Author + information
- Received November 25, 2013
- Revision received February 27, 2014
- Accepted March 13, 2014
- Published online July 1, 2014.
- Saibal Kar, MD∗,
- Dongming Hou, MD, PhD†,
- Russell Jones‡,
- Dennis Werner, BS†,
- Lynne Swanson, DVM†,
- Brian Tischler, BS†,
- Kenneth Stein, MD†,
- Barbara Huibregtse, DVM†,
- Elena Ladich, MD‡,
- Robert Kutys, MS‡ and
- Renu Virmani, MD‡∗ ()
- ∗Heart Institute, Cedars Sinai Medical Center, Los Angeles, California
- †Boston Scientific Corporation, Natick, Massachusetts
- ‡CVPath, Gaithersburg, Maryland
- ↵∗Reprint requests and correspondence:
Dr. Renu Virmani, CVPath, 19 Firstfield Road, Gaithersburg, Maryland 20878.
Objectives This study was designed for conducting a comparative evaluation of the healing response after Watchman (WM) (Boston Scientific, Plymouth, Minnesota) and Amplatzer Cardiac Plug (ACP) (St. Jude Medical, Minneapolis, Minnesota) in a canine left atrial appendage (LAA) model.
Background There is no direct comparison of the WM and ACP device in pre-clinical or clinical settings.
Methods The LAA from canine (n = 6) and human (n = 19) hearts were compared to determine the feasibility of the canine model and its relevance to clinical applications. Subsequently, implantation of WM and ACP in the canine LAA was performed (n = 3 per device) to evaluate the device conformation to the LA anatomy as well as the healing response at 28 days.
Results The LAA is a variable tubular structure in both canine and human hearts. Gross examination showed that the WM was properly seated inside the LAA ostium, in comparison to the ACP where the disk was outside of the LAA orifice and extended to the edge of the left superior pulmonary vein and mitral valve. At 28 days, complete neo-endocardial coverage of the WM was observed; however, the ACP showed an incomplete covering on the disk surface especially at the lower edge and end-screw hub regions.
Conclusions There are differences in conformation of LAA surrounding structures with variable healing response between WM and ACP after LAA closure in the canine model. WM does not obstruct or impact the LAA adjacent structures, resulting in a favorable surface recovery. In comparison, the disk of ACP could potentially jeopardize LAA neighboring structures and leads to delayed healing.
Percutaneous approaches to occlude the left atrial appendage (LAA) represent an emerging, device-based alternative to long-term pharmacologic therapy in the prevention of atrial fibrillation (AF)-associated stroke. Several transcatheter devices have been developed (1,2). Currently, the 2 most commonly implanted devices, are the Watchman (WM) (Boston Scientific, Plymouth, Minnesota), and the Amplatzer Cardiac Plug (ACP) (St. Jude Medical, Minneapolis, Minnesota).
Anatomically, the LAA has a complex geometric structure with an oval-shaped orifice. It is a gestational remnant that originates from primordial atrial tissue, whereas the rest of the left atrium, which is smooth, is formed from the absorption of the pulmonary veins and its branches (3). The atrial appendage overlaps the pulmonary trunk, bulging anteriorly, with its tip pointing cranially in most cases (4). It is in close proximity with several adjacent structures, including the lateral ridge between the LAA ostium and the left superior pulmonary vein (LSPV), mitral valve (MV) annulus, left main, left anterior descending and the left circumflex (LCX) coronary arteries, and the great cardiac vein. The LAA is a trabeculated structure that is described to be 3× larger in volume in patients with AF than in those without AF. It is believed to be the source of thrombosis in AF that may lead to systemic emboli and stroke (4,5). Therefore, we compared the LAA of human and canine anatomies to understand the relevance of the findings of WM and ACP when implanted in the LAA canine model.
Whereas both devices have shown reasonable safety and efficacy in clinical and pre-clinical settings (6–10), there has been no prospective study comparing the 2 devices. The current pre-clinical study was therefore designed to evaluate the healing response of the 2 devices in a 28-day canine model. In addition, we validated in vitro the canine model for pre-clinical testing.
Measurement of LAA and surrounding structures in canines and humans
Six healthy canine (25.2 ± 4.5 kg, 7 to 25 months of age in either sex) hearts and 19 human (101.4 ± 19.8 kg, 18 to 78 years of age in either sex) hearts from the CVPath autopsy registry were selected to compare the LAA dimension and the relationship of the LAA with the adjacent structures. Each heart was incised through the roof of the LA and opened along the posterior ventricular septum to expose the LAA and surrounding anatomic structures. Photographs of the external surface of the LAA and the opened LA were obtained. LAA is lined by pectinate muscle, whereas the LA is a smooth structure and the measurements are taken at this transition zone in the long- and short-axes of the LAA ostium (11). Measurements included: 1) external length of the LAA; 2) distance from the LAA ostium to the left main coronary artery bifurcation; 3) the closest distance from LAA ostium to the LCX; 4) LAA orifice dimension, in the long and short axes; 5) distance from LAA ostium to the LSPV; and 6) distance from LAA ostium to the MV annulus.
The WM LAA system consists of a parachute-shaped device with a self-expanding nitinol frame structure covered on the proximal half with a 160-μm permeable polyester fabric membrane (12). It has a row of fixation barbs at mid-perimeter to help secure the device in the LAA. Following successful deployment, the device acts as a filter in the initial phase allowing passage of blood but not thrombi. Subsequently over the next few months, the surface of the device is covered with neo-endocardial tissue that is impermeable to both blood and thrombi. The surface membrane can lead to a thrombus formation, thus the need for continuation of antithrombotic agents until the device is completely endothelialized. The device is available in 5 sizes ranging from 21 to 33 mm and is deployed using a 12-F delivery system. To ensure adequate closure and stability, the device that is chosen is 10% to 20% larger than the LAA ostium.
The ACP LAA system is a self-expanding device constructed from nitinol and consists of a lobe and a disk connected by a central waist. The lobe has fixation wires to ensure its stabilization, and the disk of the ACP device is placed in the outer part of the LAA ostium. The device is available in 8 different sizes based on the lobe diameter, that is, 16 to 30 mm, in stepwise 2-mm increments. The appropriate device chosen is usually 3 to 4 mm larger than the diameter of the proximal part of the LAA body (landing zone). The diameter of the proximal disk is greater than the distal lobe diameter by 4 to 6 mm and is connected by a disk end-screw hub with the device intended to cover the mouth of the ostium.
Ex vivo device implantation
Unfixed hearts from male mongrel canine (21 kg, 8 months of age) were obtained, and the right atrium and atrial septal wall were carefully trimmed to allow deployment of either the WM or ACP into the LAA under direct visualization. The smallest devices from WM (21 mm) and ACP (16-mm lobe with 20-mm disk) were selected and alternately placed in the canine hearts. Photographs of the device surfaces after each deployment were obtained. The device conformation to the LAA, and the relationship to the LSPV as well as the annulus of the MV were assessed.
In vivo device implantation
All protocols were approved by the Animal Care and Use Committee and complied with all animal use regulations as set forth in the Animal Welfare Act, 9, Code of Federal Regulations, and the American Association for the Accreditation of Laboratory Animal Care as outlined in the National Research Council's Guide for the Care and Use of Laboratory Animals. All devices were deployed by an operator experienced in the use of both the devices.
WM (n = 3) and ACP (n = 3) were implanted in 6 mongrel adult dogs (26.6 ± 1.3 kg, 8 to 15 months of age). Animals received a combination of warfarin 6 mg/day and aspirin 81 mg/day 1 day prior to the procedure and daily thereafter until termination. The international normalized ratio (INR) was monitored throughout the study and medications were adjusted to maintain the international normalized ratio between 2.0 and 3.0. The electrocardiogram, invasive blood pressure, end tidal CO2, functional O2 saturation, and body temperature were continuously monitored during the device implantation.
The devices were deployed under both transesophageal echocardiographic (TEE) and fluoroscopic guidance. Catheterization through both the femoral vein and arterial access was followed by heparinization (200 U/kg) to maintain an active clotting time of 250 to 300 s. Before device deployment, TEE was obtained to determine the LAA size (mm). Following transseptal puncture, the LAA was engaged using a pigtail catheter with the delivery sheath, and contrast was injected to define the LAA anatomy. LAA size was measured again via the right anterior oblique view. Both appropriately sized (per the manufacturers' instructions for use) WM and ACP devices were selected and implanted in the LAA, in alternating animals. Proper device positioning and stability was confirmed by TEE and fluoroscopy prior to releasing each device. Immediately post-deployment, TEE and coronary angiography were performed to evaluate device position, any leakage around the device, pericardial effusion, MV function, and LCX flow. These measurements were repeated on day 28 just prior to euthanasia.
The hearts were pressure perfusion fixed with 10% neutral buffered formalin, and imaged by Faxitron cabinet x-ray (Faxitron Bioptics, Tucson, Arizona) to assess the integrity of the implanted devices followed by gross photographs of the device surface from the LA while maintaining the relationship of the adjoining structures. The LAA device and the surrounding LA wall underwent dehydration in a graded series of ethanol, followed by infiltration, and were embedded in methylmethacrylate plastic. The device was cut along the long axis of the appendage, stained with toluidine blue and basic fuchsin, and analyzed under light microscopy. Each slide was scored for endocardial tissue growth, inflammation, granulation tissue, and fibrin/thrombus deposition using a 0 to 4 grading system. For details of methods used to grade each parameter, see the Online Appendix and Online Tables 1a and 1b.
All results are presented as mean ± SD unless otherwise indicated. Statistical analysis was performed by Sigma Stat (version 3.5, Systat Software, Inc., Germany). Measurement results from canine and human hearts were compared using the unpaired Student's t test. A p value of <0.05 was considered significant.
LAA and adjacent structures
Demographically, the canine body and heart weight ranges were 21.5 to 29.5 kg (25.2 ± 4.5 kg), and 158.9 to 254.4 g (203.4 ± 39.1 g), respectively. The human hearts included those of 11 men (6 African Americans and 5 white) and 8 women (3 African Americans, 5 white). The height and body weight ranges were 167 to 193 cm (182.4 ± 8.9 cm) and 66 to 143 kg (100.6 ± 23.8 kg) for men and 155 to 180 cm (170.2 ± 8.1 cm) and 82 to 123 kg (102 ± 14.3 kg) for women. The heart weights ranged from 350 to 877 g (538 ± 157 g) in men and 294 to 620 g (430 ± 111 g) in women. Cases were selected on the basis of absence of congenital heart disease, and the cause of death was as follows: 5 died from a noncardiac reason with normal hearts; 4 had conduction abnormalities and all were young; 4 had hypertensive heart disease; 2 had dilated cardiomyopathy; 1 had atherosclerotic heart disease; 1 had myocarditis; 1 had amyloidosis; and 1 died from thrombotic thrombocytopenic purpura. There was no significant difference in the body weight between men and women; however, height and heart weight was greater in men although both men and women had higher heart weight than seen in normal individuals (normal heart weights for men are 300 to 350 g and for women are 250 to 300 g) (13).
In general, the LAA in both the canine and the human hearts is an angled, elongated tubular structure overlying the base of the left ventricle and abutting the pulmonary artery root, and it is variable in size and shape. From the anterior view of the heart, the left main coronary and proximal left anterior descending and LCX arteries are covered by the appendage. From the LA chamber view, the ostium of the LAA is typically an oval-shaped orifice. Importantly, the LAA ostium has close anatomic relationships to LSPV, annulus of the posterior MV leaflet, and left coronary artery branches, especially the LCX, which courses in the atrioventricular sulcus in close proximity to the ostium of the LAA. Figures 1 and 2 show the LAA and its adjacent structures from both the external and internal views of the heart.
The dimensions of LAA ostium in both the long and short axes (12.7 ± 1.4 mm × 8.2 ± 2.5 mm) in the dogs are comparable to those of humans (15.3 ± 3.6 mm long axis × 8.0 ± 1.8 mm short axis) (Table 1). The closest distances from LAA ostium to both LCX and LSPV were comparable between canines and humans (p > 0.05). However, the external length of the appendage in the canines (31.7 ± 3.0 mm) is significantly shorter than that of humans (39.6 ± 6.8 mm; p = 0.014). The distance from LAA ostium to MV annulus was also less in canines (7.3 ± 2.1 mm) than in humans (10.0 ± 1.8 mm; p = 0.004). There were no significant differences among the previously mentioned parameters between men and women. When measuring the distance from the LSPV directly to the MV annulus (LSPV-to-MV) using the shortest diameter of the LAA orifice, the mean distance was 24.2 ± 3.5 mm and 27.5 ± 3.8 mm for the dogs and humans, respectively, and there was no significant difference (p = 0.088) between them even when the human hearts were enlarged.
Ex vivo study
The WM device (21 mm) conformed appropriately to the LAA orifice of the canine heart following deployment. The measured device surface area was 154 mm2. In comparison, the disk of ACP device was located outside of the LAA orifice, and both the upper and lower edges of the disk appeared in close proximity to the LSPV and the MV annulus than the WM device did. For the 16-mm ACP device, the disk is 20 mm in diameter, and its surface area is 314 mm2, which resulted in 51% greater area than that of the WM device.
In vivo study
Eight animals were enrolled in this study. Two animals were excluded due to complications of the transseptal procedure. Devices were successfully implanted in the remaining 6 dogs (n = 3 per device type) without complications until termination. The maximal orifice diameter of the LAA device to the LAA ostium ratio and percentage of device compression was similar between the 2 groups. The LAA ostium by angiography before implantation was 20.1 ± 1.0 mm for the WM device and 21.0 ± 3.0 mm for the ACP device. The device oversize was 14.8 ± 0.1% for WM and 15.0 ± 0.1% for ACP. All devices were appropriately deployed, and the representative images are shown in Figure 3. A small (<3 mm) peridevice flow disturbance post-implantation was detected by TEE in 2 WM animals and 1 of the ACP implanted animals immediately following device deployment and diminished at follow-up (Fig. 4). One animal with an ACP device had a moderate lateral mitral regurgitation caused by the inferior edge of the disk extending onto the MV (5 mm beyond the MV annulus by TEE measurement). Mitral regurgitation was also observed at 28 days and confirmed during gross examination (Fig. 5). Coronary angiography showed LCX hinge motion (Online Appendix) in this animal, however, the other LCX arteries were patent with TIMI (Thrombolysis In Myocardial Infarction) flow grade 3.
The hearts were within normal limits grossly in all animals. Faxitron radiography showed that devices maintained their integrity and expanded symmetrically without visible fractures. No instances of perforation, laceration, or erosions were observed. Grossly, the device conformed to the LAA in vivo and was similar to the ex vivo implantation (Fig. 6), with the disk of the ACP device extending over the MV annulus, as well as the lateral ridge between the LAA ostium and the LSPV. The mean length from the inferior edge of WM to the MV annulus measured 9.3 ± 1.2 mm. Histologically, WM showed tight apposition to the surrounding atrial walls and sealed the appendage lumen. The surface of WM and the central screw hub were completely covered by neo-endocardial tissue, which was incorporated with focal organized mature connective tissue with intermittent areas of granulation tissue and organizing fibrin thrombus. In comparison, the lower edges of the ACP disk extended beyond the MV annulus and its upper edge impinged the lateral ridge between the LSPV and the LAA ostium. The disk appeared to be in loose contact with the LA wall. Only a small portion of the disk surface was covered by neo-endocardial tissue, without significant coverage of the inferior disk edge and end-screw hub (Fig. 7). Histopathologically, WM scored a greater extent of healing with only focal areas of granulation tissue and less fibrin deposition and inflammation as compared to the ACP device (Table 2).
Worldwide, WM and ACP are the 2 most commonly used devices for transcatheter closure of the LAA for the prevention of thromboembolism from the LAA in patients with nonvalvular AF. Whereas the separately published clinical results appear similar for both devices, there have been no formal clinical or pre-clinical studies comparing these devices (6–8,14,15). To the best of our knowledge, this is the first translational study to evaluate the influence of the anatomical structures located adjacent to the LAA ostium and the histologic responses following deployment of WM and ACP in a canine LAA model. In addition to the observed considerable similarity of the LAA and the relation to the neighboring structures in both the canine and human LAA, we also demonstrated that the device conformation to LAA, as well as the healing of the surface of the device, was different. The most notable anatomical difference between the 2 species was shorter distance of the LAA ostium to the MV annulus in canines versus humans. The observation from the current study, therefore, confirmed that the canine model is suitable for pre-clinical evaluation of LAA occlusion devices.
LAA is a complex and variable structure in humans, which is located between the left upper pulmonary vein and the MV annulus. LAA lies in the atrioventricular groove, overlying the LCX artery and the great cardiac vein. There are several adjacent structures such as LSPV, MV, and LCX around the LAA. The left lateral ridge is a boundary between the LAA orifices and the LSPV. Su et al. (16) reported on the anatomy of the LAA and the relationship to the adjoining structures. The length from LSPV-to-MV was 30.7 ± 10.7 mm with a range of 15.7 to 57.6 mm. This range is wider than that measured in the current study. As no demographic data such as age, sex, height, and weight was reported by Su et al. (16), it is difficult to compare the 2 studies. Nevertheless, the shortest length from LSPV-to-MV in the report by Su et al. (16) was 15.7 mm when crossing short axis of the LAA ostium. This distance is shorter than the available ACP disk diameters (20 to 36 mm based on manufacturer's instructions for use). Because the disk of ACP is positioned outside of the LAA following implantation, caution should be exercised when the length of LSPV-to-MV is short. Before releasing this device, the disk should appear as concave-shaped to avoid the consequences of disk edges extending onto the adjacent structures.
Our data also demonstrates implantation of a proper size of WM resulted in better conformation to the LAA anatomy without affecting the adjacent LAA structures. Although a mild peridevice flow post-implantation was detected in 2 animals immediately after WM, the flow was diminished at follow-up. Consistent with this pre-clinical study, the PROTECT AF (Percutaneous Closure of the Left Atrial Appendage Versus Warfarin Therapy for Prevention of Strokes in Patients With Atrial Fibrillation) clinical substudy (17) also reported residual peridevice events, which occurred in 32% of implanted patients. It should be noted, however, that this residual peridevice flow (jet ≤5 mm width) has no clinical consequence, including no association with any increased risk of thromboembolism in the clinical setting (17). The incidence of thrombosis was low with or without flow, 3% and 4% in PROTECT AF, whereas in a recent small study (n = 34) with ACP implants, TEE at 3 and 6 months identified thrombosis in 17.6% patients despite dual antiplatelet therapy (18). Although another single center study reported 1 case of pericardial effusion and no cases of thrombosis or stroke out of 100 ACP implants in patients with a mean age of 73 ± 10 years (8). In comparison to the WM, the animals that received ACP devices showed that the disk interfered with the LAA adjacent structures in our pre-clinical model. First, the upper edge of the disk was found to encroach on the left lateral ridge between the LSPV and LAA ostium. Second, our data showed that the lower edge of the disk extended over the MV, which led to peridevice flow as well as LCX hinge motion in 1 animal. Lastly, the minimal distance from LSPV-to-MV was measured in both canines (24.2 ± 3.5 mm) and humans (27.5 ± 3.8 mm), it was shorter than disk diameter at least for >24 mm ACP devices based on its instructions for use (12). These findings may have significant clinical consequences especially when using bigger size ACP devices. It is important to pay attention to the disk interference with the LSPV flow or the MV during the implantation. More recently, Swaans et al. (19) reported on the feasibility and safety of performing radiofrequency catheter ablation in patients with AF followed by WM implantation. The combination of ablation and LAA closure in a single procedure may reduce the risk of repeated transseptal puncture. Because the disk of the ACP is outside of the LAA ostium with a close proximity to LSPV, the ACP device may be difficult to use in AF procedure.
A previously published pre-clinical canine study demonstrated complete endothelial cells coverage of the device with sealing of the device and LA interface at 45 days following WM placement (9). Consistent with this report, we observed that the surfaces of all WM devices were completely incorporated with organizing neo-endocardial growth consisting of well-organized fibrointimal and granulation tissue with only minimal fibrin deposition at 28 days. In comparison, a report of the ACP device (sizes 18 to 22 mm) when implanted in the canine animal model and examined grossly and histologically at 30 and 90 days had also shown that it too was covered by “mature neointima” with focal calcification and complete occlusion by angiography and color flow Doppler (10). However, 1 animal in the 30-day implant group had a mural thrombus and a cerebral infarct. The present study confirmed focal uncovered areas at the lower edge of the disk and end-screw hub at 28 days for the ACP cases. Because the mechanism of closure with the ACP depends on sealing of the disk to prevent flow into the LAA, poor apposition at the inferior edge of the LA wall may be an indicator of prolonged healing response. Delayed device surface healing could potentially be related to thrombus formation that accompanies any device implantation (20).
First, although canine LAA anatomy has many similarities to the human LAA in several parameters such as the angulation and morphology, it does not precisely represent the human nonvalvular AF status. In addition, the healthy canine model will have relatively faster healing properties than human patients would (9). Therefore, it is uncertain if the anatomy influences healing and if the model is predictive of the human diseased heart. Second, both canine and human hearts were examined following formalin fixation, resulting in tissue shrinkage, which may influence gross morphometric analysis. Lastly, this pilot study was limited by a small sample size, and statistical analysis was not performed for the in vivo study.
When designing medical devices, one should consider not only the anatomy of the target organ, but also the surrounding structures. Although differences exist, the canine LAA model closely mimics the human LAA and is the standard pre-clinical model for evaluation of LAA closure devices. Differences in conformation of the LAA surrounding structures and healing response were noted following closure of the canine LAA between WM and APC. In this small pre-clinical study, the WM device does not obstruct or affect the LAA adjacent structures, resulting in a favorable surface recovery. In comparison, the oversized ACP could potentially jeopardize the LAA neighboring structures and lead to delayed healing. These data are hypothesis generating and the clinical relevance of these differences should be tested in larger trials with longer follow-up.
The authors express gratitude to Jason Kilvington for his help performing the transesophageal echocardiography. They also acknowledge Eric Burright, PhD, and Nicole Gordon, PhD, for their excellent support.
For supplemental tables, and accompanying videos, please see the online version of this paper.
This study was funded by Boston Scientific. Dr. Kar has received research grants from Boston Scientific and St. Jude Medicalhttp://dx.doi.org/10.13039/100006279; serves on the left atrial appendage advisory committee for Boston Scientific; and also serves as the national principal investigator of the CAP2 (Continuous Access Registry) registry and as an investigator for clinical trials for both devices. Drs. Hou, Swanson, Tischler, Stein, Huibregtse, and Mr. Werner are full-time employees and shareholders of Boston Scientific. Dr. Virmani has received research support from 480 Biomedical, Abbott Vascular, Atrium, Biosensors International, Biotronik, Boston Scientific, CeloNova, Cordis Corp. (Johnson & Johnson), GlaxoSmithKlinehttp://dx.doi.org/10.13039/100004330, Medtronichttp://dx.doi.org/10.13039/100004374, MicroPort Medical, OrbusNeich, ReCor Medical, SINO Medical Technology, Terumo Corporation, and W. L. Gore; and has received consulting fees and/or honoraria from 480 Biomedical, Abbott Vascular, Biosensors International, Boston Scientific, CeloNova, Cordis Corp. (Johnson & Johnson), Lutonix, Medtronic, Terumo Corporation, Merck, and W. L. Gore. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Kar and Hou contributed equally to this work.
- Abbreviations and Acronyms
- Amplatzer Cardiac Plug
- atrial fibrillation
- left atrium
- left atrial appendage
- left circumflex artery
- left superior pulmonary vein
- mitral valve
- transesophageal echocardiography
- Thrombolysis In Myocardial Infarction
- Watchman left atrial appendage system
- Received November 25, 2013.
- Revision received February 27, 2014.
- Accepted March 13, 2014.
- American College of Cardiology Foundation
- Holmes D.R. Jr..,
- Lakkireddy D.R.,
- Whitlock R.P.,
- Waksman R.,
- Mack M.J.
- Ho S.Y.,
- Cabrera J.A.,
- Sanchez-Quintana D.
- Reddy V.Y.
- Reddy V.Y.,
- Doshi S.K.,
- Sievert H.,
- et al.,
- for the PROTECT AF Investigators
- Schwartz R.S.,
- Holmes D.R.,
- Van Tassel R.A.,
- et al.
- Sharma S.,
- Devine W.,
- Anderson R.H.,
- Zuberbuhler J.R.
- Kumar V.,
- Abbas A.K.,
- Fausto N.,
- et al.
- Reddy V.Y.,
- Möbius-Winkler S.,
- Miller M.A.,
- et al.
- Su P.,
- McCarthy K.P.,
- Ho S.Y.
- Viles-Gonzalez J.F.,
- Kar S.,
- Douglas P.,
- et al.
- Plicht B.,
- Konorza T.F.,
- Kahlert P.,
- et al.
- Swaans M.J.,
- Post M.C.,
- Rensing B.J.,
- Boersma L.V.
- Lammers J.,
- Elenbaas T.,
- Meijer A.