Article Text

Download PDFPDF

Percutaneous atrial appendage occlusion for stroke prevention in patients with atrial fibrillation: a systematic review
  1. Stine Munkholm-Larsen1,2,
  2. Christopher Cao2,
  3. Tristan D Yan2,
  4. Steen Pehrson1,3,
  5. Ulrik Dixen1,4
  1. 1The University of Copenhagen, Copenhagen, Denmark
  2. 2The Systematic Review Group, The Baird Institute for Applied Heart and Lung Surgical Research, Sydney, Australia
  3. 3Department of Cardiology, Rigshospitalet, Copenhagen, Denmark
  4. 4Hvidovre University Hospital, Copenhagen, Denmark
  1. Correspondence to Tristan D Yan, MBBS, MD, PhD The Systematic Review Group, The Baird Institute for Applied Heart and Lung Surgical Research, Sydney, Australia; tristanyan{at}


This review aims to evaluate systematically the safety and efficacy of percutaneous left atrial appendage occlusion (PLAAO) in stroke prevention for patients with atrial fibrillation (AF). A systematic review of peer-reviewed journals on PLAAO before June 2011 was performed on three electronic databases. Fourteen studies were identified for evaluation. Overall, implantation was successful in 93% of all cases. Periprocedural mortality and stroke rates were 1.1% and 0.6%, respectively. The incidences of pericardial effusion/cardiac tamponade and device embolisation were 4% and 0.7%, respectively. At the time of the latest follow-up (up to 40 months), the overall incidence of stroke among all studies was 1.4% per annum. Existing evidence suggests that PLAAO is a relatively safe treatment for patients with AF. However, there is a need for further evaluation of its efficacy in the form of large and well-designed randomised controlled trials.

  • Amplatzer cardiac plug
  • atrial fibrillation
  • coronary artery disease
  • left atrial appendage closure
  • left atrial appendage occlusion
  • minimally invasive
  • surgery—coronary bypass
  • surgery—valve
  • watchman device

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Atrial fibrillation (AF) is the most common cardiac arrhythmia encountered in clinical practice, with a projected prevalence of more than 10 million by year 2050 in the USA alone.1–4 It is associated with three serious sequelae: irregular and rapid heart rate causing haemodynamic instability; loss of atrial-ventricular synchrony, which may lead to congestive heart failure; and blood stasis in the left atrium, causing thromboembolism. Stroke is a devastating complication for patients with AF, with a reported annual incidence of 4.4%.5 For patients with cardioembolic strokes caused by AF, up to 70% will result in death or significant disability.6 To assess the risk of stroke in individual patients with non-valvular AF, the CHADS2 score was developed to determine the appropriate management.7 The CHA2DS2VASc score has recently been introduced as an attempt to improve risk stratification in the low-risk group according to CHADS2.8

Traditionally, anticoagulation therapy with warfarin has been the standard of care for stroke prevention in patients with AF.9 However, due to poor patient compliance, contraindications, a narrow therapeutic window and potential bleeding complications, warfarin has been underutilised in patients with AF, even in those who have higher risks of a cerebrovascular event.10 Antiplatelet therapy such as aspirin has been used for low-risk patients.11 12 In February 2011, the American College of Cardiology Foundation and the American Heart Association added dabigatran to their guidelines for management of non-valvular AF with a class I recommendation.13 Other emerging pharmacological agents include rivaroxaban14 and apixaban.15 Non-pharmacological methods of preventing stroke in patients with AF have also been developed in recent years. Several studies have demonstrated that up to 90% of thrombi deriving from the left atrium are formed in the left atrial appendage (LAA).16–19 Therefore, LAA occlusion to prevent cardioembolic stroke would be a strategy to minimise cerebrovascular events.

The first study on percutaneous left atrial appendage occlusion (PLAAO) in humans was conducted in 2002 by Sievert et al.20 Since then, a number of studies using three different devices have reported encouraging results in stroke prevention.21–33 Despite these findings, concerns about the safety and efficacy of this relatively novel procedure have hindered its widespread application. To date, only one randomised controlled trial (RCT) has compared PLAAO with warfarin, using the Watchman left atrial appendage system (Atritech Inc, Plymouth, Minnesota, USA).21 The current systematic review aims to evaluate the safety and efficacy of PLAAO in stroke prevention for patients with AF.


Literature search strategy

Electronic literature searches were performed to identify all peer-reviewed articles on PLAAO in patients with AF. The following electronic databases were searched from their inception until June 2011: MEDLINE, PubMed and EMBASE. The keywords used during the search included: ‘left atrial appendage occlusion’, ‘left atrial appendage closure’, ‘Watchman device’, ‘Amplatzer cardiac plug’ and ‘PLAATO’. The reference lists of all retrieved studies were manually searched for further identification of potentially relevant studies. All articles were selected with application of predefined inclusion and exclusion criteria.

Selection criteria

Studies selected for appraisal presented results of PLAAO in patients with paroxysmal, persistent or permanent AF. All studies selected were human studies, published in the English language. Studies reporting on the safety and efficacy of surgical LAA occlusion were excluded. Studies focusing on transoesophageal echocardiography (TEE), CT or MRI techniques during device implantation or follow-up, and studies focusing entirely on AF, without examining percutaneous LAA occlusion, were excluded. Abstracts, case reports, editorials, letters and expert opinions were also excluded.

Data extraction and critical appraisal

All data were extracted and tabulated from the relevant articles' texts, tables and figures. Two investigators (SML) and (CC) independently reviewed each retrieved article. Discrepancies between the two reviewers were resolved by discussion and consensus. The quality of the selected studies was evaluated according to the Oxford Centre for Evidence-Based Medicine levels of evidence34 and the RCT was assessed by the Jadad scale. The Jadad scale, also known as the Oxford quality scoring system, is a procedure to assess independently the methodological quality of clinical trials.35 Periprocedural outcomes were defined by endpoints occurring during hospital admission or up to 30 days after the procedure. It was difficult to compare the data with respect to certain outcome measures, as some studies reported their follow-up results using patient-years21–23 and others reported a mean follow-up in days or months.20 24–33 It was also impossible to establish the incidence of adverse events at fixed intervals from the available published data. The clinical endpoints were reported at the latest follow-up. The final results were reviewed by the senior investigators (TDY and UD). A meta-analysis was not appropriate because 13 out of 14 studies lacked a comparator.

Outline of the procedure

PLAAO is performed under general or local anaesthesia. A sheath is introduced into the right femoral vein. Access to the left atrium takes place via a patent foramen ovale, atrial septal defect or trans-septal puncture. After approximating the size and shape of the LAA under TEE or fluoroscopic guidance, the device is sized using a standardised chart and is then advanced into the LAA orifice. Imaging is used to confirm optimal positioning before the device is released and the delivery system withdrawn into the right atrium.


Quantity of evidence

Electronic searches using three databases identified 301 publications. After removing duplicates, and limiting the search to human studies in the English language, 125 publications were retrieved for screening. Review of the abstracts of these publications identified 77 potentially relevant articles. Manual search of the reference lists did not identify any additional relevant studies. When the inclusion and exclusion criteria were applied to these 77 publications, 14 studies were selected for evaluation.20–33 A total number of 1737 patients who underwent device implantation were included for appraisal and data extraction (table 1).

Table 1

Summary of baseline characteristics and endpoints presented in relevant publications of PLAAO for stroke prevention in AF

Quality of evidence

The 14 selected studies were published between 2002 and 2011, including one RCT,21 12 prospective non-comparative observational studies20 22 23 25–33 and one retrospective study.24 No previous systematic review or meta-analysis was identified. The only existing RCT on PLAAO, the PROTECT AF study, compared implantation of the Watchman device with the standard treatment using warfarin for stroke prevention. There were 463 patients who underwent PLAAO and 244 patients who were treated with warfarin.21 This RCT was successfully completed with the treatment allocated by an adequate process of concealed randomisation. The two arms were similar in baseline characteristics. Based on the Oxford Centre for Evidence-Based Medicine levels of evidence, the RCT was graded as level 1b evidence.34 The Jadad score was 2, which indicates an intermediate quality of study.35

The remaining 13 studies were all graded as level 4 evidence.20 22–33 The largest study, an observational multi-institutional registry, included 1002 patients.22 This study was a continuation of the interventional arm of the PROTECT AF study and presented an accumulating number of patients with an increased length of follow-up. Five studies had more than 100 patients21–24 31 and three studies had between 60 and 100 patients (range 64–75).26 27 29 The other six studies had 20 patients or fewer (range 7–20).20 25 28 30 31 33 Due to the limited numbers of patients, these six studies were unlikely to be representative of the target population.

Treatment protocols

The treatment protocols varied between institutions (table 2). Three studies used the Watchman device,21 22 29 one study used the Amplatzer septal occluder (AGA Medical Corp, Golden Valley, Minnesota, USA),33 two studies used the Amplatzer cardiac plug (AGA Medical Corp)24 25 and eight studies used the percutaneous left atrial appendage transcatheter occlusion (PLAATO) device (originally produced by ev3 Inc., Plymouth, Minnesota, USA).20 23 26–28 30–32 However, all studies shared the same concept in treatment rationale including LAA occlusion with a percutaneous device followed by administration of antiplatelet agents.

Table 2

Summary of treatment protocols and follow-up After LAA occlusion in patients with AF

It should be acknowledged that medical therapy varied among different studies. In four studies, warfarin was given following the procedure.21 22 29 33 Five studies used periprocedural antibiotics for endocarditis prophylaxis.23 25 27 32 33 Five studies reported the administration of aspirin and clopidogrel 24–48 h before device implantation.23 25 27 28 32 Eight studies used heparin during the procedure.20 22 24 25 27 29 32 33 In all but three studies, patients were treated with aspirin indefinitely.24 26 33 Clopidogrel was prescribed in 13 out of the 14 studies.20–28 30–33 In seven studies, this was administered for 6 months,20–23 26 28 31 but the duration varied from 4 weeks to 3 months in other reports.

The follow-up periods ranged from 24 h to 40 months. Four studies had a mean follow-up of more than 2 years.22 28–30 TEE was performed in nine studies at 6 months after the procedure.21 22 26–32 The mean procedure time varied from 50 to 102 min.

Baseline characteristics and periprocedural outcomes

The mean age of all patients was 70 years. Sixty-five per cent were men and the mean CHADS2 score was 2.5 (table 1). Successful PLAAO was performed in 93%. Periprocedural mortality was 1.1%. The combined haemorrhagic and embolic stroke rate was 0.6%. The rates of systemic embolism, perforation, pericardial effusion/tamponade and device embolisation were 1.6%, 0.9%, 4.0% and 0.7%, respectively. Major bleeding occurred in 0.7% of patients and 1.1% required surgical intervention (table 3).

Table 3

Summary of periprocedural outcomes in patients with AF who underwent PLAAO

Clinical outcomes at the latest follow-up

Overall mortality of the 14 studies was 6.2% at the time of the latest follow-up. Stroke occurred in 1.4%. Of all patients, the expected annual risk of stroke based on the CHADS2 score ranged from 1.9% to 8.6%, and the actual annual risk reported in the studies ranged from 0% to 3.8%. The rates of transient ischaemic attack (TIA), myocardial infarction, systemic embolism, pericardial effusion/tamponade and surgery related to implantation were 2.8%, 0.3%, 0.6%, 4.1% and 1.1%, respectively. Major bleeding occurred in 0.7% of patients (table 4).

Table 4

Summary of clinical outcomes at the latest follow-up after LAA occlusion in patients with AF

Current devices

In 2006, the manufacturer of the PLAATO device discontinued the development of this product.36 The Watchman and Amplatzer devices are currently available in clinical practice. These two devices were used in six studies published from 2002 to 2011. Three studies, including one RCT,21 one multi-institutional registry study22 and one case series29 used the Watchman device, while the remaining three studies were case series reporting on the Amplatzer device.24 25 33 Successful implantation of either the Watchman or the Amplatzer device was carried out in 1154 out of 1250 patients (92%) enrolled into the six studies. The periprocedural mortality of the six studies was 0%, and the mortality at the latest follow-up was 4%. The mean periprocedural stroke rate was 0.6%, and the stroke rate at the latest follow-up was 0.6%. There were no periprocedural episodes of TIA, but one study reported two episodes during follow-up, which resulted in a mean TIA rate of 2.1% at the latest follow-up. The expected annual stroke/TIA risk ranged from 1.9% to 5.3%, compared with an actual stroke/TIA risk of 0%. The occurrence of periprocedural pericardial effusion/tamponade was 3.9%, and at the latest follow-up the pericardial effusion/tamponade rate was 4.0%. More recently, a study examining a transcatheter patch device (Custom Medical Devices, Athens, Greece) has reported encouraging results in 20 patients.37


For patients who have contraindications for oral anticoagulation therapy, occlusion of the LAA has emerged as a stroke prevention option. Surgical ligation or amputation of the LAA was first performed by Madden38 in 1949. This procedure is now widely practised in patients with AF undergoing mitral valve surgery.39 The percutaneous approach to LAA obliteration has shown similar advantages in stroke prevention.20–33 The PLAATO device was the first to be successfully deployed in humans, but is no longer available.20 The Watchman device consists of a self-expanding nitinol frame. It has fixation barbs, but the permeable polyester fabric only covers the surface exposed to the left atrium. In contrast to the initial PLAATO device, the fabric of the Watchman device is permeable to blood, and all studies to date have used warfarin as thromboembolic prophylaxis until endothelialisation was confirmed by TEE.21 22 29 In a small case series published in 2003, the Amplatzer septal occluder was used to close the LAA.33 More recently, the Amplatzer cardiac plug was developed. The Amplatzer cardiac plug is a self-expanding device consisting of a lobe, which is positioned within the LAA, and a disc, which is designed to close the entrance to the LAA. The lobe and the disc are separated by a compliant connecting segment, known as the waist. To date, this novel device has been used in one small prospective observational study25 and one large retrospective observational study.24

A number of periprocedural complications were identified from the present systematic review. Periprocedural death occurred in five patients, with an overall rate of 1.1%. The causes of the deaths included: exacerbation of congestive heart failure aggravated by the administration of anaesthesia32; haemorrhagic shock due to perforation of the iliac artery in an attempt to remove an embolised device32; cerebral haemorrhage after anticoagulation treatment for deep vein thrombosis, which occurred secondary to cardiac surgery32; acute occlusion of the left ventricular outflow tract due to device embolisation26 and death following cardiovascular surgery.27

Pericardial effusion/cardiac tamponade was the most frequent procedure-related adverse event among all studies, reported in 69 patients (4.1%). In the RCT, this complication was reported in 30 patients (6.5%).21 Of these, 22 patients (5%) were treated with surgical drainage or pericardiocentesis. Periprocedural device embolisation occurred in 10 patients among the 14 studies (0.7%). The PROTECT AF trial reported device embolisation in three out of the 463 patients in the intervention group (0.6%).21 One event was detected during the procedure and two were detected on TEE at the 45-day follow-up. One embolised device was removed percutaneously and two were removed surgically. Device instability occurred in one patient in a case series of 73 patients.26 A case series from 2007 consisting of 75 patients reported delivery wire fractures in the first-generation devices.29 Perforation of the right femoral artery when assessing the right femoral vein was reported in another patient in a case series of 111 patients.32 Overall, urgent cardiac surgery was required in 17 patients (1.1% of the cases),22 23 26 27 29 32 33 mainly due to effusion/cardiac tamponade,22 23 32 followed by device embolisation.22 26 29 33

Mid-term outcomes varied between the studies included in the present review. Successful closure of the LAA was reported in 93% of all patients. The overall stroke rate among the 14 studies at the latest follow-up was 1.4%. The RCT included in this review reported all-cause stroke rates in the intervention group and the warfarin group of 2.3 and 3.2 per 100 patient-years, respectively (rate ratio 0.71, credible interval 0.35 to 1.64).21 Results from the PROTECT AF study showed that major bleeding and haemorrhagic stroke were more frequent in the warfarin group (4.1% and 2.5%, respectively) compared with the interventional group (3.5% and 0.2%, respectively). This indicates that PLAAO is not inferior to warfarin in terms of cardioembolic and haemorrhagic strokes, and that it is trended towards superiority.

Based on historical controls, stroke rates for patients on warfarin therapy ranged from 1.11% to 2.42%.40–42 Eight studies compared a predicted annual stroke risk calculated from the CHADS2 score with the actual annual stroke risk.23 25–30 32 This comparison indicated an overall stroke reduction from 1.9–8.6% to 0–3.8% in respective studies.

Despite identifying a number of encouraging reports from the existing literature on PLAAO, results of the present review are limited by a lack of comparative data on PLAAO versus anticoagulation therapy. In addition, there have been concerns about the design of the PROTECT AF trial, such as using a non-inferiority hypothesis to compare two different modalities of treatment. With regard to patient selection, a high proportion of patients in both groups had paroxysmal AF with a CHADS2 score of 1, indicative of a relatively low-risk population. According to the American College of Cardiology Foundation, American Heart Association and European Society of Cardiology recommendations, patients with a CHADS2 score of 1 can be treated with aspirin rather than warfarin.9 Finally, the haemorrhagic stroke rate in the control group was noted to be much higher than reported in previous trials,40 41 which may have influenced the non-inferiority results for efficacy.

In patients with an increased potential of bleeding complications, or who are unwilling to take anticoagulation, exhibit a lack of compliance or have contraindications to warfarin, PLAAO might be the only stroke-preventing option.43 Given the promising results of the RE-LY trial, patients with AF might benefit from dabigatran treatment instead of warfarin.40 Future trials should also compare PLAAO with dabigatran or other emerging pharmacological alternatives to warfarin. A non-randomised feasibility study (the ASAP study) is currently in the process of recruiting patients for PLAAO.44 The aim of that study is to test the performance of the Watchman device in patients with AF who have contraindications to long-term warfarin therapy. This study is an important step in the general acceptance and distribution of the Watchman device.

The present systematic review on PLAAO for stroke prevention in patients with AF identified a number of studies that evaluated three different devices. The Watchman occlusion device and the Amplatzer cardiac plug are still in clinical use. Existing evidence from the current literature suggests that PLAAO is a safe treatment modality for stroke prevention in patients with AF, despite increased risks associated with the learning curve effect of a novel procedure. However, there is a need for further evaluation of its efficacy, ideally in the form of large and well-designed randomized controlled trials comparing PLAAO with new warfarin alternatives. Until such trials are performed, valuable clinical evidence may be derived from national registries such as the Central Cardiac Audit Database.45 The use of PLAAO should currently be confined to institutions experienced in this procedure, with on-site surgical facilities available. Patient selection should involve a multidisciplinary approach, with cardiologists and other clinicians familiar with the management of AF.



  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.