The present study is a single-center, prospective, observational, descriptional and non-randomized registry including consecutively 37 patients with all three types of non-valvular AF, a CHA₂DS₂-VASc score ≥ 2 and, therefore, indication for OAC. Patient enrollment started in June 2014. Further inclusion criteria were age ≥ 18 years, any relative or absolute contraindication for standard OAC - i.e., major bleeding with tendency to recidivity, HAS-BLED score ≥ 3 and neurological symptoms during treatment with OAC or intolerance to OAC. Written informed consent was obtained from all study patients. Patients were excluded if at least one of the following criteria were evident: single episode of AF or treatable reason for AF, catheter ablation of AF within 30 days prior to or after potential LAA occluder implantation, electrical cardioversion within 30 days after potential occluder implantation, congestive heart failure corresponding to functional class NYHA IV, myocardial infarction within the last 3 months, atrial septum defect or interventional/surgical occlusion of ASD, mechanical heart valve, status after heart transplant, symptomatic carotid artery stenosis, transient ischemic attack (TIA) or stroke within last 30 days, intracerebral bleeding within the last 3 months, acute infection, existing or planned pregnancy, existing thrombus. The study was carried out according to the principles of the declaration of Helsinki and was approved by the medical ethics committee II of the Faculty of Medicine Mannheim, University of Heidelberg, Germany.

An electrocardiogram (ECG), standard blood analyses and transesophageal echocardiography (TEE) were performed for pre-procedural planning. LAA occlusion device implantation was performed by experienced interventional cardiologists (≥50 LAA closure device implantations each prior to the study). Patients were treated with conscious sedation using intravenous 2,6-di(propan-2-yl)phenol and midazolam. One arterial access sheath (5 French, F) for arterial blood pressure monitoring and one venous access sheath for interventional device implantation were used. Transseptal puncture was performed with a SL1 sheath and BRK1 Brockenbrough needle (St. Jude Medical, St. Paul, MN). Prior transseptal puncture heparin was administered to achieve an active clotting time of at least 250 s. A stiff guide wire (Cook Medical, Bloomington, IN) was placed in the left upper pulmonary vein (LUPV) and the transseptal sheath was removed and replaced by the Watchman™ (14 F), respectively ACP 2 (12 or 14 F) delivery sheath. Device allocation algorithm was performed according to latest consensus recommendations [21]. All patients underwent pre-procedural 2D/3D TEE to assess initially the anatomy, size and absence of thrombus or sludge on all recommended TEE views. Accordingly, peri-procedural imaging was based on fluoroscopy and 2D/3D TEE imaging in order to re-assess prior findings of the LAA and to decide which device size was appropriate in the individual patient. Fluoroscopy was performed using a 5 F pigtail catheter to visualize the LAA in at least 2 standard angulations (RAO 30°/10° cranial, RAO 30°/10° caudal). Regarding TEE, the following parameters of the LAA were assessed for appropriate sizing, as recommended for both devices [21]: LAA ostium, landing zone, angle of the LAA, depth of the LAA, differentiation of main lobes versus smaller side lobes. The device size was chosen at least 20 % larger than the measured diameters at the landing zone [22]. Post-implantation the compression of the devices was measured by 2D TEE and a device compression of at least 10 % was defined as valuable, next to lacking of relevant peri-device leaks (>5 mm), and no compression or attachment of neighboring structures, such as the circumflex coronary artery, mitral annulus or pulmonary veins. A tug test was performed repetitively before final liberation of the device as recommended. Access site was closed with an Angio-Seal™ Evolution™ (St. Jude Medical, St. Paul, MN) for the arterial access and with two ProGlide™ (Abbott Vascular, Santa Clara, CA) for the venous access and followed by a pressure band for 6 hours.

In accordance with our institutional protocol, acetylsalicylic acid (ASA) 100 mg/d was administered lifelong, clopidogrel for at least 6 months commencing the day of implantation with a loading dose of 250 mg, respectively 600 mg, if not taken before. The day after the procedure, a transthoracic echocardiogram (TTE) and a chest X-ray were performed to rule out device dislodgement and pericardial effusion, an ECG was performed to rule out a new bundle branch block (BBB) or atrioventricular (AV) block and repetitive thorough clinical examinations were carried out to rule out other clinical disorders.

The primary efficacy outcome measure of our in-hospital register was defined as technical success with a successful device implantation without relevant peri-device leaks (i.e., < 5 mm). Primary safety outcome measure was defined as the occurrence of bleeding events classified according to the BARC definition [23], pericardial effusion, device embolization, peri-procedural stroke and peri-procedural death. Events resulting in death, aggravated morbidity or prolonged hospitalization being associated with the procedure were defined as complications. Any events not being associated with the procedure were termed adverse events.

Statistical analyses were performed with SPSS Statistics (IBM, Armonk, NY). Descriptive statistics are given as medians (25^th and 75^th percentiles) or as total numbers with group-related percentages. Normal distribution of data was tested with the Kolmogorov-Smirnov test. In case of normal distribution, the t-test was applied to compare scaled data. Scaled variables not normally distributed were compared using the Mann-Whitney U test. Categorical variables were compared using the chi-squared test. Level of significance was set at p < 0.05 (two-tailed).

Baseline demographic and clinical characteristics of the study population including indications for implantation of LAA occlusion devices and risk-stratification according to CHA₂DS₂-VASc and HAS-BLED scores are shown in Table 1. Of 37 patients, 22 patients received the Watchman™ and 15 patients received the ACP 2 device. The most common indication was a prior history of bleedings (78.4 %) under treatment with OAC. 51.4 % of all patients were not treated with OAC, whereas patients with OAC were mostly treated with DOAC (66.7 %). Table 2 shows two-dimensional (2D) and functional echocardiographic measurements. As assessed by angiography, the majority of patients revealed a chicken-wing shaped LAA (67.6 %).

Procedural data related to the implantation of LAA occlusion devices are summarized in Table 3. The primary efficacy outcome measure in the Watchman™ group was 90.9 %, whereas it was 93.3 % in the ACP 2 group (p = 0.791). Only the Watchman™ device device could not be implanted in two patients (9.1 %) either due to an anatomical mismatch or due to an incomplete occlusion of the LAA landing zone. In contrast, the ACP 2 device was implanted in all patients with one patient with incomplete ostial occlusion (6.7 %). More than one transseptal puncture was needed only in only one patient (2.7 %) due to a transseptal retrieval of an intra-atrial embolized Watchman™ occluder during the procedure. Pericardial effusion and subsequent intermittent circulatory failure with rapid hemodynamic stabilization without the need for cardiopulmonary resuscitation occurred in one patient of the Watchman™ group (2.7 %). All these results showed no statistically significant difference between the Watchman™ and the ACP 2 group.

Table 4 shows all relevant data related to in-hospital outcome and peri-procedural safety events. None of the patients died due to the implantation of LAA occlusion devices. However, one patient with a preexisting highly reduced left ventricular function (ejection fraction 28 %) died from an acute heart failure linked to a severe urosepis 20 days after the implantation (i.e., adverse event). In another patient with non-ST elevation myocardial infarction (NSTEMI) and percutaneous coronary intervention (PCI) with stent implantation known clopidogrel non-response, DAPT with ASA and ticagrelor instead of clopidogrel was continued as an individual treatment attempt.

Severe conduction blocks, such as BBB or AV blocks, did not occur. Access site complications, such as groin bruise and bleedings at the access sites represented 75.0 % of all complications. Most of them were mild, corresponding to a BARC type 1 bleeding (76.5 % of all bleeding complications) and only 1 case (2.7 % of all patients) needed transfusion without the need for vascular surgery. The patient with a pericardial effusion (3 mm) in the Watchman™ group could be treated conservatively. Notably, neither peri-procedural nor in-hospital transient ischemic attack, stroke or device thrombosis occurred. Patients could have been discharged with a median of 4 days after the intervention. None of the surviving patients developed persistent neurological or heart-failure related disability.

This study is based on observational registry data concerning a relative small real-life patient population and was not intended to reliable calculate significant differences between both device types. Comparison results of both devices are for informative reasons only. The aim was to demonstrate essential peri-interventional efficacy and safety data concerning LAA occlusion device implantation in general. A subsequent enlarged set of this registry data might be able to evaluate further differences between both devices.

As this observational study focused on the peri-procedural time, the follow-up period was limited to the discharge from hospital.

The application of any of the two devices was not randomly assessed and was based on the operator’s discretion being mainly based on LAA anatomic considerations. In contrast, this more individualized and clinically driven decision algorithm might have strongly influenced the high achievement of primary efficacy outcome measure, which realistically reflects clinical and interventional practice in experienced centers.

Written informed consent was obtained from all study patients. The study was carried out according to the principles of the declaration of Helsinki and was approved by the medical ethics committee II of the Faculty of Medicine Mannheim, University of Heidelberg, Germany.

Sharing of additional data is not necessary, because all relevant patient data are presented in this article.