Author + information
- Received November 6, 2012
- Revision received January 31, 2013
- Accepted February 15, 2013
- Published online June 1, 2013.
- Bjoern Plicht, MD∗∗ (, )
- Thomas F.M. Konorza, MD∗,
- Philipp Kahlert, MD∗,
- Fadi Al-Rashid, MD∗,
- Hagen Kaelsch, MD∗,
- Rolf Alexander Jánosi, MD∗,
- Thomas Buck, MD∗,
- Hagen S. Bachmann, MD†,
- Winfried Siffert, MD†,
- Gerd Heusch, MD‡ and
- Raimund Erbel, MD∗
- ∗Department of Cardiology, West-German Heart Center Essen, University Hospital Essen, University Duisburg-Essen, Essen, Germany
- †Institute of Pharmacogenetics, University Hospital Essen, Essen, Germany
- ‡Institute for Pathophysiology, University Hospital Essen, Essen, Germany
- ↵∗Reprint requests and correspondence:
Dr. Bjoern Plicht, Department of Cardiology, West-German Heart Center, University Hospital Essen, University Duisburg-Essen, Hufelandstrasse 55, D-45122 Essen, Germany.
Objectives This study sought to identify risk factors for thrombus formation on the Amplatzer Cardiac Plug (ACP) (St. Jude Medical, St. Paul, Minnesota) after left atrial appendage occlusion.
Background Left atrial appendage occlusion with the ACP aims to reduce the risk of embolic stroke and bleeding complications associated with vitamin K antagonists in patients with atrial fibrillation.
Methods We performed transesophageal echocardiography before discharge and after 3, 6, and 12 months in 34 patients with atrial fibrillation undergoing ACP implantation and receiving dual antiplatelet therapy. Clinical, echocardiographic, and hemostaseological parameters were retrospectively analyzed to identify risk factors for thrombus formation.
Results Three patients had thrombi before discharge, 3 more at the 3-month follow-up. No differences were found in left atrial volume, left atrial appendage velocity, spontaneous echo contrast, transmitral gradient, or mitral regurgitation between patients without or with thrombi. CHADS2 (Congestion, Hypertension, Age, Diabetes, and Stroke) score (2.0 ± 1.1 vs. 4.3 ± 1.0), CHA2DS2-VASc (CHADS2 plus Vascular Disease and Sex Category) score (5.2 ± 1.3 vs. 6.8 ± 0.8), and pre-interventional platelet count (215.9 ± 63.9/nl vs. 282.5 ± 84.4/nl) were higher and ejection fraction (50.6 ± 11.4% vs. 39.7 ± 10.6%) lower in those with thrombi. Factor 2, factor 5, or methylenetetrahydrofolate reductase mutations and genetic variants associated with reduced clopidogrel activity were not more frequent in patients with thrombi.
Conclusions Transesophageal echocardiography identified 17.6% of patients with thrombus formation on the ACP despite dual antiplatelet therapy. CHADS2 and CHA2DS2-VASc scores, platelet count, and ejection fraction are risk factors for such thrombus formation.
- atrial fibrillation
- device thrombus
- left atrial appendage occlusion
- transesophageal echocardiography
Embolic stroke is the most serious complication in patients with atrial fibrillation (AF), and its annual risk ranges between 1.9 and 18.2% (1). Oral anticoagulation with vitamin K antagonists (VKA) is recommended in patients with CHADS2 (Congestion, Hypertension, Age, Diabetes, and Stroke) score ≥2 and reduces stroke risk (2). However, the therapeutic range of VKA is narrow, and severe bleeding is a serious side effect (3).
Up to 90% of emboli from nonvalvular AF originate in the left atrial appendage (LAA) (4). Therefore, interventional LAA occlusion was introduced to avoid long-term oral anticoagulation, and the Amplatzer Cardiac Plug (ACP) (St. Jude Medical, St. Paul, Minnesota) is such a device. After ACP implantation, antiplatelet/antithrombotic medication is mandatory until its endothelial coverage is complete; the manufacturer's instructions include dual antiplatelet therapy for at least 1 month with discontinuation of oral anticoagulation.
In our initial series of 34 patients undergoing ACP implantation, we surprisingly detected 6 patients with thrombus formation on the device in close follow-up by transesophageal echocardiography (TEE). We then tried to retrospectively identify clinical, echocardiographic, and hemostaseological risk factors for thrombus formation.
Between November 2009 and February 2011, 34 consecutive patients with AF, CHADS2 score ≥2, and increased bleeding risk under oral anticoagulation as reflected by HAS-BLED (Hypertension, Abnormal Liver or Renal Function, Stroke, Bleeding, Labile International Normalized Ratio, Elderly, Drug, or Alcohol Intake) score ≥3 (1) underwent ACP implantation at our institution after informed consent.
Under mild sedation, fluoroscopy, and real-time 3-dimensional TEE, the device was delivered over a 10-F or 13-F sheath into the left atrium after transseptal puncture (Baylis NRG transseptal needle, Baylis Medical, Montreal, Canada). The device was first anchored in the LAA landing zone, approximately 1 cm behind the orifice; the disk was then unfolded to cover the LAA entrance (Fig. 1), optimally covering the rim between LAA and the upper left pulmonary vein (Fig. 2). A wiggle maneuver confirmed the safe placement before its final release. Patients received up to 12,500 U of unfractionated heparin intravenously to achieve an activated clotting time >250 s. Oral anticoagulation with VKA was discontinued prior to the procedure to achieve an international normalized ratio <1.4. Dual antiplatelet therapy with 100 mg of aspirin daily and 75 mg of clopidogrel daily after a 600-mg loading dose given in the catheterization laboratory post-procedurally was administered for at least 6 months according to our clinic's standard protocol. Eight patients were pre-treated with clopidogrel before so they received a 300-mg loading dose.
From the patients' files, baseline platelet count and plasma fibrinogen; diabetes mellitus; type of AF; and CHADS2, CHA2DS2-VASc (CHADS2 Plus Vascular Disease and Sex Category), and HAS-BLED scores were retrospectively identified.
Transthoracic echocardiography and real-time 3-dimensional TEE were performed prior to the procedure to exclude intracardiac thrombi and record baseline characteristics, notably conditions possibly associated with increased risk for thrombus formation: left ventricular ejection fraction (EF), LAA flow velocities, left atrial volume, transmitral gradient, mitral valve area, and spontaneous echocardiographic contrast, which was graded semiquantitatively in a score between 0 (none) up to 4 (severe) (5). Mitral regurgitation severity was scored from 4 (mild) to 12 (severe) (6).
Intraprocedural real-time 3-dimensional TEE documented the safe placement of the device and its position. TEE follow-up was performed before discharge and after 3, 6, and 12 months to document device position, remaining flow within the LAA, and freedom from pericardial effusion or thrombus formation.
Twenty-four of the initial 34 patients gave informed consent and the local ethics committee approved a retrospective analysis of coagulation status. Five of the initial 6 patients with thrombus formation were included in this analysis. Factor 2 G20210 (7) and 5 G1691A polymorphisms (8) were identified. Methylenetetrahydrofolate reductase C677T and A1298C polymorphisms, which are related to hyperhomocysteinemia, were examined (9). Genetic mutations in the cytochrome P450 CYP2C19 locus (10) were checked for the *2, *3, and *17 variants (11). The PON1 AA variant as 1 determinant of clopidogrel resistance was examined (12). Antithrombin 3, factor 8, protein C and S activities, and the presence of lupus anticoagulant were determined.
Data were analyzed with SPSS Statistics (version 19, IBM Corp., Somers, New York) and presented as mean ± SD for normally or as medians (interquartile range) for non-normally distributed items. Univariate analysis of variance for normally or Mann-Whitney U test for non-normally distributed items were performed for continuous variables. Chi-square test was performed for dichotomous parameters. A p value <0.05 was considered significant.
Twenty patients had paroxysmal AF, and 14 of those had persisting or permanent AF (Table 1). CHADS2 score was 3.2 ± 1.2, CHA2DS2-VASc score was 5.5 ± 1.4, and HAS-BLED score was 4.2 ± 1.1. Except for 2, all patients were on oral anticoagulants prior to the procedure. Except for 1 patient on permanent AF after mitral valve reconstruction 10 years before, all patients had nonvalvular AF.
In all patients, the procedure was performed successfully within 68 ± 38 min, and the occluder size was 24.1 ± 3.1 mm (range: 18 to 30 mm). In 13 patients, the ACP disk covered the LAA rim, whereas in the other patients, a deeper implantation of the device into the LAA was necessary. Two patients experienced post-interventional pericardial effusion without hemodynamic relevance; 1 additional patient needed pericardiocentesis. No periprocedural stroke was detected. Two patients needed blood transfusions after groin bleeding.
In 32 patients, a pre-discharge TEE was performed 3.6 ± 1.7 days after the procedure; 2 patients refused the examination. In all examined patients, the ACP occluder was safely anchored in the LAA. In 3 patients, there was slight residual flow within the LAA. In 3 patients, a thrombotic mass on the LAA occluder was detected (Fig. 3); 2 patients received intravenous heparin, and complete thrombus resolution was achieved in the first patient. In the second patient, a thrombus recurred after initial partial resolution followed by an attempt of oral anticoagulation with dabigatran. Complete resolution was achieved after switching to VKA (Fig. 4). A third patient was immediately put on VKA.
The on-site follow-up rate for the cohort at 3 months was 88%, whereas the other patients were followed up by phone call. In all examined patients, the ACP was still safely anchored in the LAA. Residual flow was still detectable in the above-mentioned 3 patients. Three additional patients were found to have a thrombus on the ACP. In 2 of these patients, careful retrospective review of the pre-discharge TEE revealed very small structures on the ACP not originally identified as thrombi. The third patient had refused a pre-discharge TEE. In all patients, thrombi were resolved successfully by intravenous heparin or oral anticoagulation with VKA. Two of the 3 patients with a thrombus in the pre-discharge TEE had no recurrence, so VKA were discontinued, whereas the third patient still had thrombus remnants on the disk and VKA were continued.
The on-site follow-up rate of the cohort at 6 months was 82%. The ACP was still in place in all patients. Slight residual flow within the LAA was found in only 1 patient. No new or recurrent thrombi were detected. The follow-up in the patient with thrombus remnants at 3-month follow-up was delayed until 12 months on patient's request under continued VKA without any adverse events.
A 12-month follow-up TEE was performed in 74% of patients. All devices were found to be in place, but minimal residual flow was detected in 1 patient. No thrombus occurrence or recurrence was detected. The patient with remnants in the 3-month follow-up had no thrombi anymore.
No clinically apparent embolic event occurred during follow-up.
Echocardiographic characterization of thrombi
Except for Patient #4 (Fig. 4), in whom a massive thrombus covered the whole device, all thrombi originated from the central screw. Whereas in Patients #1 to #3, there were smaller mobile structures, Patients #5 and #6 had adherent thrombi spreading toward the rim and covering the disk to a varying extent (Online Fig. 1).
Comparison of patients without and with thrombi
Paroxysmal AF was present in 57% of patients without and in 67% with thrombi (Table 1, Online Table 1). There were no differences in the prevalence of diabetes and renal insufficiency or procedure-related factors such as occluder size or placement in relation to the rim between both groups. Apart from EF, echocardiographic parameters were not different between groups.
However, patients with thrombus had CHADS2 score ≥3 and CHA2DS2-VASc score ≥6, resulting in positive predictive values of 26% and 40%, respectively, and a negative predictive value of 100%. Pre-procedural EF was lower in patients with than without thrombi. Five thrombus patients had an EF <50%, resulting in a positive predictive value of 38% and a negative predictive value of 95%.
Pre-procedural fibrinogen concentration did not differ between groups; however, platelet count was higher in those with thrombi. The activities of antithrombin 3, factor 8, and proteins C and S were not different between groups, positive lupus anticoagulant tended to be more frequent in patients with thrombi (Table 2). Factor 2 and 5 mutations were found in 1 patient each of the nonthrombus group, both without any thrombotic events. Methylenetetrahydrofolate reductase mutations were found frequently, but only in nonthrombus patients. CYP2C19*2 mutations potentially leading to clopidogrel resistance occurred frequently, always in heterozygous form with a trend preferring the thrombus-group. The CYP2C19*3 variant was not found as expected in our Caucasian cohort. The CYP2C19*17 variant, which potentially accelerates clopidogrel metabolism with a higher risk of bleeding, was found quite often in both groups. The PON1 AA variant, possibly accounting for reduced clopidogrel activity, was identified at an equal rate in both groups.
We demonstrate unexpected thrombus formation on the ACP and, therefore, raise safety concerns for this novel treatment option in AF patients. The efficacy of LAA occlusion in patients with AF has been demonstrated in the PROTECT-AF (Watchman Left Atrial Appendage System for Embolic Protection in Patients With Atrial Fibrillation) trial (13) using the Watchman device (Atritech, Inc., Plymouth, Minnesota).
A recent device for percutaneous LAA occlusion is the ACP, which consists of a lobe anchored in the LAA and a connected disk to cover the LAA orifice, assuming that it provides better protection against thromboembolism (“pacifier principle”) (14). The disk is intended to cover the rim of the LAA at the orifice of the upper left pulmonary vein. It is uncertain whether or not the occluder can always be positioned in this way and whether or not position is a risk factor for thrombus formation.
Current registry data confirm the feasibility of ACP implantation in 137 patients with 7% serious complications (15), including 3 patients with peri-interventional stroke. Two cases of post-procedural thrombus formation were reported: 1 detected by TEE after 3 months and the cessation of clopidogrel after 2 months due to lower gastrointestinal bleeding (16); the second found by TEE 6 months after dual antiplatelet therapy for 4 weeks and then continuation of aspirin alone (17). Recently, López-Miniguez et al. (18) reported the detection of thrombi in 5 of 35 ACP implantations, but without detailed analysis of risk factors.
In the PROTECT-AF trial (13), patients received warfarin for 45 days after Watchman implantation to bridge the time until its complete endothelial coverage. The device was anchored in the LAA but did not cover its entrance; warfarin was replaced after 45 days by dual antiplatelet therapy until 6 months after implantation. Device-associated thrombi were observed in 20 of 478 successful implantations (4.2%), and thrombus-associated stroke rate was 0.3% per 100 patient-years (19). In preliminary data from the ASAP (ASA Plavix) registry (20), 4 of 116 patients having high risk for bleeding and receiving only dual antiplatelet therapy experienced thrombus formation on the Watchman device. In a meta-analysis of 54 published cases (5 with AF), thrombi were seen with all commercially available atrial septal defect/patent foramen ovale occluder devices, preferentially on their left side (21). In 20 (2%)—4 of them had AF—of 1,000 patients after atrial septal defect/patent foramen ovale occlusion with different devices, thrombi were found, again preferentially on their left side. No significant differences were found in anticoagulation/antiplatelet regimes between patients without and with thrombi; thrombus resolution was achieved with anticoagulation in 17 cases (22).
Thrombosis rate in our cohort (17.6%) exceeds those previously reported (2% to 4.2%) (19,20,22). Potential explanations relate to post-interventional anticoagulation/antiplatelet regime, implantation technique, or the ACP device itself.
Post-interventional anticoagulation/antiplatelet regime
When placed optimally, the ACP device excludes the LAA completely by covering the entrance of the LAA, possibly justifying not recommending VKA. Dual antiplatelet therapy for 1 month followed by aspirin monotherapy for 5 more months was then regarded sufficient, but it was not supported by data. In 10 dogs, the ACP was implanted with no evidence for thrombus formation on the occluder during 90 days follow-up; however, the antithrombotic or antiplatelet regime was not defined (23). Early successful ACP implantations in 2 patients with dual antiplatelet therapy had no TEE follow-up so that potential thrombus formation remained unaddressed (24). In a series of 20 patients, no thrombi were found during 12 months under dual antiplatelet therapy; however, CHADS2 was lower (2.3 ± 1.3) than in our cohort (3.2 ± 1.2) (25).
With close TEE follow-up, we identified 6 patients in whom dual antiplatelet therapy did not prevent thrombus formation. Echocardiographic parameters reflecting reduced flow velocities in the left atrium did not differ between patients with or without thrombi. Not unexpectedly, higher CHADS2 or CHA2DS2-VASc scores, along with reduced EF, were associated with thrombus formation. Reduced EF is an established risk factor for thrombus formation in the native LAA of AF patients (26–28) and was strongly associated with thrombus formation in our analysis. Also, pre-interventional platelet count was higher in the thrombus group, yet within the physiologic range. Yuce et al. (28) reported no association of platelet counts with left atrial thrombosis risk. In contrast, in acute coronary syndromes, higher platelet count is associated with worse prognosis (29,30). Possibly, the platelet response—here to implantation of foreign material into the LAA—is disproportionate even in highly normal platelet counts, emphasizing the importance of effective platelet inhibition. Retrospectively, we found no genetic risk for clopidogrel resistance or other coagulation disorders in patients with thrombi.
The high rate of post-interventional thrombi on the occluder indicates that dual antiplatelet inhibition was not sufficiently protective. This emphasizes the need for oral anticoagulation with VKA, in line with the PROTECT AF trial (13) in which all patients received oral anticoagulation for 45 days after implantation followed by dual antiplatelet therapy for 6 months and had fewer thrombi. A risk-stratified anticoagulation/antiplatelet regime appears justified according to our results, given that patients with low risk scores and good EF developed no thrombi on dual antiplatelet therapy, with negative predicted value for EF >50%, CHADS2 score <3, and CHA2DS2-VASc score <6 between 95% and 100%. This notion is supported by Lam et al. (25), who reported 20 patients with CHADS2 of 2.3 ± 1.3 and no device thrombosis, similar to our 28 nonthrombus patients with CHADS2 of 2.0 ± 1.1. Data from a larger prospective trial are needed to confirm this concept of adjusting the post-implantation antiplatelet/anticoagulation regime to the individual risk profile. As a consequence, from the current data of our first 34 patients, we now put all patients on VKA post-interventionally until the next TEE follow-up (usually 3 months) for safety reasons until more meaningful data are available.
Potential rebound effects after clopidogrel cessation are unlikely in our cohort due to reliable compliance to clopidogrel. TEE follow-up after clopidogrel cessation did not show any thrombi.
The optimal ACP position according to the manufacturer's recommendation was not achieved in more than 60% of the cases and is possibly a risk factor for thrombus formation by forming a region of slow flow in the LAA. However, the number of implants not covering the rim of the LAA was equal in patients without and with thrombi. If lack of coverage of the LAA would result in higher thrombus risk, a higher incidence would be expected from the Watchman device (19).
ACP device itself
Most thrombi originated from the central screw, suggesting it may need to be modified. The surface of the ACP is larger than that of the Watchman, possibly explaining the higher numbers of thrombi (20). Device-related factors may act in cooperation with individual risk factors in the initiation of thrombosis on the device.
Finally, our study emphasized the use of close TEE follow-up as all patients with thrombi were identified in the pre-discharge TEE or at 3-month follow-up, enabling initiation of adequate anticoagulation to avoid thrombus progression and resolve clots before apparent embolic events.
Our hemostaseological evaluation included platelet count, but no function test before ACP implantation. The study cohort is still small. Prospective studies are necessary to define the optimal risk-stratified antiplatelet/anticoagulation strategy.
Percutaneous LAA occlusion is a novel option for patients with AF and high bleeding risk under VKA. We identified several patients with thrombus formation on the ACP device and higher CHADS2 and CHA2DS2-VASc scores and platelet counts and notably reduced EF <50% as risk factors.
Our study emphasizes the need for close TEE follow-up. Post-interventional dual antiplatelet therapy does not protect from thrombus formation; therefore, VKA must be considered in the critical post-implantation phase until prospective studies have demonstrated that dual antiplatelet therapy is safe in patients with a low risk profile.
The authors thank Dr. Hilmar Kühl, MD, from the Institute for Diagnostic and Interventional Radiology of the University Hospital Essen, for the cardiac multislice computed tomography scan of 1 LAA thrombus patient (Fig. 4).
Dr. Plicht has reported that he has received speaker's honoraria and travel grants from Abbott Vascular. Dr. Kahlert has received fees as a clinical proctor for Edwards Lifesciences. Dr. Konorza has been a proctor for St. Jude Medical. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Dr. Thomas F. M. Konorza passed away in December 2012.
- Abbreviations and Acronyms
- Amplatzer Cardiac Plug
- atrial fibrillation
- Congestion, Hypertension, Age, Diabetes, and Stroke
- CHADS2 Plus Vascular Disease and Sex Category
- cytochrome P450
- ejection fraction
- Hypertension, Abnormal Liver or Renal Function, Stroke, Bleeding, Labile International Normalized Ratio, Elderly, Drug, or Alcohol Intake
- left atrial appendage
- transesophageal echocardiography
- vitamin K antagonists
- Received November 6, 2012.
- Revision received January 31, 2013.
- Accepted February 15, 2013.
- Camm A.J.,
- Kirchhof P.,
- Lip G.Y.,
- et al.,
- European Heart Rhythm Association, the European Association for Cardio-Thoracic Surgery
- Miles J.S.,
- Miletich J.P.,
- Goldhaber S.Z.,
- Hennekens C.H.,
- Ridker P.M.
- Voora D.,
- Ginsburg G.S.
- Sibbing D.,
- Koch W.,
- Gebhard D.,
- et al.
- Reddy V.Y.,
- Doshi S.K.,
- Sievert H.,
- et al.,
- PROTECT AF Investigators
- Cardona L.,
- Ana G.,
- Luísa B.,
- et al.
- Reddy V.Y.,
- Holmes D.,
- Doshi S.K.,
- Neuzil P.,
- Kar S.
- Braut A.,
- Neuzil P.,
- Schuler G.,
- et al.
- Krumsdorf U.,
- Ostermayer S.,
- Billinger K.,
- et al.