Author + information
- Received September 19, 2013
- Revision received November 20, 2013
- Accepted November 25, 2013
- Published online March 1, 2014.
- Navkaranbir Singh Bajaj, MD∗,
- Akhil Parashar, MD†,
- Shikhar Agarwal, MD, MPH†,
- Nishtha Sodhi, MD‡,
- Kanhaiya Lal Poddar, MD†,
- Aatish Garg, MD‡,
- E. Murat Tuzcu, MD† and
- Samir R. Kapadia, MD†∗ ()
- ∗Division of Cardiovascular Disease, University of Alabama, Birmingham, Alabama
- †Department of Cardiovascular Medicine, Cleveland Clinic, Cleveland, Ohio
- ‡Department of Internal Medicine, Cleveland Clinic, Cleveland, Ohio
- ↵∗Reprint requests and correspondence:
Dr. Samir R. Kapadia, Department of Cardiovascular Medicine, Sones Cardiac Catheterization Laboratories, J2-3, Cleveland Clinic, 9500 Euclid Avenue, Cleveland, Ohio 44195.
Objectives The goal of this study was to provide a systematic review and analysis of observational studies on percutaneous left atrial appendage (LAA) occlusion for stroke prophylaxis in nonvalvular atrial fibrillation (NVAF).
Background A recent randomized controlled trial in patients with NVAF suggested noninferiority of percutaneous LAA occlusion versus medical management for stroke prevention. However, the use of percutaneous devices remains controversial because of limited literature on their efficacy and safety. We performed a systematic analytical review of existing observational studies to assess the rate of neurological events for patients treated with occlusion devices.
Methods A comprehensive search of the Medline, Scopus, and Web of Science databases from inception through August 1, 2013, was conducted using pre-defined criteria. We included studies reporting implantation in at least 10 patients and a follow-up of 6 months or more.
Results In 17 eligible studies, a total of 1,052 devices were implanted in 1,107 patients with 1,586.4 person-years (PY) of follow-up. The adjusted incidence rate of stroke was 0.7/100 PY (95% confidence interval [CI]: 0.3 to 1.1/100 PY), of transient ischemic attacks was 0.5/100 PY (95% CI: 0.1 to 1.8/100 PY), and of combined neurological events (strokes or transient ischemic attacks) was 1.1/100 PY (95% CI: 0.6 to 1.6/100 PY). Access site vascular complications and pericardial effusion were the most commonly observed procedural complications at a rate of 8.6% (95% CI: 6.3% to 11.7%) and 4.3% (95% CI: 3.1% to 5.9%), respectively.
Conclusions Our systematic review suggested comparable efficacy of LAA occlusion devices compared with historical controls treated with adjusted-dose warfarin and other anticoagulation strategies for prevention of stroke in patients with NVAF.
- closure devices
- left atrial appendage occlusion
- nonvalvular atrial fibrillation
- systematic analysis
Atrial fibrillation (AF) is the most common arrhythmia, affecting more than 5.5 million people in the United States alone (1–3). The lifetime risk of developing AF is reported to be as high as 26% for men and 23% for women (4). Stroke is a major contributor to morbidity and mortality among the U.S. population, affecting almost 800,000 patients every year. AF significantly increases the risk of ischemic stroke and is an underlying factor in up to 20% of total strokes among the elderly (5).
The left atrial appendage (LAA) is a remnant structure of the embryonic left atrium and persists as an out-pouching of the adult left atrial chamber. In a fibrillating atrium, the LAA becomes a major site of blood stasis, which significantly increases the risk of clot formation. Indeed, almost 15% of all patients with nonvalvular atrial fibrillation (NVAF) develop a thrombus in their heart, and transesophageal imaging studies have implicated the LAA as being the site for thrombogenesis in >90% of these cases (6,7). For this reason, stroke prophylaxis strategies and risk schema have been developed to determine which patients are at the highest risk for thromboembolic events. Therefore, anticoagulation with adjusted-dose warfarin or the newer anticoagulants has been offered to appropriate patients based on the accepted CHADS2 (congestive heart failure, hypertension, age >75 years, diabetes mellitus, and prior stroke or transient ischemic attack) scoring system. However, anticoagulation increases the risk of both intracerebral and extracranial bleeding, and approximately 30% to 50% of patients with AF are ineligible to receive anticoagulation (5,8).
More recently, percutaneous LAA occlusion devices have been developed to overcome the aforementioned challenges in thromboembolic prophylaxis in patients with NVAF. The Percutaneous Left Atrial Appendage Transcatheter Occlusion (PLAATO) device (Appriva Medical, Plymouth, Minnesota) was the first to be tested and used in humans. Since then, multiple devices have emerged, including the Amplatzer device (AGA Medical Corporation/St. Jude Medical, Golden Valley, Minnesota), the Watchman device (Boston Scientific, Natick, Massachusetts), and the LARIAT suture delivery device (SentreHeart, Redwood City, California) (9–11) (Fig. 1). The only reported randomized clinical trial, PROTECT-AF (Watchman Left Atrial Appendage System for Embolic Protection in Patients With Atrial Fibrillation), has demonstrated noninferiority of the Watchman device to adjusted-dose warfarin therapy, and results from the CAP (Continued Access Protocol) registry following up on this trial suggest a decreasing trend of procedure-related complications as more experience is gained with the technique (1,10). Although multiple observational studies have assessed the efficacy and safety of this novel technique, large-scale comparative data from clinical trials are not yet available. The results of the PREVAIL (Prospective Randomized Evaluation of the WATCHMAN LAA Closure Device in Patients With Atrial Fibrillation Versus Long Term Warfarin Therapy) study and the ongoing AMPLATZER Cardiac Plug Clinical Trial are expected to produce more robust clinical evidence once they are reported. To that end, we performed a systematic statistical review of data pooled from the currently available observational studies to estimate the adjusted rates of stroke and major procedure-related complications. This type of analysis has the strength to provide a valuable estimate of the absolute effect of a particular intervention in the real-world scenario when enough comparator data from multiple randomized controlled trials or other comparative studies are not available (12).
With the expertise of a medical librarian, we queried the Medline, Web of Science, and SCOPUS databases until August 1, 2013, for eligible studies with the following list of MeSH (Medical Subject Headings) terms: “atrial fibrillation,” “atrial appendage,” “left atrial appendage,” “atrial fibrillation/prevention and control,” “atrial fibrillation/surgery,” “atrial fibrillation/therapy,” “non valvular,” “percutaneous,” “minimally invasive,” “transcutaneous,” and “transcatheter.”
All observational studies, including case series and abstracts presented at scientific conferences, were considered. Of these, studies that reported outcomes after percutaneous closure in at least 10 patients with a minimum of 6 months follow-up were included. In instances where research groups reported cumulative results from multiple publications, caution was exercised in data extraction, and only studies with the largest sample size and follow-up period for each outcome were included (Fig. 1).
The primary outcome was defined as the incidence of stroke in the follow-up period. Secondary outcomes included transient ischemic attacks (TIAs) and composite neurological events (defined as the combination of stroke and/or TIA) during the follow-up period. Safety outcomes including procedure failures and procedural/device-related complications were also analyzed. Procedural/device-related complications were defined as any of the following events: vascular access site hemorrhage, pericardial effusion or tamponade, device embolization, procedural mortality, and need for rescue cardiac surgery. Additionally, several subgroup analyses were performed to define the device characteristics that may predict the occurrence of stroke/TIA in the follow-up period. The number needed to treat (NNT) for stroke prevention was also calculated using the Newcombe-Wilson hybrid score (13).
The primary authors (N.S.B. and A.P.) searched all titles and associated abstracts using the described MeSH search strategy. Full-text papers for all potentially relevant studies were retrieved and reviewed with a primary focus on inclusion criteria, study outcomes, and methodological quality. If the clinical study met the inclusion criteria based on these 3 characteristics, it was included in the analysis. The data from the studies were then extracted using pre-designed structured forms and reviewed by 2 authors (S.A. and N.S.) for accuracy and validity. Disagreements, if any, were resolved by mutual consensus between the authors.
After the data elements were verified for accuracy, systematic and statistical analyses were conducted using Comprehensive Meta-Analysis version 2 (Biostat, Englewood, New Jersey) and SAS version 9.3 (SAS Institute, Cary, North Carolina), respectively. Because of a relatively small proportion of primary events in each study, the data were assumed to follow a Poisson distribution. Pooled event rates (expressed as percentage per year) for the efficacy outcomes in the LAA arm and safety outcome rates (per 100 procedures) were calculated using standard methods. Fixed effects modeling was primarily utilized to conduct systematic analysis of included studies. However, we used random effects modeling in the case of statistically significant heterogeneity. Assessment of heterogeneity was achieved by comparing the inclusion/exclusion criteria and the minor differences in the design and conduct of the studies. We assessed for heterogeneity using the I2 test (I2 >50% with p < 0.05 implies significant heterogeneity). In cases of significant heterogeneity, the heterogeneity was first explored in the studies, and subsequently, analysis using a random effects model was performed to statistically account for the heterogeneity. Publication bias was assessed using the funnel plot method and corrected using the Duval and Tweedie trim and fill method (14).
Subgroup analyses were performed to determine event rates in different strata of the study population. All subgroup analyses were pre-specified. All of the p values were 2-tailed, with statistical significance specified at p < 0.05 and confidence intervals (CIs) computed at the 95% level. The analysis has been reported in accordance with the PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) guidelines (15).
We included 17 observational studies (16–30) for the final analysis in accordance with study inclusion criteria (Fig. 2). A total of 1,052 devices were implanted in 1,107 patients. The pooled analysis yielded 1,586.4 person-years (PY) of follow-up. Table 1 demonstrates the baseline characteristics of the included studies. The follow-up varied from 7 to 29 months among the studies. The mean CHADS2 score was 2.7. Two studies had a mean CHADS2 score of <2, and another 2 studies did not report CHADS2.
The predicted stroke rate varied between 2.5 and 5.3/100 PY, assuming a CHADS2 score of 3. Nine of 17 studies reported post-procedural utilization of antiplatelet agents (aspirin ± clopidogrel) for varying time periods. There were minor differences observed among inclusion and baseline characteristics of included studies as demonstrated in Table 1.
The actual procedural success was 1,052 of 1,107 (95.1%). The adjusted procedural failure rate was 8.4% (95% CI: 6.6% to 10.6%) using random effects modeling and adjusting for publication bias. The adjusted pooled incidence rate of stroke in our systematic analyses was 0.7/100 PY (95% CI: 0.3 to 1.1/100 PY) (Fig. 3). There was no publication bias to the right of mean, which was assessed using the Duval and Tweedie trim and fill method (Fig. 4). Table 2⇓ demonstrates subgroup analyses by device type, year, and region; similar stroke rates were observed in the Watchman and PLAATO devices of 0.7/100 PY (95% CI: 0.0 to 1.5/100 PY) and 0.7/100 PY (95% CI: 0.0 to 1.6/100 PY), respectively, as compared with Amplatzer 0.9/100 PY (95% CI: 0.7 to 2.4/100 PY), although the difference was not statistically significant. There was no statistically significant difference observed in stroke rate over the year or region (Table 2). The NNT for 1 stroke was estimated at 21.7 (95% CI: 16.1 to 31.5) assuming a CHADS2 score of 3 for the pooled population.
The adjusted rate of TIAs was observed to be 0.5/100 PY (95% CI: 0.1 to 0.8/100 PY), with an estimated NNT of 23.3 (95% CI: 16.9 to 35.5). The pooled composite rate of neurological events across all studies was found to be 1.1/100 PY (95% CI: 0.6 to 1.6/100 PY) (Fig. 2, Table 3) with a NNT of 25.0 (95% CI: 17.7 to 40.4).
Tables 4 and 5 demonstrate the adjusted pooled rates of periprocedural adverse events per 100 procedures observed across the studies. We calculated a combined adverse event rate of 16.0% (95% CI: 13.6% to 18.7%). There were no statistically significant differences in the combined rate of complications over time (Table 5). Subgroup analyses revealed a higher rate of combined adverse events with both the Amplatzer and the Watchman devices (23.5% [95% CI: 15.9% to 33.2%] and 13.6% [95% CI: 9.7% to 18.8%], respectively), as compared with 5.4% (95% CI: 3.5% to 8.2%) for PLAATO and 7.3% (95% CI: 2.7% to 18.3%) for LARIAT devices, although these were not statistically significant. The most frequent complication was of the vascular access site: 8.6% (95% CI: 6.3% to 11.7%). The other frequently observed procedural complications were pericardial effusions and device embolization, which occurred at a rate of 4.3% (95% CI: 3.1% to 5.9%) and 3.9% (95% CI: 2.7% to 5.6%), respectively (Table 4).
Comparison with other anticoagulation strategies reported in the literature
Figure 5 demonstrates the rates of ischemic stroke despite therapeutic anticoagulation, as reported in the literature. In our review, the stroke rate with percutaneous LAA occlusion was estimated at 0.7/100 PY (95% CI: 0.2 to 1.2/100 PY), which is similar to recently reported results: preliminary data of the PREVAIL study (31) (reporting a stroke rate of 0.7/100 PY [95% CI: 0.1 to 5.1/100 PY]) and the PROTECT-AF study (10) (2.2/100 PY [95% CI: 1.3 to 3.5/100 PY]) and to other anticoagulation strategies, including adjusted-dose warfarin (32–35).
Stroke prophylaxis poses a major challenge for patients with NVAF who have contraindications to anticoagulation. The LAA is the most common site for clot formation; therefore, surgical ligation of the LAA is widely practiced in patients undergoing open heart mitral valve surgery (36). Physiologically, percutaneous LAA occlusion should offer the same advantage, and hence, percutaneous closure device use is an interesting alternative in these patients. However, there is a paucity of clinical evidence behind the efficacy and safety of percutaneous LAA occlusion devices. Thus, we performed the first reported systematic review of pooled observational studies to compile the available contemporary evidence.
We estimated the rate of ischemic stroke to be 0.7/100 PY, which is similar to the preliminary PREVAIL trial data, but less than the rate reported in the PROTECT-AF trial (10). The estimated stroke rate in our study was either less than or comparable to contemporary strategies utilizing pharmacotherapeutic anticoagulation, suggesting noninferiority of the percutaneous device implantation (10,32–35). We observed a procedure failure rate of 8.4% among all 4 implanted devices, which was comparable to the rate observed in the PROTECT-AF trial (9.1%), CAP registry (5.7%), and preliminary PREVAIL trial data available online (4.9%) (1,31). The rate of any complication was calculated to be 7.1% for all device types and 13.6% in the Watchman device subgroup, which seemed to be higher than what was observed in the PROTECT-AF trial (8.7%), CAP registry (4.1%), and preliminary PREVAIL trial (4.1%) data. We observed access site complications to be most common (8.6%), followed by pericardial effusion (4.1%), which is comparable to the PROTECT-AF trial data that showed these rates to be 3.5% and 4.8%, respectively. We observed no difference in pooled stroke rates with various device types, year, or region. There was no difference in procedural complications by device type. The comparison of pooled efficacy and safety outcomes by device type has not been reported in the literature thus far. The pooled rates of efficacy and safety outcomes of percutaneous LAA occlusion in observational studies look similar, if not better, than the PROTECT-AF trial data, indicating that the percutaneous LAA closure strategy could be a feasible option for stroke prophylaxis in the real world.
The choice of medical therapy versus percutaneous LAA closure is dependent on whether the patient has any contraindication for medical therapy, which inherently creates a selection bias. Systematic pooling of different observational studies with different baseline characteristics may introduce some imprecision in the results as a result of heterogeneity. Several studies included in the systematic analysis have small numbers or have short follow-up periods. Because of the relative rarity of strokes and safety outcomes after percutaneous closure, recurrent events are more likely to be observed in larger study samples with a longer follow-up period. Another limitation arises from the fact that the included studies were conducted over different time periods in different countries utilizing different devices. It has to be taken into consideration that diagnosis and treatment modalities, including referral patterns, might have changed over time. Also, the event rates in observational studies are under-reported compared with randomized controlled trials. Despite these limitations, our review seeks to clarify extensive and confusing published data, by systematic organization and aggregation of available information.
In our systematic review and analysis, percutaneous LAA occlusion for stroke prophylaxis was comparable to historical controls for adjusted-dose warfarin and comparable to other anticoagulation agents, providing evidence for device utilization in NVAF patients unable to receive long-term anticoagulation. Needless to say, more randomized trials are required to conclusively compare LAA occlusion to current medical management strategies.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Bajaj and Parashar contributed equally to this work.
- Abbreviations and Acronyms
- atrial fibrillation
- congestive heart failure, hypertension, age >75 years, diabetes mellitus, and prior stroke or transient ischemic attack
- confidence interval
- left atrial appendage
- number needed to treat
- nonvalvular atrial fibrillation
- Percutaneous Left Atrial Appendage Transcatheter Occlusion
- transient ischemic attack
- Received September 19, 2013.
- Revision received November 20, 2013.
- Accepted November 25, 2013.
- American College of Cardiology Foundation
- Reddy V.Y.,
- Holmes D.,
- Doshi S.K.,
- Neuzil P.,
- Kar S.
- Psaty B.M.,
- Manolio T.A.,
- Kuller L.H.,
- et al.
- Wolf P.A.,
- Abbott R.D.,
- Kannel W.B.
- Lloyd-Jones D.M.,
- Wang T.J.,
- Leip E.P.,
- et al.
- Landmesser U.,
- Holmes D.R. Jr..
- Gangireddy S.R.,
- Halperin J.L.,
- Fuster V.,
- Reddy V.Y.
- Bayard Y.L.,
- Omran H.,
- Neuzil P.,
- et al.
- Bartus K.,
- Han F.T.,
- Bednarek J.,
- et al.
- Agarwal S.,
- Bajaj N.S.,
- Kumbhani D.J.,
- Tuzcu E.M.,
- Kapadia S.R.
- Moher D.,
- Liberati A.,
- Tetzlaff J.,
- Altman D.G.,
- PRISMA Group
- Ostermayer S.H.,
- Reisman M.,
- Kramer P.H.,
- et al.
- Sick P.B.,
- Schuler G.,
- Hauptmann K.E.,
- et al.
- Reddy V.Y.,
- Möbius-Winkler S.,
- Miller M.A.,
- et al.
- Urena M.,
- Rodés-Cabau J.,
- Freixa X.,
- et al.
- Faustino A.,
- Paiva L.,
- Providência R.,
- et al.
- ↵PREVAIL Study: Prospective Randomized Evaluation of the WATCHMAN LAA Closure Device in Patients With Atrial Fibrillation Versus Long Term Warfarin Therapy. Boston Scientific website. Available at: http://www.bostonscientific.com/watchman-eu/.