OUP user menu

Minimal invasive surgery for atrial fibrillation: an updated review

Mark La Meir, Sandro Gelsomino, Fabiana Lucà, Laurent Pison, Andrea Colella, Roberto Lorusso, Elena Crudeli, Gian Franco Gensini, Harry G. Crijns, Jos Maessen
DOI: http://dx.doi.org/10.1093/europace/eus216 170-182 First published online: 10 July 2012


Aims Despite its proven efficacy, the Cox-Maze III procedure did not gain widespread acceptance for the treatment of stand-alone atrial fibrillation (SA-AF) because of its complexity and technical difficulty. Surgical ablation for SA-AF can now be successfully performed utilizing minimally invasive surgery (MIS). This study provides an overview of state-of-the-art MIS for the treatment of SA-AF.

Methods and results Studies selected for this review were identified on PUBMED and exclusion and inclusion criteria were applied to select the publication to be screened. Twenty-eight studies were included; 27 (96.4%) were observational in nature whereas 1 was prospective non-randomized. The total number of patients was 1051 (range 14–114). Mean age ranged from 45.3 to 67.1 years. Suboptimal results were obtained when employing microwave and high focused ultrasound energies. In contrast, MIS ablation of SA-AF achieved satisfactory 1-year results when the bipolar radiofrequency was employed as energy source, with antiarrhythmic drug-free success rate comparable to percutaneous catheter ablation (PCA). The success rate in paroxysmal was even higher than in PCA. In contrast, ganglionated plexi ablation and left atrial appendage removal seem not to influence the recurrence of AF and the occurrence of postoperative thromboembolic events.

Conclusion Minimally invasive surgery ablation of SA-AF achieved satisfactory 1-year results when the bipolar radiofrequency was employed. Nevertheless, the relatively high complication rate reported suggest that such techniques require further refinement. Finally, the preliminary results of the hybrid approach are promising but they need to be confirmed.

  • Atrial fibrillation
  • Ablation
  • Surgery


Over the last 20 years, surgeons and electrophysiologists (EPs) have been making progress towards finding a cure for atrial fibrillation (AF) that offers patients an acceptable risk/benefit profile.

The Cox-maze III procedure has been shown to be an effective, but invasive and complex procedure and, for this reason, it has not been widely accepted as a stand-alone therapy.1,2

Transvenous pulmonary vein isolation (PVI) has become the cornerstone of catheter ablation for AF and is currently the therapy of first choice if patients warrant rhythm control and fail antiarrhythmic drugs (AADs).3,4 Pulmonary vein isolation is reported to be effective in 60–85% of the patients, especially in patients with paroxysmal short-lasting episodes of AF.35 Nevertheless, in about one-third of patients, a second ablation procedure is necessary and complications may occur in up to 6% of patients.6,7

The surgical treatment of atrial fibrillation has undergone dramatic changes over the last decade. New technologies have allowed the creation of transmural lesions on a beating heart through alternative, less-invasive incisions.8 The introduction of these technologies has greatly simplified the performance of the Maze procedure and allowed for the development of minimally invasive surgery (MIS).8 Minimally invasive video-assisted techniques allow pulmonary vein island creation and left atrial appendage (LAA) removal or exclusion, usually with ganglionic plexi evaluation and destruction.9

Nevertheless, it is still uncertain whether these procedures are effective in restoring permanent sinus rhythm.10 We present a systematic literature overview of MIS for the treatment of stand-alone atrial fibrillation (SA-AF) and briefly discuss the hybrid approach which combines, in one-step, off pump epicardial ablation with PVI.

Review criteria

Studies selected for this review were identified on PUBMED (last search 30 September 2011) using the following search terms: ‘stand alone atrial fibrillation’, ‘Mini-Maze’, ‘video-assisted’, ‘epicardial ablation’, ‘endoscopic ablation’, ‘videothoracoscopy’, ‘minimally invasive’, ‘surgical ablation’, ‘bilateral thoracoscopic surgical ablation’, ‘right thoracoscopic surgical ablation’, ‘left atrial appendage exclusion’, and ‘ganglionic plexi ablation’.

Exclusion criteria were animal studies, reviews, case reports, concomitant surgery, not lone atrial fibrillation, not MIS, not performed on the beating heart, not English, no full-text availability. These criteria were chosen to make a selection based on title and/or abstract. Inclusion criteria were studies with >10 patients, follow-up of >3 months, minimal invasive beating heart surgery and surgery for lone atrial fibrillation.

The search returned a total of 631 papers. A total of 40 articles were selected. Ten met inclusion criteria but were excluded because they had <10 patients.1120 Two were excluded because ablations were performed on a cardiopulmonary bypass.21,22 The final selection included 28 studies.2350


In recent years, it has been hypothesized that SA-AF and pathological AF are different disease processes with contrasting risk predictors.51 The 2006 ACC/AHA/ESC guidelines stated that there was no standard definition for lone AF.52 The guidelines applied the term to ‘… individuals younger than 60 years without clinical or echocardiographic evidence of cardiopulmonary disease, including hypertension’.1,21 A recent international consensus on nomenclature and classification of AF mentions that only AF in the absence of heart disease is termed ‘lone’ whereas in the absence of any disease it is termed ‘idiopathic’.53 Indeed, lone AF does not necessarily mean ‘idiopathic’. By convention, the term ‘non-valvular AF’ is restricted to cases in which the rhythm disturbance occurs in the absence of rheumatic mitral valve disease, a prosthetic heart valve, or mitral valve repair.1

However, the reviewed publications were appraised employing a definition of AF as a stand-alone procedure for treating AF without concomitant surgery.

Baseline characteristics

An overview of the selected studies is presented in Table 1. Studies were published between 2004 and 2011, and 27 of 28 (96.4%) studies were observational in nature whereas one study42 was prospective non-randomized. Twenty-six studies (89.6%) were performed in a single centre, one study included five centers,36 one study included three North American Institutions,35 and one study prospectively enrolled patients from two different centres.42 Three studies of the groups of Edgerton et al.,26,31,36 Pruitt et al.,25,29 Sirak et al.,32,43 and Cui et al.41,50 may report on overlapping patients, thus, when discussing results, we refer to their most recent publications.

View this table:
Table 1

Minimally invasive surgery for lone atrial fibrillation: baseline characteristics

First authorYearU/MNumber of patientsAge (years)AF duration (years)LA diameter (mm)PCAPXPRLSPPM
Salenger232004U1460 ± 46.947.3 ± 7.311013
Wolf242005U2957.2 ± 14.97.5a21845
Pruitt252006U5059 ± 106.1b11cccc
McClelland272007U2059 ± 81.444 ± 3ns1154
Koistinen282007U2245.3 ± 10.26.642.9 ± 5.201012
Pruitt292007U10060 Embedded Image 96.0fns641125
Sagbas302007U2656 ± 112.842.2 ± 6.36818
Sirak322008U3261 ± 107.748.2 ± 8.2ns0329
Wudel332008U2262 ± 93.734.7 ± 11.57148
Bagge342009U4258 [36–71]745 [26–62]182869
Beyer352009M (3)10065 Embedded Image 114.943 ± 6NS392932
Edgerton362009M (5)11459.5 ± 10.6>1 (n = 101)47.2 ± 8.324603222
Edgerton372009U3058 ± 9>152NS1020
Han382009U4564 ± 8.76.243 ± 6123312
Klinkemberg392009U1547 ± 105.038 ± 5NS114
Castella402010U3454 ± ± 5.634g17125
Cui412010U8158 ± 105.949.7 ± 11.44491715
Edgerton422010M (2)5260 [42–79]>148 [3.7–6.0]1052
Sirak432010U4860 ± 64.345 [34–63]1348
Speziale442010U4654.9NS42 ± 5141927
Yilmaz462010U3055.6 ± 8.66.6ns181983
Krul472011U2757 ± 70.647 ± 7141632
La Meir482011U2867.1 ± ± 5.7211459
Nasso492011U10463.9 ± 8NS21.3 ± 5h24104i104i
Wang502011U83j57 ± 115.152.5 ± 11.5NS83
  • Studies were presented by year of publication and in alphabetical order.

  • Age expressed as mean ± SD or (range) and LA diameter expressed as mean ± SD or median (interquartile range).

  • U/M, unicentric/multicentric study; PX, paroxysmal atrial fibrillation; PR: persistent atrial fibrillation; PM, permanent atrial fibrillation; NS, not specified.

  • a< 60 mm (n = 26) >60 mm (n = 1).

  • b<46 mm (n = 23), 46–59 mm (n = 24), ≥60 mm (n = 3).

  • cAF classified as intermittent (n = 33) and continuous (n = 17).

  • d<6 m (n = 3), 6–12 m (n = 5), >12 m (n = 73), unknown (n = 2).

  • ePersistent and Long-standing persistent reported together.

  • f<46 mm (n = 42), 46–59 mm (n = 53), ≥60 mm (n = 5).

  • gAll patients had PCA by definition.

  • hIndexed by BSA.

  • iParoxysmal and persistent reported together.

  • jForty-one patients treated with irbesartan.

The total number of patients was 1051 (range 14–114). Mean age ranged from 45.3 to 67.1 years. Only two studies had a population with mean age of <50 years.28,39 Two studies31,45 did not report the population's age. Atrial fibrillation duration ranged from 0.6 to 12.1 years. Three papers did not report AF duration44,45,49 and one paper26 classified patients on the basis of AF duration: <6 months (n = 3), 6–12 months (n = 5), >12 months (n = 73), and unknown (n = 2).

Only eight papers (34.8%) distinguished long-standing persistent (LSP) AF as requested by recent guidelines54 and in two studies43,50 100% of patients had LSP-AF. A total of 443 patients (42.1%) with paroxysmal AF underwent MIS, 189 (18.0%) had persistent AF, 123 (11.8%) LSP, and 95 (9.0%) permanent, whereas 97 (9.2%) were not clearly defined and one paper reported paroxysmal and persistent AF together (n = 104, 9.9%).49

Finally, 218 patients (20.7%) had one or more previous percutaneous catheter ablation (PCA). In one paper40 all patients had at least one PCA before surgery.


Video-assisted thoracic surgery was employed in all studies (Table 2). In their 2006 paper Pruitt et al.25 performed a robotic-assisted pericardial ablation in 3 of 50 (6%) of patients. Edgerton et al.,26,31,36,37,42 McClelland et al.,27 Beyer et al.,35 and Stamou et al.45 carried out a small thoracothomy (4–6 cm) to introduce the endoscopic instruments into the thoracic cavity.

View this table:
Table 2

Minimally invasive surgery for lone atrial fibrillation: surgery

First authorSourceMethodAccessGPLAAAL
Salenger23MWVATSBNY (9)Y
Pruitt25MWVATS (47)/Rob (3)BNY(46)Y
Edgerton26RF(b)IThoracotomy /VATSBYYY
Mc Clelland27RF(b)IThoracotomy /VATSBYNN
Sagbas30RF(b)IVATSBNY (16)N
Beyer35RF(b)I NIThoracotomy /VATSBYYN
Edgerton36RF(b)IThoracotomy /VATSBYY(101)N
Edgerton37RF(b)IThoracotomy /VATSBYYY
Castella40RF(b) NIVATSBYYN
Cui41RF(b) NIVATSBNY (71)N
Edgerton42RF(b)IThoracotomy /VATSBYY(44)N
Stamou45RF(b) I NIThoracotomyBYYN
La Meir48RF(b)IVATSBYY (14)Y
  • Studies were presented by year of publication and in alphabetical order.

  • AL, additional lines; RF, radiofrequency; b, bipolar; m, monopolar; I, irrigated; NI, not irrigated; VATS, video-assisted thoracoscopic surgery; Rob, robotic-assisted surgery. Y/N, Yes/Not

  • aDuring reoperation for bleeding a tricuspid vena cava inferior isthmus ablation was performed in one patient.

  • bLAA was excised if it was deemed to be significantly enlarged or if the patient had suffered from a thromboembolic event.

Three studies39,44,49 employed a right monolateral approach where in the others a bilateral access was performed. These authors used monopolar radiofrequency44,49 or high-intensity focused ultrasound (HIFU). Four studies23,25,28,29 employed microwave (MW), whereas in the majority of them the energy source was bipolar radiofrequency. Excluding the three right thoracotomy accesses, patients in all studies but one26 underwent excision/ligation of the LAA. In 15 studies LAA was stapled/legated in all patients. In four studies23,25,30,36 a variable percentage of patients (16%, 8%, 39% and 11%, respectively) did not have LAA stapler/ligation for unexplained reasons. Only a few groups clearly stated the indications for LAA removal: La Meir et al.48 performed LAA stapler/clip in patients with CHADS2 score ≥1, in the presence of rapid firing coming from the LAA and when the procedure was deemed safe (50%). Cui et al.41 did not perform LAA excision in the first 10 patients (12.4%) because of the lack of the appropriate endoscopic equipment and Edgerton et al.42 did not remove LAA in eight patients (15%) because the procedure was deemed hazardous.

Seventeen groups (60.7%) carried out ganglionated plexi (GP) ablation. Additional left atrial lesions are reported in 10 studies. They consisted of a roof line connecting superior pulmonary veins,26,31,37,47,48 an inferior line connecting inferior pulmonary veins,47,48 a connecting line between the superior line (or the box) and the left fibrous trigone,26,31,32,37,43,47 a connecting line from the superior pulmonary vein and the LAA23,25,26,28,29,31,32,37,43 and a line from the right inferior pulmonary vein to the coronary sinus (CS).32,43

Additional right lesions are reported in four papers and indications were persistent/long-lasting persistent AF with LA volume >58 mL,48 permanent/persistent AF,23 continuous AF25 or it was not specified.29 Lesion consisted of a superior vena cava (SVC) to inferior vena cava (IVC) line,23,25,29,48 a line up to the tip of the right atrial appendage,23,25,29 or a line towards the CS.23

La Meir et al.48 added an SVC circumferential isolation in three cases (to be sure that the SVC-IVC line would not stop at an area of conduction (in patients with a small portion of intrapericardial IVC). These investigators also added a circumferential SVC isolation in case of a rapid firing coming from the SVC (n = 7.1). Klinkenberg et al.39 added a tricuspid vena cava isthmus line ablation in one patient during reoperation for bleeding.

Electrophysiological testing of the lesions was carried out only by Krul et al.47 In other groups entry and exit block was checked by baseline and post-isolation sensing and pacing.26,3138,4042,45,46,4850 Speziale et al.44 measured the conductance across the lesion and a value of <50% than predicted defined an adequate lesion. Finally, McClelland et al.27 performed ablations guided by epicardial mapping.

Follow-up, rhythm monitoring, anticoagulation, and antiarrhythmic therapy

One thousand onehundred ninety-nine patients (87.4%) reached the maximum follow-up which was 100% complete in 13 papers (46.4%). In the others the follow-up completeness ranged from 32% to 95% (Table 3). The longest follow-up was 36 months48 and nine studies reported a follow-up of <12 months which is the minimum recommended for evaluating procedural efficacy.4

View this table:
Table 3

Minimally invasive surgery for lone atrial fibrillation: End-points and rhythm monitoring

First authorFollow-up (months)End-pointN (%)Follow-up typeRhythm monitoringAAD (percentage of patients taking)OA (percentage of patients taking)
Salenger2312114 (100)OCNSNSNS
Wolf2246123 (79)OC-MR-TI-CV (3-5-6 M)ECG (10) OT (13)9NS
Pruitt257.6244 (88)OC-TI (NS)ECG EPS2050
Edgerton266357 (69)OC (1-3-6 M)ECG, 14–21-day event recorder (24)24 HM (24), Pm (9)34NS
McClelland2717 ± 3420 (100)OC (1-2-3-6-W 3-6 M-every 6 M)30-day HM (12 M)a25NS
Koistinen2811.5 ± 62–522 (100)OC (1-3-6 M)24 HM (6 M)20
Pruitt2923.1688 (88)OR-MR-EPS (NS) ECG-HM (NS)1154
Sagbas308 ± 3226 (100)OC (3–6 M)OT; 24 HM (NS)1954
Edgerton316366 (89)0C (1-3-6 M)ECG; 14–21-day event recorder (NS) 24 HM (NS)47NS
Sirak326119 (40)OC7-day event recorder (3-6-13-24 M)NSNS
Wudel3318 ± 4222 (100)OC (3-6-12 M)24 HM (NS)99
Bagge3412141 (78)OC (3-6-12-M)24–48 HM (NS)3224
Beyer3513 ± 87100 (100)OC (3 M to every year)24 HM (3 M)3736
Edgerton3617 ± 3.41114 (100)OC (1-3-6 M)24 HM (26) 14- to 21-day event recorder (68) Pm (15)43NS
Edgerton376830 (100)OC (1-3-6 M)ECG-event recorder Pm5347
Han3812943 (95)OC (6-12 M to yearly)30-day event recorder (6 M) TTM (12 M)3533
Klinkemberg391.3 ± 0.6 years1015 (100)OC (2-6–12–24 M)ECG 24 HM (2 M) 48 HM (6 M)40NS
Castella4016 ± 111111 (32.3)OC (1-4-6-12 M)ECG 24 HM (4-6 M to 1 Y)NS63
Cui41121149 (60)OC (1-3-6-12 M)ECG, 24-48 HM, Pm ,UCGNSNS
Edgerton4212852 (100)OC (1-3-6-12 M)24 HM 14- to 21-day event recorder Pm1119
Sirak436724 (75)OC (3-6-13 M)OT12NS
Speziale449.3 ± 3.2346 (100)OC (3-6-12 M)ECG 24 HM (6–12 M)NSNS
Stamou4512NS12 (60)NSNSNSNS
Yilmaz4611 ± 4730 (100)OC(1-3-6-W 3-6-12 M)ECG 24 HM (3-6-12 M)3552
Krul471212-922 (71)OC (3-6-12-15-18-24 M)ECG 24 HM (every 3 M)1448
La Meir48361324 (85.7)OC (2-W 1-6-12 M to yearly)ECG 7-day HM (2-W 1-6-12 M to yearly)11.315.2
Nasso4917 ± 61–5104 (100)OC (3-6-every 6 M)ECG, 24 HM (3-6 M)4960
Wang50261181 (97.5)OC (1-3-6-12 M-every 6 M)ECG 24-48 HM (3 M)21NS
  • Studies were presented by year of publication and in alphabetical order.

  • N (%), number (percentage) of patients at follow-up.

  • Endpoint: 1, freedom from AF; 2, number of patients in sinus rhythm/with no AF; 3, no episodes of AF >15 s at 6 months; 4, free of AAD (class IC and III) and AF, no >30 s of AF; 5, episodes of AF; 6, absence of AF on all serial ECG, Holter Monitoring or 30-day event –free monitoring >6 months after procedure; 7: No episode of AF o left Atrial tachyarrhythmia (AT) after blanking period of 3 months; 8, no AT, AF, or atrial flutter >30 s during monitoring at 6 months; 9, no AT, AF, or atrial flutter lasting >30 s off antiarrhythmic drugs; 10, absence of AF, atrial flutter, and AT on 96 h HM or ECG >30 s and no AF compliant; 11, no AF or atrial flutter >30 s during ECG or Holter monitoring; 12, freedom from AT, AF, or atrial flutter off antiarrhythmic drugs after 12 months;13, time-related prevalence of AF (defined as of AF or left atrial tachyarrhythmia after blanking period of 3 months.

  • Follow-up type: OC, outpatient clinic; MR, medical records; TI, telephonic interviews; CV, cardiologic visits; M, months; NS, not specified.

  • Rhythm monitoring: ECG, electrocardiograms; OT, outpatient (cardiac) telemetry; TTM, trans-telephonic (event) monitor; Pm, pace-maker; UCG, ultrasonic cardiographic evaluation; EPS, ectrophysiology studies.

  • aOne patient underwent 15-day HM.

Excluding two which did not mention any continuous rhythm monitoring23,45 and one which performed electrophysiological studies,25 all the others papers employed at least one method of long-term monitoring. Fifteen studies (53.5%) utilized 24 h Holter monitoring (HM) at different intervals,28,33,35,40,41,44,46,47,49,50 two 24–48 h monitoring,34,39 one 7-day HM48 and one 30-day HM27 and in one it was not specified. In five (17.8%) studies HM was associated with another long-term monitoring: 14- to 21-day event recorder in Edgerton et al.26,31,36,42 and outpatient telemetry (OT) in Sagbas et al.30 Among studies which did not utilize HM, OT,24,43 event monitoring,32,37and trans-telephonic monitoring associated with event monitoring were the methods for rhythm assessment.

Only a few papers clearly described the adopted protocol for AADs during the follow-up: AADs were withdrawn in patients in sinus rhythm (SR) after 3 months in seven studies,22,23,40,41,44,47,48 at 6 months in two studies,30,42 and two more studies stated that AADs were stopped after demonstration of stable SR at 3- and 6-month visits.44,49,50 Sirak et al.32 and Yilmaz et al.46 reported that AADs were withdrawn at the discretion of the referring cardiologist.

Similarly, a standard protocol was not followed for oral anticoagulation (OA) therapy: it was discontinued at three months by Salenger et al.23 and La Meir et al.48 {if CHADS2 [cardiac failure, hypertension, age, diabetes, stroke (doubled)] score was <2}, at 4 months by Cui et al.,41 at 6 months by Sagbas et al.,30 Edgerton et al.,37 Han et al. (if CHADS2 score was <2)38 and Klinkenberg et al.,39 at 6 months to 1 year by Castella et al.40 and at 1 year by McClelland et al.27 Furthermore, in Wudel et al.33 and Yilmaz et al.46 anticoagulants were discontinued at the discretion of the referring cardiologist: Speziale et al.44 and Nasso et al.49 kept patients under OA until SR was maintained for 3 months, whereas Krul et al.47 discontinued OA when CHADS2 score was <1 without AF and AAD. Finally, Wang et al.50 continued OA for at least 3 months.

Patients taking AADs at latest follow-up were shown or could be calculated in 22 of 28 (78.5%) studies. The range was from 2% to 53% (mean 26 ± 15%). Patients still under OA therapy at follow-up could be obtained from 15 of 28 (53.5%) studies. The range was from 0% to 63% (mean 39 ± 19%).

Quality-of-life assessment was carried out employing the 36-item short form health survey questionnaire by Speziale et al.44 and Nasso et al.,49 both showing an improvement in all indexes with the exception of vitality score. Pruitt et al.25,29 explored the perceived health status after surgery: 81.8% and 52.3% of patients were pleased with the outcome and 81.8% and 55.7% felt the operation had been beneficial, in the two publications, respectively.


Only six papers,27,34,38,42,47,50 all utilizing radiofrequency, clearly defined the primary efficacy endpoint of surgery as suggested by current guidelines [freedom from AF, off AADs at 1 year (Table 4)].4 The follow-up was 93.3% complete (251 of 269 patiens) and the success rate ranged from 51% to 86%. One study reported a 2-year success rate of 80%.

View this table:
Table 4

Minimally invasive surgery for lone atrial fibrillation with radiofrequency: studies reporting results according Heart Rhythm Society/European Heart Rhythm Association/ European Cardiac Arrhythmia Society consensus (freedom from atrial fibrillation, off antiarrhythmic drugs at 1 year)

MClelland27575 (15/20)75 (15/20)191 (10/11)91 (10/11)180 (4/5)80 (4/5)325 (1/4)25 (1/4)
Bagge34876 (25/33)51 (17/33)579 (19/24)NS0100 (2/2)NS354 (4/7)NS
Han381565 (28/43)65 (28/43)1165 (20/31)65 (20/31)467 (8/12)67 (8/12)
Edgerton421081 (42/52)73 (38/52)1081 (42/52)73 (38/52)
Krul47291 (20/22)86 (19/22)0100 (12/12)92 (11/12)278 (7/9)78 (7/9)0100 (1/1)100 (1/1)
Wang501580 (65/81)80 (65/81)1580 (65/81)80 (65/81)
  • Studies were presented by year of publication and in alphabetical order.

  • NS, not specified.

With specific reference to the different type of AF, AAD-free success rate at 12 months ranged from 65% to 92% in paroxysmal AF and from 67% to 80% in persistent AF. Only three studies27,47,50 reported 12-month AAD-free success rate in LSP [25% (4 patients at follow-up), 100% (1patient at follow-up), and 80% (81patients at follow-up)].

Eight studies26,3033,3537 reported outcomes as freedom from AF and AAD at 6 months (Table 5). This rate ranged from 57% to 91% in the whole population, from 72% to 100% in paroxysmal and from 35%to 78% in persistent AF. Two papers36,37 showed this outcome in LSP patients [32% (22 patients at follow–up) and 47% (15 patients at follow-up)] and one report30 showed AAD-free in permanent AF (72% (18 patients at follow-up)].

View this table:
Table 5

Minimally invasive surgery for lone atrial fibrillation with radiofrequency: studies reporting results with freedom from atrial fibrillation, off antiarrhythmic drugs at 6 months

Edgerton261574 (42/57)63 (36/57)782 (32/39)74 (29/39)856 (10/18)39 (7/18)aaa
Sagbas30581 (21/26)81 (21/26)0100 (8/8)100 (8/8)572 (13/18)72 (13/18)
Edgerton311774 (49/66)58 (38/66)784 (36/43)70 (30/43)1057 (13/23)35 (8/23)NSNSNS
Sirak32388 (21/24)88 (21/24)NSNSNSNSNSNS
Wudel33991 (20/22)91 (20/22)NSNSNSNSNSNS
Beyer351387 (87/100)63 (63/100)587 (34/39)NS0100 (29/29)NS972 (23/32)NS
Edgerton363371 (81/114)57 (65/114)887 (52/60)72 (43/60)1456 (18/32)47 (15/32)1150 (11/22)32 (7/22)
Edgerton37680 (24/30)58 (14/24)190 (9/10)78 (7/9)575 (15/20)47 (7/15)
  • Studies were presented by year of publication and in alphabetical order.

Wolf et al.24 Yilmaz et al.46 and Nasso et al.49 reported freedom from AF-AAD at different intervals (Table 6): 65% beyond 3-month follow-up, 65% at 12.6-month follow-up, and 51% at 17.1-month follow-up, respectively.

View this table:
Table 6

Minimally invasive surgery for lone atrial fibrillation with radiofrequency: studies reporting results with alternative follow-up

Wolf24291 (21/23)65 (15/23)1NSNS0NSNS1NSNS
Castella401062 (16/26)NS282 (9/11)NS460 (6/10)NS420 (1/5)NS
Cui411080 (39/49)NSNS80 (NS)NSNS75 (NS)NSNS67 (NS)NS
Sirak430100 (19/19)NS0100 (19/19)NS
Speziale44687 (40/46)NS195 (18/19)NS581 (22/27)NS
Stamou45375 (9/12)NS0100 (7/7)NS240 (3/5)NS
Yilmaz46777 (23/30)65 (19/30)384 (16/19)NS275 (6/8)NSNSNSNS233 (1/3)NS
La Meir48,a4.5%11.3%0%0%100 (8/8)8.3%50 (3/6)
Nasso49,b1189 (93/104)51 (53/104)296 (NS)Ns980 (NS)NS
  • Studies were presented by year of publication and in alphabetical order.

  • NS, not specified.

  • aPublished as 36-month time-related prevalence.

  • bPersistent and long-standing persistent presented together.

Three papers showed outcome as 12-month freedom from AF (62% and 75%, respectively)40,45 or 3-month freedom from AF (100%).43 Furthermore, in two publications41,44 80% (39 of 49) and 87% (40 of 46) of patients are in sinus rhythm at 12 and 6 months, respectively (Table 6)

View this table:
Table 7

Minimally invasive surgery for lone atrial fibrillation: freedom from atrial fibrillation recurrence and antiarrhythmic drugs after microwave ablation

Pruitt25,b979.5 (35/44)NSaaaaaaaaaaaa
Koistinen289a86 (13/22)63 (10/22)NSNSNSNSNSNS
Klinkemberg39847 (7/15)33 (5/15)554 (6/11)27 (4/15)325 (1/4)25 (1/4)
  • Studies were presented by year of publication and in alphabetical order.

  • NS, not specified.

  • aAF defined as intermittent or continuous.

  • b AF defined as intermittent and continuous.

La Meir et al.48 employed a method suggested by the Workforce on Evidence-Based Surgery of the Society of Thoracic Surgeons, reporting time-related prevalence of AF.55 (Table 6)

Four studies employing MW23,25,28,29 reported a freedom from AF success rate of 67%, 79.5%, 86%, and 42%, respectively. One group utilizing HIFU39 reported a 6-month AF-AAD-free success rate of 33%. (Table 7)

Krul et al.47 carried out a roof and inferior line with 86% 1-year freedom from AF-AAD (22 patients). It was 100% for Sirak et al.43 (19 patients all LSP) and 58% for Edgerton et al.31(24 patients) at 6 months. In three studies utilizing MW23,25,29 an additional line towards LAA was performed. Unfortunately none of these papers reports AAD-free rate.

Groups performing GP ablation reported a 6-month and 12-month AAD-free success rate ranging from 51% to 86% and 58% to 88%, respectively. Two studies employing bipolar radiofrequency30,33 without GP ablation reported an AAD-free success rate of 81% and 91%, respectively.

Adverse events

Three early deaths have been reported (Table 8): one26,31,36 occurred during the procedure as a consequence of tearing of LAA, one41 was due to cerebral infarction and for one death whose cause was unknown.36 Overall, early complications occurred in 125 patients and the complication rate ranged from 0% to 39%. Early complications were as follows: 12 conversions to sternotomy, 15 bleedings, 10 access-port complications, 11 cardiac complications, 8 cerebrovascular events, and 28 pulmonary complications. Furthermore, 38 other complications occurred: 4 renal insufficiencies, 16 diaphragmatic-phrenic nerve dysfunctions, 1 liver damage, and 4 brachial plexopathies were observed. Finally, 10 patients required a pacemaker implantation and two patients underwent PCA within 30 days (Table 8).

View this table:
Table 8

Minimally invasive surgery for lone atrial fibrillation: adverse events

First authorN%DeathConversionBleedingPortCardiacCerebralPulmonaryOtherLate
Salenger23428.5020000PN; PE0PCA [2(3-6-M)]; PM [(9 M)]
Pruitt25240000000D (2)Death (1) C-MZ (5)
Edgerton2644.81b00000HRI; BP0
Mc Clelland2710.5000000H0PCA [3(NS)]
Koistinen2852.20020000Liv.Dam D (2)PM [(NS)]
Pruitt29131300200CVA0D (7) PM (3)4 CVA, ACC C-MZ (9)
Edgerton3145.41b00000HRI; BP0
Sirak3213.101000000PCA [5(within 3 M)]
Wudel3329000Rev00PE0PCA [1(7 M)]
Bagge341739006Rib I (4)VT PPDis StrokeH (2)0PP INF Esoph He (NS)
Beyer35131300000TIAH (3) PEPM (5) Phre (3)NS
Edgerton3615132b010Pe VT (2)0RD (5)RI (2) Phre BPNS
Edgerton370000000000PM(2) PCA (NS)
Han380000000000PCA [8 ( 532 ± 73 days)] PM
Klinkemberg39320002a0Pe000PCA (6)
Castella40514.700200StrokeH PAV0NS
Cui4167.4110Rev (2)MI0Reint0NS
Edgerton4247.60000000BP, PM PCA(2)0
Stamou453150000Pe (3)000NS
Krul47622.2030000H PX PN00
La Meir4800000000000
Nasso4932.901000TIA stroke00Death (2) TEE
Wang5044.80Rev (2)StrokeReint
  • Studies were presented by year of publication and in alphabetical order.

  • N, Number of in-hospital complications; %, percentage of in-hospital complications; Port, access port complications; Rev, surgical revision; Rib, rib fracture; VT, ventricular tachycardia; Pe, pericarditis/pericardial effusion; MI, myocardial infarction; CVA, cerebrovascular accident; Dis: disorientation/confusion; TIA, transient ischaemic attack; PN, pneumonia; PE, pulmonary embolism; PX, pneumothorax; PLE, pleural effusion; H, hemothorax; RD, respiratory distress; PAV, prolonged assisted ventilation; Reint, reintubation; Phl, phlebitis; D, diafragmatic dysfunction; RI, renal insufficiency; BP, brachial plexopathy; PM, pacemaker implantation; Phre, phrenic nerve injury; PCA, percutaneous catheter ablation. RH, rehospitalization; C-MZ, Cox-Maze procedure; ACC, anticoagulation complications; PP, post-perdicardiotomy syndrome; INF, infection; Esoph, esophagitis; He, portal hernia; TEE, thromboembolic event; NS, not specified [number of events (time)]; M, months.

  • aDuring reoperation for bleeding a tricuspid vena cava inferior isthmus ablation was performed in one of these patients.

  • bReferring to the some patients.

Three deaths occurred during the follow-up and none was cardiac related: one unexplained death25 occurred in an obese patient who had severe pulmonary disease at the time of operation, one patient died from trauma and one from malignant neoplasm.46 Other complications included cerebrovascular accident (n = 4), thromboembolic event (n = 1), anticoagulation-related complication (n = 1), and post-pericardiotomy syndrome (n = 1). In addition, during the follow-up, 27 patients underwent PCA, 13 surgical Cox-Maze, 5 had pacemaker implantations, and 1 patient was rehospitalized for congestive heart failure and atrial flutter.


The key purpose of this review is to summarize and discuss published articles about MIS for the treatment of SA-AF. From this overview it is evident that results employing MW energy are rather poor. Indeed Pruitt et al.29 in the most recent paper report a 42% rate with AAD at 12 months. Microwave is an unfocused heat energy which can create transmural endocardial lesion but it is not capable of creating epicardial lesion on the beating heart8 and this may explain such unsatisfactory results. In addition, results with HIFU energy were also suboptimal: Klinkenberg et al.39 reported a 6-month AAD-free success rate of 33%. Indeed, this energy source appears capable of creating endocardial lesions on the arrested heart but has not been able to create transmural lesion on the beating heart and this explains the poor outcome of this study.

Bipolar radiofrequency is the most commonly adopted energy source for MIS as it has been demonstrated to produce permanent transmural linear lesions on the beating heart.8 Two groups performed surgery through a monolateral right thoracotomy approach44,49 with a success rate without AAD of 51% and with AAD of 87% and 89%. The monolateral approach should significantly lower the possible complication rates (bleeding, pulmonary complications), should be less painful for patients and allow potentially faster recovery. Nevertheless, only monopolar devices were used because of the lack of maneuverability of bipolar clamps which can hamper the orientation of the probe and constrain the variety of achievable lesion sets when employed from the right side. There is concern about the ability of these devices to create transmural lesion with bidirectional conduction block on the beating heart.56 Further studies are necessary to establish whether this approach may represent a reliable alternative to bilateral bipolar surgical ablation.

Among studies utilizing radiofrequency, freedom from AF-AAD at 12 months ranged from 51% to 86%; this figure was 65−92% in paroxysmal and 67–80% in persistent. Wang et al.50reported 80% AF-AAD freedom with 81 patients reaching 1-year follow-up. These results are comparable to those following PCA.57 Indeed, an updated worldwide survey7 reported a success rate free of AAD after PCA of 69.9%. It was 71.4% in paroxysmal, 59.4% in persistent, and 59.7% in LSP. Similarly, MIS was more effective in paroxysmal and four of seven studies27,38,42,47 showed an AAD-free success rate higher than PCA. However, MIS also had better results in persistent in three of three studies reporting AAD-free success rates. In contrast, the success rate was lower in two of three studies in LSP AF.

Furthermore, authors performing left additional lines had very high AAD-free success rates in persistent and LSP. This confirms that persistent and LSP AF cannot be successfully treated by PVI alone as the mechanisms for initiation and maintenance of AF lie in the changed left atrial substrate beyond the PVs. From a pathophysiological perspective, this is explained by structural and electrical remodelling of the atrial myocardium, which can then initiate and sustain AF independent of the PVs in patients with persistent/LSP AF. In these patients, the augmented number and location of drivers for fibrillation necessitates additional linear ablation strategies.

Another interesting finding of this overview is that from the analysis of the reported results, it seems that GP ablation is not an essential part of the MIS of SA-AF to prevent recurrence of AF. Indeed, groups performing GP ablation reported a 12-month AAD-free success rate lower than studies not performing GP ablation. The long-term efficacy of GP ablation has been questioned58 and this was related to the possibility of autonomic ganglia to potentially reconnect or regrow.59 Nevertheless, these data must be interpreted with caution because of the small number of studies with full results available and the short follow-up, and the role of GPs and their destruction in AF must be further investigated.

Mortality was low and comparable to PCA (0.4%).54 Postoperative complication rates are not negligible and higher than PCA.54 Pulmonary problems, bleeding, and diaphragm/phrenic nerve dysfunction were the most common complications. Nevertheless, the occurrence of transient ischaemic attack/stroke was very low and less than that reported in catheter-based PVI.54 Interestingly, the incidence of cardiovascular accident was not different with or without LAA removal, and from these findings, it seems that this procedure is not a critical component of MIS. However, this finding must be interpreted with extreme caution due to the small number of publications and the lack of uniform anticoagulation policy. Certainly, it is necessary to clearly define the role of LAA removal in MIS ablation as the LAA exclusion/closure is a potentially dangerous procedure and most of the reported bleedings were related to this maneuver.

Unfortunately, no randomized control trials have been published comparing MIS with PCA. At the moment there are two ongoing studies: the SCALAF success trial which compares the efficacy of circumferential pulmonary vein ostia ablation using surgical vs. catheter techniques in the treatment of paroxysmal atrial fibrillation, and the FAST II trial that explores whether mini invasive thoracoscopic radiofrequency is more effective compared with a percutaneous catheter-based technique in patients with symptomatic paroxysmal atrial fibrillation refractory to at least one AAD. The results of these studies will provide a great understanding of the role of both procedures in the treatment of SA-AF.

In our recent paper60 we had given an overview of medical and surgical treatments of atrial fibrillation. This study is mainly focused on providing a more detailed oversight of efficacy and complications of minimally invasive procedure. Krul et al.61 have recently published a review on the mini-Maze. They included only papers employing radiofrequency and performed an aggregate patient data meta-analysis to compare published results. Our approach was different and we choose to give to the readership as much information as possible to allow them to make their own opinion about the effectiveness/safety of these procedures. Furthermore, we believe that although a well-designed meta-analysis can provide valuable information for researchers, there are too many critical caveats in performing and interpreting it which, therefore, can yield misleading information62 in particular when individual patient data (IPD) are not available.63 Indeed, IPD meta-analysis offers advantages and, when feasible, should be considered the best opportunity for summarizing the results of multiple studies. However, the resources, time, and cooperation required for such studies will continue to limit their use in many important areas of clinical medicine which can be meaningfully and cost-effectively approached only by aggregate patient data-based meta-analyses.

Towards a multi-disciplinary approach: the hybrid procedure

With the development of new tools and advanced techniques, the treatment of SA-AF moves towards a multidisciplinary approach involving cardiac surgeons and EPs.45

The rationale of the so-called ‘hybrid’ approach is to combine, in one step, a minimally invasive epicardial technique with a percutaneous endocardial ablation in order to limit the shortcomings of both techniques and, at the same time, to combine their advantages.

The potential for improved outcomes derives from combining levels of expertise. Surgeons are very good at making linear lesions and EPs at mapping for completeness. Lesions are more likely to be transmural when burning from the inside outwards and from the outside inwards simultaneously and the EP can add an endocardial ‘touch-up’ in the case of incomplete isolation of one of the pulmonary veins or if the connecting lesions are not transmural.

Furthermore, as discussed above, a more extensive lesion set beyond PVI to include targets along the LA substrate is often necessary in persistent and LSP AF. Nevertheless, while the other lesions can be performed through the transverse sinus, the biggest challenge in the case of the full beating heart is making the connection to the mitral annulus, and this for different reasons: first, the visualization behind the LA of the full beating heart is very limited; second, there is the risk of collateral damage to the circumflex artery; third, the CS, which is used as the epicardial landmark for the mitral annulus, is unreliable and may leave a gap.64

An attempt to address this problem was made by Edgerton et al.37 who developed the ‘Dallas lesion’ in which a line was made connecting to the anterior annulus at the junction of the left and non-coronary cusps of the aortic root. Nevertheless, this line might not be transmural due to the inability of radiofrequency energy effectively to penetrate fatty tissue associated with the dome of the left atrium and the SVC. This is an indication for mapping conduction block which can be checked by using a hybrid approach.

In contrast, mitral isthmus lesion can easily be carried out (or completed) endocardially by the EP. Another potential advantage of the hybrid technique is that, from the EP's point of view, there is no longer a risk of phrenic nerve and oesophageal injury because these structures can be protected by the surgeon if necessary, as well as no risk of tamponade as the pericardium is open. Furthermore, by reducing the total number of endocardial ablations the risk of emboli during these ablations should be potentially reduced.65

A potential drawback of the single-step approach is the patient's heparinization after the transseptal puncture which could increase the risk of epicardial bleeding and the risk of measuring a temporary block.

The concept of ‘hybrid’ procedure was first published by Pak et al.,66 who combined percutaneous epicardial catheter ablation and endocardial ablation in difficult cases of AF. More recently, Krul et al.47 presented their experience with thoracoscopic PV isolation and GP ablation guided by periprocedural electrophysiological testing.

Mahapatra et al.67 have recently published their initial experience with surgical epicardial-catheter and endocardial ablation for persistent and LSP atrial fibrillation carried out in two sequential steps (during the same hospitalization). After a mean follow-up of 20.7 ± 4.5 months, 86.7% patients (persistent and LSP) were free of any atrial arrhythmia and off AADs, compared with 56.7% of patients undergoing repeat catheter ablation. More recently, Pison et al.68 published results of a one-step hybrid thoracoscopic surgical and transvenous catheter ablation of atrial fibrillation in 26 patients. One-year off-AAD success free of AF/AFL/AT was 93% for patients with paroxysmal AF and 90% for patients with persistent AF.

Nevertheless, demonstration of the effectiveness and the safety of this approach awaits the completion of studies currently underway. Figure 1 shows the conversion of AF to sinus rhythm during the one-step hybrid procedure.

Figure 1

Hybrid procedure: conversion to sinus rhythm during epicardial ablation of the right superior ganglionated plexi.


This review has some important limitations which must be pointed out. First of all, a meta-analysis could not be carried out because of the heterogeneity of the papers and the lack of IPD. Second, the unavailability of data from prospective randomized studies on the minimally invasive surgical treatment of SA-AF vs. PCA was a drawback of this systemic review. Third, not all results are reported following HRS/EHRA/ECAS expert consensus which makes a comparison of the different studies difficult.


Minimally invasive surgical ablation of SA-AF achieved satisfactory 1-year results when the bipolar radiofrequency was employed as energy source with AAD-free success rate comparable to PCA. The success rate in paroxysmal was even higher than in PCA. In persistent and LSP, results were improved by additional LA ablation lines. In contrast, GP ablation and LAA removal seem not to influence the recurrence of AF and the occurrence of postoperative thromboembolic events. Nevertheless, the relatively high complication rate reported suggests that such techniques require further refinement. Finally, the preliminary results of the hybrid approach are promising but they need to be confirmed.

Authors' contribution

Mark La Meir and Sandro Gelsomino organized Research design, drafting, approval, and writing the manuscript.

Fabiana Lucà helped in article collection, research design, drafting, and approval of the study.

Roberto Lorusso and Andrea Colella involved in research design, drafting, and approval of the study.

Elena Crudeli helped in article collection and research design.

Ludovico Vasquez and Gian Franco Gensini involved in drafting, final revision, and approval of the manuscript.

Laurent Pison did a critical revision and involved in drafting and approval of the study.

Final revision and final approval of the manuscript were carried out by Jos Maessen.

Conflict of interest: Dr La Meir is consultant/advisor for Atricure and Esthec. Other authors have no conflict of interest.


No financial support.


  • These authors equally contributed to the paper.


View Abstract