Surgery

Left Atrial Appendage Occlusion Study (LAAOS):Results of a randomized controlled pilot study of leftatrial appendage occlusion during coronary bypasssurgery in patients at risk for strokeJeff S. Healey, MD,a Eugene Crystal, MD,b Andre Lamy, MD,a Kevin Teoh, MD,a Lloyd Semelhago, MD,a

Stefan H. Hohnloser, MD,c Irene Cybulsky, MD,a Labib Abouzahr, MD,a Corey Sawchuck, MD,a

Sandra Carroll, BSc,a Carlos Morillo, MD,a Peter Kleine, MD,c Victor Chu, MD,a Eva Lonn, MD,a

and Stuart J. Connolly, MDa Toronto and Hamilton, Ontario, Canada, and Frankfurt, Germany

Aim This pilot study assessed the safety and efficacy of left atrial appendage (LAA) occlusion, performed at the time ofcoronary artery bypass grafting (CABG).

Methods and results At the time of CABG, 77 patients with risk factors for stroke were randomized to LAAocclusion or control. The LAA was occluded using sutures or a stapling device. Completeness of occlusion was assessed withtransesophageal echocardiography. There were no significant differences in cardiopulmonary bypass duration, perioperativeheart failure, atrial fibrillation, or bleeding between the 2 groups. During surgery, there were 9 appendage tears, all of whichwere repaired easily with sutures. Among patients having a postoperative transesophageal echocardiography, completeocclusion of the LAA was achieved in 45% (5/11) of cases using sutures and in 72% (24/33) using a stapler, P = .14. The rateof LAA occlusion by individual surgeons increased from 43% (9/21) to 87% (20/23) after performing 4 cases ( P = .0001).After a mean follow-up of 13 F 7 months, 2.6% of patients had thromboembolic events.

Conclusions LAA occlusion at the time of CABG is safe. The rate of complete occlusion improves, to acceptable levels,with increased experience and the use of a stapling device. A large trial is needed to determine if LAA occlusion preventsstroke. (Am Heart J 2005;150:288-93.)

Atrial fibrillation (AF) is an important risk factor for

stroke, and effective strategies for stroke prevention are

needed. In the last 10 years, the use of oral anti-

coagulants in patients with AF has lead to a significant

reduction in stroke.1 Still, stroke remains a major cause

of serious disability and death in AF,2 and additional

preventive strategies are needed. AF accounts for one

sixth of all strokes, and up to 25% of strokes in patients

aged N80 years.3 Many patients undergoing coronary

artery bypass grafting (CABG) eventually develop AF and

are at risk of AF-related stroke. Patients undergoing

CABG aged N75 years, with a history of hypertension or

From the aMcMaster University, Hamilton, Ontario, Canada, bUniversity of Toronto,

Toronto, Ontario, Canada, and cJ. W. Goethe University, Frankfurt, Germany.

Submitted May 9, 2004; accepted September 20, 2004.

Dr. Healey is supported by a research fellowship grant from the Heart and Stroke

Foundation of Canada and Astra Zeneca Canada Ltd.

Reprint requests: Jeff Healey, MD, McMaster University–General Site, 237 Barton St.

East, Hamilton, Ontario, Canada L8L 2X2.

E-mail: healeyj@hhsc.ca

0002-8703/$ - see front matter

n 2005, Mosby, Inc. All rights reserved.

doi:10.1016/j.ahj.2004.09.054

with previous stroke, have risk factors for both AF and

stroke.4 Approximately 90% of left atrial thrombi are

found in the left atrial appendage (LAA),5 and thus

surgical occlusion of this structure is an attractive

method for potentially reducing stroke. In contrast to

procedures for LAA occlusion done independently of

other surgery, occlusion at the time of CABG may be

done with little incremental time, cost, and risk.

However, the safety and feasibility of LAA occlusion, at

the time of CABG surgery, have never been evaluated in

a randomized trial. This pilot study was conducted to

assess the safety and feasibility of surgical LAA occlusion

as a concomitant procedure to CABG.

MethodsThe rationale for this trial has been previously published.6

The study complies with the Declaration of Helsinki and was

approved by the local research ethics board. Patients at a single

university hospital (Hamilton, Canada) were considered for

participation if they were undergoing elective CABG (without

concomitant valve surgery) and had any 1 of the following 4 risk

factors for AF and stroke: age N75 years, hypertension and

age N65 years, previous stroke, or a history of AF. Informed

American Heart Journal

Volume 150, Number 2Healey et al 289

consent was obtained before surgery. All patients with AF at

baseline underwent preoperative or intraoperative transeso-

phageal echocardiography (TEE) to exclude atrial appendage

thrombus. At the time of bypass surgery, the surgeon inspected

the LAA and determined, before randomization, if it could be

successfully occluded. Only suitable patients were then

randomized, using sealed envelopes, to undergo LAA occlusion

or serve as a control. Patients were randomized 2:1, favoring

occlusion. Treatment was not blinded.

In the initial portion of the study, LAA occlusion was done

with sutures, either using an encircling technique or a running

suture, with or without pledgets. As a result of a lower than

expected rate of successful occlusion observed in the first

16 patients, surgeons began to use a stapling device to close

the LAA (Ethicon TX30/TX60, using 1.0 or 1.5 mm, unbut-

tressed staples). Surgeons were free to use sutures in

conjunction with staples if needed. The completeness of

occlusion was assessed with transesophageal echocardiogra-

phy 8 weeks after surgery. The postoperative use of anti-

thrombotic therapy was documented but left to the discretion

of the individual clinicians.

Transesophageal echocardiographyEchocardiograms were performed 8 weeks after surgery and

were read centrally at a core laboratory. The LAA was imaged in

multiple views, and quantitative measurements, including

ostial diameter, length, and area, were taken in the views

where these were maximal. Measurements were performed

with screen calipers. The junction of the anterior left atrial wall

and the appendage defined the base of the appendage for the

assessment of length. In the absence of previous established

criteria, a residual appendage length of N1.0 cm was defined

as an incomplete occlusion, as this represented N50% of the

preocclusion length of the smallest appendage observed in this

study. Any Doppler flow across a suture or staple line into the

excluded portion of the appendage was also considered a

failure of occlusion. The presence or absence of LAA and/or LA

thrombus was evaluated. Patients were also evaluated for other

potential sources of embolism, including aortic atheroma,

patent foramen ovale, or other intracardiac shunts.

Data collection and statistical analysisData were collected at baseline, at the time of CABG, and

throughout the perioperative hospitalization. Patients were

contacted at 6 months, 1 year, and up to 18 months after surgery.

Their physicians were also contacted 12 to 18 months after

surgery to corroborate and clarify any clinical events. Data were

collected regarding death, hospitalizations, stroke, heart failure,

and AF. A modified version of the questionnaire for verifying

stroke-free status was used to enhance the detection of stroke

events.7 Clinical events were adjudicated in a blinded fashion.

Continuous data are expressed as means with SDs and were

compared using a t test or, in the case of nonnormal data, with

a Mann-Whitney U test. Categorical variables were compared

using Fisher exact test.

ResultsScreening and enrollment

During the enrollment period, a detailed screening log

was kept. A total of 1500 patients were screened between

June 2001 and October 2002, of which 427 met the entry

criteria (Figure 1). The main reasons for exclusion were

concomitant valve surgery, emergency surgery, or lack of

identified risk factors for stroke. There were 97 patients

who consented to participate. At the time of surgery,

20 enrolled patients were not randomized because the

surgeon decided that the LAA was unsuitable for closure

(LAA too broad, n = 11; too small, n = 5; too friable, n = 1;

could not access with stapler, n = 1; not specified, n = 2).

Thus, 77 patients were randomized (52 to LAA occlusion

and 25 to control). Characteristics of all patients are

recorded in Table I. There were no differences between

the 2 treatment groups. The average age of all patients

was 71 years, 27% were women, 81% had hypertension,

14% had a prior history of AF, and 12% had a prior history

of stroke.

Echocardiographic resultsPostoperative TEE was performed in 44 of the 52

patients in the occlusion group. After initially giving

consent, 8 patients refused the follow-up echocardio-

gram. Occlusion of the LAA was achieved in 66% of

patients, using the criteria of no flow beyond the line of

occlusion and a residual stump of b1 cm.

Mechanisms of LAA occlusion failureTwo distinct mechanisms of LAA occlusion failure

were observed. When sutures alone were used, failures

were always caused by persistent flow across the

suture line (Figure 2, D). With the stapler technique,

there were no cases of persistent flow across the staple

line, and all failures were caused by deployment of the

staples too distally from the base of the LAA, leaving a

residual stump of N1 cm in length (Figure 2, C). No

thrombus was observed in any residual LAA stump, and

none of these patients experienced any thromboem-

bolic events.

Factors influencing closure successThere was a trend for higher occlusion rates with the

use of a stapling device, which produced complete

occlusion in 24 (72%) of 33 patients, compared with

5 (45%) of 11 patients using sutures alone ( P = .14). If

one limits the definition of a failed occlusion to having

color Doppler flow traversing the separation between

the LAA and the LA body, as proposed by others,8 then

the use of a stapling device was successful in 100% (33/

33) of patients, which is significantly better than sutures

alone ( P = .0001). The use of the stapling device was

not randomized and occurred later during the study, as

surgeons were becoming more experienced.

A total of 7 surgeons performed LAA occlusion,

with individual surgeons performing between 2 and

11 (median 5) cases each. Surgeon experience was a

predictor of successful occlusion, which increased from

Figure 1

Screened 1500

Eligible 427

Consented 97

20 had unsuitable appendage

Randomised 77

Emergency Surgery, Off-Pump Surgery, Concomitant Valve Surgery, or Patients not Meeting Inclusion Criteria

Esophageal Disorder, Patient Preference or Resource Limitation Precluding TEE, Participation in Another Study, Surgeon Preference

LAA Occlusion 52

Control 25

Completed Follow-up

Survey 215

Patient flow in pilot study.

Table I. Characteristics of patients

LAA occlusion(n = 52)

Control(n = 25)

Age (y) 72 F 6 71 F 5Women (%) 27 28Hypertension (%) 75 92History of AF (%) 17 8Prior stroke (%) 17 0No. of grafts (n) 2.4 F 0.9 2.3 F 0.8Postoperative coumadin (%) 27 24

American Heart Journal

August 2005290 Healey et al

43% (9/21) success for a surgeon’s first 4 cases to 87%

(20/23) success for subsequent cases ( P = .002). The

use of staplers and increasing surgeon experience

occurred in parallel, and so it is difficult to determine the

effect of each factor in isolation. Nonetheless, one can

conclude that an experienced surgeon using a stapler

can achieve a high rate of complete LAA occlusion.

Safety of LAA occlusionPerioperative events are summarized in Table II.

Performance of LAA occlusion did not prolong cardio-

pulmonary bypass time, which was 72 minutes in the

occlusion group and 75 minutes in the control group,

and it did not increase perioperative bleeding, the

occurrence of postoperative AF, or diuretic use. There

were 9 intraoperative tears involving the left atrial

appendage (n = 8) or the left atrium (n = 1). One LAA

tear occurred in the control group (during manipulation

of the LAA to determine its suitability for occlusion), and

8 occurred in the occlusion group. Four tears were

related to the stapler (including the left atrial tear),

2 were related to manipulation of the LAA with forceps,

1 was from a suture, and 1 was not specified. The

Figure 2

A, Intraoperative TEE before LAA occlusion, demonstrating a 2.5 cm LAA. B, Same patient after successful LAA occlusion with stapler.C, Incomplete occlusion of LAA using a stapler technique, demonstrating a large residual stump (RS). The initial 5-cm LAA has been reduced to2 cm. No Doppler flow was seen across staple line. D, Incomplete occlusion of LAA using a suture technique, demonstrating persistent flow acrossthe suture line.

Table II. Perioperative events

Occlusion(n = 52)

Control(n =25) P

Cross-clamp time (min) 72 F 27 75 F 39 .63Furosemide/72 h (mg) 161 F 134 156 F 99 .87Total chest tube output (mL) 402 F 230 439 F 276 .53Postoperative AF (%) 23 16 .56

American Heart Journal

Volume 150, Number 2Healey et al 291

frequency of tears did not decrease with increasing

surgeon experience. All tears were repaired promptly

and without complication using sutures.

Thromboembolic eventsTwo (2.6%) patients had perioperative thromboem-

bolic events: one an intraoperative ischemic stroke, and

the other a TIA occurring on the third postoperative day.

The former patient was in AF and had echocardio-

graphic evidence of a patent foramen ovale and bilateral

carotid stenoses. The latter patient was in sinus rhythm

and had no other known sources of emboli. Both

patients were randomized to the occlusion group and

had a successfully occluded LAA. There were no

fatalities, and no stroke was hemorrhagic.

After surgery, randomized patients were followed for

an average of 13 F 7 months, during which time no

additional patients had stroke. Surveys were also sent to

all patients who were eligible for the study but chose not

to participate (Figure 1).7 A of total 215 questionnaires

were completed between 10 and 29 (median 12) months

postoperatively, and 25 patients self-reported either a

transient ischemic attack (TIA) (n = 13) or stroke (n = 12).7

DiscussionThis is the first randomized study of LAA occlusion. It

shows that surgical closure of the LAA is safe and can be

performed at the time of CABG, without significantly

prolonging the length of surgery. Complete occlusion of

the LAA is technically challenging, however, and can be

achieved in approximately 90% of cases performed by an

experienced surgeon, using a stapling device.

Surgical closure of the LAA has been performed for

years, without any randomized data documenting its

safety or efficacy. Although the benefit of LAA occlusion

is unknown, it clearly carries potential risks of bleeding,

fluid overload, and possibly stroke.9-11 This pilot study

found no increase in perioperative bleeding. However,

LAA tears were quite common. Although these tears

American Heart Journal

August 2005292 Healey et al

were minor and were easily repaired, they raise concern

about the use of this specific closure technique in an off-

pump setting. Limited human experience with mini-

mally invasive, off-pump LAA occlusion has also raised

concerns about potential intraoperative bleeding.9

The potential for LAA occlusion to cause heart failure

was previously described when bilateral atrial appen-

dectomies were performed as part of the MAZE

operation.11 As up to 30% of atrial natriuretic peptide is

produced in the atrial appendages,12 removal of these

structures could impair the body’s ability to regulate

volume status. The technique used in this study ligates,

rather than removes, the LAA and leaves the right atrial

appendage intact.13 Using this approach, no excess in

clinical heart failure or perioperative diuretic use

was observed.

A small cohort study has suggested that there might be

an increased risk of stroke among patients who had LAA

occlusion at the time of heart surgery.10 Another study,

based on a retrospective review of patients undergoing

mitral valve replacement, suggests a reduction in the risk

of stroke in patients who had LAA occlusion (3.4% vs

17%, P = .01).8 Although these conflicting results most

likely result from the small size and nonrandomized

design of these studies, it might possibly reflect the

higher rate of successful LAA occlusion observed in the

latter study (90% vs 64%). In the present study, there

were somewhat fewer strokes in the patients random-

ized to LAA occlusion, despite an identical rate of

bcompleteQ occlusion to that seen by Juratli et al.10

However, a large randomized trial is needed before any

conclusions about the effect of LAA occlusion on stroke

can be drawn. Until then, it is important that inves-

tigators specify the completeness of LAA occlusion and

the types of failed LAA occlusion.

Achieving complete LAA occlusion requires careful

attention to technique. A previous study of endocardial

closure of the LAA at the time of mitral valve surgery,

using a double row of prolene sutures, reported only a

64% rate of successful LAA occlusion.14 More recently,

the 90% rate of successful LAA occlusion reported by

Garcia-Fernandez et al8 was attributed to the endocardial

closure technique which used both a purse-string suture

and a running suture. Neither study commented about

the amount of residual LAA remaining proximal to their

sutures. Clearly, to be useful for stroke prevention, a

high rate of LAA occlusion success is required. In our

experience, the use of a stapling device by an experi-

enced surgeon (N4 cases) resulted in a 90% rate of

complete occlusion.

This study was not designed to evaluate stroke out-

comes. Although there were 2 strokes among 77 patients,

both occurred during the perioperative period. Although

a high rate of stroke was seen during the survey follow-

up of nonenrolled patients (5.6% stroke and 6.0% TIA

over 12 months), these numbers may overestimate the

true rate of embolic events for 2 reasons. Patients who

chose not to participate may differ from those who did,

in such a way as to place them at higher risk of embolic

events. As well, the questionnaire used for the survey is

designed to be highly sensitive for stroke events and

thus may overestimate their true frequency.7 Thus, for

the purpose of a large randomized trial evaluating the

effect of LAA occlusion in CABG patients, it may be

worthwhile to include higher risk patients than those in

this study, such as patients having CABG who also have a

history of AF.

There is considerable interest in LAA occlusion as a

means of stroke prevention in AF, including endovascular

and minimally invasive surgical approaches.9 There are

several advantages to evaluating LAA occlusion in patients

undergoing CABG. Because there is no evidence pres-

ently that LAA occlusion prevents stroke, it is preferable

to evaluate this intervention as an adjunctive therapy,

where it poses no additional risk to patients. In addition,

the incremental cost of LAA occlusion at the time of CABG

is very small in comparison to the cost of performing LAA

occlusion as an isolated procedure. More than half a

million CABG procedures are done yearly, and many

patients have AF or are at very high risk of AF. A

randomized trial of LAA occlusion at the time of CABG

could be an efficient way to prove the concept that LAA

occlusion reduces stroke. Before any technique for LAA

occlusion is recommended, a large randomized trial is

required to show that LAA occlusion reduces stroke.

Study limitationsThe main limitations of this study are its small size and

the fact that stapler use was not randomized.

ConclusionsThis is the first randomized trial of LAA occlusion.

Surgical occlusion of the LAA can be successfully

performed at the time of routine CABG, without

significantly increasing operative time, bleeding, or heart

failure. Complete occlusion of the LAA is challenging,

but a high success rate can be achieved with surgeon

experience and the use of a stapling device. A larger

randomized trial is needed to determine if LAA occlusion

reduces stroke.

This research was funded by Guidant Canada.

Some surgical staplers were provided by Ethicon

Endosurgery (Markham, Ontario, Canada).

References1. Hart RG.Antithrombotic therapy toprevent stroke inpatientswithatrial

fibrillation: a meta-analysis. Ann Intern Med 1999;131:492 -501.2. Goldstein LB, et al. Primary prevention of ischemic stroke: a

statement for healthcare professionals from the stroke council of theAmerican Heart Association. Stroke 2001;32:280 -99.

American Heart Journal

Volume 150, Number 2Healey et al 293

3. Wolf PA, Abbot RD, Kannel WB. Atrial fibrillation as an independentrisk factor for stroke: the Framingham study. Stroke 1991;22:983 -8.

4. Connolly SJ, Cybulsky I, Lamy A, et al. Double-blind, placebo-controlled, randomised trial of prophylactic metoprolol for reductionof hospital length of stay after heart surgery: the beta-Blocker LengthOf Stay (BLOS) study. Am Heart J 2003;145:226-32.

5. Blackshear JL, Odell JA. Appendage obliteration to reduce stroke incardiac surgical patients with atrial fibrillation. Ann Thorac Surg1996;61:755 -9.

6. Crystal E, Lamy A, Connolly SJ, et al. Left Atrial Appendage OcclusionStudy (LAAOS): a randomised clinical trial of left atrial appendageocclusion during routine coronary artery bypass graft surgery forlong-term stroke prevention. Am Heart J 2003;145:174 -8.

7. Jones WJ, Williams LS, Meschia JF. Validating the questionnaire forverifying stroke-free status (QVSFS) by neurologic history andexamination. Stroke 2001;32:2232 -6.

8. Garcia-Fernandez MA, Perez-David E, Quiles J, et al. Role of leftatrial appendage obliteration in stroke reduction in patients withmitral valve prosthesis. J Am Coll Cardiol 2003;42:1253 -8.

9. Blackshear JL, Johnson WD, Odell JA, et al. Thoracoscopicextracardiac obliteration of the left atrial appendage for strokerisk reduction in atrial fibrillation. J Am Coll Cardiol 2003;42:1249 -52.

10. Juratli N, Wilkoff BL, Tchou P, et al. Left atrial appendage ligationduring mitral valve surgery may increase the risk of late thrombo-embolic events. J Am Coll Cardiol 2002;39:85A.

11. Yoshihara F, et al. Atrial natriuretic peptide secretion and body fluidbalance after bilateral atrial appendectomy by the maze procedure.J Thorac Cardiovasc Surg 1998;116:213 -9.

12. Al-Saady NM, Obel OA, Camm AJ. Left atrial appendage:structure, function and role in thromboembolism. Heart 1999;82:547 -54.

13. Yoshihara F, et al. Preservation of right atrial appendage improvesreduced plasma atrial natriuretic peptide levels after the mazeprocedure. J Thorac Cardiovasc Surg 2000;119:790 -4.

14. Katz ES, Tsiamtsiouris T, Applebaum RM. Surgical left atrialappendage ligation is frequently incomplete: a transesophagealechocardiographic study. J Am Coll Cardiol 2000;36:468 -71.