James L Cox introduced the first surgical procedure for atrial fibrillation (AF) and the original Cox maze procedure has undergone several iterations [1]. The first and second iterations were discontinued because of chronotropic insufficiency, a higher rate of post-surgical pacemaker implantation, occasional left atrial (LA) dysfunction, and the difficulty of portions of the surgery [2]. The Cox maze III procedure achieved wide popularity as both a stand-alone procedure and as an add-on procedure for patients undergoing other types of open-heart surgery [3]. The Cox maze III procedure involved a box lesion consisting of a through and through surgical incision around all 4 PVs (PVs). In addition to isolating the veins, this box also isolated the posterior wall. The Cox maze III also involved a mitral isthmus line, a tricuspid isthmus line, a line from the superior vena cava to the inferior vena cava, and excision of the LA appendage. With newer ablation tools such as bipolar radiofrequency ablation, the Cox maze IV procedure evolved and many of the surgical incisions were replaced by radiofrequency or cryoablation lesions [4]. In the original Cox maze IV, the right and left PVs were encircled as two circles and an inferior line leaving most of the posterior LA electrically connected to the remainder of the atrium. Later, a superior ablation line was added for full posterior wall isolation [5]. Variations of the Cox maze IV procedure have been done at the time of open-heart surgery via sternotomy or as stand-alone procedures through mini thoracotomies or ports, both as on-pump or off-pump procedures. Procedures have used focused ultrasound [6], and surgical and catheter-based ablations have been combined for a “hybrid” procedure [7]. Many times, these surgical procedures eliminate atrial arrhythmias. However, some patients experience arrhythmia recurrence and come for catheter ablation to complete the goal of arrhythmia elimination. In this study, we examined patients undergoing catheter ablation after a failed surgical maze procedure. We compared the patients who had undergone a classic cut and sew (CS) maze to those who had undergone a non-cut and sew (NCS) maze.

The main finding of this study is that among patients undergoing a radiofrequency catheter ablation for failed surgical maze, the CS maze technique of a classic Cox maze III operation isolated nearly all the PVs whereas other modalities of NCS procedures never isolated all the PVs. In addition, the most common clinical failure rhythm for the NCS procedures was AF and for the CS procedures was atrial flutter.

There have been no randomized trials comparing a CS to a NCS procedure for AF. However, there have been reports comparing populations of patients undergoing various surgical techniques for the treatment of AF. Gaynor et al. [11] compared a series of 276 patients who had undergone various iterations of the Cox maze procedure. This study had an extraordinarily long follow-up time for earlier versions of the Cox maze procedure. With 19.5% of patients on antiarrhythmic drugs, the original Cox maze I procedure had a 10-year freedom from AF rate of 75.4%, the Cox II procedure had a 10-year 83.6% freedom from AF rate, and the Cox maze III had a 10-year 89.3% freedom from AF. Follow-up was short for the Cox maze IV procedure which included many radiofrequency lesions. However, the 1-year freedom from AF, including those on antiarrhythmic drugs, was only 89% suggesting that it was going to be inferior to the Cox maze III procedure which had a similar freedom from AF after 10 years. In that study, the predictors of long-term success were the duration of AF prior to the surgery and the type of Cox maze procedure done. McCarthy et al. [6] examined 400 patients undergoing surgical treatment of AF. They compared 5 types of surgical AF procedures. At last follow-up, freedom from both AF and the need for a subsequent ablation was 90% for patients undergoing a classic maze procedure, 43% for a high intensity-focused ultrasound procedure, 79% for a LA maze procedure, 79% for a biatrial maze procedure, and 69% for PV isolation alone. For those patients who had only a LA maze procedure or PV isolation, the RA was the source for failure in 75%.

The best comparison of CS to NCS was reported by Stulak et al. [12]. They reported a matched control series of patients undergoing traditional CS Cox maze procedure compared to patients undergoing an identical operation but only using bipolar RF energy. Patients were matched for gender, age, New York Heart Association Class, AF type, and concomitant mitral valve surgery. They found that the CS maze was superior in all regards to the radiofrequency version of the same operation. At hospital discharge, 88% of patients with the CS procedure were in sinus rhythm compared to 63% for the radiofrequency maze. At 12 months of follow-up, 91% of the CS patients were free of AF versus only 76% for the radiofrequency group and at 15 months, the numbers were 91% and 61%. In addition, new permanent pacemaker implantation occurred in 25% of the radiofrequency group compared to only 5% of the CS group (P = 0.004). Antiarrhythmic medication at last follow-up was required in 75% of the radiofrequency ablation group as opposed to only 25% of the CS group (P < 0.05). Eighty-five of the patients in the radiofrequency group were still receiving warfarin anticoagulation at last follow-up compared to only 25% of patients in the CS group (P < 0.05). These authors concluded that “because transmurality can be assured, the standard CS Cox maze procedure remains the gold standard for the surgical treatment of AF.” The only logical explanation for the high failure rate for pulmonary vein isolation in our NCS group is that they did not have transmural lesions. One would anticipate that the surgeons would have a better opportunity to create transmural lesions in the operating room than ablationists can do in the EP lab with a catheter. However, when a through and through cut-and-sew incision is not made in the operations, the NCS lesions can be subject to the same limitations of performing it with a catheter in the electrophysiology laboratory: namely that as the tissue heals, conduction can recur. It is well known that the thickness of the atrium varies considerably, and radiofrequency, microwave, or other energies applied across the atrium might result in significant gaps in the lines the surgeons are attempting to create. Even electrophysiological mapping to document vein isolation at the time of surgery cannot prevent reconduction as healing occurs.

The study of Gillinov et al. [13] does suggest that doing a NCS operation is better than doing no AF operation at all. That study randomized patients undergoing mitral valve surgery with persistent or long-standing AF to receive no AF surgery, a pulmonary vein isolation ablation alone, or a pulmonary vein isolation plus a biatrial ablation. The surgical ablations were done using bipolar or unipolar RF. At 12 months 29.4% of patients receiving no AF ablation were in sinus rhythm and 63.2% of patients receiving AF ablation were in sinus rhythm. There was no difference between the two ablation strategies. Patients undergoing ablation had a 21.5% need for pacemaker implantation. The AF free outcomes in that study are remarkably similar to the study described above by Stulak et al. [12]. In that study, 76% of patients were free of AF and 25% required a pacemaker. One wonders if this study by Gillinov [13] might not have achieved the 91% freedom from AF reported by Stulak et al. [12] with only a 5% need for pacemaker had they used a CS instead of a NCS approach for patient’s randomized to AF surgery.

Despite the apparent superiority of a full CS maze procedure, it is our impression that relatively few surgeons still perform this operation. One reason may be the perceived difficulty of the procedure, but, in experienced hands, it can be done safely and efficiently. We have found that leaving the small 2-cm “button” of tissue near the LA appendage (which is subsequently cryoablated) greatly facilitates closure of the otherwise through and through box lesion around the PVs. Most centers seem to continue to perform the inferior alternatives to the standard Cox maze CS.

Many studies have examined the types of arrhythmias seen at catheter ablation following failed surgical maze procedures. None have directly compared the findings in patients undergoing CS to those undergoing a NCS maze procedures. In addition, many of the published studies are only reports of the patients with regular atrial tachycardias or flutters and do not include patients with AF recurring after surgery. Wazni et al. [14] examined 23 patients, all of whom underwent a classic CS maze procedure. Like our study, they found a high incidence of atrial tachycardias and flutters as the clinical failure rhythm. Sixty-five percent had atrial tachycardia or flutter, including 5 atrial tachycardias, 6 LA flutters, and 4 RA incisional flutters. In comparison to our study where we only had 10% with PV reconnection, they did have a 35% incidence of AF due to recovered PV conduction. At 1 year of follow-up, 82.6% were free of atrial arrhythmias off antiarrhythmic drugs after the ablation procedure. Magnano et al. [15] examined 20 patients who had only atrial tachycardia or flutter following a failed maze operation. They found 9 LA tachycardias, 6 typical and 1 atypical RA flutter, and 3 focal atrial tachycardias. They underwent a total of 26 electrophysiologic studies, and, at follow-up, 75% were in sinus rhythm. None of their patients underwent a CS maze, and a variety of energies including microwave, radiofrequency, cryoablation, and laser were used for the surgical maze. Since they excluded patients with AF as the failure rhythm after surgery, their findings could not be directly compared to those of our study. McElderry et al. [16] examined 22 patients who developed sustained atrial tachycardia after a surgical maze procedure. That study did not include any patients who received a pure CS procedure. The surgery was done using a combination of surgical incisions, radiofrequency ablation, and/or cryoablation. Half of the patients underwent a true CS box lesion around the veins and the other half had radiofrequency ablation. A total of 25 separate atrial tachycardias were mapped which included 15 in the RA and 10 the LA. Of the RA rhythms, 7 were isthmus-dependent atrial flutter, 7 non-isthmus-dependent flutters, and 1 focal atrial tachycardia. Seven LA tachycardias involved the roof and three involved the mitral annulus. All were ablated successfully, and they had no recurrences.

Our study has several limitations. The number of patients is small. None of the surgical maze procedures utilized electrophysiologic evaluation to document PV isolation or the integrity of the lines. The surgical procedures were done by many different surgeons at different institutions, and techniques may not have been fully standardized, especially for the NCS procedures. Our knowledge about the precise ablation protocol utilized for the NCS ablations was limited to what was in the operative notes and often did not include the number of applications or duration of ablation. However, they represent a “real-world” picture of what is being done by cardiac surgeons. Our findings could not be extrapolated to all patients undergoing a surgical maze as we only examined those patients referred for radiofrequency ablation after a failed surgical procedure.

After a failed maze operation, the CS surgical procedure isolated nearly all PVs and the NCS procedure never isolated all PVs. The clinical failure rhythm was more frequently AF for NCS and AFl for CS. If the primary goal of the surgical procedure is to eliminate AF by PV isolation, then the CS procedure would appear to be the “gold standard.” More careful surgical attention to the mitral and tricuspid isthmus lesion sets would probably reduce many of the late surgical failures which are commonly atrial flutters involving these anatomical areas.