Original Contribution

Occurrence of Incomplete Endothelialization Causing Residual Permeability After Left Atrial Appendage Closure

Mathieu Granier, MD, MSc1;  Guillaume Laugaudin, MD2;  Fran√ßois Massin, MD2;  Stephane Cade, MD2;  Pierre Fran√ßois Winum, MD1;  Cornelia Freitag, MD1;  Jean-Luc Pasquie, MD, PhD2

Mathieu Granier, MD, MSc1;  Guillaume Laugaudin, MD2;  Fran√ßois Massin, MD2;  Stephane Cade, MD2;  Pierre Fran√ßois Winum, MD1;  Cornelia Freitag, MD1;  Jean-Luc Pasquie, MD, PhD2

Abstract: Aims. Percutaneous left atrial appendage (LAA) occlusion is occasionally incomplete, with residual permeability of the LAA on cardiac computed tomography. The cause for this is unclear. Our objective was to determine if residual permeability was related to incomplete endothelialization. Methods. A total of 35 consecutive patients contraindicated for anticoagulant therapy admitted for LAA occlusion were included; 12 patients received a Watchman device (Boston Scientific) and 23 patients received an Amplatzer Cardiac Plug (St. Jude Medical). Incomplete endothelialization was defined as residual permeability on cardiac computed tomography without peridevice leak on transesophageal echocardiography at follow-up. Results. Five patients did not receive cardiac computed tomography. After 10 ± 6 months of follow-up, residual permeability of the LAA (at least partial) was recorded on cardiac computed tomography in 21 of 30 patients (70%). Seven of 30 patients presented with a peridevice leak on transesophageal echocardiography. Among the remaining 23 patients, 14 (61%) presented with incomplete endothelialization and 9 (39%) presented with complete endothelialization. There was no statistical difference between the patients presenting with complete vs incomplete endothelialization. Conclusion. We found that incomplete endothelialization, defined as residual permeability on cardiac computed tomography without peridevice leak on transesophageal echocardiography, occurred in 61% of the patients after 10 ± 6 months of percutaneous LAA closure. Predisposing factors and appropriate monitoring of LAA patients remain to be determined in larger cohorts. 

J INVASIVE CARDIOL 2018;30(7):245-250. Epub 2018 May 15.

Key words: left atrial appendage, percutaneous left atrial appendage occlusion, left atrial appendage occlusion, stroke prevention, atrial fibrillation

Atrial fibrillation (AF) is the most frequent supraventricular tachycardia, responsible for approximately 20% of ischemic strokes.1 Anticoagulant therapy is the cornerstone of preventive treatment of stroke for AF patients, but both vitamin K antagonists (VKAs) and novel oral anticoagulants are associated with major or minor bleeding risk of around 15% per year.2-5 The concept of percutaneous left atrial appendage (LAA) occlusion is to prevent embolic ischemic stroke while avoiding bleeding complications from anticoagulants. There is a growing interest in this technique, as it has demonstrated non-inferiority compared to VKA in stroke prevention.6 Nevertheless, residual permeability of LAA after device implantation can reach 60%,7 and is usually attributed to peridevice leak with an unclear clinical relevance.8 However, little is known about device endothelialization. Animal studies have reported that the devices are totally endothelialized within 3 months,9 but some cases of incomplete endothelialization have been reported in man.10 We hypothesized that transesophageal echocardiography (TEE) and cardiac computed tomography (CCT) scan comparison could allow the identification of incomplete endothelialization among patients with residual permeability.


Study objectives. The main objective of the study was to identify patients with peridevice leak that may be related to incomplete endothelialization, which was defined as residual LAA permeability assessed by CCT but without peridevice leak as assessed by residual flow in color Doppler by TEE. Patients with peridevice leak assessed by TEE were excluded from the comparison. Secondary objectives were to determine the rate of incomplete LAA occlusion after CCT analysis and predisposing factors associated with incomplete occlusion.

This retrospective study was conducted in Montpellier and Nimes (France) University Hospitals. Consecutive patients admitted for LAA occlusion from July 2013 to February 2015 were included in the study. The research complied with the Declaration of Helsinki and was approved by the local ethics committee of our institution. All patients signed a written informed consent.

Patients. All patients were >18 years old and had non-valvular AF (paroxysmal or permanent), a CHA2DS2-VASc score >2, and a formal contraindication for oral anticoagulation (assessed by two physicians from different specialties). Exclusion criteria for LAA closure were valvular AF, presence of recent LAA thrombus at the time of intervention, a non-formal contraindication for anticoagulants, or a non-formal indication.

Procedure. Before intervention, TEE and CCT were performed to assess vacuity and anatomy of the LAA (depth, number of lobes) and to ensure its compatibility with the implantation of a closure device. Device allocation (either Amplatzer Cardiac Plug [ACP; St. Jude Medical] or Watchman [Boston Scientific]) was unrelated to anatomical considerations, with each brand allocated alternatively for consecutive patients.

Intervention was TEE guided under general anesthesia. After a right venous femoral puncture, transseptal puncture was performed as low and posterior as possible according to TEE imaging and directed under continuous pressure guidance. All patients received intravenous (IV) heparin before transseptal puncture (initial IV bolus of 500 IU/kg) to achieve a target activated clotting time (ACT) >250 seconds. The delivery sheath (12 to 14 Fr depending on the device size) was introduced through transseptal puncture in the LAA ostium to perform LAA angiogram in two orthogonal projections. Angiogram measurements were recorded and compared to CCT and TEE measurements. Periprocedural TEE measurements were assessed after an LAA pressure >10 mm Hg.

Three physicians (two senior in device implantation and one senior echocardiographer) and a technical support specialist provided by the manufacturer selected the device size by comparing TEE, angiography, and CCT measurements. 

Implantation was attempted under both fluoroscopic and echocardiographic guidance. Once the device was released, echocardiographic measurements and fluoroscopic optimal position of the device were assessed and discussed between the three operators. On TEE, compression rate of the device has to be 10%-20%, and position regarding circumflex coronary artery, mitral valve leaflet, and ridge with left pulmonary vein has to be considered optimal in multiple views (0°, 30°, 60°, 90°, and 135°). On angiogram, depth of the device in the LAA and angulation between long axis of the device and the LAA have to be considered optimal by the operators (right anterior oblique 30°/caudal 15° and right anterior oblique 30°/cranial 15°). If position of the device was not optimal, recaptures of the device and repositioning were performed. Once positioned, a tug test was performed followed by a waiting period of a few minutes before the device was released. 

TEE was performed on day 1 post procedure, prior to discharge, to verify absence of device dislodgment. After the procedure, single-antiplatelet therapy (either aspirin or clopidogrel 75 mg daily) was usually continued, or antithrombotic therapy adapted to the patient according to the physician’s discretion. Follow-up visit with TEE and CCT occurred at 12 ± 4 months.

Follow-up. TEE was performed by an echocardiographer trained in device follow-up and blinded to procedural information and result of CCT. Examination of the LAA was performed at 0°, 30°, 60°, 90°, and 120°, and focused on residual LAA permeability, determined by a persistent low-speed color flow inside the LAA (measured by lowering speed color scale to 20 cm/sec), peridevice leak (defined by the presence of a peridevice flow in color Doppler), and thrombus on the luminal side of the device.

CCT included a first electrocardiography-gated sequence covering the entire heart without injection to locate the prosthesis. High-pitch spirals covering only the LAA were performed with administration of a contrast agent (60 mL of iomeprol 816 mg/mL). Reconstructions were made at 70% RR interval. Scan initiation was triggered by bolus tracking with attenuation threshold in the ascending aorta set at 150 Hounsfield units. CCT analysis was focused on permeability of the LAA, defined by persistent LAA opacification. Permeability was classified into three grades: grade I = complete occlusion (absence of LAA opacification); grade II = partial occlusion (partial LAA opacification); and grade III = absence of occlusion (opacification of the entire LAA).

Serious adverse events were recorded at clinical follow-up visits after 45 days, 6 months, and 12 months. 

Statistics. Statistical analyses were performed with biostaTGV online software (http://marne.u707.jussieu.fr/biostatgv/). For between-group comparisons, continuous variables are expressed as mean ± standard deviation and compared with Wilcoxon-Mann-Whitney test. Qualitative variables are expressed as percentage of the effective, and compared with Chi-square test or Fischer’s exact test depending on the effective size (expected effective n>5 for Chi-square validation). 


Population. Thirty-five patients were admitted for LAA closure at the University Hospital of Nimes and Montpellier during the inclusion period. Patient characteristics are presented in Table 1. Mean age was 74.9 ± 6.4 years. The mean CHA2DS2-VASc score was 4.3 ± 1.4, with a mean HAS-BLED score of 3.4 ± 1.0. Patients had permanent AF in 40% of cases and paroxysmal AF in 60% of cases.

Procedures. Interventions were carried out in the conditions described above. Procedural characteristics are presented in Table 2. Failure of  Watchman device implantation occurred in 2 patients despite attempts using different device sizes. For both patients, an ACP device was successfully implanted during a second intervention. One intracardiac thrombus formation occurred on the distal part of the transseptal sheath at the beginning of the procedure without clinical consequences. Two groin hematomas without the need for surgical repair were also recorded (Table 3).

Clinical follow-up. One patient died of pulmonary infection 6 weeks after implantation, and 1 patient suffered from an ischemic stroke 2 weeks after implantation. One patient presented with a thrombus formation on the luminal side of the device without clinical consequences after resumption of full-dose anticoagulant therapy. Six patients were hospitalized for heart failure. Of these, 1 patient had previous systolic dysfunction, 1 patient suffered from rapid atrial tachycardia after AF ablation, 1 patient presented with cardiorenal syndrome, and 3 patients exhibited diastolic dysfunction with preserved ejection fraction. No acute renal failure or device embolization was recorded. Results are presented in Table 3.

CCT follow-up. Thirty out of 35 patients underwent CCT after a mean follow-up of 10 ± 6 months. CCT was not performed in 5 patients (1 patient died of pulmonary infection, 3 patients had severe renal failure [glomerular filtration rate <15 mL/min], and 1 patient refused CCT). Overall, 9/30 patients (30%) exhibited complete LAA occlusion without residual permeability (grade 1): 2/12 Watchman patients (16.7%) vs 7/18 ACP patients (38.9%); P=NS. The global permeability rate (grade 2 and grade 3) after 10 ± 6 months was 70%. CCT results are presented in Table 4.

TEE follow-up. Thirty-two patients underwent TEE after a mean follow-up of 8 ± 4 months. TEE was not performed in 3 patients (1 patient died and 2 patients refused TEE), all of whom had similarly not received CCT. A peridevice leak was diagnosed by aliasing in color Doppler mode in 7/32 patients (21.8%). One peridevice leak was >5 mm, 5 had a visible defect in device apposition, and all had grade 2 or 3 residual permeability on CCT (representative example in Figure 1). The TEE results are presented in Table 4.

Incomplete endothelialization. Among our cohort of 35 patients, a total of 5 did not undergo CCT during follow-up (see CCT follow-up section). A peridevice leak was found at TEE in 7 patients. The remaining 23 patients were divided into two groups. Patients presenting with incomplete endothelialization (n = 14; 60.8%), defined as LAA residual permeability on CCT but without peridevice leak on TEE (representative example in Figure 2), were compared to patients with complete endothelialization (n = 9; 39.2%), defined as absence of peridevice leak on TEE and residual permeability on CCT (grade 1) (Figure 3). Comparative results of the two groups are summarized in Table 5. Time from procedure to TEE or CCT was not statistically different between groups, but tended to be longer in the incomplete endothelialization group. There was no statistical difference between the two populations, but patients with incomplete endothelialization tended to be more likely to have diabetes (36% vs 11%) and permanent AF (57% vs 22%), and to have been implanted with larger devices (86% vs 44%). The ratio of Watchman to ACP device was non-significantly different between the two groups, although there was a trend for a higher ratio of Watchman devices in the complete endothelialization group (7:7 complete vs 2:7 delayed). 


In this study, we combined CCT and TEE for the follow-up of LAA closure with ACP or Watchman device. We observed a residual permeability rate of the LAA in 70% on CCT. Among these patients, we identified that incomplete endothelialization, defined as residual permeability on CCT without peridevice leak on TEE, could occur in 61% after 10 ± 6 months.

 Our result of residual permeability rate on CCT is consistent with the results of Jaguszewski et al, who found a permeability rate of 62% in a study including only ACPs.7 We found that CCT was the most objective marker to assess residual permeability, and TEE was more suitable to detect peridevice leak in aliasing color mode. Lowering speeds in color mode can help to detect residual permeability in our experience, but remains less accurate than CCT (sensitivity 90% and specificity 89% in our cohort).

The consequences of incomplete occlusion of the LAA have been a concern after surgical ligation,11,12 leading to recent European Society of Cardiology guidelines that recommend the continuation of anticoagulation treatment in AF patients at high risk for stroke after surgical LAA ligation.1 Conversely, Viles-Gonzalez et al8 did not find any clinically relevant risk of embolization in LAA devices implanted with small peridevice leaks (and so an incomplete sealing), but as mentioned by the authors, the study was too under-powered to conclude a formal absence of risk. Altogether, our study suggests, in agreement with previous studies, that the possible consequences of residual permeability should be investigated by further studies with more patients.

Incomplete endothelialization leads to residual filling of the LAA with very low velocity, which could theoretically trigger thrombus formation behind (and ultimately on) the LAA closure device. Patients with such an incomplete endothelialization should be carefully monitored in future large-cohort studies to investigate whether this might be a risk factor for embolic stroke. If so, this could lead to a recommendation for continued antiplatelet therapy.

Two reasons can explain residual permeability: apposition defect on the LAA ostium and lack of device endothelialization. We assumed that apposition defect was correctly detected by TEE through detection of peridevice leak. We found that 21.8% of our patients presented such a visible leak by TEE after 8 ± 4 months. This is comparable to the results of Viles-Gonzalez et al,8 who found a peridevice leak rate of 32% after 12 months. This was far from the 62% rate of residual permeability visible on CCT. We hypothesize that the remaining patients presented with an incomplete endothelialization. We cannot fully exclude that some patients of this group presented with apposition defect not detected at TEE, but it is unlikely that occurred for the majority. Following our definition of incomplete endothelialization, we found that around 60% of the patients conformed to this classification. This is of interest because dual-antiplatelet therapy following Watchman device implantation was intended to cover the endothelialization period, based on a study in dogs.9 It is reasonable to think that the endothelialization process could be longer in aged humans with comorbidities than in young animal models. This finding deserves further investigation, although data from large retrospective studies seem to indicate that postprocedural antiplatelet regimen does not have a strong impact on stroke/transient ischemic attack.13

We failed to identify predictive factors of incomplete endothelialization. This was probably due to our small number of patients, and observed tendencies deserve further investigation in larger populations with appropriate power. Diabetes, larger devices, and permanent AF (and thus larger atria?) are potential predictive factors. Identifying predictive factors of incomplete endothelialization may have clinical implications, especially for the follow-up and postoperative medications for these patients. 

Our study included both Watchman and ACP devices. Complete occlusion of the LAA, as assessed by CCT, was 22% for the Watchman group and 50% for the ACP group, but the small number of patients does not allow determination of superiority of one device over another. Further comparisons of these devices will be of interest for the future, as will careful monitoring of new-generation devices, which will hopefully have higher complete occlusion rates.

The population of our study differs from patients included in previous randomized control trials, especially Protect-AF, and better reflects the real-life patients implanted and contraindicated for long-term anticoagulant therapy, following the European Society of Cardiology guidelines. This could explain some differences in the procedural and follow-up outcomes, as well as differences in residual permeability and peridevice leak. 

This data set represents the initial experience of our center. All operators performed the same number of cases and were present at every procedure, as was the echocardiographer. We did not study the learning-curve effect on residual permeability and apposition defect due to our small number of patients, which is a limitation of our study. We used the term “incomplete endothelialization” in reference to the putative endothelialization time based on initial animal studies, and thus the duration of dual-antiplatelet regimen recommended for the Watchman device. Indeed, “normal” delay of endothelialization in humans has never been studied. Our cohort was too small and follow-up was too short to address this question, but emphasizes the need for a larger cohort.

Study limitations. In the absence of an available tool as a gold standard, we determined “incomplete endothelialization” as derived by multimodal imaging. It is likely that after validation in animal models, new imaging techniques such as three-dimensional TEE could help to define such a gold standard to monitor the endothelialization process in the future. 

The small number of patients and limited follow-up duration do not allow us to fully answer the questions raised by our study, especially the clinical relevance of incomplete endothelialization and thus the appropriate antithrombotic regimen in this population contraindicated to anticoagulant therapy and the superiority of one device over another.


Our study showed that 9 months after occlusion, the LAA remains permeable in nearly 70% of cases. The causes of residual permeability are twofold: apposition defect represented by peridevice leak, and incomplete endothelialization. By comparing TEE and CCT results, we found that incomplete endothelialization (according to our definition) may be the cause of nearly 61% of residual permeability cases. The consequences of these incomplete occlusions and the predictive factors of incomplete endothelialization must now be determined in a larger cohort.

In this study, we addressed the question of LAA device endothelialization. In the future, new imaging tools will be needed to monitor the endothelialization process. The pending question will be to determine the prognosis of patients with delayed or incomplete endothelialization.


1.     Kirchhof P, Benussi S, Kotecha D, et al. 2016 ESC guidelines for the management of atrial fibrillation developed in collaboration with EACTS. Europace. 2016;18:1609-1678. 

2.     Connolly SJ, Ezekowitz MD, Yusuf S, et al. Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med. 2009;361:1139-1151. 

3.     Patel MR, Mahaffey KW, Garg J, et al. Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med. 2011;365:883-891. 

4.     Granger CB, Alexander JH, McMurray JJV, et al. Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2011;365:981-992. 

5.     Giugliano RP, Ruff CT, Braunwald E, et al. Edoxaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2013;369:2093-2104. 

6.     Holmes DR, Reddy VY, Turi ZG, et al. Percutaneous closure of the left atrial appendage versus warfarin therapy for prevention of stroke in patients with atrial fibrillation: a randomised non-inferiority trial. Lancet. 2009;374:534-542. 

7.     Jaguszewski M, Manes C, Puippe G, et al. Cardiac CT and echocardiographic evaluation of peridevice flow after percutaneous left atrial appendage closure using the AMPLATZER cardiac plug device. Catheter Cardiovasc Interv. 2015;85:306-312. 

8.     Viles-Gonzalez JF, Kar S, Douglas P, et al. The clinical impact of incomplete left atrial appendage closure with the Watchman Device in patients with atrial fibrillation: a PROTECT AF (Percutaneous Closure of the Left Atrial Appendage Versus Warfarin Therapy for Prevention of Stroke in Patients With Atrial Fibrillation) substudy. J Am Coll Cardiol. 2012;59:923-929. 

9.     Schwartz RS, Holmes DR, Van Tassel RA, et al. Left atrial appendage obliteration: mechanisms of healing and intracardiac integration. JACC Cardiovasc Interv. 2010;3:870-877. 

10.     Schiettekatte S, Czapla J, Nijs J, La Meir M. Unmasking a naked left atrial appendage closure device: a case of a silent embolic threat. Heart Rhythm. 2014;11:2314-2315. 

11.     Katz ES, Tsiamtsiouris T, Applebaum RM, Schwartzbard A, Tunick PA, Kronzon I. Surgical left atrial appendage ligation is frequently incomplete: a transesophageal echocardiographic study. J Am Coll Cardiol. 2000;36:468-471. 

12.     García-Fernández MA, Pérez-David E, Quiles J, et al. Role of left atrial appendage obliteration in stroke reduction in patients with mitral valve prosthesis: a transesophageal echocardiographic study. J Am Coll Cardiol. 2003;42:1253-1258. 

13.     Tzikas A, Shakir S, Gafoor S, et al. Left atrial appendage occlusion for stroke prevention in atrial fibrillation: multicentre experience with the Amplatzer cardiac plug. EuroIntervention. 2016;11:1170-1179.

From the 1Service de Cardiologie Centre Hospitalier Universitaire de Nimes, Nimes, France; and 2Département de Cardiologie Centre Hospitalier Universitaire de Montpellier, Montpellier, France.

Disclosure: The authors have completed and returned the ICMJE Form for Disclosure of Potential Conflicts of Interest. The authors report no conflicts of interest regarding the content herein.

Manuscript submitted November 3, 2017, provisional acceptance given December 19, 2017, final version accepted January 25, 2018. 

Address for correspondence: Dr Mathieu Granier, Service de Cardiologie CHU Caremeau, 1, pl. R. Debré, 30900 Nimes, France. Email: mathieu.granier@chu-montpellier.fr