Risk factors analysis on failure of maze procedure: mid-term results

Objective: Since the late 1980s, surgical ablation of atrial ﬁbrillation (AF) has been one of the most effective means of curing this arrhythmia. However, about 20% of patients who underwent maze procedures have shown recurrence of AF during the follow-up periods. The aim of this study is to evaluate our result of maze procedures in last decade and to analyze the risk factors of maze failure. Methods: Between July 1997 and July 2007, 560 consecutive patients underwent maze procedures for AF by a single surgeon. Demographics showed that average age of the patients is 51.3 years,with a slightfemale predominance(M:F = 248:312). Most ofthe maze procedureshadbeenperformedin conjunction with mitralvalve ( n = 494, 88.6%), while only six cases (1.1%) were performed with isolated maze procedure. The maze failure was deﬁned as showing any rhythm besides normal sinus rhythm at the last follow-up. Univariate and multivariate analysis for the risk factors of maze failure were identiﬁed. The survival impact of maze failure was also evaluated. Results: The in-hospital mortality (1.6%) was acceptable. During the 29.7 months of median follow-up period, the late mortality rate was 3.8% and permanent pacemaker insertion was necessary in 2.3% ( n = 13) of the patients. The success rate of maze was 84.1% (471/560) and effective left atrial contraction was identiﬁed in 97.2% (458/471) of these patients. In multivariate analysis, the size of left atrium larger than 60 mm, cardiothoracic ratio over 60%, ﬁne AF wave in preoperative ECG, no early normal sinus restoration and simpliﬁed surgical ablation were found as an independent predictor of maze failure. Furthermore, the patients with successful maze showed better long-term survival rates. Conclusions: The results of our maze procedure during the last decade showed an acceptable success rate and the patients who were restored to sinus rhythm after maze procedures showed better long-term survival rates. For the patients who have independent biological risk factors, more thorough ablation lesion set is recommended for better long-term results.


Introduction
Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia that increases morbidity and mortality [1], and presents in up to 60-80% of patients undergoing surgery for mitral valve disease [2]. Since Cox Maze procedure was introduced by James L. Cox in 1987, surgical ablation of AF has been one of the most effective means of curing this arrhythmia. The advantage of maze procedure has been reported that it shows a better survival rate and lower stroke incidence after combining with mitral valve surgery [3,4].
Because of a complexity of the classical cut and sew maze procedure, lots of modifications with various energy sources have been invented to perform maze procedures more rapidly and safely [5][6][7][8][9]. However, about 20-30% of the patients who underwent maze procedures have shown recurrence of AF during the follow-up periods [10], and there are few reports showing mid-to long-term results and reporting predictors of the maze failure [11][12][13].
The aim of this study is to evaluate results of 560 consecutive cases of maze procedure in the last decade and to identify the risk factors of the failed maze procedures.

Patients and definitions
We analyzed 560 consecutive patients (248 male and 312 female patients, the mean age was 51.3 AE 13.1 years) who underwent maze procedures for AF by a single surgeon between July 1997 and July 2007. As the goal of maze procedure is the restoration to normal sinus rhythm (NSR), the success of maze procedure was exclusively defined as the sinus restoration after the maze procedures in this study. Therefore, all patients with recurring atrial tachyarrhythmia (such as AF, atrial flutter), junctional rhythm and cardiac rhythm of permanent pacemaker were categorized into the failed maze group (Fig. 1).
The fine AF wave was defined when the voltage of f-wave in V1 lead is lesser than 0.1 mV in preoperative ECG. The size of the left atrium (LA) was measured preoperatively by parasternal long axis view of transthoracic echocardiography. As the method to determine the size of the LA in this study was restricted to only the antero-posterior diameter of LA, the cardiothoracic ratio was assessed for better evaluation lateral enlargement of the LA. Sustained AF was defined as AF that does not self terminate and paroxysmal AF was defined as intermittent or recurrent AF. After maze procedure, the day of sinus restoration was recorded for every patient. The early NSR return was defined as sinus restoration within the day of operation. To evaluate the changes of the success rate during follow-up period, the day of AF recurrence was documented by various methods, such as ECG, echocardiography or Holter monitoring. All data were collected prospectively and stored in a specially designed and regimented database for uniformity, accuracy, and objectivity of the generated data by using Access 2000 (Microsoft Inc.) program for later assessment.
This study was approved by our institution's ethical committee/institutional review board. The informed consent was waived by our institution's ethical committee/institutional review board owing to the retrospective nature of our study.

Surgical procedures
During the last decade, three different types of maze procedures have been performed. Before January 1999, the conventional Cox Maze III procedure was performed on 30 patients. Between January 1999 and July 2001, 136 patients had received maze operations according to the modification devised by Lee and associates. Detailed surgical techniques of Cox Maze III and modified maze by Lee have been reported previously [5]. Since July 2001, 394 patients have undergone simplified surgical ablation (SSA) with various energy sources (cryothermy, microwave and radiofrequency). Compared to the original Cox Maze III procedure, the resection of right atrium auricle and incision from auricle to tricuspid valve were omitted, and most of the cut and sew lesions in the right atrium were changed to ablation lines except the long oblique incision to enter the right atrium. Three ablation lines for the right atrium were proceeded on intercaval, carvo-tricuspid and free wall of right atrium.
The lesion sets of the LA consisted of inferior extension of left atriotomy to orifice of left inferior pulmonary vein and four ablation lines; inferior and superior ablation lines for pulmonary vein isolation, endocardial LA isthmus ablation, and epicardial coronary sinus ablation. Resection of inferior wall of LA was performed as much as possible, whenever the size of the LA was larger than 60 mm by the preoperative echocardiogram or its enlargement was definite in operative findings. The area of resection depended on the size of the LA.

Clinical follow-up
Preoperative and intraoperative data were prospectively collected and patients were followed up at regular intervals by cardiologists of our institute or referring physicians. Postoperative rhythms were checked daily during their postoperative hospitalization by using standard 12-channel surface ECG and electroatriogram atrial pacing wire, which were routinely applied after the maze procedure in our institute. Follow-up ECGs of the outpatient clinic were checked postoperatively at 3 months, 6 months, and then annually in all patients. The basic rhythms were classified into sinus rhythm, junctional rhythm, AF, atrial flutter and pacing rhythm. To evaluate the restoration of mechanical function in LA, an assessment of transmitral A-wave with transthoracic echocardiography was conducted during the week after surgery, at postoperative 6 months, and then annually during the follow-up. Whenever a patient complained of intermittent palpitation, a Holter monitoring was recommended to disclose the paroxysmal AF.
For the patients who showed sinus rhythm on ECG and transmitral A-wave was documented with postoperative echocardiography, no preemptive antiarrhythmic medication was prescribed at the time of discharge. If postoperative AF or atrial flutter was noted before discharge, intravenous administration of amiodarone with a loading dosage (up to 1350 mg/day) was initiated and 400 mg of amiodarone was given every 12 h. If a patient failed to restore sinus rhythm with medications, an electrical cardioversion was aggressively attempted.
In cases of recurrent atrial arrhythmia under amiodarone administration, we changed antiarrhythmic drug regimens to sotalol or flecinide and attempted electrical cardioversion. We attempted rate control on all of our patients who did not maintain sinus rhythm with the use of beta-and calcium channel blockers.

Statistical analysis
Continuous variables were presented as mean AE SD, and categorical variables as percentages or numbers. For univariate analyses, preoperative and operative variables were analyzed using Kaplan-Meier method or Cox regression Downloaded from https://academic.oup.com/ejcts/article-abstract/36/2/272/517541 by guest on 30 July 2018 model to investigate the influences of these variables on the maze failures during follow-up period. Independent predictors were determined by using the Cox multivariable analysis involving a backward elimination procedure. The proportional hazards assumption was confirmed by examining the log(Àlog[survival]) curves and testing partial (Schoenfeld) residuals [14], and no relevant violations were found. Kaplan-Meier survival curve with log-rank test was conducted to evaluate the survival benefit of patients who had sinus restoration.
All p values were two-sided and a probability value of p < 0.05 was considered to indicate a significant difference. Statistical analysis was performed by using SPSS version 12.0 for Windows (SPSS Inc, Chicago, IL).

Perioperative demographics
In these patients, 55.7% of the cases (n = 312) were female and sustained AF were 86.6% (n = 485). The mean duration of AF was 5.4 AE 6.6 years, mean LA size was 58.4 AE 10.3 mm, and mean CTR was 59.4 AE 7.8%. Mitral valve repair was performed in 52.8% of mitral valve surgery and rheumatic mitral valve disease was the most common (68.6%) cause of mitral pathology. Our study populations consisted of 166 cases (29.6%) of cut and sew mazes and 394 cases (70.4%) of SSAs, which included 79 cases (14.1%) of right minithoracotomy approaches. The mean aortic cross-clamp time was 111.4 AE 38.2 min, and the mean cardiopulmonary bypass time was 160.8 AE 51.3 min. The common energy sources of ablations were cryothermy (76.3%) and microwave (22.7%).
To analyze the preoperative risk factors for maze failures, the patients were divided into two groups; patients with sinus rhythm restored (group S: 84.1%, n = 471) and failed maze (group F: 15.9%, n = 89), according to the documented rhythm at the last followup. Preoperative and operative characteristics of the patient are summarized in Table 1. In the present study, 99.8% of the preoperative size of the LA, 99.8% of AF duration, 96.8% of AF wave pattern and 96.4% of CTR were recorded. There were no significant differences between the successful and failed maze groups in terms of follow-up duration, female predominance, preoperative left ventricular ejection fraction, cardiopulmonary bypass time, aortic cross-clamping time, incidence of paroxysmal AF, rheumatic etiology of MV disease, choice of energy sources, prevalence of mitral valve repair and concomitant tricuspid annuloplasty. Patients in the failed group showed older age, longer history of AF, larger LA, bigger cardiothoracic ratio and more incidence of fine AF ( p < 0.001).

Operative results
Despite 98.9% of the maze procedures being combined with cardiac surgery, the early mortality happened in only 9 (1.6%) patients. Causes of the early mortality included low output syndromes with multiple organ failure (n = 4), sudden death after discharge with unknown etiology within 30 days (n = 2), left ventricular rupture after mitral valve replacement (n = 2), stroke and esophageal bleeding (n = 1).
Postoperative complications included 13 cases (2.3%) of pacemaker implantation after the maze procedure, 4 cases of mild reversible cerebral accident, 5 cases of tracheostomy for prolonged ventilation, 8 cases of acute renal failure required dialysis, and 31 (5.5%) cases of re-exploration for bleeding.
At the last follow-up, sinus restoration was shown in 84.1% (471/560) of our patients and effective left atrial contraction was identified in 97.2% (458/471) of these patients. At the time of hospital discharge, 76.4% (n = 428) of patients showed NSR and 7.7% (n = 33) of those patients were reverted to AF at their last follow-up. Among 23.6% (n = 132) of patients who showed any rhythm besides NSR at the time of discharge, 57.6% (n = 76) were restored NSR at the last examination and most of them showed junctional rhythm at the time of discharge. In the present study, cumulative maze success rates in all patients were 82.2% at 5 years and 69.8% at 9 years ( Fig. 2A).
The overall survival rates after the operation at 3, 5, and 9 years were 95.8%, 94.1%, and 88.9% respectively. Actuarial survival curves showed significant better long-term survival in successful maze group (Fig. 2B).
The univariate analysis of the maze failure in terms of preoperative and operative variables is shown in Table 2. The Kaplan-Meier curves with log-rank test showed significant difference on freedom from maze failure rate following dichotomous variables (Fig. 3).
Although the rate of sinus restoration at last follow-up was similar according to the modifications of surgical technique, the Kaplan-Meier curves with log-rank test revealed a significant difference of success rate between cut and sew maze and SSA (Fig. 3F).
In multivariate analysis, preoperative size of the LA larger than 60 mm, cardiothoracic ratio over 60%, fine AF wave in preoperative ECG, no early sinus restoration and SSA were found as independent predictors of maze failure in mid-term follow-up period (Table 3).

Impact of the simplified surgical ablation comparing the cut and sew maze
As the significant difference of the sinus restoration during follow-up period was disclosed, the comparison between simplified surgical ablation and cut and sew maze (Cox Maze III and Lee's modified surgical maze) was conducted (Table 4). Although most of the preoperative variables were comparable between groups, the patients in SSA group show significantly earlier sinus restoration. The shorter bypass time and aortic cross-clamping time was due to the simplicity of SSA, and the earlier sinus restoration resulted in the shorter postoperative ICU and hospital stay.

Discussion
Since the Cox Maze procedure was innovated as a surgical therapy for AF by Cox [15], the basic concept has been widely adopted by many surgeons and at the same time, various modifications over the past decade have been produced to simplify and easily perform the maze procedures [5][6][7][8][9]. This has enabled a wide acceptance among surgeons all around the world. However, there are few reports of intermediate or long-term results of modified maze procedure comparing to the original cut and sew maze procedures [13,16]. This study demonstrates well observed mid-term follow-up data of a considerable number of patients. With a single surgeon experience, a consistent surgical indication of maze  procedure and uniform surgical technique during the study period may increase the reliability of the presenting data. As various methods and different definitions of reporting the success of the maze procedures have been made, to evaluate the outcomes of maze procedure is difficult. Outcomes of the maze procedure are influenced by a thoroughness of follow-up as well as the method of a rhythm assessment. Considering the method of a rhythm assessment, 'last follow-up rhythm' may underestimate the recurrence rate of AF, which then overestimates the success rate of the procedure. Conversely, actuarial methods used to delineate time-related events, 'AF recurrence-free rate' define any recurrent AF as a failure of the procedure, which may underestimate the actual clinical success rate. Since the most favorable method of the reporting the success rate is 'rhythm at last follow-up' [17][18][19], the method was adopted to evaluate the success and failure of maze procedures in this study. Looking at the definition of maze success, the reported clinical results of maze procedures were variously expressed, such as 'freedom from AF recurrence', and 'normal sinus rhythm restoration rate'. In this study, the success of the maze procedure was exclusively defined as the sinus restoration after the maze procedure, On the other hand patients with recurring atrial tachyarrhythmia, junctional rhythm and cardiac rhythm of permanent pacemaker were categorized into the failed maze group.
Regarding to the long-term results of the maze procedure, Prasad and associates demonstrated that overall freedom from AF at 14 years is 92% and the freedom from AF rate was well maintained during long-term follow-up period [17]. On the contrary, many other long-term results of the maze procedure showed a gradual attrition rate of freedom from AF during the follow-up period [13,18,20]. This study also showed a progressively decreased maze success rate at 3, 5, 7, and 9 years after operation; 86.5%, 82.2%, 75.4%, and 69.8% respectively. Although the differences of the long-term success rate could be influenced by the differences among intraoperative variables, the long-term success rate of maze mostly varied from preoperative variables, such as the concomitant valvular surgery and rheumatic valvular heart disease. In this study, only 1.1% of the isolated maze procedure and 68.6% of the rheumatic mitral valve disease could be a possible explanation of relatively low maze success rate in a long-term follow-up period.
The well-known risk factors of maze failure in previous publications have included old age, larger LA diameter, longer history of AF, lower amplitude f-wave, having a rheumatic mitral valve disease, permanent AF, and lesion sets of maze procedures [11][12][13]21,22]. In this study, an LA size, a CTR, a fine AF, no early NSR return, and SSA maze procedure were identified as independent predictors of maze failure. The clinical applications of this study are as follows: first, we need to change our surgical approach from the right mini-thoracotomy to the median sternotomy for the patients who have three biological risk factors (LA size >60 mm, CTR >60%, fine AF). The sternotomy enables us to resect the LA auricle and to ensure the transmurality of ablation line. Second, close observation is necessary for the patients who returned to NSR after the day of operation.
There are still controversies whether the success rate of cut and sew maze is superior to other surgical modifications and ablations using alternative energy sources [23]. Some authors insist that more extensive lesion sets were not shown to improve outcomes of the surgery and rather biology of patients might be a major determinant of surgical AF ablation success [7,22]. However, others argued that more extensive lesion sets were associated with a greater success rate in their study population [10,24]. In this study the ablation line of the SSA reproduced most lesion sets of the Cox Maze III procedure except the resection of a right atrium auricle and an incision between right atrium auricle and tricuspid annulus. Although the SSA enabled earlier sinus restoration and shorter hospital stay, the mid-term success rate of the SSA was significantly lower than cut and sew maze. Furthermore nontransmural lesions of SSA associated with various condition could be a possible explanation of these difference.
This study has several limitations. First, this study was subjected to limitations inherent to a nonrandomized, retrospective, and observational data. As a result, operative techniques and energy sources were decided by surgeon's preference. As the present study was based on a single surgeon's experience, however, such bias might be minimized. Second, although close follow-up was maintained on most of the patients, none of the patients had continuous cardiac monitoring such as transtelephonic ECG devices. As the symptomatic patients seem to have better follow-up with intermittent 12-lead ECG, the overestimation of AF at the long-term observation might be possible. On the contrary, underestimation of postoperative AF by missing the paroxysmal AF could also be possible. Third, the significant difference of follow-up duration between cut and sew maze and SSA was a major limitation on comparison of these two techniques. Therefore, longer duration of observation is needed. Fourth, as the two groups were divided according to the rhythm of the latest follow-up, these groups do not actually exist and patients in both groups can be altered in the next follow-up. This limitation weakened the statistical credibility of this study.
In conclusion, our maze procedure during last decade showed an acceptable success rate and the patients with sinus restoration after maze procedure showed better longterm survival rates. The close observation is necessary for the patients who returned to NSR after the day of operation. Although the SSA showed favorable early results, the longterm success rate needs to be observed precisely. For the patients who have independent biological risk factors such as LA size larger than 60 mm, CTR over 60% and fine AF, more thorough ablation lesion set is recommended for better longterm results.