This copy is for personal use only. To order printed copies, contact reprints@rsna.

org
680
CARDIAC IMAGING

Pre- and Postprocedural CT of

Transcatheter Left Atrial Appendage
Closure Devices
Prabhakar Rajiah, MBBS, MD,
FRCR, FSCMR Transcatheter left atrial appendage (LAA) closure is an alternative
Mohamad Alkhouli, MD to long-term anticoagulation therapy in selected patients with non-
Jeremy Thaden, MD valvular atrial fibrillation who have an increased risk for stroke. LAA
Thomas Foley, MD closure devices can be implanted by means of either an endocardial
Eric Williamson, MD or a combined endocardial and epicardial approach. Preprocedural
Praveen Ranganath, MD imaging is key to identifying contraindications, accurately sizing the
device, and minimizing complications. Transesophageal echocar-
Abbreviations: LA = left atrium, LAA = LA diography (TEE) has been the reference standard imaging modality
appendage, LAAC = LAA closure, LSPV = left
superior pulmonary vein, TEE = transesopha-
to assess the anatomy for LAA closure and to provide intraprocedur-
geal echocardiography, 3D = three-dimensional, al guidance. However, CT has emerged as a less-invasive alternative
2D = two-dimensional to TEE for pre- and postprocedural imaging. CT is comparable to
RadioGraphics 2021; 41:680–698 TEE for exclusion of thrombus but is superior to TEE for the de-
https://doi.org/10.1148/rg.2021200136 lineation of complex LAA anatomy, measurement for device sizing,
and evaluation of pulmonary venous and extracardiac structures.
Content Codes:
CT provides accurate measurements of the LAA ostial diameter,
From the Department of Radiology (P. Rajiah,
T.F., E.W.) and Department of Cardiology
landing zone diameter, and LAA length, which are vital for accu-
(M.A., J.T.), Mayo Clinic, 200 1st St SW, Roch- rate sizing of the device. CT allows evaluation of the relationship
ester, MN 55905; and Department of Radiol- with the pulmonary veins and other adjacent structures that can be
ogy, UT Southwestern Medical Center, Dallas,
Tex (P. Ranganath). Recipient of a Magna Cum injured during the procedure. CT also simulates procedural fluo-
Laude award for an education exhibit at the roscopic angles and provides evaluation of the interatrial septum,
2019 RSNA Annual Meeting. Received May
25, 2020; revision requested November 2 and
which is punctured during LAA closure. CT also provides a more
received February 9, 2021; accepted February convenient method for the evaluation of postprocedural complica-
12. For this journal-based SA-CME activity, tions such as incomplete closure, peridevice leaking, device-related
the authors, editor, and reviewers have disclosed
no relevant relationships. Address correspon- thrombus, and device dislodgement.
dence to P. Rajiah (e-mail: [email protected]).
©
Online supplemental material is available for this article.
RSNA, 2021
©
RSNA, 2021 • radiographics.rsna.org
SA-CME LEARNING OBJECTIVES
After completing this journal-based SA-CME
activity, participants will be able to:
„ Identify the role of CT in the evalu-
Introduction
ation of patients for transcatheter left Patients with atrial fibrillation have a fivefold higher risk of stroke
atrial appendage (LAA) closure (LAAC). due to thromboembolism when compared with healthy patients
„ Describe the optimal CT protocol for (1,2). The left atrial (LA) appendage (LAA) is the source of a
pre- and postprocedural evaluation for thrombus in up to 91% of patients with nonvalvular atrial fibrillation
transcatheter LAAC.
(3), which results in poor LA contractility, increased blood stasis,
„ Discuss the common postprocedural
increased thrombogenicity, and LA remodeling (4). Oral anticoagu-
complications of LAAC.
lants including warfarin and direct agents such as rivaroxaban and
See rsna.org/learning-center-rg. apixaban are the first-line therapy for prevention of stroke in these
patients. However, many patients with nonvalvular atrial fibrillation
are not eligible to take lifelong anticoagulation medications because
they have a prohibitive risk of bleeding or other medical conditions.
Furthermore, less than 50% of patients who are prescribed antico-
agulation therapy are found to be adherent to therapy at 1 year (5).
Transcatheter LAA closure (LAAC) has emerged as a feasible al-
ternative to address the unmet clinical need for prevention of stroke in
selected patients with nonvalvular atrial fibrillation. Landmark random-
ized controlled trials PROTECT AF (Watchman Left Atrial Appendage
System for Embolic Protection in Patients with Atrial Fibrillation) and
PREVAIL (6) were instrumental in the determination of procedural
RG • Volume 41 Number 3 Rajiah et al 681

cedure. The Watchman FLX (Boston Scientific),

TEACHING POINTS Amplatzer Cardiac Plug (Abbott), Amplatzer
„ Transcatheter LAA closure (LAAC) has emerged as a feasible
Amulet (Abbott), Wavecrest (Coherex), Lambre
alternative to address the unmet clinical need for prevention
of stroke in selected patients with nonvalvular atrial fibrillation.
(Lifetech Scientific) and Ultrasept (Cardia) are
„ CT is comparable to TEE for evaluation of a thrombus but is
examples of endocardial devices. The Amplatzer
superior to TEE for providing comprehensive information on Cardiac Plug and Amulet have an anchoring
the complex and variable size and shape of the LAA, for accu- lobe and a proximal occluder; other devices are
rate device sizing, and for evaluation of extracardiac structures. anchored only by occluders with fixation barbs.
„ A thrombus in the LAA is an absolute contraindication for Commonly used hybrid endocardial-epicardial
LAAC. devices include the Lariat (SentreHeart) and Si-
„ For the Watchman device, the length of the dominant lobe erra (Aegis Medical). Only some of these devices,
is measured from the landing zone plane to the tip of the such as the Watchman FLX and Lariat devices,
dominant lobe.
are approved by the U.S. Food and Drug Admin-
„ At CT, a normal LAAC device is seated in the LAA and sealed to
istration (FDA) (7,8,10,13).
its wall, without any rotation around its central axis.

Role of Imaging in Transcatheter

LAAC Planning
success and the safety of transcatheter LAAC (7). The main goals of preprocedural imaging in
The efficacy of LAAC was further supported in transcatheter LAAC are to evaluate for contra-
several observational studies (7–10). As a result, indications, assess the shape of the LAA, size the
transcatheter LAAC for prevention of stroke has device accurately, and minimize procedural ma-
been assigned a Class IIb recommendation, with a nipulations and complications. The ideal LAAC
level of evidence of B per the 2019 American Heart procedure uses one catheter, one device, and one
Association, American College of Cardiology, and deployment. Transesophageal echocardiography
Heart Rhythm Society guidelines for the manage- (TEE) is the established standard for preproce-
ment of nonvalvular atrial fibrillation (11). dural imaging for LAAC. TEE adequately allows
exclusion of an intracardiac thrombus, evalua-
LAA Anatomy tion of LAA size and shape, characterization of
The LAA is a fingerlike extension of the antero- the atrial septum for transseptal puncture, and
lateral LA that is located in the left atrioventricu- characterization of any pericardial effusion (14).
lar groove between the pulmonary trunk superi- Nonetheless, TEE is an invasive procedure, re-
orly and the left ventricle inferiorly. The LAA has quires sedation, and is often not well tolerated by
three regions: the ostium, which is the opening elderly patients. MRI is equally effective as TEE
from the LA; the neck; and the body (Fig E1). in the evaluation of cardiac thrombi (15), but
The LAA has one or more lobes, which are pat- limited data on its use for LAAC preprocedural
ent outpouchings from the main body that are evaluation exist (16). Lower spatial resolution
demarcated by an external crease. These lobes and longer examination times are the challenges
may be located in a different plane or direction of MRI.
than that of the body (12). The LAA has a thin
wall (approximately 1 mm) and a rough inter- CT for Planning of Transcatheter
nal surface because of thick (>1 mm) pectinate LAAC
muscles or trabeculations. This reflects the devel- CT has emerged as an important imaging modal-
opment of the LAA from the primordial LA, un- ity in preprocedural evaluation for transcatheter
like the smooth-walled LA, which develops from LAAC. High isotropic spatial resolution, good
the outgrowth of pulmonary vein buds from the temporal resolution, a large field of view, mul-
lungs (4). The LAA has contractility and quadri- tiplanar reconstruction capabilities, and rapid
phasic blood flow in a normal rhythm, which are turnaround time are the strengths of CT. CT is
impaired in patients with atrial fibrillation. comparable to TEE for evaluation of a thrombus
but is superior to TEE for providing comprehen-
Transcatheter LAAC Devices sive information on the complex and variable size
Several transcatheter LAAC devices are avail- and shape of the LAA, for accurate device sizing,
able, each with a different deployment approach, and for the evaluation of extracardiac structures.
anchoring mechanism, and anchoring position Unlike TEE, CT is a noninvasive technique and
(Table E1) (Fig 1). These devices can be placed does not require sedation.
by means of either an endocardial or a hybrid
endocardial-epicardial approach (Fig 2, Fig E2). CT Protocol
The hybrid devices have the advantage of not A multidetector CT scanner with at least 64
requiring anticoagulation therapy after the pro- detector rows is required for a high-quality scan.
682 May-June 2021 radiographics.rsna.org

Figure 1. Commonly used transcatheter LAAC devices. Endocardial devices include the Watchman, Watchman FLX, and Amplatzer
Amulet. The Lariat is the commonly used combined endocardial-epicardial device.

Oral hydration with 500 mL of water can be

given to the patient before image acquisition to
increase LA preloading. The scan is most com-
monly performed with retrospective electrocar-
diographic gating because of dynamic changes
in the size of the LAA during the cardiac cycle
(17,18). Some centers perform prospective
electrocardiographic triggering in the ventricular
systolic phase. A biphasic protocol of contrast
material injection is used. Bolus tracking in the
ascending aorta is used to trigger the imaging ac-
quisition. Images are acquired from the carina to
the diaphragm at expiratory breath hold. An ad-
ditional delayed phase scan is performed through
the heart 60 seconds after the arterial phase,
either with prospective electrocardiographic gat-
ing or helical high-pitch acquisition to evaluate
for a thrombus. Further details are provided in
Table E2.

Critical Preprocedural CT Parameters

Evaluation of Contraindications.—A thrombus Figure 2. Schematic illustration of the endocardial LAAC ap-
in the LAA is an absolute contraindication for proach. A sheath is introduced through the femoral vein and
LAAC. Patients with atrial fibrillation have an advanced to the right atrium (RA). Through a transeptal punc-
ture, the sheath is advanced to the LAA, where the device is
inherent increased risk of a thrombus. In addi-
deployed under fluoroscopic and echocardiographic guidance.
tion, they may also have a pacemaker or defibril- AO = aorta, LV = left ventricle, RV = right ventricle.
lator, which is prone to thrombus formation and
can serve as a potential source for paradoxical
embolism across the site of transseptal puncture. slow flow include imaging with the patient in a
At CT, an LAA thrombus is seen as a well- prone position, reconstructing an iodine map
defined hypoattenuating filling defect. A similar from spectral CT data, and using contrast mate-
appearance may be seen owing to incomplete rial injection techniques such as a split bolus or
mixing of contrast material and blood because double injection (21–24).
of slow flow (19). A thrombus generally shows Specific to hybrid LAAC techniques, peri-
lower attenuation (<100 HU) than that of slow carditis and pericardial adhesions from previous
flow and persists during the delayed phase, surgery or radiation are contraindications. At CT,
whereas slow flow does not persist (Fig 3). pericardial thickening, calcifications, effusions,
Combined arterial and delayed phase CT has and contrast enhancement may be seen. Addi-
higher accuracy for detection of a thrombus tional contraindications for epicardial techniques
compared with that of TEE (sensitivity, 100%; include pectus excavatum (Fig E3), a posteriorly
specificity, 99%; positive predictive value, 92%; rotated heart, and an LAA positioned behind
negative predictive value, 100%) (20). Other CT the pulmonary trunk or adjacent to a coronary
strategies for distinguishing a thrombus from bypass graft (25) (Fig 4).
RG • Volume 41 Number 3 Rajiah et al 683

Figure 3. Contraindication for LAAC

in a 63-year-old woman who was un-
der evaluation for transcatheter LAAC.
(a) Two-chamber reconstructed CT
image shows a hypoattenuating fill-
ing defect (arrow) in the tip of the LAA.
(b) Two-chamber reconstructed CT im-
age obtained 60 seconds later shows
persistence of the filling defect (arrow),
which is consistent with a thrombus.

Figure 4. Contraindications for placement of hybrid endocardial-epicar-

dial devices in three patients. (a) Short-axis reconstructed CT image in a
47-year-old man shows circumferential pericardial calcification (arrows).
(b) Axial CT image in a 43-year-old man shows a severe pectus excava-
tum deformity (arrow) with a Haller index score of 4.5. (c) Coronal recon-
structed CT image in a 72-year-old man shows a venous bypass graft from
the aorta to the obtuse marginal artery (straight arrow), which is situated
directly adjacent to the LAA (curved arrow).

muscle wall of the LA vestibule separates it from

the mitral annulus (4). Hence, it is ambiguous to
define the ostial plane along the LA vestibule.
To ensure accurate and reproducible measure-
ments, two-dimensional (2D) oblique transverse
measurement is recommended over 2D orthogo-
nal or three-dimensional (3D) measurements
LAA Measurements.—The LAA ostium, device (26). Since the LAA ostium size is dynamic
landing zone, and LAA length are key determi- (Movies 1, 2), ostium measurements are obtained
nants of sizing and seating an LAAC device. The during the cardiac phase at which the LAA os-
LAA ostium is roughly planar and typically of an tium is the largest. This is typically in the ventric-
elliptical shape (ie, in 69% of patients) (4). Occa- ular end systolic phase (30%–40% R-R interval).
sionally the ostium is round (6%), triangular (8%), From the axial CT images, an oblique section is
shaped like a water drop (8%) or a foot (10%), or created perpendicular to the long axis of the left
irregular (4,26,27). The ostium margins are well- circumflex artery. In this long-axis two-chamber
defined posterosuperiorly by the tip of the Cou- view of the heart and the LAA, the ostium is
madin ridge, which is called the limbus and sepa- defined as the plane between the left circumflex
rates the ostium from the left superior pulmonary artery and the Coumadin ridge. An en-face view
vein (LSPV) (28) (Fig 5). However, the ostium is generated at this location for measurement
is ill-defined anteroinferiorly, where the smooth of maximal and minimal diameters, area, and
684 May-June 2021 radiographics.rsna.org

Figure 5. LAA ostium in a 59-year-old man.

Oblique 3D volume-rendered endocardial view
CT image of the LA (tan) shows that the extent
of the ostium of the LAA (blue) is well defined at
the posterior and superior aspect, where it is bor-
dered by the Coumadin ridge (straight arrow)
that separates the LAA ostium from that of LSPV.
Anteroinferiorly, the ostium is ill-defined (curved
arrow) at the region of the LA vestibule.

Figure 6. Measurement of LAA ostium. (a) Axial CT image of the heart is used to identify the left circumflex coronary artery (arrow).
A plane that is perpendicular to the long axis of the left circumflex artery is identified. Multiplanar reformat axes are shown on the
long-axis images (blue and green lines). (b) Long-axis-view CT image of the left ventricle and LAA shows the LAA ostium, which is
defined in the plane between the left circumflex artery (straight arrow) and the Coumadin ridge (curved arrow). (c) En-face view of
the ostial plane in b shows the LAA ostium in cross-section (arrow), where it is measured. The maximum diameter, minimum diam-
eter, perimeter, and area can be measured.

perimeter (Fig 6, Movie 3). There is a wide range Movie 4). For the Amplatzer Cardiac Plug and
of ostial diameters, from 12.1 mm to 38.8 mm, Amulet devices, the landing zone is located 10
with a mean ± standard deviation of 21.9 mm ± mm and 12–15 mm inward of the LAA orifice,
4.1, and the ostium is generally larger and round respectively, and the diameters are measured on
in patients with atrial fibrillation (29). an en-face-view CT image that is obtained at
Optimal LAAC device seating is achieved these positions perpendicular to the walls of the
when the atrial end of the device is flush with LAA (17) (Fig 8). The landing zone for the Am-
the plane of the ostium. To accomplish this, the platzer Cardiac Plug and Amulet devices should
LAAC device is delivered at a certain distance in be greater than or equal to 10 mm in width,
the LAA from the ostial plane that is referred to and the distal lobe should be 1.5–3.4 mm larger
as the landing zone. Each specific LAAC device in diameter than that of the landing zone (30).
has a different optimal landing zone, according The maximum landing zone diameter is used for
to manufacturer guidelines. At CT, the landing the Watchman device, and a perimeter-derived
zone is identified on the long-axis LAA view. For diameter is used for the Amplatzer Cardiac
the Watchman device, the landing zone is located Plug and Amulet devices. Watchman devices are
10–20 mm inside the LAA from the Coumadin available with maximum landing zone diameters
ridge. The landing zone diameters for the Watch- of 21–33 mm, and Watchman FLX devices are
man device are measured on an en-face-view CT available with maximum diameters of 20–35 mm.
image obtained at a line that connects the LAA Amplatzer Cardiac Plug and Amulet devices are
adjacent to the left circumflex artery to the point available with perimeter-derived diameters of
10–20 mm inward of the Coumadin ridge (Fig 7, 16–34 mm and 11–31 mm, respectively.
RG • Volume 41 Number 3 Rajiah et al 685

Figure 7. Measurement of landing zone and length of LAA for the Watchman device in a
52-year-old man. (a) Vertical long-axis-view CT image of the left ventricle (LV) shows how
the landing zone for a Watchman device is identified at a distance of 10–20 mm (red dotted
line) distal to the Coumadin ridge (arrow). The pink line is the LAA ostium. The landing zone
diameter is measured in the plane between this point and the point where the left circumflex
artery is located (green line). The length of the LAA is also measured on the same image from
the landing zone to the tip of the dominant lobe (blue line). (b) En-face-view CT image shows
the cross-sectional view of the landing zone (arrow), where the maximum diameter, minimum
diameter, and perimeter are measured.

length of the dominant lobe is measured from

the landing zone plane to the tip of the domi-
nant lobe (Fig 7). This length should be larger
than the maximum diameter of the landing zone,
because the sheath-constrained length of the
Watchman device is often similar to the uncon-
strained diameter of the deployed device (32).
For the Amplatzer Cardiac Plug and Amulet
devices, the depth of the LAA is measured from
the LAA orifice to the roof of the LAA (Fig 8).
This length should be greater than 10–12 mm for
safe implantation of an Amplatzer Cardiac Plug
Figure 8. Measurements for the Amulet device in
or Amulet device.
a 52-year-old man. Long-axis-view CT image of the
left ventricle shows the landing zone for the Amulet Device Sizing.—Accurate measurements are
device, which is measured 12–15 mm distal (dotted imperative to optimal sizing of the LAAC device.
red line) to the LAA ostium (pink line). The tip of
the Coumadin ridge (arrow) demarcates the poste-
The consequences of inaccurate device sizing can
rior boundary of the LAA ostium. The landing zone be devastating. Undersizing can lead to device
(green line) is perpendicular to the walls of the LAA. embolization and peridevice leaks, whereas over-
The depth of the LAA (blue line) is measured as a sizing can result in device erosion and rupture.
perpendicular line from the center of the LAA ostium
to the roof of the LAA.
Most centers use the landing zone measurements
at CT in conjunction with those at TEE to select
the appropriate device size. If there is discrepancy
LAA length ranges from 24.9 mm to 85.7 between TEE and CT measurements, the larger
mm (mean length, 49.4 mm ± 9.1) (31). The of the two diameters is used. Measurements
length of the dominant lobe is essential for de- at TEE are typically smaller (diameters and
vice sizing. A minimum length is required to en- lengths 0.7–5.2 mm smaller) than those at CT
sure that the atrial end of the implanted device (29,33,34). There are several potential reasons
does not extend into the LA cavity. Insufficient for this difference between modalities. TEE mea-
lengths increase the risk of device embolization surements are made oblique to the true maximal
and pericardial effusion. Device-specific mini- diameter at four angles: 0°, 45°, 90° and 135°,
mum LAA lengths are given in manufacturer with the largest diameter selected (Fig 9, Fig E4).
guidelines. Because of the curvilinear nature of the LAA, the
At CT, the length is measured as the long-axis length of the LAA is often underestimated with
view of the LAA. For the Watchman device, the TEE. In addition, LAA measurements are often
686 May-June 2021 radiographics.rsna.org

Figure 9. TEE and CT measurements of the LAA ostium. (a) Illustration shows the measurement of the LAA
ostium at TEE at four different angles: 0°, 45°, 90°, and 135°. (b) Illustration shows the measurement of the
diameter at CT.

smaller at TEE because the patient is dehydrated

as a result of preprocedural fasting.
A single-center study (35) of transcatheter
LAAC showed that the use of TEE alone would
have allowed the inappropriate exclusion of 23%
of patients because of undersizing. The use of CT
has a device-sizing accuracy of 100% compared
with the use of TEE, which is 53% accurate. Ac-
curacy is higher with 3D TEE than it is with 2D
TEE (29,36).
In specialty centers in which intracardiac
echocardiography is used for procedural guid-
ance, preprocedural TEE is not required, and Figure 10. Windsock-type LAA in a 49-year-old
man. Coronal reconstructed CT image shows the
only CT-based measurements are used (2). CT windsock type of LAA (arrow), which has a single
is associated with higher procedural success dominant lobe larger than 4 cm long, with an angle
than that of TEE (100% vs 82%, respectively) of more than 100° from the ostium. Tiny accessory
and procedural efficiency, with a lower average lobes are seen.
number of devices used (1.2 vs 1.8, respectively)
(10), fewer guide catheters, shorter procedure
times, and lower radiation and contrast mate- the number of lobes, the bend angle, and the
rial doses (37). The optimal device size is the presence of a dominant lobe (Table E3). The
next size up (3–6 mm larger) from the maxi- “windsock” type of LAA is seen in 19% patients
mum diameter for the Watchman device and and has a single long dominant lobe (Fig 10, Fig
from the perimeter-based mean diameter for E5). It is the type of LAA that allows the easiest
the Amplatzer Cardiac Plug or Amulet devices. measurement and implantation of LAAC devices.
For example, if a 24-mm-diameter device is The “chicken wing” type, which is the most
identified on the basis of CT, a 27-mm device common type (48% of patients) of LAA, has a
is implanted. This intentional oversizing by sharp bend (<100°) in its proximal or middle
10%–15% reduces the risk of a peridevice leak. portion or is folded back on itself (Fig 11, Fig
However, it also increases the risk of damage to E6) (27). Patients with the chicken wing type of
adjacent structures. LAA have the lowest incidence of stroke (40).
This type of LAA increases procedural complex-
LAA Size, Shape, and Variations.—LAAs vary ity because of the short LAA length proximal to
widely in the size, shape, and number of lobes, the bend and the difficulty of navigating the cath-
which has implications for device selection and eter around the sharp bend (32). For this reason,
procedure planning (17). These characteristics can the linear distance from the ostium to the sharp
be evaluated at 2D or volume-rendered 3D CT. bend is reported at CT (Fig 11b). Because of the
The majority (54%) of LAAs have two lobes, 25% sharp bend, the chicken wing tip is located close
have three lobes, and 20% have a single lobe (12). behind the pulmonary trunk, which is a contrain-
Four main types of LAA anatomy have been dication to using devices with a hybrid approach.
described (27,38,39) and are distinguished by However, there are multiple potential approaches
RG • Volume 41 Number 3 Rajiah et al 687

Figure 11. Chicken wing type

of LAA in a 56-year-old woman.
(a) Coronal 2D image of a chicken
wing–shaped LAA (arrow), in which
there is a sharp bend (<100°) in
the proximal segment. (b) Mea-
surement of distance from the LAA
ostium (pink dotted line) to the
bend (red solid line) in a chicken
wing type of LAA. Note that the
LAA is located directly posterior to
the pulmonary trunk (PT), which
makes this a contraindication for
the hybrid endocardial-epicardial
approach. AO = aorta.

The “cauliflower” type is the least common

type (3% of patients) of LAA. It is short, mul-
tilobed, and irregular, with thick trabeculations
and complex architecture and no dominant lobe
or bends (Fig 13, Fig E8). Patients with this
type of LAA have an eightfold higher rate of
stroke than do those with the chicken wing type
of LAA (27). Because of the absence of a domi-
nant lobe and unclear shoulders, identification
of the landing zone may be challenging. Also,
because of their short length, cauliflower-type
Figure 12. Cactus-type LAA in a 77-year- LAAs may not accommodate LAAC devices,
old woman. Coronal reconstructed CT portending a theoretically higher risk of device
image shows the cactus type of LAA (ar- embolization.
row), which has a short dominant lobe Other notable types of LAAs include the cone
and small accessory lobes.
and bilobed types. The cone type of LAA has a
single short dominant lobe, with a rapidly taper-
ing transverse diameter from the ostium to the
tip (Fig 14, Fig E9). Because of the steep gradi-
ent in size, even a minor alteration in position of
the landing zone substantially changes the device
size. In addition, the tapering increases the pres-
sure on the distal end of the LAAC device, po-
tentially increasing the risk of inadequate seating
and device embolization (27,40). The bilobed (ie,
“whale tail”) type has two large codominant lobes
of similar sizes (Fig 15, Fig E10), which provides
Figure 13. Cauliflower-type LAA in a
two feasible axes of implantation. This type is
61-year-old woman. Reconstructed 2D technically challenging because the common
coronal CT image shows the cauliflower proximal portion of the LAA is often too short to
type of LAA, which is short and multi- accommodate a device; thus, implantation of two
lobed, without a dominant lobe or bend.
distal devices is potentially required (41).
LAA types are complex and diverse, and an
individual LAA may not fit perfectly into one
for implantation, and hence, caution should be type. Other classification systems exist, includ-
exercised when correlating CT and fluoroscopic ing one in which eight types are described: hook,
measurements. sea horse, finger, knob, wing, spiral, flame, and
The “cactus” type of LAA (30% of patients) arrowhead types (42). Regardless of the spe-
is short, is multilobed (more than two lobes that cific type, a complex shape is often difficult to
each measure >1 cm in length), has a dominant measure and is associated with a higher risk of
lobe, and does not have a bend (Fig 12, Fig E7). a thrombus and procedural complexity (Fig 16)
As with the windsock type, measurement and (40,41). LAAs with complex shapes, large sizes
implantation are straightforward. (>4 cm), and large lobes and those that extend in
688 May-June 2021 radiographics.rsna.org

Figure 15. Bilobed-type LAA in a 72-year-old

Figure 14. Cone type of LAA in a 63-year-old woman. 2D reconstructed coronal CT image
woman. 2D reconstructed coronal CT image shows shows an LAA with a bilobed or whale tail anat-
an LAA with a cone-type anatomy (arrow), in which omy (arrow). Because of the short proximal por-
there is a rapidly tapering transverse diameter from tion of the LAA, it is often challenging to implant
the base to the apex. Because of this steep gradient, a device in the common proximal portion. Im-
a minor alteration in measurement location can lead plantation of two devices may be needed distally
to a big change in device size. in the two lobes.

Figure 16. Complex LAA anatomy types.

(a) Coronal reconstructed CT image in a
69-year-old woman shows a complex multi-
lobed anatomy that is extending in different di-
rections. Note that there are multiple implanta-
tion angles. (b) Axial CT image in a 63-year-old
man shows a large (approximately 6 cm) LAA
(arrow), which makes the Lariat device unsuit-
able for use in this patient.

multiple directions with lobes greater than 4 cm (43). The width of the ridge has implications for
apart are not suitable for the epicardial approach. device sizing and potential complications for disk-
In LAs without a single long-axis plane in and lobe-type devices. The ridge is usually narrow
the dominant lobe that is perpendicular to both (≤5 mm), especially superiorly, particularly in
the ostium and landing zones (ie, chicken wing, a cauliflower type of LAA (43). Thinner ridges
cauliflower, bilobed), a modified curved planar are thought to increase the risk of LSPV ostial
reformation technique can be used to measure impingement by the device (26). Conversely,
the landing zone and length (Fig E11, Movie 5). wider ridges (>5 mm) can mitigate this risk and
Any variant accessory lobe that is located proxi- allow more flexibility with device oversizing. The
mal to the estimated landing zone should be re- Coumadin ridge width is best measured as a
ported, because it can result in a peridevice leak linear distance on the 3D volumetric endocardial
if it is not excluded at the time of device deploy- view CT images (Fig 18) (26). An alternative
ment (Fig 17). Large muscular trabeculations technique with easier postprocessing is to measure
extending inferiorly from the LAA to the mitral the 2D linear distance from the transverse ostium
vestibule have a higher risk of a peridevice leak plane image (Fig 19).
and perforation, whereas small pits in the antero-
inferior ostium may entrap catheters and may not Pulmonary Venous Anatomy and Relation-
be occluded by the LAAC (4,30). ships.—In most transcatheter LAAC techniques,
the delivery catheter is anchored in the LSPV
Coumadin Ridge.—The Coumadin ridge (ie, the because this stabilizes the guidewire across which
left lateral ridge) is an infolding of the lateral atrial the delivery sheath is introduced into the LAA.
wall between the LAA and LSPV that encom- Thus, the anatomy of the left pulmonary veins
passes the ligament and oblique vein of Marshall and the relationship to the LAA ostium have a
RG • Volume 41 Number 3 Rajiah et al 689

Figure 17. Accessory lobes proximal to the

landing zone in a 71-year-old man. Two-cham-
ber CT image shows the presence of a large
accessory lobe (arrow) proximal to the landing
Figure 18. Measurement of the Coumadin ridge
zone (black line) with a chicken wing shape. The
in a 39-year-old man. 3D endocardial view of the LA
pink line is the LAA ostium. Implantation of the
shows the Coumadin ridge (arrow) between the LAA
device in this landing zone would result in in-
and the LSPV, measuring 6 mm.
complete LAA occlusion.

view from the LA parallel to the Coumadin ridge

(44). On these views, the superior margin of the
LAA ostium is compared with the superior mar-
gin of the LSPV ostium. Three LAA relationships
to the LSPV have been described: (a) the middle
type is the most common (60%–65% of patients),
in which the superior margins of the LAA and
LSPV are both at the same level (Fig E13); the
high type (25%–30% of patients), in which the
LAA superior margin is cranial to the LSPV
superior margin (Fig E14); and the low type
(5%–10%), in which the LAA superior margin is
caudal to the LSPV superior margin (Fig E15)
Figure 19. Measurement of the Coumadin ridge in (4,45). LAAC device placement in an LAA os-
a 39-year-old man. Axial CT image shows the mea- tium that is close to the LSPV may interfere with
surement (arrow) of the Coumadin ridge in between subsequent pulmonary vein ablation and venous
the LAA and LSPV.
return (32). At some centers, practitioners place
a precurved wire in the LA rather than anchoring
substantial effect on procedural success (26). At it in the LSPV to perform catheter exchanges and
CT, assessment of the pulmonary veins can be avoid damage to the pulmonary vein. With this
done rapidly, with multiplanar reformation or procedural modification, preprocedural evaluation
volumetric views of the LA. A spectrum of varia- of the relationship of the LAA with the LSPV is
tion exists in the number, sizes, and branching less important.
patterns of the pulmonary veins. The standard
number of pulmonary veins is four: two right Relationship to Other Adjacent Structures.—CT
veins and two left veins. Any deviations from stan- is the optimal modality for providing a 3D assess-
dard anatomy, particularly a common left pulmo- ment of important adjacent structures that may be
nary vein trunk or vein stenosis, are reported (Fig injured during the LAAC. The pulmonary trunk
E12). With a common left ostium, both the left and left pulmonary artery are posteroinferior to
superior and inferior veins can be affected after the LAA and are separated by the transverse sinus
the procedure. LAAC can exacerbate any preex- of the pericardium (Fig E16). The left circumflex
isting left pulmonary vein stenosis. When CT is artery and great cardiac vein are located in the left
performed for planning pulmonary vein isolation atrioventricular groove, inferior to the LAA (Fig
and ablation and LAAC, a more comprehensive 20a, 20b), and have a higher potential for injury
assessment of pulmonary vein sizes and branching when an Amulet device is used (30). Occasionally
pattern is needed. (in 14% of patients), an S-shaped sinoatrial nodal
The relationship between the LSPV and the artery originating from the left circumflex artery
LAA can be assessed at CT, either on multiplanar may course between the LAA and the LSPV (30)
reformation views or a volumetric endocardial or adjacent to the LAA (Fig 20c). A persistent
690 May-June 2021 radiographics.rsna.org

Figure 20. Relationship of the LAA to adjacent structures in three patients. (a) Two-chamber reconstructed CT image in a 37-year-
old woman shows the left circumflex coronary artery (straight arrow) and the great cardiac vein (curved arrow), which are located
inferior to the LAA. (b) Axial oblique CT image in an 82-year-old man shows the sinoatrial nodal artery (straight arrow), originat-
ing from the left circumflex artery (curved arrow) and coursing adjacent to the LAA. (c) Axial oblique reconstructed CT image in a
39-year-old man shows a left-sided superior vena cava (arrow), which is located between the LAA and the LSPV.

left superior vena cava, which is seen in less than with the exact location dependent on the cra-
1% of the population, also runs between the LAA niocaudal tilt of the LAA and the specific device
and the LSPV (30,43). The Bachmann bundle, (48). The anterosuperior approach has the risk of
an integral part of the atrial conduction system, is accidental puncture of the aortic root. From the
located slightly lateral to the ligament of Marshall estimated puncture site or fossa ovalis (Fig 22a),
and encircles the LAA neck (4). Venous coronary lateral, anterior, and superior distances to the LAA
bypass grafts to diagonal or marginal branches orifice can be estimated at CT (49). Transseptal
may be located close to the LAA (30). The left puncture can be simulated at CT with a virtual
phrenic nerve is located in the fibrous pericardium guidewire directed from the posteroinferior inter-
posterolateral to the LAA, overlying the tip of the atrial septum through the LA and into the LAA
LAA in 59% of patients and the neck in 23% of ostium along the LAA long axis (Fig 22b, 22c).
patients (46). In the epicardial approach, the left These images can help the proceduralist visualize
internal mammary artery and inferior epigastric the catheter approach and select an appropriate
artery may also be injured (32). delivery sheath.

Interatrial Septum.—The interatrial septum is Left Atrium.—The size of the LA is a reasonable

evaluated for abnormalities that preclude LAAC surrogate for the severity of LA myopathy, which
or make it challenging. Such abnormalities include is often present in patients who are undergoing
atrial septal defects, a patent foramen ovale, septal LAAC. LA size can help proceduralists ap-
aneurysms, and lipomatous hypertrophy (47) (Fig proximate the catheter travel distance from the
21). With a patent foramen ovale, a lower trans- interatrial septum to the LSPV and the LAA. The
septal puncture may be needed because of the LA volume can be calculated with 2D biplane
absence of sealing of the anterosuperior rim (30). area-length and prolate ellipse methods (Fig
Lipomatous hypertrophy results in thick antero- E17) or more sophisticated 3D threshold-based
superior and inferior rims, with constriction of methods (50). The presence, position, and size
the fossa ovalis. Indwelling septal occluder devices of any LA diverticula or accessory LAAs are also
(Fig 21e) and fibrosis from surgical patches make reported because they can theoretically compli-
transseptal puncture challenging (30). cate catheter manipulation and device delivery.
Furthermore, site-specific puncture has been Diverticula and accessory LAAs are seen in up to
increasingly recognized as a key element for 15% of patients and are associated with embolic
procedure success. A puncture location that al- strokes (30,51). Atrial diverticula are slightly
lows a straight trajectory along the long axis of larger than accessory LAAs, have a smooth wall,
the LAA perpendicular to its ostium is needed to and are more common on the anterosuperior
minimize catheter exchange and mitigate the risk wall, whereas accessory LAAs have rough inter-
of device tilting or a peridevice leak (48). This nal architecture and are more common in the left
usually requires a posteroinferior septal puncture, lateral wall (Fig E18) (52).
RG • Volume 41 Number 3 Rajiah et al 691

Figure 21. Abnormalities of the interatrial septum in five patients. (a) Axial CT image in a 55-year-old man shows a large secundum
type of atrial septal defect (arrow) located in the mid interatrial septum. (b) Short-axis CT image in a 71-year-old man shows a patent
foramen ovale, with a small jet of contrast material from the left to the right (arrow). (c) Four-chamber CT image in a 49-year-old man
shows a large atrial septal aneurysm (arrow), which is diagnosed when the combined excursion of the interatrial septum in systole
and diastole is greater than 1.5 cm. (d) Axial CT image in a 36-year-old woman shows lipomatous hypertrophy of the atrial septum
(arrows) measuring larger than 20 mm, with characteristic sparing of the fossa ovalis. (e) Sagittal maximum intensity projection CT
image in a 51-year-old man shows an atrial septal occluder (arrow).

Figure 22. Interatrial septal puncture planning in a 54-year-old man. (a) Four-chamber reconstructed CT image shows the normal
appearance of a thinned fossa ovalis (arrow) in the mid atrial septum. (b) Coronal oblique CT image shows a virtual guide of the tra-
jectory of the sheath from the septal puncture into the LAA (orange line). (c) Volume-rendered 3D CT image of the left atrium shows
the virtual guide (solid orange line) and the corresponding fluoroscopic angle (dotted orange line). CRA = cranial, IAS = interatrial
septum, RAO = right anterior oblique.

Simulated Fluoroscopic Projections.—The is difficult to assess the exact anatomy with 2D

LAAC procedure is usually performed with cine fluoroscopy and contrast material injec-
fluoroscopy at a standard angle such as right tion. Views that reduce foreshortening of the
anterior oblique (RAO) 30/caudal (CAU) 20. length of the LAA do not necessarily correspond
This view does not always show the maximal to the coplanar angle for the ostium. Hence,
ostium diameter or the LAA length, because it standardized angles have a high potential for
692 May-June 2021 radiographics.rsna.org

Figure 23. Coplanar fluoroscopic

angles in a 57-year-old man. A copla-
nar fluoroscopic angulation of the right
anterior oblique (RAO) 40/cranial (CRA)
0 view was derived for the plane that
is perpendicular to the maximum diam-
eter of the LAA ostium.

Figure 24. 3D printing. Flex-

ible 3D printed models of
the left atrial appendage in a
52-year-old man before (a)
and after (b) placement of a
simulated Watchman device
(open arrow in b). The ade-
quacy of the Watchman device
sizing and positioning can be
assessed with the models. The
left atrial appendage (white * in
a) , the inferior pulmonary vein
(black solid arrow in a and b),
superior pulmonary vein (white
solid arrow in a and b), and the
mitral valve orifice (black * in a
and b) are also shown.

incorrect measurement and device sizing and life-size physical model of the LAA. Different
also make LAA cannulation and device delivery devices of varying sizes and catheters can be
challenging. Simulated procedural fluoroscopic tested with these 3D-printed models to evaluate
coplanar views can be created from volumetric positioning (ie, implantation depth and angula-
CT images to help with the procedure. With tion), anchoring, sealing, and stability (Fig 24).
commercially available software, personalized Deformation maps can be used to evaluate the
coplanar fluoroscopic angles that correspond to affect on the LAA wall (53). The benefits of 3D
the maximal ostial diameter, maximal landing modeling and printing for accuracy in device
zone diameter, and maximal LAA length can be selection and procedural efficiency have been
generated. The ostial plane view can be used to demonstrated in multiple studies (54–56). The
measure the ostium and navigate the catheter use of 3D-printed models correlates better with
engagement of the ostium (Fig 23). The land- the final implanted device size than does the use
ing zone plane and the plane of the longest LAA of CT or TEE (54,55), resulting in less device
length view can be used to measure the length waste, shorter procedure times, lower contrast
and to guide device deployment (35). Use of material and radiation doses, and fewer post-
these personalized angles improves device sizing procedural complications such as peridevice
and reduces device waste, procedural time, and leaks and incomplete closures (56). 3D printing
contrast material and radiation doses. models can shorten the learning curve for early
operators (35). However, 3D printing uses only
3D Modeling and Printing: Virtual static images and is not widely available.
Procedural Planning
3D models of the LA and LAA can be created CT Report Template
from CT data to plan LAAC procedures. With A standardized reporting process is strongly en-
the use of dedicated software, the LA and LAA couraged for preprocedure CT for transcatheter
are segmented from CT images to create a LAAC. This report should contain the key elements
virtual 3D model. With the use of a stereolitho- we have discussed: LAA measurements; LAA
graphic file, the virtual model is then printed shape, size, and relationship to adjacent structures;
with a 3D printer to generate a personalized the anatomy of the Coumadin ridge and the inter-
RG • Volume 41 Number 3 Rajiah et al 693

atrial septum; and fluoroscopic angles. A content after the procedure, TEE is performed to exclude
report template is shown in Appendix E1. an intracardiac or device-related thrombus or a
substantial leak (>5 mm). If there is no complica-
Procedural Imaging tion, warfarin is discontinued and clopidogrel is
TEE and fluoroscopy are the principle intraproce- added. Contrast-enhanced MRI can allow local-
dural imaging techniques for transcatheter LAAC. ization of the LAA occluder devices and detec-
In specialized centers, intracardiac echocardiogra- tion of leaks but is limited by artifacts and long
phy is used instead of TEE for procedural guid- imaging times (61). CT is increasingly used to
ance. The advantages of intracardiac cardiography clarify equivocal TEE findings or to characterize
include its ability to allow direct visualization of the complications that are identified at TEE. CT
the LAA and the septum and to avoid the need can replace TEE for characterization of postpro-
for general anesthesia and additional operators cedural complications, specifically for thrombi
(57,58). Via femoral venous access, the guide- and leaks (62). Higher tube voltages are used at
wire and delivery sheath are advanced to the postprocedural CT to mitigate device-related
right atrium. Under TEE or intracardiac echo- metallic artifacts. The device is evaluated in mul-
cardiographic guidance, transseptal puncture is tiple dedicated planes, with images obtained by
performed (Movies 6, 7), the LA is accessed, the centering on the screw hub and generating planes
sheath is anchored in the LSPV, and a cannula is perpendicular to the coves of the parachute in the
placed in the LAA. Fluoroscopy with cine angiog- Watchman device and perpendicular to the disk
raphy and TEE are then used to examine the LAA in the Amplatzer Cardiac Plug device (62–64)
size and shape and fine tune the catheter position (Fig 25, Movies 15, 16). At CT, a normal LAAC
(Fig 3, Movies 8, 9). The real-time intraproce- device is seated in the LAA and sealed to its
dural fluoroscopic images can be correlated or wall, without any rotation around its central axis.
fused with preprocedural 3D CT data to improve The Watchman device has a compression size
guidance. The device delivery sheath is then posi- of 10%–20% compared with the original device
tioned, and the device is deployed in the landing size. The lobe of the Amplatzer Cardiac Plug de-
zone with the use of simultaneous fluoroscopy and vice is located two-thirds of the way distal to the
multiplanar TEE (Movies 10–12). After deploy- left circumflex artery, with its axis in line with the
ment, device compression of 8%–20%, with the LAA neck axis (63).Typically, after 45 days, there
device shoulders in the LAA ostium should be is no contrast material opacification either in or
seen in all TEE views (Movies 13, 14). A tug or around the device or distally in the LAA due to
push-pull test is performed with TEE or fluoros- endothelialization. Equal contrast enhancement
copy to ensure that the device is anchored. Peride- between the LAA and LA indicates the absence
vice leaks can be seen as a Doppler TEE color jet of endothelialization. Incomplete endothelializa-
adjacent to the device. For hybrid devices such as tion is diagnosed when the contrast enhancement
the Lariat, a magnet-tipped endocardial guidewire in the LAA is lower than that in the LA by at
is placed via a transseptal approach at the LAA least 50 HU (63). However, recent studies have
apex, and a balloon occlusion catheter is placed at shown that complete thrombosis in the device is
the LAA ostium. In addition, through pericardial seen in only 27% of patients, even at 6 months,
puncture and epicardial access, a magnet-tipped and is seen more often in windsock- and cauli-
epicardial guidewire with a pretied suture loop flower-type LAAs (62). Endothelialization cannot
is introduced. The endocardial and epicardial be assessed if there is a peridevice leak. The
guidewires act as rails for snaring the LAA os, with presence of any remnant LAA or a lobe between
release of a pretied suture (25). the sealing part of the device and the plane of the
LAA ostium is assessed. For successfully placed
Postprocedural CT endocardial and epicardial devices, a “waist” is
Complications related to vascular access and seen in the LAA, with no contrast material in the
transseptal puncture (eg, cardiac tamponade, LAA (Fig 26).
aortic perforation) are more common in the
immediate periprocedural period (59). Compli- Pericardial Effusion
cations related to the device include stroke, air Pericardial effusion may be due to either direct
embolism, major bleeding, pericardial effusion, pericardial injury overlying the thin-walled LAA
a device-related thrombus, a moderate leak, during deployment or injury to the adjacent
device embolization, and systemic embolization structures (eg, LSPV), which may then bleed into
(60). Traditionally, patients start or continue oral the pericardium (65). Ninety percent of pericar-
anticoagulation therapy for 45 days after trans- dial effusions occur within 24 hours after LAAC
catheter LAAC, which is thought to be the time (65). Although they are typically evaluated with
needed for device endothelialization. At 45 days echocardiography, pericardial effusions can be
694 May-June 2021 radiographics.rsna.org

Figure 25. Normal CT appearance after

placement of a Watchman device in a 73-year-
old man. (a) Volume-rendered 3D recon-
struction CT image obtained 60 days after
implantation shows the Watchman device
(arrow). (b) Coronal contrast-enhanced CT
image shows no enhancement in the device
(arrow), which indicates that endothelializa-
tion has occurred, and no peridevice leak.
(c) Cross-sectional reconstructed CT image
shows the device hubs and normal appear-
ances of a thrombus (arrow) in the device.

Figure 27. Device-related thrombus in

Figure 26. Normal postprocedural CT a 64-year-old woman. Axial CT image
appearance after placement of the Lariat shows a hypoattenuating lesion on the
device in a 59-year-old woman. Axial CT surface of the Watchman device (arrow)
image shows ligation of the LAA (arrow) that persisted in the delayed phase (not
and no contrast material opacification of shown), which is consistent with a device-
the LAA. related thrombus.

characterized at CT by observing the volume and noncoaxial alignment, and shallow placement.
character of the fluid in the pericardial sac (Fig Complex LAA shapes and sizes, a wide ostium,
E19). High-attenuation pericardial fluid suggests a large landing zone, and a shallow LAA depth
hemopericardium. Treatment options include are LAA-related risk factors. Device emboli-
performing pericardiocentesis or a pericardial zation typically occurs during the procedure;
window, suturing an LAA tear, or redeploying an thus, CT is rarely required for diagnosis. Larger
LAAC device (65). devices embolize to the LA or left ventricle or
get trapped in the mitral or aortic valve, whereas
Device Embolization smaller devices embolize to the descending
Device embolization is a rare complication (66). aorta (66). Embolized devices can be retrieved
Risk factors include undersizing, oversizing, percutaneously (65).
RG • Volume 41 Number 3 Rajiah et al 695

Figure 28. Peridevice leak in four patients. (a) Coronal reconstructed CT im-
age in a 59-year-old man with a Watchman device shows incomplete sealing
of the device in the LAA wall, resulting in a peridevice leak lateral to the de-
vice (arrow), which results in free flow of contrast material inside the device.
(b) Reconstructed coronal CT image in a 63-year-old man with a Watch-
man device shows off-axis placement of the device (white arrow) with a peri-
device leak laterally (black arrow). There is flow of contrast material in the
device (*) and distally in the LAA apex. (c) Axial CT image in a 47-year-old
man shows a large leak measuring 11 mm around the lateral aspect of the
device (arrow). Contrast material is also seen in the distal LAA beyond the
device. (d) Fabric leak with two contrast channels is seen within the device
material (arrows).

Device-related Thrombus seen in the device and in the distal LAA. Com-
A thrombus on the LA surface of the device is mon causes of peridevice leaks are a peridevice
an uncommon complication of LAAC (65). Risk gap, which is the failure of the Watchman device
factors include incomplete LAA occlusion, deep to expand against the LAA in the landing zone
implantation, large LAA diameter, high CHADS2 (Fig 28a); off-axis positioning of the Amplatzer
(congestive heart failure, hypertension, age ≥75 Cardiac Plug or Amulet devices (Fig 28b); and
years, diabetes mellitus, stroke or transient isch- failed endothelialization (70,71). Tissue remod-
emic attack) score or CHA2DS2-VASc (conges- eling, a highly elliptical orifice, a large landing
tive heart failure, hypertension, age ≥75 years, dia- zone, large LA size, and a non–chicken-wing
betes mellitus, stroke or transient ischemic attack, type LAA (especially windsock type) are other
vascular disease, age 65–74 years, sex category [fe- risk factors (25,64,65,72). A peridevice leak can
male]) score, and a low ejection fraction (64). With be minimal (<1 mm), mild (1–3 mm), moder-
the Amulet device, thrombus commonly develops ate (4–5 mm), or severe (>5 mm). A peridevice
near the superior edge of the disk, adjacent to the leak larger than 5 mm is considered significant
Coumadin ridge (65). At CT, the thrombus is seen and is also called incomplete closure of the LAA
as a hypoattenuating lesion on the surface of the (Fig 28c). CT is more sensitive than echocar-
device (Fig 27). The persistence of the filling de- diography for detection of peridevice leaks (68),
fect in the delayed phase helps in distinguishing a especially for detection of submillimeter marginal
thrombus from stasis. Although early studies (67) leaks, transfabric leaks, and defects of endotheli-
showed no increased risk of stroke with a device- alization (25) (Fig 28d). With incomplete closure
related thrombus, recent studies (68,69) suggest or endothelialization, there is free flow of contrast
a four- to fivefold increase in thromboembolic material into the device and distal LAA, without
events. Oral anticoagulants or low-molecular- thrombus formation (7,73) (Fig 29, Fig E20).
weight heparin are used for treatment (65). Defects of endothelialization disappear within
45 days to 6 months after LAAC, whereas other
Peridevice Leaks and Incomplete LAAC mechanisms persist. Hence, follow-up CT later
A peridevice leak is commonly seen in up to is prudent to identify persistent peridevice leaks.
68.5% of patients after LAAC (7,8,13,25). At There is an increased risk of thrombus formation
CT, a peridevice leak is defined as the presence with a peridevice leak, but there is no significant
of a contrast-enhancement trail adjacent to the correlation between peridevice leaks and throm-
device (64). Contrast enhancement may also be boembolic events (67,70,71,74).
696 May-June 2021 radiographics.rsna.org

Incomplete closure and leaks can be seen in

5%–14% of procedures in which the Lariat device
was used (25,75). Acute leaks (during the proce-
dure) most commonly occur because of a subop-
timally tightened suture (25). Early (<6 months)
and late leaks are due to slippage of the knot,
misalignment of the magnets or suture, and selec-
tive ligation of one lobe in a multilobar anatomy
(25). At CT, incomplete ligation is characterized
by persistent free flow of contrast material into the
LAA, typically in a central location (ie, the “gunny
sack” effect) (25) (Fig 30, Fig E21). Lariat leaks
have been shown to cause a higher risk of stroke, Figure 29. Incomplete closure due
particularly when the leak is small (<5 mm) to absence of endothelialization in an
(25,76). 80-year-old man 60 days after implanta-
Peridevice leaks smaller than 5 mm can resolve tion of a Watchman device. Axial CT im-
age shows the absence of a thrombus and
spontaneously with endothelialization (65), but extensive contrast material in the device
patients with leaks that are larger than 5 mm need (arrow), which is consistent with a com-
oral anticoagulation therapy. Irrespective of the plete absence of neoendothelialization.
size of a peridevice leak, an incomplete or absent
thrombus inside the device also warrants antico-
agulation therapy (25). A large peridevice leak in
endocardial devices is closed with endovascular
coils or vascular plugs (25), whereas concentric
leaks after LAAC with a Lariat device are closed
by repeating the procedure with a Lariat device or
using an Amplatzer septal occluder (Abbott) (77).

Device Erosion
Erosion of the device through the LAA and into
the adjacent pulmonary artery is extremely un-
common. It can be seen several weeks or months
Figure 30. Incomplete closure in a
after the procedure and can lead to cardiac
53-year-old man after placement of a Lar-
tamponade (78). Device oversizing and increased iat device. Two-chamber CT image shows
proximity of the pulmonary artery to the LAA are free flow of contrast material in the LAA
thought to precipitate device erosion. CT can be (arrow) as the result of incomplete liga-
tion of the LAA.
used to identify the location of the device and the
location of bleeding.
3. Blackshear JL, Odell JA. Appendage obliteration to reduce
Conclusion stroke in cardiac surgical patients with atrial fibrillation. Ann
Transcatheter LAAC is a rapidly growing option Thorac Surg 1996;61(2):755–759.
4. Naksuk N, Padmanabhan D, Yogeswaran V, Asirvatham
for prevention of stroke in patients with nonvalvu- SJ. Left atrial appendage: Embryology, anatomy, physiol-
lar atrial fibrillation. Although TEE has been the ogy, arrhythmia and therapeutic intervention. JACC Clin
reference standard imaging modality for pre- and Electrophysiol 2016;2(4):403–412.
5. Holmes DR Jr, Alkhouli M, Reddy V. Left Atrial Appendage
postprocedural imaging, cardiac CT allows more Occlusion for The Unmet Clinical Needs of Stroke Preven-
accurate delineation of the LAA anatomy and tion in Nonvalvular Atrial Fibrillation. Mayo Clin Proc
sizing of the device, while providing an excellent 2019;94(5):864–874.
6. Holmes DR Jr, Kar S, Price MJ, et al. Prospective randomized
ability to detect contraindications before the pro- evaluation of the Watchman Left Atrial Appendage Closure
cedure. CT is also helpful in examining the device device in patients with atrial fibrillation versus long-term
integrity and complications during follow-up after warfarin therapy: the PREVAIL trial. J Am Coll Cardiol
2014;64(1):1–12 [Published correction appears in J Am Coll
the procedure. Cardiol 2014;64(11):1186.].
7. Holmes DR Jr, Doshi SK, Kar S, et al. Left Atrial Appendage
References Closure as an Alternative to Warfarin for Stroke Prevention
in Atrial Fibrillation: A Patient-Level Meta-Analysis. J Am
1. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an Coll Cardiol 2015;65(24):2614–2623.
independent risk factor for stroke: the Framingham Study. 8. Reddy VY, Doshi SK, Kar S, et al. 5-Year Outcomes After
Stroke 1991;22(8):983–988. Left Atrial Appendage Closure: From the PREVAIL and PRO-
2. Alkhouli M, Noseworthy PA, Rihal CS, Holmes DR Jr. Stroke TECT AF Trials. J Am Coll Cardiol 2017;70(24):2964–2975.
Prevention in Nonvalvular Atrial Fibrillation: A Stakeholder 9. Busu T, Khan SU, Alhajji M, Alqahtani F, Holmes DR,
Perspective. J Am Coll Cardiol 2018;71(24):2790–2801. Alkhouli M. Observed versus Expected Ischemic and Bleed-
RG • Volume 41 Number 3 Rajiah et al 697

ing Events Following Left Atrial Appendage Occlusion. Am 28. Beigel R, Wunderlich NC, Ho SY, Arsanjani R, Siegel
J Cardiol 2020;125(11):1644–1650. RJ. The left atrial appendage: anatomy, function, and
10. Reddy VY, Holmes D, Doshi SK, Neuzil P, Kar S. Safety noninvasive evaluation. JACC Cardiovasc Imaging
of percutaneous left atrial appendage closure: results 2014;7(12):1251–1265.
from the Watchman Left Atrial Appendage System for 29. Nucifora G, Faletra FF, Regoli F, et al. Evaluation of the
Embolic Protection in Patients with AF (PROTECT AF) left atrial appendage with real-time 3-dimensional trans-
clinical trial and the Continued Access Registry. Circulation esophageal echocardiography: implications for catheter-
2011;123(4):417–424. based left atrial appendage closure. Circ Cardiovasc Imaging
11. Writing Group Members; January CT, Wann LS, et al. 2019 2011;4(5):514–523.
AHA/ACC/HRS focused update of the 2014 AHA/ACC/ 30. Cabrera JA, Ho SY, Climent V, Sánchez-Quintana D. The
HRS guideline for the management of patients with atrial architecture of the left lateral atrial wall: a particular anatomic
fibrillation: A Report of the American College of Cardiol- region with implications for ablation of atrial fibrillation.
ogy/American Heart Association Task Force on Clinical Eur Heart J 2008;29(3):356–362.
Practice Guidelines and the Heart Rhythm Society. Heart 31. Kanmanthareddy A, Reddy YM, Vallakati A, et al. Embry-
Rhythm 2019;16(8):e66–e93. ology and anatomy of the left atrial appendage: why does
12. Veinot JP, Harrity PJ, Gentile F, et al. Anatomy of the nor- thrombus form? Interv Cardiol Clin 2014;3(2):191–202.
mal left atrial appendage: a quantitative study of age-related 32. Ismail TF, Panikker S, Markides V, et al. CT imaging for
changes in 500 autopsy hearts: implications for echocardio- left atrial appendage closure: a review and pictorial essay. J
graphic examination. Circulation 1997;96(9):3112–3115. Cardiovasc Comput Tomogr 2015;9(2):89–102.
13. Lakkireddy D, Windecker S, Thaler D, et al. Rationale and 33. Saw J, Fahmy P, Spencer R, et al. Comparing Measurements
design for AMPLATZER Amulet Left Atrial Appendage of CT Angiography, TEE, and Fluoroscopy of the Left
Occluder IDE randomized controlled trial (Amulet IDE Atrial Appendage for Percutaneous Closure. J Cardiovasc
Trial). Am Heart J 2019;211:45–53. Electrophysiol 2016;27(4):414–422.
14. Chue CD, de Giovanni J, Steeds RP. The role of echocar- 34. Bai W, Chen Z, Tang H, Wang H, Cheng W, Rao L. Assess-
diography in percutaneous left atrial appendage occlusion. ment of the left atrial appendage structure and morphology:
Eur J Echocardiogr 2011;12(10):i3–i10. comparison of real-time three-dimensional transesophageal
15. Rathi VK, Reddy ST, Anreddy S, et al. Contrast-enhanced echocardiography and computed tomography. Int J Cardio-
CMR is equally effective as TEE in the evaluation of left vasc Imaging 2017;33(5):623–633.
atrial appendage thrombus in patients with atrial fibrillation 35. Wang DD, Eng M, Kupsky D, et al. Application of 3-Di-
undergoing pulmonary vein isolation procedure. Heart mensional Computed Tomographic Image Guidance to
Rhythm 2013;10(7):1021–1027. WATCHMAN Implantation and Impact on Early Operator
16. Burrell LD, Horne BD, Anderson JL, Muhlestein JB, Learning Curve: Single-Center Experience. JACC Cardio-
Whisenant BK. Usefulness of left atrial appendage volume vasc Interv 2016;9(22):2329–2340.
as a predictor of embolic stroke in patients with atrial fibril- 36. Yosefy C, Laish-Farkash A, Azhibekov Y, Khalameizer
lation. Am J Cardiol 2013;112(8):1148–1152. V, Brodkin B, Katz A. A New Method for Direct Three-
17. Korsholm K, Berti S, Iriart X, et al. Expert recommendations Dimensional Measurement of Left Atrial Appendage
on cardiac computed tomography for planning transcatheter Dimensions during Transesophageal Echocardiography.
left atrial appendage occlusion. JACC Cardiovasc Interv Echocardiography 2016;33(1):69–76.
2020;13(3):277–292. 37. Eng MH, Wang DD, Greenbaum AB, et al. Prospective,
18. Rizvi A, Weber NM, Sheedy EN, et al. Dynamic evaluation randomized comparison of 3-dimensional computed tomog-
of the left atrium using multiphase ECG-gated cardiac CT: raphy guidance versus TEE data for left atrial appendage
Implications for left atrial appendage occlusion planning. occlusion (PRO3DLAAO). Catheter Cardiovasc Interv
SCCT, 2018. 2018;92(2):401–407.
19. Pathan F, Hecht H, Narula J, Marwick TH. Roles of 38. Wunderlich NC, Beigel R, Swaans MJ, Ho SY, Siegel
Transesophageal Echocardiography and Cardiac Computed RJ. Percutaneous interventions for left atrial appendage
Tomography for Evaluation of Left Atrial Thrombus and exclusion: options, assessment, and imaging using 2D
Associated Pathology: A Review and Critical Analysis. JACC and 3D echocardiography. JACC Cardiovasc Imaging
Cardiovasc Imaging 2018;11(4):616–627. 2015;8(4):472–488.
20. Romero J, Husain SA, Kelesidis I, Sanz J, Medina HM, Garcia 39. Kimura T, Takatsuki S, Inagawa K, et al. Anatomical
MJ. Detection of left atrial appendage thrombus by cardiac characteristics of the left atrial appendage in cardio-
computed tomography in patients with atrial fibrillation: a genic stroke with low CHADS2 scores. Heart Rhythm
meta-analysis. Circ Cardiovasc Imaging 2013;6(2):185–194. 2013;10(6):921–925.
21. Romero J, Cao JJ, Garcia MJ, Taub CC. Cardiac imaging 40. Lupercio F, Carlos Ruiz J, Briceno DF, et al. Left atrial
for assessment of left atrial appendage stasis and thrombosis. appendage morphology assessment for risk stratification of
Nat Rev Cardiol 2014;11(8):470–480. embolic stroke in patients with atrial fibrillation: A meta-
22. Kalisz K, Halliburton S, Abbara S, et al. Update on car- analysis. Heart Rhythm 2016;13(7):1402–1409.
diovascular applications of multienergy CT. RadioGraphics 41. Alkhouli M, Chaker Z, Mills J, Raybuck B. Double device
2017;37(7):1955–1974. closure for large or bilobar left atrial appendage anatomy.
23. Hur J, Kim YJ, Lee HJ, et al. Cardioembolic stroke: EuroIntervention 2020;16(12):e1039–e1040.
dual-energy cardiac CT for differentiation of left atrial 42. Beutler DS, Gerkin RD, Loli AI. The morphology of left
appendage thrombus and circulatory stasis. Radiology atrial appendage lobes: A novel characteristic naming scheme
2012;263(3):688–695. derived through three-dimensional cardiac computed
24. Hur J, Kim YJ, Lee HJ, et al. Dual-enhanced cardiac CT for tomography. World J Cardiovasc Surg 2014;4(3):17–24.
detection of left atrial appendage thrombus in patients with 43. Rajiah P, MacNamara J, Chaturvedi A, Ashwath R, Fulton
stroke: a prospective comparison study with transesophageal NL, Goerne H. Bands in the heart: Multimodality imaging
echocardiography. Stroke 2011;42(9):2471–2477. review. RadioGraphics 2019;39(5):1238–1263.
25. Sahore A, Della Rocca DG, Anannab A, et al. Clinical 44. Lindner S, Behnes M, Wenke A, et al. Relation of left
Implications and Management Strategies for Left Atrial Ap- atrial appendage closure devices to topographic neigh-
pendage Leaks. Card Electrophysiol Clin 2020;12(1):89–96. boring structures using standardized imaging by car-
26. Wang Y, Di Biase L, Horton RP, Nguyen T, Morhanty P, diac computed tomography angiography. Clin Cardiol
Natale A. Left atrial appendage studied by computed tomog- 2019;42(2):264–269.
raphy to help planning for appendage closure device place- 45. Syed FF, Noheria A, DeSimone CV, Asirvatham SJ. Left
ment. J Cardiovasc Electrophysiol 2010;21(9):973–982. atrial appendage ligation and exclusion technology in the
27. Di Biase L, Santangeli P, Anselmino M, et al. Does the left incubator. J Atr Fibrillation 2015;8(2):1160.
atrial appendage morphology correlate with the risk of stroke 46. Sánchez-Quintana D, Ho SY, Climent V, Murillo M,
in patients with atrial fibrillation? Results from a multicenter Cabrera JA. Anatomic evaluation of the left phrenic nerve
study. J Am Coll Cardiol 2012;60(6):531–538. relevant to epicardial and endocardial catheter ablation:
698 May-June 2021 radiographics.rsna.org

implications for phrenic nerve injury. Heart Rhythm imaging proposal evaluating implanted left atrial appendage
2009;6(6):764–768. occlusion devices by cardiac computed tomography. BMC
47. Rajiah P, Kanne JP. Computed tomography of septal de- Med Imaging 2016;16(1):25.
fects. J Cardiovasc Comput Tomogr 2010;4(4):231–245. 64. Lindner S, Behnes M, Wenke A, et al. Assessment of peri-
48. Alkhouli M, Rihal CS, Holmes DR Jr. Transseptal Tech- device leaks after interventional left atrial appendage closure
niques for Emerging Structural Heart Interventions. JACC using standardized imaging by cardiac computed tomography
Cardiovasc Interv 2016;9(24):2465–2480. angiography. Int J Cardiovasc Imaging 2019;35(4):725–731.
49. Krishnaswamy A, Patel NS, Ozkan A, et al. Planning left 65. Wilkins B, Fukutomi M, De Backer O, Søndergaard L. Left
atrial appendage occlusion using cardiac multidetector Atrial Appendage Closure: Prevention and Management of
computed tomography. Int J Cardiol 2012;158(2):313–317. Periprocedural and Postprocedural Complications. Card
50. Mahabadi AA, Samy B, Seneviratne SK, et al. Quantitative Electrophysiol Clin 2020;12(1):67–75.
assessment of left atrial volume by electrocardiographic-gated 66. Alkhouli M, Sievert H, Rihal CS. Device Embolization
contrast-enhanced multidetector computed tomography. J in Structural Heart Interventions: Incidence, Outcomes,
Cardiovasc Comput Tomogr 2009;3(2):80–87. and Retrieval Techniques. JACC Cardiovasc Interv
51. Lee WJ, Chen SJ, Lin JL, Huang YH, Wang TD. Images 2019;12(2):113–126.
in cardiovascular medicine. Accessory left atrial appendage: 67. Cochet H, Iriart X, Sridi S, et al. Left atrial appendage
a neglected anomaly and potential cause of embolic stroke. patency and device-related thrombus after percutaneous left
Circulation 2008;117(10):1351–1352. atrial appendage occlusion: a computed tomography study.
52. Abbara S, Mundo-Sagardia JA, Hoffmann U, Cury RC. Eur Heart J Cardiovasc Imaging 2018;19(12):1351–1361.
Cardiac CT assessment of left atrial accessory appendages 68. Alkhouli M, Busu T, Shah K, Osman M, Alqahtani F,
and diverticula. AJR Am J Roentgenol 2009;193(3):807–812. Raybuck B. Incidence and clinical impact of device-related
53. Otton JM, Spina R, Sulas R, et al. Left atrial appendage thrombus following percutaneous left atrial appendage
closure guided by personalized 3D-printed cardiac recon- occlusion: A meta-analysis. JACC Clin Electrophysiol
struction. JACC Cardiovasc Interv 2015;8(7):1004–1006. 2018;4(12):1629–1637.
54. Hell MM, Achenbach S, Yoo IS, et al. 3D printing for sizing 69. Dukkipati SR, Kar S, Holmes DR, et al. Device-Related
left atrial appendage closure device: head-to-head compari- Thrombus After Left Atrial Appendage Closure: Inci-
son with computed tomography and transoesophageal echo- dence, Predictors, and Outcomes. Circulation 2018;
cardiography. EuroIntervention 2017;13(10):1234–1241. 138(9):874–885.
55. Hachulla AL, Noble S, Guglielmi G, Agulleiro D, Müller 70. Nguyen A, Gallet R, Riant E, et al. Peridevice leak after left
H, Vallée JP. 3D-printed heart model to guide LAA closure: atrial appendage closure: Incidence, risk factors, and clinical
useful in clinical practice? Eur Radiol 2019;29(1):251–258. impact. Can J Cardiol 2019;35(4):405–412.
56. Bieliauskas G, Otton J, Chow DHF, et al. Use of 3-di- 71. Saw J, Tzikas A, Shakir S, et al. Incidence and Clinical Impact
mensional models to optimize pre-procedural planning of of Device-Associated Thrombus and Peri-Device Leak Fol-
percutaneous left atrial appendage closure. JACC Cardiovasc lowing Left Atrial Appendage Closure With the Amplatzer
Interv 2017;10(10):1067–1070. Cardiac Plug. JACC Cardiovasc Interv 2017;10(4):391–399.
57. Gilhofer TS, Saw J. Periprocedural Imaging for Left Atrial 72. Alkhouli M, Chaker Z, Clemetson E, et al. Incidence, char-
Appendage Closure: Computed Tomography, Transesopha- acteristics and management of persistent peri-device flow
geal Echocardiography, and Intracardiac Echocardiography. after percutaneous left atrial appendage occlusion. Struct
Card Electrophysiol Clin 2020;12(1):55–65. Heart 2019;3(6):491–498.
58. Alkhouli M, Hijazi ZM, Holmes DR Jr, Rihal CS, Wieg- 73. Lakkireddy D, Afzal MR, Lee RJ, et al. Short and long-
ers SE. Intracardiac echocardiography in structural term outcomes of percutaneous left atrial appendage suture
heart disease interventions. JACC Cardiovasc Interv ligation: Results from a US multicenter evaluation. Heart
2018;11(21):2133–2147. Rhythm 2016;13(5):1030–1036.
59. Bajaj NS, Parashar A, Agarwal S, et al. Percutaneous left atrial 74. Viles-Gonzalez JF, Kar S, Douglas P, et al. The clinical
appendage occlusion for stroke prophylaxis in nonvalvular impact of incomplete left atrial appendage closure with
atrial fibrillation: a systematic review and analysis of observa- the Watchman Device in patients with atrial fibrillation: a
tional studies. JACC Cardiovasc Interv 2014;7(3):296–304. PROTECT AF (Percutaneous Closure of the Left Atrial
60. Yerasi C, Lazkani M, Kolluru P, et al. An updated systematic Appendage Versus Warfarin Therapy for Prevention of
review and meta-analysis of early outcomes after left atrial Stroke in Patients With Atrial Fibrillation) substudy. J Am
appendage occlusion. J Interv Cardiol 2018;31(2):197–206. Coll Cardiol 2012;59(10):923–929.
61. Levine GN, Gomes AS, Arai AE, et al. Safety of magnetic 75. Pillarisetti J, Reddy YM, Gunda S, et al. Endocardial (Watch-
resonance imaging in patients with cardiovascular devices: man) vs epicardial (Lariat) left atrial appendage exclusion
an American Heart Association scientific statement from devices: Understanding the differences in the location and
the Committee on Diagnostic and Interventional Cardiac type of leaks and their clinical implications. Heart Rhythm
Catheterization, Council on Clinical Cardiology, and the 2015;12(7):1501–1507.
Council on Cardiovascular Radiology and Intervention: 76. Mohanty S, Gianni C, Trivedi C, et al. Risk of thrombo-
endorsed by the American College of Cardiology Founda- embolic events after percutaneous left atrial appendage
tion, the North American Society for Cardiac Imaging, ligation in patients with atrial fibrillation: Long-term results
and the Society for Cardiovascular Magnetic Resonance. of a multicenter study. Heart Rhythm 2020;17(2):175–181.
Circulation 2007;116(24):2878–2891. 77. Pillai AM, Kanmanthareddy A, Earnest M, et al. Initial
62. Dieker W, Behnes M, Fastner C, et al. Impact of left atrial experience with post Lariat left atrial appendage leak closure
appendage morphology on thrombus formation after suc- with Amplatzer septal occluder device and repeat Lariat ap-
cessful left atrial appendage occlusion: Assessment with plication. Heart Rhythm 2014;11(11):1877–1883.
cardiac-computed-tomography. Sci Rep 2018;8(1):1670. 78. Sepahpour A, Ng MK, Storey P, McGuire MA. Death
63. Behnes M, Akin I, Sartorius B, et al. LAA Occluder View from pulmonary artery erosion complicating implantation of
for post-implantation Evaluation (LOVE): standardized percutaneous left atrial appendage occlusion device. Heart
Rhythm 2013;10(12):1810–1811.

TM
This journal-based SA-CME activity has been approved for AMA PRA Category 1 Credit . See rsna.org/learning-center-rg.