Incomplete exclusion of the LAA from blood flow is quite common after surgical LAA ligation and is associated with a higher risk of thromboembolic events. Two studies reported successful surgical LAA ligation/closure in only 40–70 % of patients [4, 5]. Incomplete closure was associated with spontaneous echo contrast or thrombus formation and went along with a higher rate of thromboembolic events during follow-up. Evidence of LAA thrombus was found in one-third of patients with incomplete surgical LAA exclusion.

A routine practice to ligate the LAA during open-heart surgery is supported by another study [6]; however, the absence of LAA ligation during mitral valve surgery was a predictor of future thromboembolic events. After a mean follow-up of 69 months, 17 % of patients without LAA ligation suffered a thromboembolic event as opposed to 3.4 % in patients who underwent LAA ligation (odds ratio 6.7). This was true despite the fact that >10 % of LAA ligations were incomplete. If only complete LAA ligations were included in the analysis, the odds ratio for a thromboembolic event without LAA ligation was as high as 11.9, indirectly suggesting a higher risk for thromboembolic events if the LAA ligation is not complete.

Current American Heart Association (AHA)/American College of Cardiology (ACC)/Heart Rhythm Society (HRS) atrial fibrillation (AF) guidelines state that surgical excision of the LAA may be considered in patients undergoing cardiac surgery (Class IIB, Level of evidence C). In our opinion, routine LAA closure during open-heart surgery makes sense and is supported by the published data. However, incomplete surgical LAA closure is a very common finding and goes along with a higher thromboembolic risk. Furthermore, data suggest an association of residual leaks with the occurrence of LAA thrombus.

Our “Bern single-center experience” with dedicated Amplatzer LAA devices Amplatzer Cardiac Plug (ACP) and Amulet in 210 patients (own unpublished data) found residual leaks in 5 % of patients. Three quarters of these leaks were minor (<5 mm). Device-related thrombus was found in 6 % of patients whereof 1 % were mobile thrombi. Interestingly, the incidence of residual leaks was much higher (25 %) in patients with device-related thrombi. Such leaks were >5 mm in most patients.

The large European multicenter registry of LAA occlusion with the ACP [7] found comparable rates of peri-device leaks (11.6 %, with relevant leaks >3 mm in 1.9 % of patients) and device-related thrombus (4.4 %). In this large registry, neither peri-device leaks nor the presence of device-related thrombus translated into adverse outcomes.

A slightly higher incidence of device-thrombus was found in a smaller TEE study (17.6 %) [8]. Patients who developed device-related thrombus had a significantly higher CHADS₂-score, CHA₂DS₂-Vasc-score, platelet count, and lower ejection fraction as compared to patients without device thrombus.

In a substudy of the PROTECT-AF trial [9], a much higher rate of peri-device leak is reported (32 %), comparable to that of surgical results. The presence of such a peri-device leak was, as opposed to surgical data, not associated with worse clinical outcome. Furthermore, there was no difference in outcome if patients with a peri-device leak continued oral anticoagulation with Vitamin K antagonists (VKA) or not, suggesting that the presence of a peri-device leak does not warrant continued oral anticoagulation. In this study, device thrombus was found in 3.4 %.

Data on the epicardial/endocardial LARIAT technique indicate a high closure rate of 95 % at 3 months [10]. No patient was left with a leak >3 mm. No thrombus formation at the ligation site was detected. Of note, of the 119 patients, 30 were excluded due to anatomical reasons; therefore only 75 % of patients could be treated with this device.

Several case reports also documented the presence of stump thrombus after the Lariat procedure, but the absolute incidence is not known.

In summary, the rate of incomplete LAA closure seems lower with percutaneous LAA closure as compared to surgical ligation. Data on percutaneous closure suggest that peri-device leaks (unless large) have no clinical consequences. Large leaks appear to be linked to a higher risk for thrombus formation and may need to be addressed.

Fibrin deposition is an important step during device endothelialization and occurs in all patients after device implantation (Fig. 20.2). Thrombus formation is usually followed by organization of the thrombus, some inflammation, formation of granulation tissue, and finally endothelialization. Therefore, thrombus formation is an important part during endothelialization.

During follow-up TEE, it is therefore important to differentiate “physiologic” thrombus formation/fibrin deposition of the occluder and “tissue-filling” of niches from “pathologic” thrombus formation.

Device-related thrombi that protrude into the left atrium and show mobile components have to be considered “high-risk thrombi” (Fig. 20.3). Differentiation from endothelialization is unambiguous. Predilection sites are niches around the device and protruding device structures such as screws. Intentions to minimize these predilection sites therefore make sense (e.g., internalization of the disc-screw with the Amulet second-generation ACP device).

Tissue-filling starts at the same predilection sites and progressively covers the entire device. It consists of a thin tissue layer without protruding components and needs no further measures.

A residual leak is defined as residual flow into the LAA after LAA closure. In the aforementioned substudy of the PROTECT-AF trial [9], peri-device leaks were classified as none, minor (<1 mm), moderate (1–3 mm), and severe (>3 mm). No difference in thromboembolic events was found during follow-up between the four subgroups.

From our own data on 210 patients using Amplatzer devices, the incidence of peri-device leak was 5 %, with small leaks (<5 mm) being responsible for the vast majority of the leaks (73 %) (Bern LAA registry, unpublished data).

A large residual flow into the LAA (e.g., entire LAA lobes that are not excluded, Fig. 20.4a) most likely results in immediate intra-procedural treatment measures (e.g., implantation of a second device, Fig. 20.4b). Peri-device leaks diagnosed during follow-up are therefore most likely smaller. Classification according to the size is somewhat arbitrary, with >3 mm being considered a relevant leak in the Watchman trials (while warfarin was continued post-device implantation if the leak exceeded 5 mm) and >5 mm being considered significant for Amplatzer devices.