Long-Term Safety and Efficacy in Continued Access Left Atrial Appendage Closure Registries

JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY

VOL. 74, NO. 23, 2019

ª 2019 PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION

Long-Term Safety and Efficacy in Continued Access Left Atrial Appendage Closure Registries David R. Holmes, JR, MD,a Vivek Y. Reddy, MD,b Nicole T. Gordon, BSEE,c David Delurgio, MD,d Shephal K. Doshi, MD,e Amish J. Desai, MD,f James E. Stone, JR, MD,g Saibal Kar, MDh

ABSTRACT BACKGROUND Long-term data on the safety and efficacy of left atrial appendage closure (LAAC) for stroke prevention in patients with nonvalvular atrial fibrillation remain limited. OBJECTIVES The purpose of this study was to evaluate 4.5- to 5-year data in 2 U.S. Food and Drug Association LAAC mandated registries (CAP [Continued Access to PROTECT-AF] and CAP2 [Continued Access to PREVAIL]) for safety and efficacy. METHODS Two registries of patients implanted with LAAC devices provide the largest source of follow-up data. Both accompanied their respective randomized clinical trials, PROTECT-AF (Watchman Left Atrial Appendage System for Embolic PROTECTion in Patients With Atrial Fibrillation) and PREVAIL (Prospective Randomized Evaluation of the WATCHMAN LAA Closure Device In Patients with Atrial Fibrillation versus Long Term Warfarin Therapy), which used the same endpoints (primary efficacy of composite of stroke, systemic embolism, cardiovascular/unexplained death, and safety). RESULTS CAP included 566 patients with an average follow-up of 50.1 months (2,293 patient-years), and CAP2 included 578 patients with an average follow-up of 50.3 months (2,227 patient-years). CAP2 patients were significantly older and had higher CHA2DS2-VASc (congestive heart failure, hypertension, age $75 years, diabetes mellitus, stroke/transient ischemic attack, vascular disease, age 65 to 74 years, sex category) scores (4.51 vs. 3.88; p < 0.001). Procedural success was similar in both (94%). The primary composite endpoint occurred at a rate of 3.05 per 100 patient-years in CAP and 4.80 per 100 patient-years in CAP2; events contributing to this endpoint were most commonly cardiovascular/unexplained death (1.69 per 100 patient-years for CAP and 2.92 per 100 patient-years for CAP2). Hemorrhagic stroke was significantly less than ischemic stroke (0.17 per 100 patient-years in CAP and 0.09 per 100 patient-years in CAP2), and total stroke rates were significantly less than predicted by CHA2DS2-VASc score (78% reduction with CAP, 69% reduction with CAP2). CONCLUSIONS These registries, which contain the longest and largest follow-up data of patients with the Watchman device, support LAAC as a safe and effective therapy for long-term anticoagulation in patients with nonvalvular atrial fibrillation, and document the lowest rate of hemorrhagic stroke identified in this population. (J Am Coll Cardiol 2019;74:2878–89) © 2019 Published by Elsevier on behalf of the American College of Cardiology Foundation.

From the aMayo Clinic, Rochester, Minnesota; bIcahn School of Medicine at Mount Sinai, New York, New York; cBoston Scientific, St. Paul, Minnesota; dEmory University, Atlanta, Georgia; ePacific Heart Institute, Santa Monica, California; fLegacy Health Cardiology, Portland, Oregon; gNorth Mississippi Medical Center, Tupelo, Missouri; and the hCenter of Advanced Cardiac and Vascular Interventions, Los Angeles, California. Dr. Holmes is a member of the advisory board (unreimbursed) for Boston Scientific. Dr. Reddy has received research grant support from and served as a consultant to Boston Scientific, Abbott, and Biosense Webster (significant); and has served as a consultant for and has equity in Surecor (significant). Ms. Gordon is a salaried employee Listen to this manuscript’s

of Boston Scientific. Dr. Delurgio has received speaking fees and research support from and served as a consultant to Abbott

audio summary by

Medical and Boston Scientific. Dr. Doshi has received research grant support from, served as a consultant for, and served as

Editor-in-Chief

National Principal Investigator for CAP2 for Boston Scientific (significant). Dr. Desai has served as an Advisory Board Member and

Dr. Valentin Fuster on

Proctor for Boston Scientific. Dr. Stone has performed consulting work in the form of physician training and proctoring cases for

JACC.org.

Boston Scientific. Dr. Kar has received research grant support from, served on the Advisory Board for, served as National Principal Investigator for CAP and CAP2, and served as proctor for left atrial appendage closure for Boston Scientific (significant); has received research grant support from and served as a member of the Advisory Board for structural heart disease for Abbott Vascular (significant); and has received personal fees from Valcore, W.L. Gore, and Medtronic. Lucas V. Boersma, MD, PhD, served as Guest Associate Editor for this paper. Manuscript received June 18, 2019; revised manuscript received August 29, 2019, accepted September 16, 2019.

ISSN 0735-1097/$36.00

https://doi.org/10.1016/j.jacc.2019.09.064

Holmes Jr. et al.

JACC VOL. 74, NO. 23, 2019 DECEMBER 10, 2019:2878–89

T

he increasing use of left atrial appendage

PROTECT-AF.

closure (LAAC) devices for stroke prevention

required

Patients

prevention

had of

NVAF

in nonvalvular atrial fibrillation (NVAF) ad-

embolization, were eligible for warfarin, and had a CHADS2 (congestive heart failure, hy-

who are at risk for stroke are either not treated or are

pertension, age >75 years, diabetes mellitus,

not compliant with anticoagulation (1–7). A collection

and

of longer-term data using LAAC devices is important

attack) stroke risk stratification score $1. The

previous

ABBREVIATIONS

and

AND ACRONYMS

stroke/systemic

dresses the unmet clinical need that many patients

stroke/transient

DRT = device-related thrombus GI = gastrointestinal LAA = left atrial appendage

ischemic

LAAC = left atrial appendage closure

to document continued safety and efficacy (2,8–10).

CHADS2 score initially used has been sup-

The initial randomized clinical trials (RCTs) (11–14) of

planted by CHA2DS2-VASc (congestive heart

the Watchman LAAC device (Boston Scientific, Dublin,

failure, hypertension, age $75 years, diabetes

Ireland) versus anticoagulation with warfarin—PRO-

mellitus, stroke/transient ischemic attack,

TECT-AF (Watchman Left Atrial Appendage System

vascular disease, age 65 to 74 years, sex

for Embolic PROTECTion in Patients With Atrial Fibril-

category). The CHA2DS2-VASc was retrospec-

lation) and PREVAIL (Prospective Randomized Evalu-

tively

ation of the WATCHMAN LAA Closure Device In

collected elements that compromise that score.

Patients with Atrial Fibrillation versus Long Term

Important exclusion criteria included a contraindi-

Warfarin

registries

cation to warfarin and the presence of intracardiac

designed to continue accrual of data on longer-term

thrombus or dense spontaneous echo contrast in the

outcomes and allowed continued access to the tech-

left atrial appendage (LAA). Exclusion of warfarin

nology prior to U.S. Food and Drug Administration

contraindicated patients was necessary due to the

(FDA) approval (12). These registries (CAP [Continued

randomization requirement, as well as the need for

Access to PROTECT-AF] and CAP2 [Continued Access

short-term warfarin after device implantation to

to PREVAIL]) represent the largest number and longest

optimize device endothelialization.

Therapy)—had

2879

Continued Access Left Atrial Appendage Registries

accompanying

calculated

using

NVAF = nonvalvular atrial fibrillation

RCT = randomized clinical trial RRR = relative risk reduction TEE = transesophageal echocardiography

prospectively

follow-up of patients implanted with the only FDA-

Both primary efficacy and safety endpoints were

approved device in the context of a clinical trial. This

identical to the RCT. Efficacy was the composite of

paper documents the final 5-year total experience of

stroke (ischemic and hemorrhagic), cardiovascular

CAP and the 4-year follow-up of CAP2.

(CV) death (CV and unexplained), and systemic em-

SEE PAGE 2890

METHODS

bolism. The safety endpoint was the occurrence of life-threatening events (as adjudicated by the clinical events committee) over the entire course of followup,

including

device

embolization

requiring

Device characteristics and implant technique for the

retrieval,

Watchman device have been previously described (11).

requiring drainage, intracranial hemorrhage, gastro-

Briefly, the self-expanding device available in 5 sizes

intestinal bleeding requiring transfusion, and any

has a nitinol frame with fixation barbs and a permeable

bleeding

such

as

pericardial

effusion

bleeding related to the device or procedure that

polyester fabric. After standard transseptal puncture,

necessitated a surgical procedure. This endpoint was

the device is percutaneously delivered using a 12-F

intended to capture all types of serious bleeding,

sheath typically with transesophageal echocardio-

regardless of relationship to the procedure, medica-

graphic (TEE) guidance. After implantation, patients

tions, or unrelated events, but was different than the

were treated with warfarin with an International

time-bound

Normalized Ratio goal of 2.0 to 3.0 and aspirin (81 mg)

employed in later studies.

7-day

procedural

safety

endpoint

for 45 days, then clopidogrel (75 mg) plus aspirin (81 to 325 mg) if no peridevice leak >5 mm was observed during the 45-day follow-up post-TEE, and finally,

CAP2

REGISTRY. Similarly,

the

nonrandomized

institutional review board-approved CAP2 Registry

aspirin (325 mg) was continued indefinitely beginning

(NCT01760291)

6 months post-implant.

Enrollment

CAP REGISTRY. The nonrandomized CAP registry

commenced September 25, 2012, and completed

(NCT00129545) (12) supplements the original RCT

March 21, 2014, with follow-up data available through

PROTECT-AF; enrollment commenced August 7,

4 years on all patients remaining in the trial.

of

supplements 578

subjects

the

PREVAIL

from

48

trial.

centers

2008, and was completed June 30, 2010, and included

As with CAP, CAP2 used identical inclusion/exclu-

566 patients who continued follow-up through their

sion criteria as PREVAIL, and had endpoint adjudi-

5-year visit or until study exit. Patients met the

cation as documented in the previous text for CAP.

identical inclusion/exclusion criteria and were fol-

Differences

lowed with adjudication of endpoints as used in

PROTECT-AF and CAP included an antiplatelet

in

inclusion/exclusion

criteria

from

2880

Holmes Jr. et al.

JACC VOL. 74, NO. 23, 2019 DECEMBER 10, 2019:2878–89

Continued Access Left Atrial Appendage Registries

washout period $7 days prior to implant and a

not having an event or who were lost to follow-up be-

requirement for a CHADS 2 score of 2 unless a patient

ing censored at the time of last documented follow-up.

had a score of 1 and at least 1 of the following: female

Kaplan-Meier curves are presented for graphical

age $75 years, baseline left ventricular ejection

assessment of time-dependent events, and Cox re-

fraction <35%, diabetes or coronary artery disease,

gressions with proportional hazards were applied to

age $65 years, or documented congestive heart

adjust for risk score differences across trials.

failure.

To provide more consistent comparisons across

CAP2 included 2 primary efficacy endpoints and 1

trials, the more contemporary safety endpoint defi-

safety endpoint consistent with PREVAIL (14). The

nitions from CAP2 were utilized for reporting and

first primary efficacy endpoint was identical to CAP

analysis because they were more discrete and mirror

and was defined as a composite of stroke, systemic

the endpoints used in the subsequently implemented

embolism, and CV/unexplained death. A second pri-

Watchman NESTed SAP (New Enrollment PoST

mary efficacy endpoint was included to evaluate the

Approval Surveillance Analysis Plan) (15), which is

ischemic

rate

derived from a subset of the American College of

excluding events occurring within 7 days post-

Cardiology NCDR LAAC Registry. Mortality was

stroke

and

systemic

embolism

procedure. Finally, the primary safety endpoint was

identified from multiple sources including death re-

modified to include only procedure-related events

cords, and correspondence with the families. Due to

within the first 7 days after implant.

the multiple sources, the exact cause of death could

As previously mentioned, CHA 2DS 2-VASc was

frequently not be determined even after review of the

retrospectively calculated. Similarly, the HAS-BLED

information available. Multivariate analysis was used

score did not exist at the time of study initiation;

to evaluate relationships between baseline factors

however, with the exception of labile International

and subsequent mortality using a Cox regression

Normalized Ratio and liver disease in CAP, all other

model and a p value cutoff of 0.05.

components were prospectively collected. A modified HAS-BLED score (for CAP) was calculated retrospectively by assigning 0 points to the 2 missing elements. This conservative estimate for the patient’s bleeding risk on warfarin was calculated, understanding that this approach may represent an underestimate of bleeding hazard. Both registries enrolled patients from a subset of centers participating in the accompanying RCTs, so all operators had experience with both the device and procedure. Procedural success was defined as delivery and release of the device into the LAA. STATISTICAL METHODS. Comparison of both regis-

tries versus the parent trials was facilitated by including patients with the same inclusion/exclusion criteria (excepting the 7-day antiplatelet washout period prior to implant in CAP2), procedural perfor-

Several post hoc analyses were performed to further characterize and contextualize outcomes. 1. Comparison of long-term outcomes of CAP and CAP2 to their prior respective RCTs. 2. Comparison of follow-up events versus expected event rate based on baseline CHA 2DS 2-VASc scores. 3. Device related thrombus. 4. Timing of follow-up stroke rates. 5. Comparison of the periprocedural event rates in CAP and CAP2 with another large contemporary registry: EWOLUTION (Registry on WATCHMAN Outcomes in Real-Life Utilization). 6. Comparison of the procedure safety and efficacy endpoints from the device limbs of the total of both RCTs and registries.

RESULTS

mance by experienced operators, and the same monitoring and similar follow-up schedules. Descriptive

BASELINE CHARACTERISTICS AND PROCEDURAL

statistics are used to present results, as there was no

SUCCESS. Patients in CAP2 were significantly older,

active control in the registries. For continuous vari-

particularly more were age $75 years (p ¼ 0.0083),

ables, mean
SD, median, range, and 95% confidence

and they had higher CHA2DS 2-VASc scores (4.51 vs.

intervals (CIs) are reported. Numbers (%) of patients

3.88; p < 0.001) (Table 1). In CAP, only 4.1% of pa-

experiencing events are given with chi-square tests to

tients had a CHA 2DS2-VASc score of 1, whereas there

assess difference between studies and trends over

were none in CAP2. In contrast, CAP2 patients had

time. Baseline variables across the studies are

lower HAS-BLED scores (2.0 vs. 2.30; p < 0.001).

compared with chi-square or Student’s t-tests as

There were also significant differences in the atrial

appropriate.

and

fibrillation pattern and atrial anatomy with multiple

All

p

values

are

2-sided,

p values <0.05 are considered statistically significant,

lobes present less often in CAP2 patients. (Compari-

without adjustment for multiple comparisons. Time-

sons of registry patients to the predicate RCT are

to-event analyses were performed with all subjects

provided in Online Table 1.)

Holmes Jr. et al.

JACC VOL. 74, NO. 23, 2019 DECEMBER 10, 2019:2878–89

2881

Continued Access Left Atrial Appendage Registries

T A B L E 1 Baseline Characteristics

CAP

CAP2

p Value (Overall, 2-Sided)

73.96
8.27 (566) (44.00, 94.00)

75.28
7.99 (576) (33.00, 94.00)

0.0065

Age $75 yrs

51.8 (293/566)

59.5 (344/578)

0.0083

Age >65 yrs

86.4 (489/566)

91.0 (526/578)

0.0138

Male

65.5 (371/566)

60.4 (349/578)

0.0704

Age, yrs

2.5
1.19 (566) (1.00,6.00)

2.71
1.07 (576) (1.00, 6.00)

<0.0001

3.88
1.50 (565) (1.00, 9.00)

4.5
1.32 (576) (2.00, 9.00)

<0.0001

CHF

19.1 (108/566)

27.1 (156/576)

0.0013

LVEF

56.51
8.93 (565) (29.00, 82.00)

56.34
9.31 (575) (30.00, 82.00)

0.7571

89.0 (503/566)

92.5 (533/576)

0.0413 0.0011

CHADS2 score continuous CHA2DS2-VASc score (continuous)

Hypertension Diabetes

24.9 (141/566)

33.7 (194/576)

History of TIA/stroke

30.4 (172/566)

29.0 (167/576)

0.9315

Vascular disease

45.1 (255/566)

45.8 (265/578)

0.7872

History of PVD

10.4 (59/566)

8.5 (49/578)

0.2603

History of CAD

42.6 (241/566)

42.6 (246/578)

0.9948

2.30
1.08 (564) (0.00, 6.00)

2.00
0.94 (578) (0.00, 5.00)

<0.0001

53.2 (301/566)

53.1 (307/578)

0.9821

0.0 (0/566)

6.2 (36/578)

<0.0001

Paroxysmal AF

42.8 (242/566)

53.5 (309/578)

0.0003

Permanent AF

24.0 (136/566)

14.4 (83/578)

<0.0001

Persistent AF

30.2 (171/566)

25.6 (148/578)

0.0824

Unknown AF

3.0 (17/566)

0.0 (0/578)

<0.0001

HAS-BLED score* (continuous) History of smoking Paced AF

54.4 (308/566)

37.9 (218/575)

<0.0001

LAA length, mm

29.04
6.15 (551) (13.20, 51.00)

28.94
8.59 (574) (1.70, 90.00)

0.8088

LAA diameter, mm

21.11
3.40 (551) (10.00, 36.00)

21.02
6.01 (573) (1.40, 81.50)

0.7499

Multiple lobes

Values are mean
SD (n) (minimum, maximum) or % (n/N). *Score is a modified HAS-BLED retrospectively calculated; 2 components (labile International Normalized Ratio and liver function) not collected in the trials. AF ¼ atrial fibrillation; CAD ¼ coronary artery disease; CAP ¼ Continued Access to PROTECT-AF; CAP2 ¼ Continued Access to PREVAIL; CHADS2 ¼ congestive heart failure, hypertension, age >75 years, diabetes mellitus, and previous stroke/transient ischemic attack; CHA2DS2-VASc ¼ congestive heart failure, hypertension, age $75 years, diabetes mellitus, stroke/transient ischemic attack, vascular disease, age 65 to 74 years, sex category; CHF ¼ congestive heart failure; HAS-BLED ¼ hypertension, abnormal renal or liver function, stroke, bleeding, labile international normalized ratio, elderly, drugs or alcohol; LAA ¼ left atrial appendage; LVEF ¼ left ventricular ejection fraction; PVD ¼ peripheral vascular disease; TIA ¼ transient ischemic attack.

Procedural success rates were similar in both reg-

or other in 3.5% (20 of 566), the latter of which

istries: 94.3% in CAP and 94.3% in CAP2. Using the

included patients who withdrew permission and

procedure safety definition from CAP2 as detailed in

consent.

the previous text, the 7-day event rates were 1.2% and

Ability to discontinue warfarin was documented in

1.4% for CAP and CAP2, respectively. In CAP2, cardiac

95.8% of patients at 45 days when the mandated

perforation, ischemic stroke, and death within 7 days

TEE examination was performed and documented

of the index procedure were 0.5%, 0.2%, and 0.2%,

successful device closure of the LAA. At 60 months,

respectively, with no device embolizations, whereas

94.8% remained off warfarin.

in CAP the incidence of both cardiac perforation and embolization was 0.2% but there were no ischemic

T A B L E 2 Composite Efficacy and Component Details

strokes or deaths.

CAP

In CAP at the time of study closure, the average follow-up was 50.1 months (Q1, median, Q3: 47.6, 59.9, 61.0 months, respectively) with a total follow-up of 2,293 patient-years. Overall protocol scheduled visit compliance was 98%, and the full 5-year followup was completed in 68% of patients (384 of 566). Inability to obtain full 5-year follow-up was related to

Primary efficacy

CAP2

Rate per 100 Patient-Yrs (n Events/Patient-Yrs)

95% CI

Rate per 100 Patient-Yrs (n Events/Patient-Yrs) 95% CI

3.05 (70/2,292.5)

2.4–3.9

4.80 (102/2,125.8)

1.48 (34/2,296.0)

1.1–2.1

2.25 (48/2,131.1)

1.7–2.9

1.30 (30/2,300.1)

0.9–1.9

2.20 (47/2,135.4)

1.7–2.9

Hemorrhagic stroke

0.17 (4/2,359.9)

0.06–0.5

0.09 (2/2,221.3)

0.0–0.3

Systemic embolism

0.04 (1/2,359.8)

0.01–0.3

0.09 (2/2,221.9)

0.0–0.3

CV/unexplained death

1.69 (40/2,363.2)

1.2–2.3

2.92 (65/2,227.2)

2.3–3.7

All stroke Ischemic stroke

4.0–5.8

mortality, which occurred in 101 patients (17.8%);

Primary safety

3.05 (66/2,160.9)

2.4–3.9

N/A

N/A

initial failure to implant the device, which as per

All-cause mortality

4.27 (101/2,363.0)

3.5–5.2

6.24 (139/2,227.2)

5.3–7.3

protocol led to exit from the study in 5.7% of subjects (32 of 566); being lost to follow-up in 5.1% (29 of 566);

CI ¼ confidence interval; CV ¼ cardiovascular; other abbreviations as in Table 1.

Holmes Jr. et al.

JACC VOL. 74, NO. 23, 2019 DECEMBER 10, 2019:2878–89

Continued Access Left Atrial Appendage Registries

C E N T R A L IL L U ST R A T I O N Relative Ischemic Stroke Reductions Compared With Expected Rates Based on CHA 2 DS 2 -VASc Scores*

Thromboembolic Events per 100 pt-yrs

2882

12% 10% 8% 6%

7.10% 5.86% RRR 69%

RRR 78%

4%

2.20% 2%

1.30%

0% Ischemic Stroke Continued Access to PROTECT-AF

Ischemic Stroke Continued Access to PREVAIL Observed

Expected, Based on CHA2DS2-VASc if Untreated Holmes, Jr., D.R. et al. J Am Coll Cardiol. 2019;74(23):2878–89.

Compared with the predicted ischemic stroke rate based on CHA2DS2-VASc (congestive heart failure, hypertension, age $75 years, diabetes mellitus, stroke/transient ischemic attack, vascular disease, age 65 to 74 years, sex category) scores, left atrial appendage closure patients had relative reductions of 78% (4.56 per 100 patient-years) and 69% (4.7 per 100 patient-years) in CAP (Continued Access to PROTECT-AF) and CAP2 (Continued Access to PREVAIL), respectively. *Effectiveness in stroke reduction versus estimated in the absence of therapy for comparable CHA2DS2-VASc scores based on Friberg et al. (24). PROTECT-AF ¼ Watchman Left Atrial Appendage System for Embolic PROTECTion in Patients With Atrial Fibrillation; PREVAIL ¼ Prospective Randomized Evaluation of the WATCHMAN LAA Closure Device In Patients with Atrial Fibrillation versus Long Term Warfarin Therapy; RRR ¼ relative risk reduction.

In CAP2, the average follow-up at the time of this

primary efficacy event included most commonly CV/

analysis was 50.3 months (Q1, median, Q3; 37.9, 50.4,

unexplained death at a rate of 1.69 per 100 patient-

59.7, months, respectively) with a total of 2,227

years (40 per 2,363.2 patient-years; 95% CI: 1.2 to 2.3

patient-years. At 48 months, follow-up had been

per 100 patient-years), and all stroke at 1.48 per 100

completed in 94.9% of patients expected, and 87.4%

patient-years (34 per 2,295.0 patient-years; 95% CI:

of visits expected were completed at 60 months.

1.1 to 2.1 per 100 patient-years). The specific type of

Reasons for exit from the study prior to completion of

adjudicated stroke was different, with hemorrhagic

all follow-up visits include death in 21.8% (126 of

stroke occurring at a rate 0.1 per 100 patient-years

578), failure to implant the device in 4.8% (28 of 578),

compared with ischemic stroke at a rate of 1.30 per

consent withdrawal or withdrawal from study in 4.2%

100 patient-years. Compared with predicted rates of

(24 of 578), lost to follow-up in 2.4% (14 of 578), and

untreated patients with similar CHA2DS2 -VASc risk

the remainder due to other reasons such as investi-

scores, the observed rate of ischemic stroke was 1.30

gator discretion, surgical excision of the LAA, or

compared with the expected rate of 5.86 per

adverse events. Similar to CAP, 92.6% of CAP2 pa-

100

tients were able to discontinue warfarin at 45 days,

(4.56 per 100 patient-years absolute reduction)

while 96.9% discontinued by 12 months.

(Central Illustration).

PRIMARY EFFICACY. C A P r e g i s t r y . The primary ef-

CAP2

ficacy composite endpoint occurred in 70 patients

endpoint occurred in 102 patients, a rate of 4.80 per

(12.4%) a rate of 3.05 per 100 patient-years (70 per

100 patient-years (102 per 2,125.8 patient-years;

2,292.5 patient-years; 95% CI: 2.4 to 3.9 per 100

95% CI: 4.0 to 5.8 per 100 patient-years). As in CAP,

patient-years) (Table 2). Events contributing to this

the events contributing to this event rate included

patient-years,

a

78%

relative

reduction

r e g i s t r y . The primary efficacy composite

Holmes Jr. et al.

JACC VOL. 74, NO. 23, 2019 DECEMBER 10, 2019:2878–89

2883

Continued Access Left Atrial Appendage Registries

in a rate of 3.05 per 100 patient-years (95% CI: 2.4 to

T A B L E 3 Primary Safety (CAP Only)

3.9 patient-years) (Table 3). The most frequent event No. of Events (% of Subjects)

Event Type

Gastrointestinal bleeding

33 (5.8)

was GI bleeding (46 of 578, 8.1%), and of those, 33 (5.8%)

qualified

as

a

primary

safety

event.

Approximately one-half of the 33 GI bleeds occurred

Pericardial effusion with cardiac tamponade

7 (1.2)

Other study-related

5 (0.9)

within the first 45 days (9) and 6 months (7), while

Major bleed requiring transfusion

4 (0.7)

subjects were either on warfarin therapy or dual

Stroke—hemorrhagic

4 (0.7)

antiplatelet therapy. As previously mentioned, CAP2

Stroke—ischemic

2 (0.5)

did not have a similar pre-defined endpoint. Device-

Pseudo aneurysm

2 (0.4)

related thrombus (DRT) rates were similar (Table 4).

Ventricular tachyarrhythmia

2 (0.4)

Prolonged bleeding from a laceration

1 (0.2)

In CAP, 14 subjects had 21 events of device

Rectal bleeding

1 (0.2)

Device embolization

1 (0.2)

month TEE. Of these subjects, 2 had subsequent

Cardiac perforation

1 (0.2)

stroke

Cranial bleed

1 (0.2)

identification within 6 months of the stroke. In

Anemia requiring transfusion

1 (0.2)

CAP2, 25 patients had a DRT. Four (4) of those

thrombus during the trial, most noted during the 12events,

1

of

whom

had

thrombus

patients experienced 5 ischemic strokes; 3 of the 4

CAP ¼ Continued Access to PROTECT-AF.

DRTs occurred within 3 months of implant, and 1 patient had a DRT identified at 3 separate visits (1, 3, and 5 years post-implant).

most commonly CV/unexplained death at 2.90 per 100

patient-years

(65

per

2227.2

patient-years;

95% CI: 2.3 to 3.7 per 100 patient-years), followed by ischemic stroke at 2.20 per 100 patient-years (47 per 2,135.4 patient-years; 95% CI: 1.7 to 2.9 per 100 patient-years) (Table 2). The second primary efficacy endpoint of ischemic stroke and systemic embolism post-procedure was 2.20 per 100 patient-years. As in CAP, adjudicated hemorrhagic stroke was markedly less at 0.09 per 100 patient-years (2 per 2,221.3 patient-years; 95% CI: 0 to 0.3 per 100 patient-years) and based on CHA 2 DS2-VASc risk scores, the expected rate of 7.1% was higher than the observed rate of ischemic stroke of 2.20 per 100 patient-years, demonstrating a 69% relative reduction (4.70 per 100

patient-years

absolute

reduction)

(Central

Illustration). Per Cox regression analysis, baseline factors associated with mortality were age (continuous), age (both $75 and >65 years), CHA2DS2 -VASc score (continuous), baseline serum creatinine level, left ventricular ejection fraction, congestive heart failure, renal disease, history of vascular disease, diastolic

RELATIONSHIP BETWEEN THE PRIMARY RCTS AND REGISTRIES. Registry data for primary efficacy and

CV mortality was plotted in Kaplan-Meier curves of the device limbs and overlaid with the predicate RCT for context (PREVAIL/CAP2: Figures 1A and 1B; PROTECT-AF/CAP: Figures 2A and 2B). Log-rank p value tests show no significant difference in events within the matched registry and RCT. However, there were significant differences in the baseline clinical patient demographics between 2 sets of studies

(PROTECT-AF/CAP

and

PREVAIL/CAP2)

(Table 1), although these were expected due to the changes in the inclusion/exclusion criteria designed to recruit higher-risk patients. Additional analyses were performed to compare primary efficacy in all 4 studies (Figure 3). Unadjusted comparisons of primary efficacy yielded significant differences between the trials (hazard ratio: 1.48; 95% CI: 1.0 to 2.1; p ¼ 0.036), but after adjusting for CHA2D 2-VASc, so the outcome differences were no longer significant

(hazard

ratio:

1.23;

95%

CI:

0.85

to

1.79; p ¼ 0.27).

blood pressure, history of peripheral vascular disease, history of coronary artery disease, history of gastro-

T A B L E 4 Device Thrombus

intestinal (GI) bleed, body mass index, HAS-BLED score (continuous), diabetes, prior major bleeding or

Thrombus subjects

predisposition to bleeding, paroxysmal atrial fibrilla-

Thrombus events

CAP (n ¼ 534)

CAP2 (n ¼ 545)

14 (2.6)

21 (3.9)

21

25

tion (AF), history of carotid artery disease, and per-

Experienced ischemic stroke

2

4

manent AF (using a univariate Cox regression model

Experienced serious adverse event

10

6

0.1

0.2

and p value cutoff at 0.05). PRIMARY SAFETY EVENTS. C A P r e g i s t r y . Primary

safety events were identified in 66 subjects, resulting

Device thrombus-related stroke rate (per 100 patient-yrs) Values are n (%) or n, unless otherwise indicated. Abbreviations as in Table 1.

Holmes Jr. et al.

JACC VOL. 74, NO. 23, 2019 DECEMBER 10, 2019:2878–89

Continued Access Left Atrial Appendage Registries

F I G U R E 1 PREVAIL/CAP2 Primary Efficacy and Cardiovascular Mortality Kaplan-Meier Curves

A

Efficacy 0.25

Event Probability

0.20

0.15

0.10

0.05

0.00 0

6

12

48

54

60

269 238 234 222 215 209 205 196 191 578 531 515 496 473 446 415 390 335

185 252

126 106

B

18

24

30 36 42 Time (Months)

PREVAIL CAP2

Cardiac or Unexplained Death 0.25

0.20 Event Probability

2884

0.15

0.10

0.05

0.00 0

6

269 242 578 542

12

18

237 531

231 513

54

60

227 222 217 208 204 198 495 470 443 417 357 271

134 116

24

36 42 30 Time (Months)

PREVAIL

48

PREVAIL CAP2

CAP2

Kaplan-Meier plots of the device limbs comparing PREVAIL (Watchman LAA Closure Device in Patients With Atrial Fibrillation Versus Long Term Warfarin Therapy) and CAP2 (Continued Access to PREVAIL). There is no statistically significant difference in the composite efficacy endpoint (A) (log rank p value ¼ 0.20). Significant differences in cardiac/unexplained death (B) (p ¼ 0.032) may relate to differences in baseline characteristics.

Holmes Jr. et al.

JACC VOL. 74, NO. 23, 2019 DECEMBER 10, 2019:2878–89

Continued Access Left Atrial Appendage Registries

F I G U R E 2 PROTECT-AF/CAP Primary Efficacy and Cardiovascular Mortality Kaplan-Meier Curves

A

Efficacy 0.25

Event Probability

0.20

0.15

0.10

0.05

0.00 0

6

12

18

24

463 566

397 518

382 503

370 488

360 468

B

36 42 30 Time (Months)

48

54

60

345 454

321 401

314 391

235 246

337 436

327 420

PROTECT AF CAP

Cardiac or Unexplained Death 0.25

Event Probability

0.20

0.15

0.10

0.05

0.00 0

6

12

18

24

30 36 42 Time (Months)

48

54

60

463 566

403 524

389 512

381 497

372 479

360 465

333 424

324 412

243 263

352 449

PROTECT AF

340 439

PROTECT AF CAP

CAP

Kaplan-Meier plots of the device limbs comparing PROTECT-AF (Watchman Left Atrial Appendage Closure Technology for Embolic Protection in Patients With Atrial Fibrillation) and CAP (Continued Access to PROTECT-AF) in the (A) primary composite efficacy endpoint or (B) cardiac or unexplained death. Neither endpoint was statistically different (p ¼ 0.19 and p ¼ 0.12, respectively) for efficacy in cardiac/unexplained death.

2885

Holmes Jr. et al.

2886

JACC VOL. 74, NO. 23, 2019 DECEMBER 10, 2019:2878–89

Continued Access Left Atrial Appendage Registries

F I G U R E 3 All Studies Primary Efficacy Kaplan-Meier Curves

Efficacy 0.25

Event Probability

0.20

0.15

0.10

0.05

0.00 0

6

12

18

24

463 566 269 578

397 518 238 531

382 503 234 515

370 488 222 496

360 468 215 473

PROTECT AF

30 36 42 Time (Months)

48

54

60

337 436 205 415

321 401 191 335

314 391 185 252

235 246 126 106

345 454 209 446

CAP

327 420 196 390

PREVAIL

PROTECT AF CAP PREVAIL CAP2

CAP2

Kaplan-Meier plots of the composite of primary efficacy of the device limbs of both randomized controlled trials and their respective registries unadjusted comparisons documented that PREVAIL/CAP2 had worse outcome than PROTECT/CAP. When adjusting for differences in patient characteristics, the differences were not significant (p ¼ 0.27). Abbreviations as in Figures 1 and 2.

DISCUSSION

adverse events with more severe and extensive comorbidities.

Following each RCT of the device, 2 matched sequential registries

were

performed (CAP

and

CAP2) to collect ongoing safety and efficacy data

2. Procedural success rates were excellent, and complications in this higher-risk group of patients were lower than in the respective RCTs.

while the device approval was under review by the

3. During 1-year follow-up, the ability to safely dis-

FDA. These 2 registries included 566 and 578 pa-

continue anticoagulation was high at approxi-

tients, respectively, representing more device pa-

mately 95%.

tients in each study than were in the RCTs, and

4. The rate of hemorrhagic stroke is the lowest re-

they have the longest follow-up reported to date.

ported, with a rate of 0.17 per 100 patient-years in

The major findings of these 2 registries are as

CAP and 0.09 per 100 patient-years in CAP2. The

follows.

rate of ischemic stroke was also very low at 1.30 per

1. Compared with the accompanying RCTs, the reg-

100 patient-years and 2.20 per 100 patient-years,

istry patient population was at higher risk of

respectively.

Holmes Jr. et al.

JACC VOL. 74, NO. 23, 2019 DECEMBER 10, 2019:2878–89

Continued Access Left Atrial Appendage Registries

5. Compared with the predicted ischemic stroke rate

lower rates than the underlying stroke risk that is the

based on CHA2DS2 -VASc scores, LAAC patients had

focus of treatment. The comparable stroke rates to

relative reductions of 78% (4.56 per 100 patient-

the randomized patients, despite the increased un-

years) and 69% (4.90 per 100 patient-years) in

derlying risks, further underscores the potential

CAP and CAP2, respectively.

benefit of closure therapy in these patients.

6. At a mean of 48.8 months in the 1,144 patients of

Although U.S. labeling requires patients to be

whom 57.2% were $75 years at baseline, mortality

eligible

(all-cause) occurred in 20.8% of patients.

accommodate post-procedure healing, it is not an

for

short-term

oral

anticoagulation

to

unreasonable hypothesis that registry patients may There is increasing information about the inter-

have had relative contraindications to long-term

of

anticoagulation. Perhaps a more similar set of sub-

stroke, and increasing age, with AF being associated

jects would be found in the European EWOLUTION

with a 5-fold increase in stroke risk (16–18). There is

Registry of 1,020 patients treated in routine clinical

also increasing awareness about the complexities of

practice where the median follow-up is 2 years (2).

optimal stroke prevention. Given the fact that with

The majority of patients were felt to have a relative or

NVAF, the pathophysiology of stroke has been the

absolute contraindication to anticoagulation and

LAA as the source of thrombus in approximately 90%

were typically treated with a variety of antiplatelet

of cases, focus has been centered on mitigating that

regimens. However, in comparing the baseline char-

(3). Although anticoagulation has been the standard

acteristics from the registries to EWOLUTION, several

of care, due to the issues of its application in clinical

similarities are seen. Patient ages were similar, as

practice either because of a variety of relative or ab-

were risk profiles based on CHA 2DS2-VASc scores:

solute contraindications and patient noncompliance

EWOLUTION and CAP2 patients were at 4.5, but CAP

rates of 50% to 60% by 1.5 to 2.0 years, local therapy

registry patients were slightly lower at 3.88. There

with

frequency

were differences in mortality across all studies. At 2

(6,17–20). Although there are 2 RCTs and multiple

years, the Kaplan-Meier rate of mortality in EWOLU-

registries (1), LAAC still has a Class IIb indication in

TION was 16.4% whereas the CAP and CAP2 yielded

the 2016 European Society of Cardiology guidelines,

Kaplan-Meier rates at the same time point of 7.2% and

as well as the 2019 American College of Cardiology/

7.6%, respectively. The differential in mortality rates

Heart Rhythm Society guidelines, which indicate that

may be in part related to differences in baseline

relationship

LAAC

among

is

used

AF,

increased

with

incidence

increasing

it may be considered in patients who are at increased

characteristics (e.g., CHA 2DS2-VASc) and fundamental

of stroke and who have contraindications to long-

differences in other patient characteristics (e.g., 73%

term anticoagulation (Level of Evidence: B-NR) (18).

of patients in EWOLUTION were contraindicated for

The field continues to grow with multiple devices

anticoagulation), and that the U.S. registries excluded

available globally, but only Watchman is approved in

patients with life expectancies <2 years. In EWOLU-

the United States. Longer-term data is needed on

TION, although specific details of the process of

both safety and efficacy both from registries as well as

ascertainment of cause of death are not presented, it

the current new RCTs.

is reported that the most common causes of death

When comparing the registries to the predicate

were noncardiovascular. The difference between CAP

RCTs, patients were slightly older and at a signifi-

and CAP2 may relate to the higher risk based on the

cantly higher risk of stroke. Although not statistically

CHA 2DS2-VASc score at baseline.

significant, registry patients had more prior strokes

In the CAP and CAP2 registries, a primary com-

and less prior treatment with warfarin than their

posite efficacy endpoint consisting of stroke, sys-

randomized counterparts. This is not surprising as the

temic embolism, and CV/unexplained death was

risk of randomization may have dissuaded partici-

used. That endpoint was seen in 12.4% of CAP pa-

pants from participation if they were further along on

tients with a rate of 3.05 per 100 patient-years,

the spectrum of being poor long-term anticoagulation

which

candidates. In addition, as all operators in the regis-

PROTECT-AF at 2.24 per 100 patient-years. The

tries gained implant experience in the RCTs, major

most common event was CV/unexplained death at a

complication rates were <1.5%, which is consistent

rate of 1.69 per 100 patient-years in CAP as opposed

with post-approval rates of 1.5%. These low rates of

to PROTECT, where the most common event was

procedure-related complications further enhance the

stroke at 1.46 per 100 patient-years. For CAP2, the

viability of the therapy, as procedural events occur at

primary composite efficacy rate was 4.80 per 100

was

higher

than

the

device

arm

of

2887

2888

Holmes Jr. et al.

JACC VOL. 74, NO. 23, 2019 DECEMBER 10, 2019:2878–89

Continued Access Left Atrial Appendage Registries

patient-years versus 3.65 per 100 patient-years in

However, once DRT is identified, more frequent im-

PREVAIL, with CV/unexplained was most common

aging surveillance is indicated.

at 2.92 per 100 patient-years in CAP2 compared with

STUDY LIMITATIONS. Single-arm registries lack pre-

the same measure in PREVAIL at 1.88 per 100 pa-

cisely matched control arms with which to compare

tient-years.

performance for outcome events. We have attempted

Stroke as part of the composite efficacy endpoint

to find similar trials to test consistency of results in

was low in both registries. In CAP, all stroke was 1.48

CAP and CAP2, but results should be seen as contex-

per 100 patient-years. The specific type of adjudi-

tual and directional when evaluating the efficacy of

cated stroke was markedly different, with hemor-

the therapy. In addition, we have attempted to con-

rhagic stroke at 0.17 per 100 patient-years compared

trol for biases between the registries and their asso-

with ischemic stroke at 1.30 per 100 patient-years,

ciated

which were both lower than the same rates in the

contributing to the result differences (such as death)

device limb of PROTECT-AF (0.16 and 1.34 per 100

beyond risk scores that may be more subtle than are

patient-years, respectively). With CAP2, similarly,

captured in baseline data. Finally, the lack of cause of

adjudicated hemorrhagic stroke was very low at

death information prohibits further study into the

0.090 per 100 patient-years and ischemic stroke at

mortality findings.

2.20 per 100 patient-years. Ischemic stroke in the device arm of PREVAIL was lower at 1.67 per 100 patient-years, and hemorrhagic slightly higher at 0.18

RCTs,

but

there

may

be

other

factors

CONCLUSIONS

per 100 patient-years. Despite slight differences in

The final 5- and 4-year data presented from the CAP

rates, the reductions in stroke were consistent with

and CAP2 registries adds to increasing information

device therapy across all 4 trials and significant im-

that local site therapy is an effective and safe alter-

provements versus expected rates in untreated pa-

native to long-term anticoagulation in patients with

tients of similar risk.

NVAF who are at increased risk for stroke. The cur-

As there are no active control groups for comparison in the registries, observed versus expected event

rent data documents the lowest rate of hemorrhagic stroke yet documented in this patient population.

rates were calculated using imputed rates based on

ACKNOWLEDGMENT The authors thank Erin Hynes

risk scores. With ischemic stroke rate of 1.3 per 100

from BSC for her quality check of all numerical values

patient-years in EWOLUTION, there was an 83%

provided in the manuscript.

relative risk reduction (RRR) from expected (7.2% to 1.3%). This magnitude of reduction is consistent with

ADDRESS FOR CORRESPONDENCE: Dr. David R.

CAP, in which a similar analysis documented a 78%

Holmes, Jr., Mayo Clinic, 200 First Street SW,

RRR from expected to observed (5.86% to 1.30%),

Rochester, Minnesota 55905. E-mail: holmes.david@

while CAP2 showed a 69% RRR from expected to

mayo.edu. Twitter: @MayoClinicCV.

observed (7.10% to 2.20%). The more striking finding has been the reduction in hemorrhagic strokes to the lowest that has been reported at 0.17 per 100 patientyears in CAP and 0.09 in CAP2. DRT has been of increasing concern (21–23). Early in the development of these registries, the issue of device thrombus formation had not been highlighted. The etiology still remains unclear but is thought to relate to either failure of endothelialization of the surface or the point of the connector pin. Because of differences in antithrombotic/anticoagulant strategies, it is unclear whether there would be a difference between the 3 different studies. In EWOLUTION, the incidence was 4.1%, whereas in CAP and CAP2, it was 2.6% and 3.9%, respectively. The clinical importance is also not fully explored. Given different treatment regimens and surveillance strategies, the specific timing and frequency of DRT remains unclear and deserves further study with well-designed prospective clinical trials.

PERSPECTIVES COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: Percutaneous deployment of the LAA occlusion device may be beneficial for some patients with AF by eliminating the need for lifelong anticoagulation. Outcomes observed in clinical registries during 5-year follow-up after device deployment document sustained efficacy with very low rates of ischemic and hemorrhagic stroke. TRANSLATIONAL OUTLOOK: Further studies are needed to define optimum post-procedural antithrombotic strategies and compare the outcomes of LAA occlusion against target-specific oral anticoagulant therapy in patients with AF.

Holmes Jr. et al.

JACC VOL. 74, NO. 23, 2019 DECEMBER 10, 2019:2878–89

Continued Access Left Atrial Appendage Registries

REFERENCES 1. Holmes DR Jr., Alkhouli M, Reddy V. Left atrial appendage occlusion for the unmet clinical needs of stroke prevention in nonvalvular atrial fibrillation. Mayo Clin Proc 2019;94: 864–74. 2. Boersma LV, Ince H, Kische S, et al. Evaluating real-world clinical outcomes in atrial fibrillation patients receiving the Watchman left atrial appendage closure technology. Circ Arrhythm Electrophysiol 2019;12:e006841.

Amulet device: one-year follow-up from the prospective global Amulet observational registry. EuroIntervention 2018;14:e590–7.

fibrillation developed in collaboration with EACTS. Revista espanola de cardiologia (English ed) 2017; 70:50.

10. Nielsen-Kudsk JE, Johnsen SP, Wester P, et al. Left atrial appendage occlusion versus standard medical care in patients with atrial fibrillation and intracerebral haemorrhage: a propensity scorematched follow-up study. EuroIntervention 2017;

19. Oldgren J, Healey JS, Ezekowitz M, et al. Variations in cause and management of atrial fibrillation in a prospective registry of 15,400 emergency department patients in 46 countries: the RE-LY Atrial Fibrillation Registry. Circulation

13:371–8.

2014;129:1568–76.

11. Holmes DR, Reddy VY, Turi ZG, et al. Percuta-

20. Hsu JC, Maddox TM, Kennedy KF, et al. Oral

3. Alkhouli M, Noseworthy PA, Rihal CS, Holmes DR Jr. Stroke prevention in nonvalvular atrial fibrillation: a stakeholder perspective. J Am Coll Cardiol 2018;71:2790–801.

neous closure of the left atrial appendage versus warfarin therapy for prevention of stroke in patients with atrial fibrillation: a randomised noninferiority trial. Lancet 2009;374:534–42.

anticoagulant therapy prescription in patients with atrial fibrillation across the spectrum of stroke risk: insights from the NCDR PINNACLE Registry. JAMA Cardiol 2016;1:55–62.

4. Yao X, Abraham NS, Alexander GC, et al. Effect

12. Reddy VY, Holmes D, Doshi SK, Neuzil P, Kar S. Safety of percutaneous left atrial appendage closure: results from the Watchman Left Atrial Appendage System for Embolic Protection in Patients with AF

21. Pracon R, Bangalore S, Dzielinska Z, et al. Device thrombosis after percutaneous left atrial appendage occlusion is related to patient and procedural characteristics but not to

(PROTECT AF) clinical trial and the Continued Access Registry. Circulation 2011;123:417–24.

duration of postimplantation dual antiplatelet therapy. Circ Cardiovasc Interv 2018;11: e005997.

of adherence to oral anticoagulants on risk of stroke and major bleeding among patients with atrial fibrillation. J Am Heart Assoc 2016;5: e003074. 5. Vinereanu D, Lopes RD, Bahit MC, et al. A multifaceted intervention to improve treatment with oral anticoagulants in atrial fibrillation (IMPACT-AF): an international, clusterrandomised trial. Lancet 2017;390:1737–46.

13. Reddy VY, Doshi SK, Kar S, et al. 5-year outcomes after left atrial appendage closure: from the PREVAIL and PROTECT AF Trials. J Am Coll Cardiol 2017;70:2964–75.

6. Yao X, Shah ND, Sangaralingham LR, Gersh BJ,

14. Holmes DR Jr., Kar S, Price MJ, et al. Prospective randomized evaluation of the Watchman Left Atrial Appendage Closure device in patients with

Noseworthy PA. Non-vitamin K antagonist oral anticoagulant dosing in patients with atrial fibrillation and renal dysfunction. J Am Coll Cardiol 2017;69:2779–90. 7. Reddy VY, Mobius-Winkler S, Miller MA, et al. Left atrial appendage closure with the Watchman device in patients with a contraindication for oral anticoagulation: the ASAP study (ASA Plavix Feasibility Study With Watchman Left Atrial Appendage Closure Technology). J Am Coll Cardiol 2013;61:2551–6. 8. Boersma LV, Ince H, Kische S, et al. Efficacy and safety of left atrial appendage closure with WATCHMAN in patients with or without contraindication to oral anticoagulation: 1-year follow-up outcome data of the EWOLUTION trial. Heart Rhythm 2017;14:1302–8. 9. Landmesser U, Tondo C, Camm J, et al. Left atrial appendage occlusion with the AMPLATZER

atrial fibrillation versus long-term warfarin therapy: the PREVAIL trial. J Am Coll Cardiol 2014;64:1–12. 15. Varosy P, Masoudi F, Reisman M, Shipley R, Stein K, Ellenbogen K. Procedural Safety of WATCHMAN implantation: The US Nested Post Approval Study. J Am Coll Cardiol 2018;11 Suppl: A320. 16. Alkhouli M, Alqahtani F, Aljohani S, Alvi M, Holmes DR. Burden of Atrial Fibrillation-Associated Ischemic Stroke in the United States. J Am Coll

22. Kubo S, Mizutani Y, Meemook K, Nakajima Y, Hussaini A, Kar S. Incidence, characteristics, and clinical course of device-related thrombus after Watchman left atrial appendage occlusion device implantation in atrial fibrillation patients. J Am Coll Cardiol EP 2017;3:1380–6. 23. Fauchier L, Cinaud A, Brigadeau F, et al. Device-related thrombosis after percutaneous left atrial appendage occlusion for atrial fibrillation. J Am Coll Cardiol 2018;71:1528–36. 24. Friberg L, Rosenqvist M, Lip G. Evaluation of risk stratification schemes for ischaemic stroke and bleeding in 182,678 patients with atrial fibrillation: the Swedish Atrial Fibrillation cohort study. Eur Heart J 2012;33:1500–10.

Cardiol Clin Electrophysiol 2018;4:618–25. 17. Hayden DT, Hannon N, Callaly E, et al. Rates and determinants of 5-year outcomes after atrial fibrillation-related stroke: a population study. Stroke 2015;46:3488–93. 18. Kirchhof P, Benussi S, Kotecha D, et al. 2016 ESC guidelines for the management of atrial

KEY WORDS atrial appendage, atrial fibrillation, LAAC, NVAF, stroke, Watchman

A PP END IX For a supplemental table, please see the online version of this paper.

2889