OUP user menu

A randomized double-blind study of epicardial left atrial cryoablation for permanent atrial fibrillation in patients undergoing mitral valve surgery: the SWEDish Multicentre Atrial Fibrillation study (SWEDMAF)

Carina Blomström-Lundqvist, Birgitta Johansson, Eva Berglin, Leif Nilsson, Steen M. Jensen, Stefan Thelin, Anders Holmgren, Nils Edvardsson, Göran Källner, Per Blomström
DOI: http://dx.doi.org/10.1093/eurheartj/ehm378 2902-2908 First published online: 5 November 2007


Aims The efficacy of epicardial left atrial (LA) cryoablation in eliminating atrial fibrillation (AF) in patients undergoing mitral valve surgery (MVS) is unknown. We hypothesized that MVS combined with LA cryoablation is superior to MVS alone.

Methods and results Sixty-nine patients with permanent AF, included at four centres, underwent MVS with or without epicardial LA cryoablation. The primary endpoint was regained sinus rhythm. Risk factors for failed AF cryoablation were elucidated.

Sixty-five out of 69 patients reached the primary endpoint. At 6 and 12 months follow-up, 73.3% of patients who underwent cryoablation had regained sinus rhythm at both follow-ups, compared with 45.7 and 42.9% of patients, respectively, who underwent MVS alone (group differences, at 6 months P = 0.024, after 12 months P = 0.013). The in-hospital complication rate was 11.4% in the MVS group and 26.5% in the cryoablation group (P = 0.110). Risk factors for failed elimination of AF by cryoablation were duration of permanent AF (P = 0.012) and presence of coronary artery disease (P = 0.047), according to multiple logistic regression analysis.

Conclusion This first prospective randomized study showed that combining MVS with epicardial LA cryoablation is significantly better in eliminating pre-operative permanent AF than MVS alone.

  • Ablation
  • Fibrillation
  • Mitral valve
See page 2827 for the editorial comment on this article (doi:10.1093/eurheartj/ehm511)


Atrial fibrillation (AF) is associated with decreased quality of life (QoL) and substantial morbidity.1,2 Previous observations that AF will persist in the majority of patients after mitral valve surgery (MVS) have urged researchers to perform a concomitant Cox Maze III surgery or modifications of that procedure.3 Two small randomized studies suggested that the ‘cut and sew’ Maze procedure more effectively eliminates pre-operative AF compared to MVS alone,4,5 but the complexity of such surgical approach has limited a wide spread use of the technique.3,6 As a consequence, simplified surgical procedures using ablation lesions in place of surgical incisions have emerged. Non-randomized trials using either radiofrequency or cryoenergy for ablation lesions have reported comparable rates of freedom from AF as the ‘cut and sew’ Maze surgery79 whereas randomized trials1012 applying radiofrequency ablation lines endocardially reported outcomes varying from 44.4 to 80%. Although endocardial cryoablative surgery has successfully eliminated AF in 61 to 98% of patients,7,13,14 the lack of randomized trials and doubts concerning achieved transmural lesions,15 urged us to perform a multicentre, prospective, controlled, double-blinded, and randomized trial to clarify whether epicardial left atrial cryoablation combined with MVS would result in a better elimination of pre-operative permanent AF than MVS alone.



The study was conducted between November 2003 and May 2005 at four university centres in Sweden. Patients aged 18 to 80 years with permanent AF and mitral valve disease requiring MV surgery were randomly assigned to MVS combined with epicardial left atrial cryoablation or to MVS alone (controls). Permanent AF was defined as AF that had been present for at least 3 months with failed or not attempted cardioversion. The exclusion criteria were heart failure in New York Heart Association (NYHA) function class IV, previous cardiac surgery other than coronary artery bypass graft (CABG) surgery, planned MVS combined with other surgical procedures other than CABG, and tricuspid valvuloplasty, conditions that would impose an increased risk for prolonged surgical procedure, permanent pacemaker secondary to AV block, hyperthyroidism, geographical reasons, or unwillingness to participate. Patients, personnel, and all physicians (excluding the operating team) were blinded to the allocated surgery, which was recorded separately from the patient’s surgical notes.

Each patient signed a written informed consent. The study was approved by the institutional ethics and review committees.

Study design and investigations

The primary endpoint of the study was regained sinus rhythm without documented episodes of AF recurrence at 6 months after surgery. The secondary endpoints were maintained sinus rhythm after 12 months without recurrences of AF during the preceding 6 months, QoL, morbidity, and the incidence of predefined adverse events. Potential risk factors for failed cryoablation were analysed.

Demographics and clinical data, including NYHA function class, were prospectively recorded. The length of pre-operative AF history was obtained by patient medical history and files. The duration of permanent AF was verified by the earliest recorded 12-lead electrocardiogram (ECG) documenting AF. All 12-lead ECG recordings were reviewed by one experienced electrophysiologist for the differentiation between AF and atrial tachycardia.

Coronary artery disease (CAD) was defined as present if the patient had a history of previous myocardial infarction, coronary revascularization procedure, or significant coronary artery stenosis on pre-operative coronary angiography.

Two-dimensional and Doppler echocardiography was scheduled prior to surgery and at 6 months of follow-up to evaluate left atrial dimensions, left ventricular ejection fraction (LVEF), and cardiac valve function. Effects on QOL and detailed echocardiographic data are reported elsewhere.

Surgical procedure

All cryoablation lines were applied epicardially during cardiopulmonary bypass on the beating heart before the aortic cross-clamp (AOC) and cardioplegic arrest. Cryoablation lesions of 90 s duration were applied in an overlapping fashion around each pulmonary vein (PV) pairs with three extra linear lesions; one between the right and left PV pairs, one connecting the left lower PV to the mitral annulus, and one connecting the left upper PV to the rim of the left atrial appendage (Figure 1). The cryosurgical probe had a variable freezing segment (4–60 mm long) and an integrated thermocouple for temperature monitoring, capable of reaching a temperature of −160°C using an argon-based cooling system (SurgiFrostTM CryoAblation System, CryoCath Technologies Inc., Quebec, Canada). Lesion lines were not assessed for conduction block in order to not prolong the surgical procedure.

Figure 1

Schematic drawing of the left atrium in a postero-anterior view with the cryoablation lesions (hatched lines) and the suture lines. Abbreviations: IVC, inferior vena cava; LAA, left atrial appendage; SVC, superior vena cava.

After cold cardioplegic arrest, the left atrium was opened through a left paraseptal vertical incision unless a transseptal approach was preferable. After the mitral valve had been repaired or removed, the left atrial appendage was excluded with a purse string suture placed endocardially in all patients. The valve procedures were performed according to routine. After completion of the surgery and weaning from bypass epicardial electrical cardioversion was performed in patients if AF was present. The surgical procedure, apart from the cryoablation, was identical for patients allocated to MVS only.

Post-operative care and follow-up

All patients were monitored by telemetry while in hospital. All antiarrhythmic drugs were discontinued from the day of surgery. Beta-blocking agents were administered from the day of surgery to prevent post-operative AF. Post-operative AF was treated with sotalol or amiodarone infusion. Electrical cardioversion was recommended on at least two occasions in case AF recurred during the hospital stay. Patients with AF at discharge were scheduled for a last attempt of electrical cardioversion 1 month after surgery. Prophylactic antiarrhythmic drugs (sotalol, flecainide, propafenone, disopyramide, or amiodarone as the last resort), administered to patients with post-operative AF that required cardioversion, were continued for the first 3 months after surgery and then withdrawn in the absence of AF recurrence. Patients with AF recurrence after discharge were advised to undergo repeated electrical cardioversions on at least three different occasions followed by oral prophylactic antiarrhythmic drugs as needed. Warfarin was advised from the day of surgery and for at least 3 months or longer if patients had mechanical valve prosthesis or recurrence of AF.

After hospital discharge, the patients were evaluated at 1, 2, 3, 6, and 12 months after surgery by medical history, clinical examination, 12-lead ECG, medications, previous cardioversions, and adverse events. The patient’s heart rhythm and type of AF were defined 6 and 12 months after surgery. The type of AF was defined as paroxysmal, persistent, or permanent based on subjective symptoms, ECG recordings, and cardioversions performed in between or at follow-up visits. Continuous rhythm monitoring was not performed.

Statistical analysis

Sinus rhythm was assumed to be present at 6 months follow-up in 60% of patient after MVS combined with left atrial cryoablation. The estimated minimum number of patients required to show a difference of 40% with a statistical power of 90% (alpha value of 0.05), was 60 patients (30 in each arm). The type-I error rate was controlled for by using a P-value less than 0.05 or 95% confidence intervals. Fifteen patients were added to compensate for possible drop outs; the final sample therefore comprised 75 patients. The results were analysed on an intention-to-treat basis.

Block randomization stratified by hospital ensured a balanced number of cryoablation procedures for each hospital. Related to difficulties in recruiting patients in one hospital, the allocated block was redistributed to the two hospitals with the largest number of included patients. The number of included patients at each centre were: Uppsala: 28 patients, Gothenburg 26 patients, Umeå 12 patients, and Stockholm 3 patients.

Continuous variables are reported as mean ± one standard deviation and were, when appropriate, compared with the use of Student’s t-test, or otherwise Mann-Whitney’s U test. Logistic regression models were used to assess the influence of various continuous and categorical variables on the ability to maintain sinus rhythm. The number of considered predictors was, to begin with, 5. The best subsets were used to obtain the final model. Categorical variables were compared using Fisher’s exact test except for NYHA class, when Pearson χ2 analysis was used. All P-values were two sided, and a P-value less than 0.05 was regarded as statistically significant.


A total of 71 patients were randomized. Two patients randomized to the combined cryoablation procedure were excluded related to defect cryoablation system at surgery and a left-sided superior vena cava precluding left atrial ablation, respectively. Combined left atrial cryoablation and MVS was performed in 34 patients and MVS alone in 35 patients. Four of these patients could not be evaluated at follow-up because of death in ovarian cancer, death in heart failure peri-operatively, development of a senile psychotic state, and patient refusal. The analysis for primary and secondary events is based on the 65 patients who reached the primary endpoint, although all patients included were evaluated for adverse events.

The two patient groups did not differ in terms of their baseline characteristics and pre-operative medications were similar except for the warfarin treatment (Table 1).

View this table:
Table 1

Baseline characteristics

CharacteristicsCryoablation + MVS (n = 30)MVS alone (n = 35)
Males, n (%)25 (83.3)26 (74.3)
Age (years)69.5 ± 7.965.6 ± 8.8
AF history (months)49 ± 57 (3–240)69 ± 74 (3–264)
AF duration (ECG verified) (months)26 ± 33 (3–120)33 ± 54 (3–240)
Previous electrical cardioversion, n (%)15 (50)17 (48.6)
History of embolic stroke/TIA, n (%)1 (3.3)3 (8.6)
Left atrial diameter (cm)6.1 ± 1.1 (4.1–9.4)5.8 ± 0.7 (4.3–7.5)
LVEF (%)53.6 ± 9.1 (29–67)57 ± 12 (20–77)
LVEF <40%, n (%)5 (16.7)2 (5.7)
Pre-operative heart failure, n (%)8 (26.7)3 (8.6)
NYHA function class, n (%)
 I2 (6.7)1 (2.9)
 II8 (26.7)10 (28.6)
 III20 (66.7)24 (68.6)
Coronary artery disease6 (20.0%)9 (25.7)
Hypertension, n (%)9 (30.0)11 (31.4)
Diabetes mellitus, n (%)2 (6.7)6 (17.1)
Lung disease, n (%)2 (6.7)6 (17.1)
Medication, n (%)
 Antiarrhythmic drugs03 (8.6)
 Beta-blocking agents25 (83.3)22 (62.9)
 Verapamil or diltiazem1 (3.3)2 (5.7)
 Digitalis17 (56.7)23 (65.7)
 ACE inhibitors or ARB20 (66.7)25 (71.4)
 Diuretics21 (70.0)19 (54.3)
 Warfarin30 (100)30 (85.7)
 Aspirin or other anticoagulants2 (6.7)5 (14.3)
  • Figures are mean ± 1 SD with ranges in brackets, unless otherwise stated. n indicates number of patients, ACE, angiotensin converting enzyme; ARB, angiotensin receptor blockers; antiarrhythmic drugs (Vaughan Williams class I or III) for prevention of AF.

There were no evidence for a difference between the two groups with regard to the number and types of operations (Table 2). All patients underwent MVS for mitral valve regurgitation except one patient in the cryoablation group who had mitral valve stenosis. The additional extra-corporal circulation (ECC) time in the cryoablation group was 27.4 min.

View this table:
Table 2

Operative data

VariableCryoablation + MVS (n = 30)MVS alone (n = 35)P-value
Mitral valve replacement, n (%)9 (30.0)7 (20.0)0.351
Mitral valve repair, n (%)21 (70.0)28 (80.0)0.351
Concomitant surgery
 CABG, n (%)6 (20)5 (14.3)0.742
 TVAP, n (%)1 (3.3)3 (8.6)0.618
 Closure of ASD, n (%)1 (3.3)1 (2.9)0.912
Interatrial access, n (%)23 (76.7)20 (57.1)0.097
ECC (min)146.6 ± 27.9 (95–215)119.2 ± 33.0 (73–193)0.00052
AOC (min)87.4 ± 95.2 (53–181)84.4 ± 23.3 (52–142)0.649
Cryoablation duration (min)21.6 ± 5.6 (14–37)
Hospital stay, median days8.0 ± 7.0 (4–44)8.0 ± 3.9 (4–22)0.479
  • Figures are mean ± 1 SD with ranges in brackets, unless otherwise stated. n indicates number of patients, TVAP, tricuspid valve annuloplasty; ASD, atrial septal defect.

The rate of in-hospital complications, excluding pacemaker implantations, was 11.4% (4 patients) in the MVS group and 26.5% (9 patients) in the cryoablation group (P = 0.110) (Table 3). Low cardiac output syndrome occurred while weaning from bypass in three patients, was related to right heart failure and pre-operative myocardial infarction in two patients, respectively, and required temporary assist by intra-aortic balloon pump in three patients. All except one patient, who died on the third post-operative day, recovered. The in-hospital mortality rate was 2.9% (1/34 patients) in the cryoablation group and zero in the MVS alone group. The permanent pacemaker implantation rate did not differ between the two treatment groups (P = 0.583) (Table 3).

View this table:
Table 3

In-hospital and late complications

ParametersCryoablation + MVS (n = 34)MVS alone (n = 35)
Low cardiac output syndrome411.80
Bleeding requiring re-exploration25.925.7
Intraoperative bleeding (LAA perforation)12.90
Congestive heart failure, pleural effusion12.90
Myocardial infarction and VT12.90
Paralytic ileus12.90
Sternal wound skin injury012.9
Post-operative permanent pacemaker723.3411.4
 AV block I, II, or III25.938.6
 Sinus bradycardia/AF with slow rate514.712.9
Operative mortality12.90
Late complications
Mitral valve prosthesis thrombosis12.90
Congestive heart failure12.939.0
Pleural effusion12.925.7
Tamponade, pericardiocentesis012.9
Sternal wound infection12.90
Myocardial infarction012.9
Atrial tachycardia012.9
  • n indicates number of patients, LAA, left atrial appendage; VT, ventricular tachycardia.

Follow up

Twenty-two of 30 patients (73.3%) who underwent combined left atrial cryoablation and MVS, maintained sinus rhythm at 6 month follow-up, compared to 16 of 35 patients (45.7%), who underwent MVS alone (P = 0.024). After 12 months, 22/30 (73.3%) in the cryoablation group and 15/35 (42.9%) patients in the MVS group were still in sinus rhythm and no patient had required cardioversion during the preceding 6 months. Of the patients in sinus rhythm, 72.7% of patients in the cryoablation group and 53.3% of patients in the MVS group were without antiarrhythmic drug therapy at 12vmonths (Table 4).

View this table:
Table 4

Post-operative medication at discharge and 12 months follow-up

Cryoablation + MVS (n = 30)MVS alone (n = 35)P-value
At discharge, n (%)
 Antiarrhythmic drugs15 (42.9)16 (53.3)0.730
 Beta-blocking agents7 (23.3)6 (17.1)0.534
 Verapamil or diltiazem1 (3.3)1 (2.9)1.000
 Digitalis3 (10.0)9 (25.7)0.104
 ACE inhibitors14 (46.7)21 (60.0)0.282
 Diuretics21 (70.0)15 (42.9)0.028
 Warfarin26 (86.7)30 (85.7)0.912
 Aspirin, other anticoagulants0 (0)4 (11.4)0.056
At 12 months FU, n (%)
 Antiarrhythmic drugs8 (26.7)11 (31.4)0.787
 Beta-blocking agents14 (46.7)19 (54.3)0.540
 Verapamil or diltiazem1 (3.3)1 (2.9)1.000
 Digitalis8 (26.7)4 (11.4)0.199
 ACE inhibitors or ARB19 (63.3)22 (62.9)0.968
 Diuretics21 (70.0)17 (48.6)0.081
 Warfarin18 (60.0)27 (77.1)0.135
 Aspirin, other anticoagulants8 (26.7)2 (5.7)0.020
  • Figures are number of patients with percentage in brackets. Abbreviations as in Tables 1.

Late complications, occurring in six patients (17.1%) in the MVS group and in eight patients (23.5%) in the cryoablation group, did not differ between the two groups (P = 0.510) (Table 3). Transient ischaemic attacks (TIA) occurred in four cryoablated patients within 3 months after discharge; two patients, who were off warfarin at discharge, had risk factors for stroke and recurrence of AF episodes, one patient on warfarin had risk factors for stroke including prior embolism, and another patient suffered from TIA in relation to thrombolysis of mitral valve prosthesis thrombosis. In the MVS group, one patient on warfarin experienced a small embolic stroke in relation to myocardial infarction despite normal pre-operative coronary angiography, and another patient suffered from a small pons bleeding. One patient developed atrial tachycardia resulting in heart failure, which both resolved after medical treatment. All atrial tachycardias had resolved at 12 months follow-up. One patient died from ovarian cancer within 3 months after surgery.

Multiple logistic regression analysis identified longer duration of ECG verified pre-operative permanent AF (P = 0,012) and presence of CAD (P = 0,047) as independent risk factors for failed cryoablation in eliminating AF at 12 months follow-up (Figure 2).

Figure 2

Plot of predicted means from a multiple logistic regression analysis showing that sinus rhythm could only be achieved in 50% of patients with CAD compared with 80% of patients without CAD after concomitant MVS and cryoablation. The efficacy of cryoablation was, however, of the same magnitude in both patient groups. The vertical bars indicate 95% confidence intervals and the scaling on the y-axis is the probability of sinus rhythm.

For patients with a permanent AF duration equal to or exceeding 40 months, the probability of maintaining sinus rhythm at 12 months after cryoablation was less than 50%. None of the other variables tested outlined in Table 1 were statistically significant predictors for failed elimination of AF by combined surgical procedure.


This is the first prospective, randomized, multicentre, controlled study to demonstrate that MVS combined with epicardial left atrial cryoablation is more efficacious than MVS alone, in restoring and maintaining sinus rhythm in patients with permanent AF.

Recent meta-analysis reported that Maze surgery was more effective in restoring sinus rhythm compared to MVS alone (86.7 vs. 17.3%)6 and that the post-operative rate of regained sinus rhythm was higher with the ‘cut and sew’ Cox-Maze III surgery compared with alternate energy sources, 84.9 vs. 78.3%, respectively.3,16 A higher proportion of paroxysmal and lone AF and younger age was suggested to account for the higher success rate in patients who underwent the ‘cut and sew’ Maze surgery.16 The paucity of large randomized trials using a concurrent control arm has, however, made it difficult to discern the absolute effect of Maze procedures, and to what extent MVS alone may eliminate AF.

In the five randomized trials published, the absolute reduction in AF prevalence with the ‘cut and sew’ Cox Maze III surgery or Maze-like intraoperative ablation procedure during MVS vs. MVS alone4,5,1012 have varied from 30 to 53.3% at 12 months. The efficacy rate in eliminating AF in our study (30.4%) is comparable to the rates (30 and 36%) in the two studies using the biatrial ‘cut and sew’ Maze III approach,4,5 thus supporting previous meta-analysis that there is no difference in efficacy rates between the classical ‘cut and sew’ Maze III surgery and ablation procedures with alternative energy sources.3 Direct comparison of our outcome with those of the ‘cut and sew’ Maze III surgery is, however, difficult as their patient series were small and had a higher proportion of patients with paroxysmal AF (50%), shorter AF duration, and lower prevalence of concomitant cardiovascular diseases,4,5 which may explain a higher sinus conversion rate after both combined Maze surgery (80 and 92% of patients, respectively) and MVS alone (50 and 56%, respectively) at 12 months.4,5

In the two randomized studies using biatrial radiofrequency ablation during MVS,11,12 the absolute reduction in AF prevalence was higher (52.5 and 53.3%, respectively) than in our study (30.4%) at 12 months. Whether the higher efficacy was related to the biatrial approach is purely speculative, since such high efficacy rates were not observed in the ‘cut and sew’ Maze surgery trials, which also used a biatrial approach.4,5 The lower rates of regained sinus rhythm (26.7 and 26.9%, respectively) observed after MVS alone in these studies11,12 are difficult to explain since the proportion of patients with underlying heart disease seemed lower. None of the two studies seemed blinded like ours and the routine use of amiodarone for at least 3–6 months in one study may have influenced the different outcomes.12

Another study reported freedom from AF in only 44.4% of patients undergoing left atrial radiofrequency ablation during MVS, and in only 4.5% of the controls,10 which are lower rates than in our study (73.3 and 42.9%, respectively). The efficacy rate of nearly 40% was slightly higher than in our study. The remarkably low rate of regained sinus rhythm in that study,10 may partly be related to their less intensive post-operative treatment strategy of AF. In our study, all patients with post-operative AF underwent, in addition to pharmacological treatment, repeated electrical cardioversions up to 3 months after surgery per protocol, a strategy not adopted by Doukas et al.10 A more intensive post-operative treatment of AF and an earlier time for valve interventions in relation to disease progression, as in our study, may in part account for the higher rate of return to sinus rhythm after MVS.

Since AF burden was not analysed, episodes of asymptomatic paroxysmal AF could not be excluded in the present study. Apart from the lack of continuous monitoring of rhythm by event recorders, which was not yet available, our study adhered to recently issued guidelines for reporting data on surgical treatment of AF.17

Direct comparison of our findings with the other randomized studies is difficult related to the different tools and lesions sets used.1012 The use of radiofrequency energy or point by point cryoablation aiming for continuous linear lesions may carry a higher risk for conduction gaps than the use of a cryo-probe creating continuous freezing segments. Transmurality of radiofrequency ablation lesions could only be found in 75% of intraoperatively applied lesions in post-mortem studies.18 Non-randomized studies using point by point cryoablation intraoperatively, reported conduction gaps in 35% of patients according to electro-anatomic mapping performed after surgery.15 All previous randomized studies applied lesions from the endocardium. Epicardial, beating heart cryoablation as used in our study may be limited by heat from intracardiac blood flow, which is supported by experimental studies showing a higher rate of transmurality with cryoclamp lesions compared with linear ablations using a malleable probe (93 and 84%, respectively).19

The higher but non-significant overall in-hospital complication rate in the ablation group vs. the MVS group may be of clinical relevance and should be interpreted with caution as the left atrial appendage perforation and low cardiac output syndrome seemed connected to the ablation procedure. The observation that TIA developed in cryoablated patients who were off warfarin at discharge further emphasizes the importance of continuous anti-thrombotic therapy even if sinus rhythm has been achieved.

A novel finding in our study was that CAD imposed an increased risk for failed cryoablation. Still, cryoablation was as effective in eliminating AF even in these patients, who without such procedure would be expected to regain sinus rhythm in less than 20% of cases. One may speculate whether the lower sinus conversion rate in patients with CAD is related to atrial scar formations precluding transmural lesions, which warrants larger studies of patients with CAD. Our observation that long AF duration is a predictor of failed elimination of AF has been reported by others,20,21 whereas we could not confirm pre-operative large left atrial diameters and older age to be predictors.9,20,21

This first prospective randomized study showed that combining MVS with epicardial left atrial cryoablation is significantly better in eliminating pre-operative permanent AF than MVS alone. It clearly demonstrated that a high proportion of patients may regain and maintain sinus rhythm despite a long history of permanent AF.

Although our findings suggest that routine use of epicardial left atrial cryoablation is justified during MVS in patients with permanent AF, larger studies are needed to confirm its safety and efficacy on a long-term basis.


The Swedish Heart-Lung Foundation, who also provided funding to pay the Open Access publication charges for this article.

Conflict of interest: none declared.


Nurses and technicians contributing with data handling were Annika Persson and Anna Lindberg, Uppsala and Monica Dahlin, Gothenburg.


View Abstract