Article Text


Interventional cardiology
Occluding the left atrial appendage: anatomical considerations
  1. P Su1,
  2. K P McCarthy2,
  3. S Y Ho1,2
  1. 1
    National Heart and Lung Institute, London, UK
  2. 2
    Imperial College and Royal Brompton Hospital, London, UK
  1. Dr S Y Ho, National Heart and Lung Institute, Imperial College, Guy Scadding Building, Dovehouse Street, London SW3 6LY, UK; yen.ho{at}


Background: Occlusion of the left atrial appendage (LAA) is thought to reduce the risk of thromboembolic events in patients with atrial fibrillation.

Objective: To examine the LAA and its relationship to neighbouring structures that may be put at risk when intervening to occlude its os.

Methods: 31 heart specimens were examined grossly. Four of the LAAs were processed for histological examination and endocasts were made from 11 appendages. The diameters of the LAA os and proximity to the left superior pulmonary vein, mitral valve and left anterior descending artery were measured and areas of thin atrial wall in the vicinity were noted.

Results: The LAA orifice was oval shaped in all cases with a mean (SD) diameter of 17.4 (4) mm (range 10–24.1). The mean (SD) distances of the LAA orifice to the left superior pulmonary vein and mitral valve were 11.1 (4.1) mm and 10.7 (2.4) mm, respectively. The left anterior descending, circumflex artery and, in 6 cases, the sinus node artery, were in close proximity to the LAA. Pits or troughs and areas of thin atrial wall were found in 57.7% of hearts within a 20.9 mm radius from the os. Histology showed small crevices and areas of very thin wall within the trabeculated appendage.

Conclusions: The LAA orifice is oval shaped and thin areas of appendage wall and atrial wall are common. Potentially, the left superior pulmonary vein, mitral valve and anterior descending coronary artery can be at risk during occlusion of the os.

Statistics from

The left atrial appendage (LAA) is a highly complex structure which has received little attention until recent years. There has been increasing interest in it, primarily because over 90% of thrombi in patients with atrial fibrillation occur in the LAA1 and thrombus formation in the LAA increases the risk of stroke threefold.2 Elimination of the LAA to reduce the risk of stroke was previously done as an adjunct to coronary bypass or mitral valve operations.3 4 In recent years, several trials have used a percutaneous approach to occlude the os (entrance) of the LAA.57 It is suggested that this method is less invasive than surgical ligation and early short-term studies have yielded promising results. However, there have been reports of residual flow around the percutaneous device.57 which may predispose to further thromboemboli.4 There are also concerns about device embolisation and damage to surrounding structures. Currently, most occluders take account of the diameter of the os when selecting the appropriate device size for implant.

Anatomical studies have demonstrated the variation in LAA dimensions and shape in individual subjects.8 9 However, important structures in the neighbourhood of the LAA and its os have been ignored. With this study, we aimed to provide a better understanding of the morphology of the human LAA and investigate its spatial relationship with surrounding structures which are relevant to interventional cardiologists implanting devices to occlude the appendage.


This research project was approved by our local ethics committee.

Gross examination

From the archive of the Royal Brompton Hospital, 31 heart specimens from adults were retrieved for gross examination. In the majority, age, gender and clinical information were not available. None had congenital heart malformations. We considered the LAA os to be the opening to the appendage as viewed from the endocardial aspect of the left atrium. Using a pair of callipers, we measured the diameters of the os of the LAA orifice. The longest diameter was designated as the “long diameter” and the diameter perpendicular to this as the “short diameter” (fig 1). To assess left atrial size, we measured the maximal left atrial length transversely from the septum to the lateral wall. The minimal distance between the os and the left upper pulmonary vein, and the distance between the os and the hinge line (annulus) of the mitral valve were both measured (fig 1). The distance between the base of the LAA that corresponds to the os and the left anterior descending artery was measured on the epicardial surface after deflecting the appendage backwards (fig 1). The epicardial site corresponding to the os was determined by pushing pins through from the endocardial surface and the shortest distance to the artery was recorded. Not all measurements could be made in every heart owing to the way the specimens had been dissected previously.

Figure 1 (A) Internal view of the left atrium showing the long and short diameters of the os of the appendage (double arrows) and the measurements for the distances to the left superior pulmonary vein (broken line) and to the mitral valve (dotted line). (B) Epicardial aspect showing the anterior descending coronary artery (triangles) and the great cardiac vein (v) passing underneath the left atrial appendage (LAA). The arc marks the site of the os. (C) The atrial appendage deflected backwards revealing the artery and vein running near the os (arc). (D) View of the cardiac base showing the relationship between the os (double arrows) and the left aortic sinus and coronary artery (LCA). LI, left inferior pulmonary vein; LS, left superior pulmonary vein; MV, mitral valve; PT, pulmonary trunk.

We also noted areas of thin atrial wall in the vicinity of the os by placing the atria at a fixed distance in front of the same light source. Thin areas, deficient in myocardium, became transilluminated. The number of pits or troughs in these areas was counted and their maximum diameters were measured with a pair of callipers. The minimum thickness of the atrial wall in these areas and distance from the appendage orifice were recorded.

Silicone casts were made from 11 specimens. With each, we counted the number of branches and twigs using definitions similar to that used in a previous study.8 We measured the length of the appendage and the distance between the orifice and the point where the appendage first showed directional change. The angle of the first bend was measured with a protractor.

Histological analysis

Blocks of atrial and ventricular wall incorporating the LAA orifice were excised from four hearts. Each block was divided into pieces of approximately 5 mm thick by cuts perpendicular to the atrioventricular junction. The subdivided pieces were processed for serial histological sectioning. Cut sections were stained with Masson’s trichrome technique and with elastic van Gieson stain.

Statistical analysis

Measurements are expressed as mean (SD). Results were compared using a paired t test. Correlations between quantitative variables were performed by standard linear regression. A p value <0.05 was considered significant.


LAA orifice and left atrial size

On the endocardial aspect, the superior and posterior borders of the os are well demarcated by a ridge-like fold of the atrial wall that separates the os from the left upper pulmonary vein. Usually lacking a ridge, the anterior and inferior borders of the os are poorly defined and the margins were taken to be in the plane of the atrial wall directly across from the better defined superior and posterior margins. The rim of the os was smooth and pectinate muscles were clearly visible within the body of the LAA (fig 1). The shape of the os was oval in all hearts. The range of the long diameter was 10–24.1 mm (mean (SD) 17.4 (4) mm) while the short diameter was 5.2–19.5 mm (mean (SD) 10.9 (4.2) mm) (table 1). The plane of the orifice was obliquely rather than vertically orientated relative to the mitral annulus (fig 2).

Figure 2 (A) In this heart there is a cluster of pits (small arrows) in the lateral atrial wall proximal to the os of the appendage (double-headed arrow). Note that the fold in the atrial wall (triangle) that separates the orifice of the left superior pulmonary vein and the appendage is narrow. (B) Transillumination shows areas of thin atrial wall (small arrows) to either side of the os (double arrows). (C) This cut through the middle of the os (double headed arrow) shows the circumflex artery (a) and great cardiac vein (v) within the fatty tissues of the atrioventricular groove. (D) Histological section shows the relationship of the circumflex artery and great cardiac vein to the os (arrow). The enlargement shows the thin wall of the appendage in between the pectinate muscles. LAA, left atrial appendage; LI, left inferior pulmonary vein; LS, left superior pulmonary vein; MV, mitral valve.
Table 1 Mean distances in millimetres (with range) of left atrial appendage (LAA) os dimensions and proximity to surrounding structures

A large variation in left atrial size was observed. The transverse dimension ranged between 20 and 56 mm (table 1). There was no significant correlation between atrial size and size of LAA os (correlation coefficient  =  0.32; p = 0.118).

Areas of thin atrial wall on transillumination

During gross examination of 26 specimens, we noted areas of atrial wall that could be transilluminated in the vicinity of the os. These, representing areas deficient in myocardium, were manifested as shallow pits or deep and narrow troughs in 15 specimens (57.7%). In 11/15 cases (73%) these areas were situated on the anterolateral or lateral left atrial wall, or both, in the vestibule leading to the mitral valve. The pits or troughs occurred in isolation, or as a cluster or a linear array leading to the appendage os (fig 2A). The mean (SD) maximum diameter of the pits or troughs was 3.6 (3.8) mm (range 0.5–10.3). The mean (SD) atrial wall thickness of these areas was 1 (0.5) mm (range 0.4–1.5). The pits or troughs were located within a radius ranging between 1.4 and 20.9 mm from the LAA orifice. Paper-thin atrial walls distal to the appendage orifice were also apparent in most specimens on transillumination (fig 2B).

Proximity to surrounding structures

The LAA is in close proximity to the left superior pulmonary vein (fig 2). The distance between the rim of the os and the left superior pulmonary vein was measured in 22 specimens and this showed a wide range of 5.8–23.7 mm (table 1). Most values fell into one of two arbitrary ranges from 5.1 to 10 mm (45.5%) and 10.1 to 15 mm (40.9%). The average distance between the rim of LAA orifice and left superior pulmonary vein was similar to that between the LAA orifice and mitral valve (table 1). There was no significant difference in distances between the LAA and left upper pulmonary vein and LAA and mitral valve (p = 0.7).

The left anterior descending coronary artery was revealed by dissection in 28 heart specimens. In most cases, the artery was found to be located close to the os of the LAA when epicardial fat was removed. The distance was <10 mm in 46.4% of the specimens. The circumflex artery ran close to the epicardial aspect of the LAA os (fig 2). In six cases, the circumflex artery was seen to give rise to the sinus node artery. The os of the LAA was closely associated with the left aortic sinus from which the left coronary artery normally arises (fig 1). Left atrial arteries and veins and the great cardiac vein were also in close proximity to the LAA (fig 1).

Cast measurements

The mean (SD) length of the LAA was 44.9 (9.6) mm (range 27–60). The appendages had 2–5 branches and 0–12 twigs. A bend was encountered between 7 and 12 mm from the os in 10 specimens (figs 3A and B). The remaining appendage appeared straight with no evidence of directional change (figs 3C and D). The mean (SD) angle of the first bend was 133.9 (13.5)° (range 108–153). Thus, the morphological appearance of the LAA could be classified into two types: slender like a crooked finger (81.8%) and stump-like (18.2%).

Figure 3 (A) and (B) are superior and inferior views of an appendage which has a finger-like shape. The double headed arrows show the measurements made for the total length and the angle of the first bend. The solid white arrows indicate branches and the dashed white arrows indicate twigs. (C) and (D) show a straight, stump-like appendage.


Histological analysis showed the internal structure of the appendage and its close proximity to adjacent structures. Small crevices were visible within the body of the trabeculated appendage (fig 2D). The appendage wall lining these crevices was very thin (fig 2D). The close proximity of coronary arteries and veins seen macroscopically was confirmed on histology (figs 2C and D).


Elimination of the LAA may be an alternative to anticoagulation therapy to prevent stroke or embolism, although the pathophysiological consequences of losing an atrial appendage remain to be fully investigated.10 Early surgical approaches to LAA ligation included stapling or suture closures. Purse-string sutures have been associated with a “puckered” suture line, which is thought to precipitate thrombus formation on the endocardial surface.11 Other epicardial approaches using the stapling method have led to LAA tears.1 12 Alternative methods of LAA occlusion, such as the percutaneous approach to implant occluders, have developed in recent years with a range in designs of the devices.13 Ostermayer et al highlighted the potential advantages of this method over conventional surgical techniques, especially in patients in whom anticoagulation therapy is contraindicated.14 These include the less invasive nature of the procedure and faster recovery. There are still controversies about which method of LAA occlusion is safer or is associated with decreased embolic risk. Whichever technique is used, not only the anatomy of the LAA but its spatial relationship to neighbouring structures must be taken into account to avoid any complications. High-resolution imaging techniques can now be used to study the anatomy of the LAA15 but do not provide information about wall thickness and neighbouring structures.

LAA orifice

We have observed a wide range of LAA os diameters, confirming the findings of previous studies.8 9 The range of diameters in our study (10–24.1 mm) corresponds well with the range of sizes available for the occluders. These percutaneous devices/systems,13 however, have a round shape to fill or cover the os. A previous study9 and our study show that the shape of the os is consistently elliptical rather than round. This suggests that to seal the LAA orifice adequately without oversizing, devices may need to be elliptical for a snug fit. A round implant over an oval-shaped orifice may leave crevices on either side of the implant, leading to incomplete sealing of the orifice. The significance of residual leaks after device implantation has been questioned. Schneider et al commented that residual flow around the percutaneous left atrial appendage transcatheter occlusion does not preclude thrombus formation in the LAA.16 However, residual leaks in the presence of thrombi may be a risk for embolic events in patients with partially occluded LAAs.2

Ramondo et al commented that the length of the LAA in relation to the diameter of the os is a crucial measurement in order to permit complete deployment of the device into the LAA.17 Perhaps more critical is the distance between the LAA orifice and the point at which the LAA first deviates from its original course. In all but one of the 11 endocasts in this study, the appendage first changed its course at a distance of 7–12 mm from the appendage orifice. This observation is particularly important during entry of the distal end of the deployment catheter. If advanced too far, it may exit the appendage into the pericardial space, especially since parts of its wall in between pectinate muscles are paper thin. The risk of a haemopericardium should not be overlooked. There have been reports of pericardial effusions and haemopericardiums during device implantation.6 7 14

Areas of thin atrial wall in the vicinity of the LAA

The morphology of the atrial wall in the vicinity of the LAA orifice is an important consideration when deploying catheters in the left atrium. More than half the specimens in which atrial wall measurements were possible had evidence of pits or troughs. The pits or troughs tend to occur in isolation or in clusters within a radius of 20.9 mm from the LAA orifice and are located usually on the anterolateral and lateral atrial wall. During manoeuvres, catheters and delivery sheaths may become lodged in the pits/troughs. This could potentially increase the risk of perforation since the atrial wall is extremely thin inside the depressions, comparable to the paper-thin area within the LAA.

The surrounding structures

Despite the potential risk of pulmonary vein stenosis or impairment of mitral leaflet excursion owing to the proximity of the occlusion device to the left upper pulmonary vein and mitral valve, clinical studies have shown no evidence of disruption to pulmonary vein inflow or mitral valve function.6 7 14 However, other neighbouring structures such as the left anterior descending coronary artery and the circumflex artery should be considered. Our study shows that these are in close proximity to the LAA or its os and can be vulnerable to trauma during implantation of percutaneous devices, especially when devices are sized to be 20–40% larger than the os.

It is well established that the sinus node artery arises from the right coronary system in about 60% of cases. But, there is still a significant proportion of cases in which the sinus node artery originates from the left coronary system. In a study conducted by Busquet et al, the sinus node artery was seen to arise from the circumflex artery and from the left lateral atrial artery in 30% and 8% of the cases, respectively.18 When these run in the myocardium of the LAA around the os, they can be at risk of trauma from devices. Furthermore, our histological study confirmed the presence of small coronary arteries in close proximity to the os.

A previous anatomical study on the course of the phrenic nerves has demonstrated that the left phrenic nerve runs along the pericardium overlying the left atrial appendage.19 This nerve may also be at risk if epicardial approaches are used. Moreover, on the epicardial aspect the anterior interventricular trunk and the obtuse marginal trunk join to form the left coronary lymphatic channel that passes beneath the left atrial appendage and close to the os.20 The great cardiac vein on its ascent to the atrioventricular groove also passes underneath the appendage but its course tends to veer away from the os.

Nevertheless, early clinical experience with percutaneous devices appears promising with trials reporting no evidence of device embolisation or interference with surrounding structures.6 7 The follow-up periods in these studies were short and the subject population small. There is an isolated case report of device embolisation after detachment from the catheter.21 The long-term safety of such devices has yet to be evaluated and a registry with longer follow-up will help towards detecting any complications.


The heart specimens were preserved in formalin, which causes tissue shrinkage by about 10%. This may lead to an underestimation of the dimensions measured. A previous anatomical study found differences in dimensions of the appendage between hearts from patients with sinus rhythm and those with atrial fibrillation.8 We were limited by the availability of organs for research. Most of them had been in the archive for decades and had no clinical information. Owing to incomplete patient information, we could not relate our measurements to age or to any cardiac causes of death in our study. Nevertheless, our series showed there was no significant correlation between atrial size and size of the os. This suggests to us that the observations concerning spatial relationships of structures and atrial wall are still highly relevant and important when contemplating closing the LAA os.


We highlight the shape of the LAA orifice and neighbouring structures including thin areas of atrial wall that are important considerations for designing percutaneous occlusion devices and for interventionists deploying these devices. Owing to the elliptical shape of the LAA orifice, round devices may need to be oversized considerably in order to achieve complete occlusion. But, oversized devices risk overlapping or impinging upon neighbouring structures, including the left upper pulmonary vein, the coronary arteries and the mitral valve.


View Abstract


  • See Editorial, p 1113

  • Funding: From the Royal Brompton and Harefield Hospital Charitable Fund.

  • Competing interests: None.

  • Ethics approval: Approved by the local ethics committee.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles