Device Thrombosis After Percutaneous Left Atrial Appendage Occlusion Is Related to Patient and Procedural Characteristics but Not to Duration of Postimplantation Dual Antiplatelet Therapy

This is a prospective, single-center, ANIN-LAAO Registry of 102 consecutive patients with atrial fibrillation who underwent LAAO with AMPLATZER Cardiac Plug (ACP)/Amulet (n=59; ACP in 25, Amulet in 34 patients) or WATCHMAN (n=43) device between October 2011 and May 2016. The registry has been conducted in agreement with the Declaration of Helsinki and approved by the Institutional Review Board. Device type was left to the operator’s discretion. We excluded 3 patients (1 ACP/Amulet and 2 WATCHMAN) because of chronic oral anticoagulation in a patient with mechanical valve (n=1) and refusal of follow-up imaging (n=2).

The data, analytic methods, and study materials will not be made available to other researchers for purposes of reproducing the results or replicating the procedure.

Procedures were performed under general anesthesia and guided with transesophageal echocardiography (TEE). Loading dose of unfractionated heparin was given to achieve activated clotting time >250 s. Patients were preloaded with acetylsalicylic acid. Three doses of subcutaneous low–molecular weight heparin at 12-hour intervals were given starting 4 to 6 hours after LAAO. Loading dose (600 mg) of clopidogrel was given the day after the procedure.

Follow-up was done at 1.5, 3 to 6, and 12 months post-procedure with left atrial imaging. Dedicated questionnaires were used to obtain patient’s baseline and follow-up data. The primary imaging modality was TEE; however, 29 (29.3%) participants underwent cardiac computed tomography (CCT) instead.⁷ The choice to perform CCT was left to the attending physician’s discretion; estimated glomerular filtration rate <40 mL min⁻¹ 1.72 m⁻² was deemed contraindication to CCT.

TEE imaging was done in at least 3 planes (0–30°, 60–90°, and 120–180°) as recommended in appropriate guidelines.^8,9 CCT was performed with a dual-source scanner (Somatom Force; Siemens Healthcare, Forchheim, Germany) using a dedicated protocol with an ECG-gated, contrast-enhanced scan (0.6 mm slice, 0.4 mm increment, and 70 kV tube voltage), with a total estimated radiation dose of 1.9 to 3.4 mSv, and 45 to 55 mL of iopromide contrast (Ultravist 370; Bayer Schering Pharma, Leverkusen, Germany). Images were interpreted on a SyngoVia workstation (Siemens, Forchheim, Germany).⁷

DRT was defined as an echo density on the left atrial aspect of the device: (1) not explained by imaging artifact; (2) inconsistent with normal healing/device incorporation; (3) visible in multiple TEE/CCT planes; and (4) in contact with the device.¹⁰ Peridevice leak was defined as a gap or jet into the appendage ≥5 mm. In patients undergoing TEE, color-flow Doppler imaging was used to identify peridevice leaks, with measurements taken in multiple planes to evaluate for the maximal jet dimension. In patients undergoing CCT, the gap between the device and the left atrial appendage wall with contrast filling was measured in 2 planes in search for the maximal gap dimension. Deep implantation (residual appendage) was defined as incomplete coverage of the pulmonary ridge between the left atrial appendage and the upper left pulmonary vein, leaving an uncovered rim of at least 10 mm length as measured in 0° to 60° TEE projection or corresponding CCT images (Figure). The 10 mm distance of uncovered rim takes into account the rounded proximal edges of the WATCHMAN device, which unlike the disc of the ACP/Amulet device is not designed to cover the tip of the pulmonary ridge completely. Two independent observers assessed all follow-up TEE and CCT images before a final adjudication of DRT was made. In cases with disparate opinions, final adjudication was reached by consensus.

The team of physicians performing follow-up visits was given a general recommendation to prescribe P-DAPT for 1 to 6 months. Decisions on the duration of P-DAPT were made according to the physician’s judgment. Antiplatelet monotherapy was recommended for at least 6 months post-procedure.

The main end point of the study was DRT. On the basis of the time of detection, DRT was classified as early (at the 1.5-month follow-up), late (at the 3- to 6-month follow-up), or very late (at the 12-month follow-up). Thromboembolic complications are reported descriptively because the sample size did not allow for adequately powered statistical analysis.

Continuous variables are presented as mean±SD or as median with interquartile range (IQR). Categorical variables are presented as frequencies and percentages. The Shapiro–Wilk test was used to test variables for normality, and the hypothesis of normal distribution was rejected in all tested variables. Differences between continuous variable were therefore determined by Mann–Whitney test. Differences between categorical variables were determined by Fisher exact test. Given the small number of events, multivariable analysis was not performed. P value of <0.05 was considered statistically significant. Interaction analysis between the device type and predictors of DRT was performed. All statistical analyses were performed using SPSS software, version 17 (SPSS, Inc, Chicago, IL).

Baseline and periprocedural characteristics are presented in Table 1. Median CHA₂DS₂-VASc score was 4.0 (IQR, 3.0–5.0). One third of patients had prior thromboembolic episode. Deep implantation was present in 42.4% of patients (33.3% WATCHMAN and 77.2% ACP/Amulet). Median P-DAPT duration was 13.0 weeks with a wide IQR between 7.3 and 26.0. No uncovered crypts or side lobes were seen in any patients after the procedure.

DRT was diagnosed in 7 (7.1%) patients, 5 (71%) with TEE and 2 (29%) with CCT. The diagnoses were made at the first follow-up in 2 (28.6%) patients (early DRTs), second follow-up in 2 (28.6%) patients (late DRTs), and third follow-up in 3 (42.9%) patients (very late DRTs). Mobile thrombus was seen only in 1 patient (case No. 1, Table 2). DRT-directed treatment with outcomes is presented in Table 2. The only stroke reported in our cohort was in the DRT group (n=1, 14.2%) and was followed by death a couple of months later. No DRT was diagnosed during follow-up imaging in the remaining 5 (5%) patients who died.

Patients with DRT more often presented with a history of thromboembolic episodes (71.4% versus 30.4%; P=0.04) and had lower median ejection fraction (50.0% [IQR, 35.0–55.0] versus 60.0% [IQR, 55.0–66.0]; P<0.01] when compared with patients without DRT. No other baseline differences were observed between the 2 groups. Patients with DRTs more often had deep implantations (85.7% versus 39.1%; P=0.04) and were implanted with larger occluders (30.0 mm [IQR, 27.0–33.0] versus 25.0 mm [IQR, 24.0–28.0]; P<0.01) when compared with patients without DRT. Peridevice leaks were found in 8 (8.1%) patients (6 [14.6%] after Watchman and 3 [5.2%] after ACP/Amulet implantation]. Neither device type nor postprocedural peridevice leaks were related to DRT. All 3 DRTs in ACP/Amulet group were found on the second-generation, Amulet device. There was no difference in the length of P-DAPT between patients with and without DRT (12.4 weeks [IQR, 6.0–49.7] versus 13.0 weeks [IQR, 7.3–26.0]; P=0.77). Interaction analysis did not show any impact of the device type on the observed differences (P<0.1 for each variable).

Prior studies on DRT did not evaluate the impact of P-DAPT on its incidence. From the original cohort of 1047 participant from European ACP/Amulet registry, Saw et al¹¹ selected 339 patients who had TEEs available for central adjudication and reported DRT incidence of 3.2%. However, the number of TEE follow-ups and intervals between them were not standardized, postimplantation antithrombotic treatment was not detailed, and prior thromboembolism or deep implantation was not accounted for. Plicht et al¹² followed up 34 patients after LAAO with ACP/Amulet up to 1 year with systematic left atrial imaging. P-DAPT was recommended for at least 6 months. DRT was found in 6 (17.6%) patients. Similar to our results, reduced left ventricular ejection fraction, but not left atrial size or mitral regurgitation, was associated with DRT. Moreover, no correlation between genetic mutations associated with reduced clopidogrel reactivity and DRT was found, which is in line with our results where P-DAPT duration was not related to DRTs diagnosis. Other factors related to DRT were CHA₂DS₂-VASc score and preprocedural platelet count. Deep implantation was present in 62%, but its impact on DRT formation was not found in this small sample of patients. Main et al¹⁰ reported a 5.7% DRT rate based on 485 patients from PROTECT AF trial (Watchman Left Atrial Appendage System for Embolic Protection in Patients With Atrial Fibrillation). Incident DRT had higher prevalence at 6 and 12 months when compared with 1.5 months post-procedure. Kaneko et al¹³ analyzed predictors of DRT in 78 patients after WATCHMAN implantation. DRT was diagnosed in 5% of patients. Follow-up TEE was performed 45 days after the procedure in all studied patients and only in a subgroup at 6 months. DRT was predicted by CHA₂DS₂-VASc and deep implantation. Similar to our study, prior stroke or TIA was numerically more frequent in the DRT versus no DRT group (50% versus 19%); however, the difference did not reach statistical significance.

Our report is unique because of a wide variation in the length of P-DAPT, which allowed for analysis of its impact on the DRT incidence. To the best of our knowledge, no such analysis is currently available in the literature. Moreover, contrary to the other reports, it incorporates 2 different, most widely used devices. For the first time, interaction analysis was used to see if DRT predictors differ between them. Importantly, we used consistent imaging at 3 time points up to 1 year post-procedure. Feasibility of short-term P-DAPT of <6 months after WATCHMAN device has also been shown for the first time.

DRT seems to be related to global prothrombotic milieu represented by high CHA₂DS₂-VASc score or prior thromboembolism. Local mechanical factors seem to play a role too. Impaired left ventricular ejection fraction worsens left atrial emptying conditions, which may lead to blood stasis and thrombus formation. Deep implantations leave residual appendage with potential blood swirling and stasis in that area. We used at least 10 mm distance of uncovered pulmonary ridge to define deep implantation. This cutoff seemed reasonable given the design of the WATCHMAN device that plugs the appendage without the disk to cover the ridge completely. Also, it might be more difficult for left atrial side of the occluders’ surface to completely cover with neoendothelium when it is large. These are all hypothetical mechanisms that remain to be proven.

No randomized data are available to guide the scope and length of postimplantation antithrombotic treatment. A wide range of 1 to 6 months of P-DAPT is recommended after ACP/Amulet implantation.⁶ WATCHMAN implantation is followed by 6 weeks of warfarin and subsequent 5 months of P-DAPT or otherwise 6 months P-DAPT alone.^2,5 This is intended to buffer against clot formation on the device before its endothelialization. The healing response to nitinol-based devices is initiated by thrombotic material formation and then its transformation into connective tissue within 4 weeks.^14,15 However, the coverage may never be complete.^16,17 Considering the complexities of healing responses, it may even be hypothesized that antiplatelets inhibit fibrin formation at early stages of device healing and thus delay its endothelialization.¹⁸ To date, no evidence suggests that prolonged P-DAPT prevents DRT, and our report constitutes an argument to confirm this.

DRTs were classified based on time to first diagnosis as early (at 1.5 months post-procedure), late (at 3–6 months post-procedure), and very late (at 12 months post-procedure). Whether the DRT timing results from varying thrombotic mechanisms remains unproven. Besides, DRT identification may not always be unequivocal.¹⁰ Accumulation of plane thrombus is a process in time, initialized by desirable, fibrin-based healing that may turn into excessive, unwanted thrombus formation. DRT may initially be too thin to be detected by TEE or CCT and become apparent at later stages of follow-up. Relatively high rates of late and very late DRTs evidenced by our study are in line with the study by Main et al.¹⁰

Little is known about safety and efficacy of DRT treatment. In our cohort, DRT was plane and adherent to the device in all but 1 case in which it was mobile and pedunculated. Up until the currently presented analysis, treatment aimed at DRT resolution was successful in 44.4% of patients from our cohort.

Significant progress in device design has been made because the time initial type of LAAO occluders was in clinical use.¹⁹ At the current stage of LAAO technology, it still seems difficult to eliminate deep implantations, but efforts should be made to cover the entire pulmonary ridge before device release. The design of the ACP/Amulet device is intended to fully cover the left atrial appendage ostium with the disc; however, in our cohort, deep implantations were frequent in the ACP/Amulet group. These results are consistent with the findings of Wolfrum et al,²⁰ where in a cohort of 169 patients after ACP implantation, complete coverage was seen in a minority of patients (45%). Whether implantation technique and device sizing influence the percentage of incomplete ostial left atrial appendage coverage remains to be studied. Procedure planning based on CCT or 3-dimensional–printed heart models might potentially be helpful.^21,22 Future efforts in device design should aim at enabling complete closure of left atrial appendage, avoiding need for deep implantations. Patients with DRT risk features should be considered candidates for longer-term imaging surveillance after LAAO. It seems that P-DAPT after LAAO may be reduced to a minimum of 1 month given well-documented bleeding risks associated with prolonged DAPT and lack of evidence of its protective role against DRT formation. Such approach is novel with regard to WATCHMAN device and thus should be confirmed in larger prospective series. However, optimal postimplantation treatment regimen remains to be defined in future randomized trials.

This report is based on prospective registry with all limitations inherent to such study design. DRT may have been related to factors not accounted for in our analysis. Randomized data are needed to confirm these study findings. The study population is rather small; however, it is the second largest real-world series published to date. No potential impact of the device type on characteristic related to DRT was confirmed by interaction analysis; however, device-related differences might become apparent in larger series. Although the studies were read by experienced TEE and CCT readers with expertise in imaging of intracardiac devices, the diagnosis of DRT remains subjective. To reduce some of this limitation, each study was independently read by 2 reviewers who were blinded to each others finding. Despite this, the current imaging techniques are less sensitive to distinguish between thrombus versus a normal healing pattern, especially if the thrombus is thin and layered.

In this first real-world series studying impact of P-DAPT duration on DRT incidence after LAAO, DRT was observed early (2%), late (2%), and very late (3%) postprocedure with an overall incidence of 7.1%. Patients with DRT were more likely to have prior thromboembolism, lower left ventricular ejection fraction, deep device implantation, and use of larger occluders. DRT was unrelated to P-DAPT duration. On the basis of these observations, longer surveillance of patients with at-risk features for DRT together with avoidance of deep implantation should be considered.