Warfarin has been available for 6 decades. In a metaanalysis of 29 trials and 28,044 patients with non-valvular AF, warfarin compared to placebo decreased strokes by 64% (95% confidence interval [CI] 49–74%) and mortality by 26% (6% to 35%). Compared to antiplatelet therapy with aspirin, warfarin reduced strokes by 37% (23% to 48%).34 In recent times, drugs with a predictable dose-dependent anticoagulant effect have been developed. Direct thrombin (factor IIa) inhibitor dabigatran was the first such drug approved by the US Food and Drug Administration (FDA) followed by activated factor X (factor Xa) inhibitors rivaroxaban and apixaban.35-38 Another factor Xa inhibitor edoxaban is awaiting marketing approval.39 These newer
agents have respectively demonstrated non-inferiority to warfarin
for stroke prevention in head-to-head randomized controlled trials
(RCTs) each enrolling in excess of 14,000 patients. Although overall
bleeding rates with the newer oral anticoagulants are similar to
warfarin, intracranial and life-threatening bleeding is lower. On the
other hand, dabigatran, rivaroxaban, and edoxaban have a higher
risk of gastrointestinal bleeding compared to warfarin.35-39 Apixaban
has bleeding rates comparable to aspirin alone and is superior in
preventing strokes.40 Therefore, therapy with warfarin (target INR
2.0-3.0) or the newer anticoagulants is the standard for stroke
prevention in non-valvular AF and CHA2DS2-VASc score ≥2. The
newer anticoagulants are, however, not recommended in patients
with severe renal or hepatic dysfunction nor in those with prosthetic
heart valves.38
Endovascular Occlusion Of Left Atrial Appendage
Many observational studies have evaluated percutaneous LAA
occlusion in patients with non-valvular AF. In a meta-analysis of
17 retrospective studies and 1052 device implantations, the pooled
incidence for stroke at follow-up was 0.7 per 100 patient-years,
and transient ischemic attack (TIA) was 0.5 per 100 patient-years.
Access site complications occurred in 8.6% (95% CI 6.3–11.7%)
and pericardial effusion in 4.3% (3.1–5.9%).41 Several devices for
endovascular implantation have been developed.
• Plaato (Percutaneous LAA Transcatheter Occlusion; ev3,
Plymouth, Minnesota) was the first device specifically designed for
LAA occlusion, but has been abandoned due to lack of financial
sponsorship. Plaato sealed the LAA with a polytetrafluoroethylene
(PTFE) covered self-expanding nitinol cage (diameter range 15-
32 mm).19, 42, 43 A study in CHADS2 ≥1 patients demonstrated
successful implantation in 108 of 111 patients and 2.2% annual stroke
rate over 9.8 month follow-up. Adverse events included 4 deaths,2 strokes, 3 pericardiocentesis, and one case each of emergency
cardiac surgery, hemothorax, brachial plexus palsy, and deep venous
thrombosis.43 Another study in CHADS2 ≥2 patients had successful
implantation in 162 of 180 patients, LAA occlusion was confirmed
in 126 of 140 patients with 2-month TEE and stroke rate was 2.3%
per year. Major adverse events occurred in 12 patients including 2
periprocedural deaths, 6 pericardial tamponades (2 required emergent
surgery), and one device embolism.44
• Watchman (Boston Scientific, St. Paul, Minnesota) is a permeable,
polyester covered, self-expanding nitinol frame (diameter range 21-
33 mm) with fixation barbs and is positioned in the LAA using
transseptal access with a 12-Fr sheath. The entire device sits within
the LAA without projecting out of the ostium.19, 45 Design changes
in the fourth generation Watchman include more spines for better
radial strength, increased stability, and ability to recapture-redeploy
the device.
In the ASA Plavix Feasibility Study With Watchman Left Atrial
Appendage Closure Technology (ASAP), 150 non-valvular AF
patients with CHADS2 score ≥1 and a contraindication to warfarin
underwent Watchman implantation and received dual antiplatelet
therapy for 6 months and aspirin thereafter. There were 13 (8.7%)
serious adverse events and, during the mean 14.4-month follow-up,
there were 3 ischemic and 1 hemorrhagic strokes, while 6 (4%) had
device related thrombi.46 In another study, 59 patients were treated
with Watchman (device was oversized by 15% to 30%) and received
dual antiplatelet therapy for 45 days followed by aspirin alone – there
were 2 pericardial effusions, 3 device thrombi, and 1 thromboembolic
event.47 Two RCTs have evaluated clinical outcomes with the
Watchman device
o Protect-AF: Non-valvular AF patients with CHADS2 ≥1 (mean
2.2)48 were randomized to Watchman LAA closure (n=463) or longterm
warfarin (target INR 2.0-3.0; n=244). Watchman patients were
treated with warfarin for 45 days, followed by aspirin and clopidogrel
for 6 months and subsequently aspirin alone. The Watchman strategy
had 99.9% probability of being non-inferior to warfarin for primary
composite outcome of stroke, cardiovascular death, or systemic
embolism (3.0 versus 4.9 per 100 patient-years respectively at 18
months; 3.0 versus 4.3 per 100 patient-years on extended mean 2.3
year followup). However, the serious adverse events were higher
with Watchman (7.4 versus 4.4 per 100 patient-years) including
major bleeding (3.5%), pericardial effusion (4.8%), and device
embolization (0.6%).48, 49 Device related thrombus occurred in 20
of 478 (4.2%) Watchman patients.50 Following PROTECT-AF,
a non-randomized continued access registry with 460 Watchman
implantations showed improved outcomes with increase in operator
experience – higher implantation success (from 89.5% to 95.0%) and
fewer procedural complications (from 7.7% to 3.7%, including serious
pericardial effusions from 5.0% to 2.2% and procedural strokes from
0.9% to 0%).50
o Prevail: Non-valvular AF patients with CHADS2 score 2.6±1.0
were randomized to Watchman (n=269) and warfarin (n=138).
Watchman patients received short-term warfarin followed by dual
antiplatelet and then aspirin alone similar to PROTECT-AF.
At 18 months, the composite of stroke, systemic embolism, and
cardiovascular/ unexplained death was 6.4% in the Watchman group
versus 6.3% in the warfarin group (relative risk 1.07, 95% CI 0.57–
1.89), though not reaching the non-inferiority criterion. The rate of
stroke or systemic embolism >7 days after randomization was 2.5% versus 2.0% (RR 1.6, 95% CI 0.5–4.2). For Watchman implantations,
the periprocedural composite safety endpoint of all-cause death,
ischemic stroke, systemic embolism, or need for cardiovascular
surgery or major endovascular intervention occurred in 6 of 269
(2.2%; 2 device embolizations, 1 cardiac peroration, 1 pericardial
tamponade).51
• Amplatzer Cardiac Plug (St. Jude Medical, St. Paul, Minnesota) is
not cleared for use in the USA. It is a nitinol device that comprises a
lobe with barbs (shallower than the body of Watchman or Plaato) that
lodges in the body of the LAA to prevent migration. This connects
across a waist to an interconnecting disk that occludes the LAA
ostium (diameter range 16–30 mm). The device can be recaptured
and redeployed. Following implantation, dual antiplatelet therapy
for 1 month and subsequently aspirin alone is recommended.19
A second generation of the Amplatzer Cardiac Plug called the
Amplatzer Amulet (St. Jude Medical, St. Paul, Minnesota) has been
designed with the intention to facilitate the implantation process and
reduce complications.
There have been multiple retrospective reports showing a 95% to
99% implantations success with Amplatzer Cardiac Plug in patients
not suitable for anticoagulation. Procedural complications include
stroke (0–2.3%), device embolism (0–2.3%), and cardiac tamponade
(0–3.5%). Strokes have been reported in 0–2.8% patients in followups
ranging 6-21 months.19, 45, 52 There was a 16% rate of mild peridevice
leak on TEE evaluations among 52 patients from 7 Canadian
centers,53 whereas high rates of device-related thrombus were
reported from Brazil (6 of 85, 7%)54 and Spain (5 of 35, 14%).55
As opposed to Watchman there are no RCT data available for
the Amplatzer Cardiac Plug. The ACP trial (Amplatzer Cardiac
Plug clinical trial; NCT01118299) comparing LAA closure versus
anticoagulation with warfarin or dabigatran has been withheld after
failure to procure the investigational device exemption.20
• Transcatheter Patch (Custom Medical Devices, Athens, Greece)
is used for occlusion of heart defects and comprises a frameless
bioabsorbable device. Balloon inflation is used to appose the device
within the LAA, which then adheres to cardiac tissues by formation
of fibrin over 48 hours. Innovations are being made to accelerate the
adhesion process and optimize percutaneous catheter delivery. In the
initial report on 17 patients undergoing LAA closure, the patch did
not adhere in 3 patients, was placed beyond the ostium of the LAA
in one, and led to sheath thrombosis in one although no strokes were
reported at follow-up of one year.56
• Lambre (Lifetech Scientific, Shenzhen, China) device in placed
in the LAA and articulates at the waist with a component that can
self-orient itself flush with the LAA ostium. The device has been
engineered to be retrievable, and enable repositioning, and it has been
tested in canines.57
Epicardial Closure Of Left Atrial Appendage
Epicardial LAA closure obviates some of the risks associated
with endovascular closure of LAA related to transseptal puncture,
thromboembolism (due to exposure of tissue factor from transseptal
puncture and foreign material of the catheters and implanted device to
systemic circulation), need for intraprocedural and post implantation
anticoagulation, and risk of device dislodgement, erosion and
infection. Epicardial LAA closure can be performed during open
surgery, for example in combination with valve surgery and atrial
maze. Dedicated epicardial LAA closure was initially accomplished with video-assisted thoracoscopic access with selective collapse of the
left lung and surgical pericardiotomy.58 Novel approaches using only
subxiphoid pericardial access have also been developed.59
Surgical Epicardial Left Atrial Appendage Occlusion
AF patients undergoing cardiac surgery can have their LAA ligated.
In a series of 205 surgical mitral valve replacements,58 also had LAA
ligation with a reduction in embolic complications independent of
other predictors.60 In a propensity-score matched cohort of patients
operated by a cardiac surgeon, LAA ligation was associated with
fewer post-operative strokes [0 of 145 (0%) versus 7 of 115 (6.1%)
without LAA ligation].61
It is not uncommon to have incomplete LAA occlusion with
surgical closure. A TEE evaluation published in 2000 showed
incomplete LAA occlusion in 18 of 50 (36%) surgical closures.
Further, 9 of these (50%) had LAA thrombus, and 4 (22%) sustained
a clinical embolic event.58 In another study 94 patients with surgical
LAA closure who underwent TEE prior to electrical cardioversion
for post-operative AF, left atrial thrombus was much more likely with
incomplete LAA occlusion (16 of 34, 47%) versus complete LAA
occlusion (7 of 60, 12%). Suture closure as opposed to oversowing
and amputation of LAA was more likely to have residual flow in the
LAA (51% versus 17%) and have left atrial thrombus (33% versus
14%).62 The pilot Left Atrial Appendage Occlusion Study (LAAOS)
showed suture ligation having a residual leak on TEE in 6 of 11
(55%) cases and staple closure having a residual stump of LAA in 9
of 33 (27%) cases.63 Another study on137 surgical LAA exclusions
similarly showed 77% of suture ligations having residual flow and
27% of LAA excisions having a residual stump.64
LAAOS II incorporated measures to improve efficacy of LAA
closure (1) amputation or stapling of the LAA instead of simple
oversowing or ligation, (2) intraoperative TEE to evaluate successful
closure, and (3) goal for any residual LAA stump to be smaller than
1 cm.19, 65 Overall, surgical excision of the LAA appears to be the
most successful technique.45, 66 The inconclusive success of surgical
LAA exclusion and the potential for a high risk of thromboembolism
with incomplete exclusion makes it difficult to recommend it for all
AF patients undergoing cardiac surgery.38
LAAOS III is an ongoing Canadian multicenter trial due in
2019 with 4-year follow-up on 4700 cardiac surgical AF patients
randomized to LAA occlusion or no occlusion (NCT01561651).45
Specifically designed devices can be used to facilitate quick
occlusion of the LAA during open cardiac surgery
• AtriClip Pro (AtriCure, West Chester, Ohio) can be used to clip the base of the LAA from the epicardial aspect. It has been reported to
be effective in LAA occlusion in ≥96% cases in small series, without
associated complications.67, 68
• Tigerpaw System II (Maquet, Rastatt, Germany) uses a delivery
forceps to place the device, with an opposing series of barb connectors
in a compliant silicone housing, at the base of the LAA. Connectors
on one side have a needle that punctures through the LAA tissue and
catches the receptor mechanism on the other side. In a prospective
60-patient study, the reported mean application time was 27 seconds,
and two patients required adjunctive sutures. No leaks were seen on
90-day TEE among 54 patients, though residual LAA stump was ≥6
mm in 5 patients.69
Percutaneous Epicardial Left Atrial Appendage Occlusion
Aegis Medical (Vancouver, Canada) has developed a percutaneous
subxiphoid epicardial approach with a tool to record bipolar electrograms from its jaws to identify and grab the LAA. A
preloaded suture with a flexible-size loop and a support wire for
fluoroscopic visualization is positioned around the LAA and is
tightened and locked. Loss of LAA electrical activity on the bipolar
electrograms confirms adequate occlusion of the LAA.70, 71 The loop
can be undone and repositioned, or additional loops placed over the
first one if needed.19, 59 Over time the LAA shrinks and atrophies.
Epitek (Bloomington, Minnesota) created a fiber-optic endoscope
for visualizing the LAA to facilitate grasping and closure, but further
development has been abandoned.
Hybrid Epicardial-Endovascular Approach For Left Atrial Appendage Occlusion
Lariat (SentreHeart, Palo Alto, California) suture delivery device
uses a hybrid endocardial-epicardial strategy. An endovascular sheath
is placed across the interatrial septum in the LAA ostium. Following
contrast angiography to define the LAA anatomy, a magnet-tipped
wire is positioned in the LAA. Using percutaneous access a second
magnet-tipped wire in the pericardial space attaches to this wire
to form a rail across the LAA muscle. A preformed suture loop is
positioned epicardially and locked down. Lariat is not feasible when
the LAA diameter measures ≥40 mm or the LAA has a superiorly
directed body or lobe.19
A retrospective series showed successful Lariat placement in 85 of 89 (96%) patients who had a favorable LAA anatomy on CT scan. Four
patients had ≤3 mm residual leak. Complications occurred related
to transseptal puncture in one and pericardial access in 2 patients.
Two patients had severe post-procedural pericarditis, one developed
late pericardial effusion, 2 had late non-embolic strokes, and there
were 2 sudden deaths. Though ineligibility for anticoagulation was
the criterion for Lariat, at 1-year follow-up 55% of patients were
on warfarin.72 In another retrospective study Lariat was placed
successfully in 25 of 27 patients, and in 22 there was no residual
LAA flow at 4-month TEE. There was one LAA perforation, 3
pericarditis, 1 periprocedural stroke, and 1 late non-embolic stroke.73