , patients were usually started on oral antiarrhythmic agents , mainly procainamide or quinidine , until the rhythm was considered stable , unless intolerance developed .
Anticoagulation with warfarin was instituted in patients with mechanical valves and those with persistent AF .
Patients who regained atrial rhythm and contraction after reparative operation were maintained with anticoagulation for the initial 3 to 6 months , at which time anticoagulation was discontinued , and antiarrhythmic agents , if any , were tapered .

Data collection .

Information collected for comparison included duration of cardiac arrest and cardiopulmonary bypass time in the operating room .
Early after operation , patients were evaluated for hemodynamic variables , requirement for catecholamines , cardiac index , pulmonary artery pressures and pulmonary artery wedge pressure 12 h after admission to the intensive care unit ( ICU ) .
Time until extubation ( extubation ) and before discharge from the ICU ( ICU stay ) were also recorded .
Chest tube drainage ( blood loss ) and transfusion requirement during the ICU stay were calculated .
For patients with the maze procedure , cardiac rhythm was closely monitored after cardioversion during cardiopulmonary bypass immediately before and after operation and continuously thereafter .
After chest closure , atrial rhythm was checked with atrial pacemaker wires daily in the ICU and with decreasing frequency in the ward until removal of the wires before discharge from the hospital .
Postoperatively , early diastolic ventricular filling and the presence of atrial A wave during transtricuspid and transmitral flow were documented by Doppler echocardiography scheduled 1 , 3 , 6 and 12 months after operation .
Other measurements included left atrial dimension ( LAD ) and left ventricular dimensions .
Chest X-ray films were taken on the same schedule for cardiothoracic ratio ( CTR ) .
Cardiac rhythm and functional class were assessed at discharge and at every ambulatory visit thereafter .

Statistics .

Continuous variables with equal variance were compared by two-tailed t test .
When variance was considered unequal , two sample t tests with Welch 's correction was used .
Discrete variables were analyzed by contingency table analysis .
Freedom from postoperative AF was analyzed by Kaplan-Meier actuarial curves .
Changes in CTR , LAD and functional class between the groups were analyzed with analysis of variance combined with a multiple comparison procedure .
Results are presented as average value + -SD .
Differences were considered statistically significant at p < 0.05 .

Patient groups .

Because no early or late deaths occurred in the maze group , control patients were selected from surviving patients with the same diseases and undergoing the same procedures .
Thus , the case-matched nonmaze control group included the same number of patients with the same underlying diseases , history of previous operation ( 24 % ) and surgical procedures as the maze group ( Table 1 ) .
Among 43 patients with valvular disease , both the maze and control groups included 36 patients with rheumatic disease and 7 with degenerative lesions ( pure mitral regurgitation in 6 , including 2 with repeat plasty , and bioprosthetic valve failure in 1 ) .
Among 36 patients with rheumatic etiology , 15 had predominantly stenotic lesions , 5 had predominantly regurgitant hemodynamic variables , and the other 16 had combined diseases , including 9 with reoperation .
Preoperatively , the control group was similar in average age , f wave voltage , left ventricular dimensions , cardiomegaly and functional class to the maze group ( Table 2 ) , except for an average date of operation 29 months earlier .
Although average duration of AF was comparable , the maze group included five patients ( 9.8 % ) with a history of AF < 1 year compared with nine ( 17.6 % ) in the control group .
Although the maze group included no patient with a history of AF < 6 months , the control group had three such patients , including one with AF for only 2 months before operation .

Operative data .

Intraoperatively , the maze group required significantly longer cardiac arrest ( 41 min longer on average ) , cardiopulmonary bypass ( 69 min ) and operative time ( 153 min ) than the control group ( Table 2 ) .
Although the maze group had significantly greater blood loss , they did not require more transfusions .
Intraaortic balloon pumping was required in four patient in the maze group versus one in the control group .
Although no differences were found in cardiac index or catecholamine requirements , pulmonary artery and central venous pressures were higher in the maze group 12 h after admission to the ICU ( Table 2 ) .
The maze group required significantly longer respiratory care , resulting in a prolonged ICU stay ( Table 2 ) .

Postoperative course .

Electrical cardioversion was performed in 11 control group patients mainly to control tachyarrhythmia early after operation and in 17 maze group patients mainly to reverse AF before discharge , with 11 responding .
All six patients with refractory AF had mitral valve disease as the underlying pathology ( rheumatic in four , degenerative in two ) .
Cardiac rhythm was stabilized by the time of discharge , when postoperative rhythm was defined , except for two maze group patients ( one experienced return of AF ; the other had spontaneous ablation of AF ) ; three control patients experienced return of AF after discharge .
Sustained AF was much less frequent in the maze group ( Fig .
2 , Table 2 ) .
Three patients in the control group who regained sinus rhythm without the maze procedure had had AF for 7 , 7 and 17 months , respectively , before mitral valvuloplasty for ruptured chordae tendineae in two and ASD closure in the other .
Although three control patients resumed and tolerated junctional rhythm , sinus node dysfunction required atrial pacemaker implantation in three patients undergoing the initial modification ( 21.4 % ) .
In two of them , preoperative sinus node function could not be determined , and one regained sinus rhythm later , overdriving the pacemaker .
No need for permanent pacing occurred in the control group and in the last 37 patients undergoing the current modification .

Atrial contraction .

An A wave was documented in 41 ( 80 % ) of 51 patients during transtricuspid flow and in 40 ( 78 % ) of 51 during transmitral flow after the combined operations .
Whereas early diastolic ventricular filling remained high , the transtricuspid A wave significantly increased from 30 +-9 cm/s at 1 month to 39 +-14 cm/s at 3 months after operation ( p = 0.024 ) .
Similarly , the transmitral A wave tended to increase from 46 +-22 cm/s at 1 month to 54 +-25 cm/s at 3 months after operation , comparable to the normal value for age [REF]9 , and leveled off thereafter .

Cardiac size , contraction and functional class .

Left ventricular dimensions and fractional shortening did not change before and after operation or differ between the groups ( Table 2 ) .
For CTR , LAD and functional class , observation time points were combined as early ( 1 to 3 months ) and late ( 1 to 2 years ) after operation ( Fig .
3 ) .
Although cardiac size as assessed by CTR remained unchanged in the control group after an initial decrease , it continued to decrease only in the maze group ( p = 0.011 ) .
Both groups showed reduced LAD after operation ; however , in the control group , LAD increased significantly ( p = 0.03 ) , returned to preoperative levels and became significantly larger than that in the maze group ( p < 0.0015 ) , in whom LAD was essentially unchanged ( Fig .
3 ) .
Both atria appeared to contract more efficiently over time in sinus rhythm , with decreasing LAD .
Increased physical activity and load after discharge resulted in less fatigue and dyspnea in the maze group , indicating a significantly improved functional capacity in these patients ( p < 0.009 , Fig .
3 ) .

Medications .

Among 22 patients after reparative operation ( 8 after ASD closure , 14 after mitral valvuloplasty ) ( Table 1 ) , anticoagulation could be discontinued in 16 patients ( 73 % ) in the maze group .
Medication was totally discontinued in two control group patients and in seven maze group patients who regained atrial rhythm and contraction ( four after tricuspid annuloplasty with ASD closure , three after mitral valvuloplasty ) .

Late complications .

Despite warfarin anticoagulation , one transient neurologic ischemic attack occurred in a maze group patient with normal sinus rhythm and contraction 4 months after mechanical mitral valve replacement , and cerebral infarction occurred in one control patient with persistent AF 6 months after aortic valve replacement .
Intracranial bleeding requiring admission to the hospital occurred in one maze group patient and two control group patients ; all were receiving warfarin for anticoagulation .

Study design and limitations .

This retrospective study was carried out to evaluate the risks and benefits of combining the maze procedure with surgical intervention for underlying disorders causing AF ; however , a prospective , randomized study should have been performed .
After being informed of the initial results , every patient requested the combined approach despite potentially increased risks and undetermined efficacy .
Although case-matched control patients were included , obvious differences in the date of operation and treatment of postoperative AF could not be eliminated , undermining the inferences derived from the results .
The decision to add the maze procedure rest on whether the potentially greater risks of increased complexity are outweighed by the benefits from regained atrial rhythm and contraction .
Thus , the risk-benefit balance may help to determine indications for the combined approach , which should be different from indications for the isolated maze procedure for lone AF [REF]2 .

Risks .

Risks associated with the combined approach include the complexity of the maze procedure , which requires prolonged cardiac arrest and cardiopulmonary bypass with subsequent delayed postoperative recovery .
Among the first 14 patients undergoing the initial modification , 3 ( 21 % ) required intraaortic balloon pumping , and 3 ( 21 % ) required atrial pacemaker implantation .
Although the incidence of sinus dysfunction requiring atrial pacing was lower than in the series of Cox et al .
[REF]2,10 , frequent complications prompted us to further modify the procedure [REF]7 , resulting in no need for artificial pacing among the next 37 patients undergoing the current modification .
Times required for the combined operations were ~ 40 , 70 and 150 min longer for cardiac arrest , cardiopulmonary bypass and the complete operation , respectively , than for the control operations alone .
Nevertheless , these times were significantly shorter than those required for the maze III procedure of Cox et al .
[REF]10 combined with operations for organic lesions in a similar cohort [REF]11 , suggesting the simplicity of the current modification .
Preoperative and postoperative left ventricular dimensions and contraction were similar between the groups , suggesting that longer cardiac ischemia does not result in persistent or significant ventricular dysfunction .
Despite initial complications attributable to the combined maze procedure , subsequent modifications and increased experience improved results and allowed uneventful recovery in the rest of the maze group despite extended indications .

Benefits .

Even when the combined approach was considered to have failed to restore sinus rhythm or to defibrillate in all patients who required electrical cardioversion ( n = 17 ) , AF was surgically ablated in the remaining two thirds of patients ( 34 of 51 ) after the combined approach .
This number is much higher than that in current control groups including twice as many patients with a brief history of AF , who were more likely to undergo defibrillation [REF]12 .
Sato et al .
[REF]6 reported that AF was abolished in only 28 % of patients late after operation with aggressive treatment using repeated cardioversion after repair of underlying lesions alone .
Even lower long-term maintenance of sinus rhythm was reported by Hansen [REF]5 , who identified preoperative AF < 12 months in duration as the only variable affecting the results of electrical cardioversion after mitral valve surgery .
These reports [REF]5,6 discouraged the control patients from undergoing cardioversion , resulting in a significant difference in treatment of postoperative AF and earlier operation .
However , these factors alone may not account for the fact that 34 patients had no arrhythmias requiring early cardioversion , and 88 % remained AF free late after the combined approach ; although addition of the maze procedure alone may not be solely responsible for these advantages either .
In contrast , the rate of regaining atrial rhythm and contraction after the combined operations was still lower than after lone AF with the isolated maze procedure [REF]2,10 .
Results of our recent review [REF]11 and analysis [REF]13 indicated that the difference appeared to result from the duration of the arrhythmia and LAD rather than the technical modification itself .
Despite concern about recovery of atrial contractility , the majority of patients were found to have significant right ( n = 41 [ 80 % ] ) and left atrial contraction ( n = 40 [ 78 % ] ) for late diastolic ventricular filling .
Moreover , atrial contraction ( A wave ) appeared to improve over time in sinus rhythm , with decreasing LAD , approaching the normal value for age [REF]9 by 3 months after operation .
Thus , not only atrioventricular synchrony , but also active diastolic ventricular filling was improved .
This finding may account for continued reduction of cardiac size in the maze group , as reported by Gosselink et al .
[REF]14 , who observed reduction in left atrial size only in patients remaining in sinus rhythm after cardioversion .
Improved ventricular filling and reduced atrial size may account for improved functional capacity after the combined procedure .
Left atrial isolation has been reported [REF]15 to be less complicated yet comparably effective in restoring regular ventricular contraction or the RR interval in 71 % of patients late after operation .
Nonetheless , persistent AF in both atria after the procedure may limit hemodynamic improvement , which was not demonstrated , and require continued anticoagulation because the increased risks inherent to AF persist [REF]4 .
Of 22 patients undergoing repair with our combined approach , anticoagulation was no longer necessary in 16 ( 73 % ) , and medication was totally discontinued in 7 ( 32 % ) .
Although there were no significant differences in the incidence of thromboembolic or bleeding complications in the current review with limited numbers and follow-up , restoration of atrial rhythm and contraction by the maze procedure may make a long-term difference in this cohort with underlying organic lesions and a greater risk of stroke [REF]4 .

Conclusions .

Because of the lack of prospective randomization , the present results are suggestive and inferential .
Nonetheless , results of the current study convinced us that the substantial benefits from regained atrial rhythm and contraction outweigh the potential risks of adding the maze procedure , justifying the combined approach in patients with established AF and organic disorders , except for a few patients who were considered unable to tolerate the procedure .
Prospective randomization seems warranted to examine the value of adding the maze procedure in patients with AF of recent onset [REF]12 and in those with presumed maze-refractory AF [REF]13 undergoing operation for organic lesions .
Acknowledgments

We appreciate the editorial assistance of Leonard M.
Linde , MD , Professor of Pediatrics ( Cardiology ) , University of Southern California , Los Angeles .