Author + information
- Received May 4, 2009
- Accepted May 20, 2009
- Published online August 1, 2009.
- Panagiotis G. Kyrtatos, BMedSci⁎,†,§,
- Pauliina Lehtolainen, PhD⁎,
- Manfred Junemann-Ramirez, MD‡,
- Ana Garcia-Prieto, PhD∥,
- Anthony N. Price, PhD⁎,
- John F. Martin, MD‡,
- David G. Gadian, DPhil†,
- Quentin A. Pankhurst, PhD§∥ and
- Mark F. Lythgoe, PhD⁎,†,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Mark Lythgoe, Centre for Advanced Biomedical Imaging, University College London, 72 Huntley Street, WC1E 6DD London, United Kingdom
Objectives We sought to magnetically tag endothelial progenitor cells (EPCs) with a clinical agent and target them to a site of arterial injury using a magnetic device positioned outside the body.
Background Circulating EPCs are involved in physiological processes such as vascular re-endothelialization and post-ischemic neovascularization. However, the success of cell therapies depends on the ability to deliver the cells to the site of injury.
Methods Human EPCs were labeled with iron oxide superparamagnetic nanoparticles. Cell viability and differentiation were tested using flow cytometry. Following finite element modeling computer simulations and flow testing in vitro, angioplasty was performed on rat common carotid arteries to denude the endothelium and EPCs were administered with and without the presence of an external magnetic device for 12 min.
Results Computer simulations indicated successful external magnetic cell targeting from a vessel with flow rate similar to a rat common carotid artery; correspondingly there was a 6-fold increase in cell capture in an in vitro flow system. Targeting enhanced cell retention at the site of injury by 5-fold at 24 h after implantation in vivo.
Conclusions Using an externally applied magnetic device, we have been able to enhance EPC localization at a site of common carotid artery injury. This technology could be more widely adapted to localize cells in other organs and may provide a useful tool for the systemic injection of cell therapies.
One of the current challenges in the biomedical sciences is the localization of stem cells to sites of interest for the repair of tissue damage. Cellular therapies are increasingly applied in clinical trials, and in recent years the use of hematopoietic progenitors as repairing modules in the compromised cardiovascular system has been the focus of considerable attention (1–3). In the context of ischemic heart disease, these efforts have led to modest success (4,5). However, delivery to specific targets within the body generally remains a difficult task, in part due to low uptake at the site of injury. As such, cell therapies would greatly benefit from methodologies aimed at targeting and monitoring cell trafficking.
Superparamagnetic iron oxide nanoparticles (SPIO) offer attractive possibilities in biomedicine as they can be incorporated into cells affording a controllable means of “tagging” (6). These particles lead to a marked decrease in the magnetic resonance imaging (MRI) parameter T2* and the possibility of visualizing their localization noninvasively on T2*-weighted MRI (7–9). Furthermore, the magnetic properties of SPIOs allow them to be manipulated mechanically by a magnetic field gradient. This “action at a distance,” combined with the intrinsic penetrability of magnetic fields into human tissue, opens up potential applications involving the transport of magnetically tagged biological entities. Such tagging has to date mainly been the focus of drug delivery systems (10–12), including drug targeting in humans (13). Recent efforts on cell targeting in the arterial circulation have been limited to animal models; to achieve cell capture, these have necessitated the use of large nonbiodegradable micron-sized beads (14–16) or nanoparticle composites fabricated in-house (17), and in most cases, the additional introduction of permanent intravascular metallic devices not currently approved for human use (15–17). It may also be possible to attract gene-carrying cells to post-capillary venules of tumors (18). However, the targeted delivery of progenitor cells using external magnetic devices and clinically approved iron-bearing agents has not yet been accomplished.
Early endothelial progenitor cells (EPCs) derived from CD34+/CD133+ cells (19,20) have been shown to be involved in post-ischemic neovascularization (21) and re-endothelialization, reducing neointima formation following arterial injury (22). We chose to target EPCs to a site of rat common carotid artery (CCA) injury, as localization of such progenitors to sites of vascular catheterization may help prevent post-angioplasty restenosis (23,24). Although various SPIO formulations have been used in the past, their future use as clinical agents is not straightforward due to safety issues raised by cell labeling and the use of transfection agents. To obviate these concerns, we have chosen to label human EPCs with the U.S. Food and Drug Administration (FDA)-approved nano-sized SPIO compound Endorem (Guerbet, Paris, France), the only agent to date that has been used to monitor cells in humans using MRI (8).
Human mononuclear cells (MNCs) were collected by leukapheresis from peripheral blood of donors stimulated with granulocyte colony-stimulating factor (25). We magnetically isolated the CD133+ progenitor cells using antihuman CD133 (epitope 1) microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany). These beads are biodegradable and typically disintegrate after a few days in culture. Then, EPCs were derived by culturing CD133+ cells for up to 21 days on fibronectin-coated plates (BD, Franklin Lakes, New Jersey) in growth medium consisting of 20% fetal bovine serum/endothelial basal medium 2 (Lonza, Basel, Switzerland) and supplemented with fibroblast, insulin-like, and endothelial growth factors; ascorbic acid; heparin; and recombinant human vascular endothelial growth factor 165 (25 ng/ml) (R&D Systems, Minneapolis, Minnesota) (26). Growth media were supplemented with recombinant human stem cell factor (100 ng/ml), rhFlt-3/Flk-2 ligand (50 ng/ml), and recombinant human interleukin 3 (10 ng/ml) (R&D Systems) for the first 72 h. MNCs were grown for 2 h in serum-free Dulbecco's modified eagle medium (Gibco, Invitrogen, Carlsbad, California) followed by removal of nonadherent cells and culture of the adherent fraction in 10% fetal bovine serum/Dulbecco's modified eagle medium. All blood samples were obtained with written consent and were approved by local ethics committees.
Labeling of cells with SPIO
At day 9 of culturing, the CD133+ cells were labeled with the FDA-approved SPIO Endorem (also known as Feridex in the U.S.) (Guerbet) using 0.5 mg/ml SPIO in growth medium. Suspension CD133+ cells were labeled for 24 h and adherent cells for 1 h. Additionally, adherent cells were labeled for 2 or 24 h to increase the amount of iron in the cells. Mononuclear cells were labeled with SPIO by overnight incubation. Labeled cells were washed twice in 15 ml phosphate-buffered saline.
Quantification of cellular SPIO uptake
We used a superconducting quantum interference device (27) to measure the amount of Fe3O4 in the cells. To separate SPIO signal from background signal coming from the cells and media, measurements were performed at –263°C. The samples were saturated in an applied field of 2-T, which was subsequently removed to leave the SPIO particles in a magnetized state. Comparison of this remnant signal with that of an undiluted SPIO sample of known Fe3O4 concentration allowed quantification of Fe3O4 per cell.
Neodymium iron boron permanent disk magnets (1.195-T, grade N35SH, Magnet Applications Ltd., Hertfordshire, United Kingdom) 22 mm in diameter and 16 mm in height were used to test cell viability and differentiation following magnetic attraction. The SPIO-labeled EPCs were placed 1 mm from the magnet surface (maximum: 19.8 T2/m). For the flow system and in vivo magnetic targeting, a magnetic actuator was constructed using 5 10 × 10 × 25-mm3 NdFeB grade N35 blocks, attached together inside a machined aluminum casing in a configuration of rotating magnetization (Halbach array), which concentrates the force on 1 side of the actuator (28).
The CellTiter 96 AQueous One Solution Assay (Promega, Madison, Wisconsin) was performed after 24 h of magnetic attraction of labeled CD133+ cells. The tetrazolium compound in this assay undergoes reduction to a colored formazan product by viable cells, primarily mediated by reduced nicotinamide adenine dinucleotide phosphate (29). Colorimetric quantification of viability of samples of 3 × 104 CD133+ cells was performed using a spectrophotometric microplate reader (GENios, Tecan Group Ltd., Männedorf, Switzerland) (490 nm, at 4 h). Sample activity was normalized to the mean absorbance of control (no magnet). Cell apoptosis was analyzed using an annexin-V-fluorescein isothiocyanate/propidium iodide flow cytometric assay (ApoTarget, Invitrogen).
Immunophenotypic characteristics of CD133+ cells labeled with SPIO at day 10 were analyzed with a fluorescence-activated cell sorter (FACSCalibur, BD) at day 21 of cultivation. Cells pre-treated with crystallizable fragment receptor blocking reagent (Miltenyi Biotec) were stained using the following antihuman antibodies: phycoerythrin (PE)-CD304 BDCA-4/neuropilin-1 (30), biotin-CD144/vascular endothelial cadherin (Bender MedSystems Inc., Burlingame, California) with streptavidin- fluorescein isothiocyanate for endothelial differentiation; PE-CD14 and CD11b (BD Biosciences, San Jose, California) for monocytic differentiation. Fluorochrome- and dose-matched isotypes were used as controls. Cells were suspended in 20 mmol/l N-2-hydroxyethylpiperazine-N-2-ethanesulfonic acid/0.5% bovine serum albumin/phosphate-buffered saline containing 10 μg/ml 7-aminoactinomycin D (Sigma Aldrich, St. Louis, Missouri). The 7-aminoactinomycin D–positive, nonviable cells were excluded and 1 × 104 viable cells were scored per analysis (CellQuestPro, BD).
The SPIO-labeled CD133+ cells were embedded in low melting point agarose (0.5%, 37°C) at concentrations of 745, 165, 49.5, 16.5, and 0 × 103 cells/ml in 250-μl Eppendorf tubes (Eppendorf North America, Westbury, New York) and were cooled to 4°C. Suspension in agarose allowed for homogeneous distribution of the cells. The tubes were imaged in a 9.4-T horizontal bore Varian MRI system (VNMRS, Varian Inc., Palo Alto, California) using a 39-mm radiofrequency coil (RAPID Biomedical GmbH, Rimpar, Germany). The T2*-weighted images were acquired using a spoiled gradient echo sequence (echo time [TE] = 25 ms, repetition time [TR] = 200 ms, flip angle [FA] = 30, 5122 matrix). The T2* maps were obtained from 20 echo times (TE = 2.5 to 30 ms, TR = 200 ms, FA = 30, 2562 matrix) and produced using the ImageJ MRI Analysis Calculator plug-in (K. Schmidt, ImageJ, U.S. National Institutes of Health, Bethesda, Maryland).
Computer simulations and in vitro flow system
Magnet designs for acquiring optimal magnetic field strength to capture intravascular SPIO-labeled cells were modeled using Opera-2d and -3d (Cobham Technical Services, Aurora, Illinois) and COMSOL Multiphysics 3.1 (COMSOL Inc., Burlington, Massachusetts) software. Fluid flow rate, cell size and Fe3O4 content, and distance of magnet from the vessel were considered by solving for the Khan and Richardson hydrodynamic drag force and the attractive force from the magnet, in parabolic nonpulsatile laminar flow. Simulated cells were 10 μm in diameter, 3.9 pg Fe3O4/cell, within a circulating blood volume of 20 ml. Cell capture was defined as approximation and contact with the vessel wall within the 50-mm length of the array. The Halbach magnet design described earlier was assessed in an in vitro flow system by placing 1 × 106/ml SPIO-labeled blood MNCs (3.5 pg Fe3O4/cell) in peristaltic pump-driven flow (10 ml/min, mean velocity: 0.4 m/s) using biocompatible (ISO 10993) Tygon S-54 elastic microtubing (Saint-Gobain Performance Plastics, Courbevoie, France). The magnet array was placed at distances of 1 and 5 mm from the tubing. The mean of 5 hemocytometer cell counts was calculated. Digital images were acquired using a standard dissection microscope and a digital camera.
Vascular injury and in vivo magnetic targeting
Male Sprague-Dawley rats (n = 13, 380 to 420 g) (Charles River Laboratories International, Wilmington, Massachusetts) were anaesthetized using midazolam (625 μg/100 g) and fentanyl (40 μg/100 g) supported by halothane at 0.5% to 2% at 2 l/min oxygen. The left CCA was exposed at the bifurcation to the external and internal carotid artery and a 2-F embolectomy catheter (Edwards Lifesciences Corp., Irvine, California) was advanced proximally through an arteriotomy into the CCA. The catheter was inflated with 100 μl of air and rotated as it was being retracted (31). The SPIO-labeled CD133+ cells were labeled with 5-μmol/l CellTracker Green 5-chloromethylfluorescein diacetate (Invitrogen-Molecular Probes). This dye is activated by intracellular esterases and only stains viable cells. We administered 5 × 105 cells in the CCA, which was temporarily clamped for 10 min with the magnet array placed at a 5-mm distance. The circulation was restored and the magnet removed after a further 2 min of magnetic actuation (n = 5). Two types of controls were used: 1) CCA injury was performed and labeled cells were injected without magnetic actuation (n = 6); and 2) magnetic attraction of labeled cells was performed on uninjured vessels (n = 2). Tissues were retrieved 24 h later and fixed with 1% p-fluoro-phenylalanine/phosphate-buffered saline. All experimental procedures complied with the institutional guidelines for animal experiments.
Confocal microscopy and quantitation of cell engraftment
Arteries were opened longitudinally and counterstained with 4′,6-diamidino-2-phenylindole (Vector Labs, Burlingame, California). Total arterial surface confocal microscope images were acquired using a Leica TCS SP2 confocal microscope. Reflectance scanning (32) was used to detect intracellular SPIO. Image processing was performed using the Volocity imaging software (Improvision, Waltham, Massachusetts). Background correction was performed using the “rolling ball” algorithm implemented in ImageJ software. Signal outliers were removed by application of a despeckle median filter (W.S. Rasband, ImageJ, U.S. National Institutes of Health). To analyze cell engraftment, x-y-z scans were rendered into a 3-dimensional vector graphic format; this enabled quantification the total volume of CellTracker-labeled cells per artery. Cell counts were calculated from the total volume based on a 10-μm cell diameter and are presented per square millimeter of arterial surface area.
We used SPSS version 11.5.1 (SPSS Inc., Chicago, Illinois) and Prism version 4.02 (Graphpad Software Inc., San Diego, California). Data were inspected visually and with the Shapiro-Wilk test for small samples to assess normality. The homogeneity of variance assumption was evaluated with the Levene test. Mean values are quoted with the standard error of the mean, medians are quoted with the interquartile range (IQR). The p values reported are 2-tailed and were considered statistically significant at the 5% level (p < 0.05). In the apoptosis assays, equivalence testing was considered appropriate because the null hypothesis was nondifference; 2 1-sided t tests were used with a 3% equivalence limit at the 0.05 level (Schuirmann test). A maximum acceptable difference of 3% was set a priori based on 24-h serum-deprived positive control cells (3% mean increase in early apoptosis, 16% increase in late apoptosis). Other statistical tests are specified in context.
Cell labeling and biological assays
To demonstrate that the magnetic force following Endorem labeling was large enough to attract cells to the vessel wall through a liquid, we recorded the motion of labeled MNCs in real time as a magnetic force was applied (Fig. 1). Rapid movement toward the magnet was observed as the cells responded to the applied force.
The CD133+-derived EPC viability was normal following labeling and magnetic attraction (Fig. 2A) and no increase in apoptosis was observed (Fig. 2B) when the amount of SPIO in the cells was controlled. Increased labeling resulted in a decrease in viability following magnetic attraction (Fig. 2A). This may be explained by force-induced mechanical damage of intracellular structures containing the iron nanoparticles (33). As shown in Figure 2C, there was normal expression of endothelial (vascular endothelial cadherin, CD304/NP-1) and monocytic (CD14, CD11b) markers of differentiation, 10 days after labeling. Using in vitro MRI, cells were visible on T2*-weighted images at dilutions as low as 25 cells/mm2 and caused a marked reduction in T2* values from 8 cells/mm2 (Fig. 2D). Using a superconducting quantum interference device, we calculated that labeled cells contained on average 3.9 pg Fe3O4 per cell. Given the magnetic field applied in these experiments and the iron content of the cells, it is possible to estimate a “cell-safe” magnetic force for EPCs above which there is a risk of compromising viability; this was 9.3 pN per cell.
External magnet design
An important consideration in our experimental model is that the rat CCA is a nonsuperficial structure approximately 5-mm from the skin in which there are large hydrodynamic forces acting on the cells, with peak systolic blood flow rate up to 10 ml/min (34,35). However, there is a rapid decline of magnetic force at increased target tissue distances, which poses a challenge to magnetic targeting using external devices in vivo (6,11). To overcome this issue, we chose a permanent magnetic actuator design that increases the force magnitude at large distances (Halbach array). With this actuator placed at a 5-mm vertical distance from the target tissue, a computed maximum 8.7 pN force was applied on each cell, which is below the 9.3 pN limit.
We proceeded by modeling the distribution of magnetic forces at distances of 1 and 5 mm from the magnetic actuator (Fig. 3A) and simulated the hydrodynamic and magnetic forces on intraluminal CD133+ cells using finite element modeling. With the vessel placed 1 mm away from the actuator, 11% of flowing cells made contact with the endothelial surface at each pass through the vessel. With the actuator at 5 mm, 2.3% of cells made contact. If the cells adhered at the contact site, approximately 50% of cells would be extracted from the circulation within 15 min at 1 mm, or 60 min at 5 mm.
In vitro flow system
We investigated the targeting ability of the actuator in vitro at the 2 distances using a 10-ml/min flow system. At 1 mm, cells accumulated above the points of predicted maximum force within 1 min after initiation of flow (Fig. 3B). These cell aggregates dissociated a few seconds after removal of the actuator. Targeting for 15 min at 1 mm resulted in 252-fold increase in capture (mean = 41 × 103 ± 6 × 103 cells) versus control (mean = 163 ± 41 cells) specimens. At 5 mm, there was a 6-fold increase in capture (mean = 1,630 ± 276 cells) versus nontargeted control (mean = 281 ± 21 cells) specimens after 1 h of flow (Fig. 3C).
External targeting in a small animal model
To investigate the in vivo feasibility of magnetic targeting using external magnets, we attempted to increase SPIO-CD133 cell engraftment to a site of rat CCA injury by placing the actuator adjacent to the ventral aspect of the animal's neck. Images in Figures 4A and 4B display representative (median) confocal scans of the arterial surface 24 h after cell delivery; there is a notable increase in cell retention following magnetic targeting. High power magnification (Fig. 4D) confirmed the presence of nucleated cells (using 4′,6-diamidino-2-phenylindole) and the presence of intracellular clusters of SPIO nanoparticles in CD133+ cells was confirmed using reflectance confocal microscopy (Fig. 4E). Application of an external magnetic force using the actuator during and after cell delivery increased CD133+ cell engraftment to the injured vascular surface by a factor of 5.4 (median = 134, IQR: 47 to 362 cells/mm2) compared with cell delivery without the magnet (median = 25, IQR: 17 to 42 cells/mm2). In the absence of vascular injury (control studies), no cells engrafted to the endothelial surface when the magnetic force was applied (cell count = 0).
The concept of magnetic tagging and targeting could play an important role for future advances in delivery and noninvasive monitoring of cell-based therapeutic interventions. This is the first attempt to use a clinically approved agent to externally target EPCs in a high-flow system such as the arterial vasculature. We chose to use EPCs for this study as these may hold high therapeutic value for the future. Endothelial damage is an initiating event in many cardiovascular disorders. Transplantation of these early progenitor cells, which resemble hemangioblasts in their properties of marker expression, proliferation, migration, and differentiation, may expand the therapeutic arsenal to repair vessels or endothelium and improve perfusion of ischemic tissues by providing new progenitor cells capable of responding to angiogenic stimuli, and have recently been used for cardiac neovascularization in clinical trials (36–40). In the majority of these and other studies on cardiac regeneration, cells were administered via intracoronary injection with “flow-stop” periods of several minutes to increase engraftment; yet it has been shown that despite this maneuver, only 1% to 3% of the cells remain on site (1,41), and there is a pressing need for targeting strategies (42).
We have based our cell targeting technique on a biodegradable iron oxide nanoparticle preparation, which should facilitate transition to human studies, in contrast to other nonbiodegradable agents used in cell tracking (43,44) and targeting applications (14–16). Previous investigations have indicated that application of a magnetic force on labeled cells can distort intracellular membranes (33). We noted that the application of a magnetic force for 24 h had the potential to compromise cell viability, an effect that was dependent on the extent of iron loading. Although we did not evaluate this in the current study, we anticipate that higher magnetic forces will be tolerable if applied for shorter lengths of time. By controlling iron uptake and the specific design of magnetic devices, we have been able to safely attract cells without affecting cell viability, apoptosis, or differentiation in culture.
Most importantly, our data suggest that by tagging human EPCs with an FDA-approved agent, it is possible to safely target them in static and flowing conditions in vitro and in vivo. We have assembled and tested a specialized external magnetic actuator using an in vitro flow system. Using this actuator, we have also been able to enhance cell localization following common carotid artery injury in an animal model; the increase in engraftment was 5-fold higher than in control (nontargeted) cells. In this study, a flow interruption period was implemented, which is similar to those employed in clinical trials and is a favorable condition for cell enrichment. Our computer simulations and in vitro data suggest that in the presence of high flow rates at the target area, very few cells are captured under nontargeted conditions, which allows up to 252-fold enhancement in vitro using magnetic attraction, depending on the distance of the target tissue. This highlights the potential for extensive targeting in the cardiovascular system, even in areas of high blood velocity. Nevertheless, further in vivo studies are warranted to assess EPC engraftment in a variety of flowing conditions following peripheral intravenous injection. Our results also emphasize a key issue regarding external targeting in vivo; invariably, cells flowing in vessels closer to the magnetic source than the target tissue will also be targeted (6). However, such cells should be released following removal of the magnetic source in the absence of an underlying biological mechanism to keep them engrafted, as observed both in our in vitro flow experiments and our noninjured control subjects in vivo.
In the injured arteries, we were not able to provide a circumferential external magnetic force; previously this problem has been partially circumvented by axial rotation of the animal (14). Despite this fact, we did not observe any consistent preferential targeting of the cells in the arterial specimens. This may be due to a combination of factors. The broad distribution of magnetic targeting forces due to the large size of the magnetic actuator compared with the size of the artery (at least 100° arc) is likely to disperse the targeted cells. Small within-vessel variations in the extent of damage, unrelated of the direction of the magnetic force, could further facilitate nondirectional engraftment. In addition, the flow during the 24-h recirculation period may enable cells to migrate away from the targeted regions to areas of more pronounced damage. Indeed, in a recent study, despite the presence of an intravascular permanently magnetized device for 24 h, iron-labeled cells were present in the damaged areas adjacent to the device, as well as on it (15).
The magnetic control of cells inside the vasculature is a fascinating and promising concept. In this preliminary report, we have outlined the feasibility of external targeting in the arterial circulation using a clinical-grade agent, bringing this technology a step closer to intervention in humans. Long-term studies will be necessary to investigate the retention of the cells at the site of engraftment, confirming a functionally beneficial increase in healing following magnetic targeting. Ultimately, we envisage concurrent MR imaging and MR-based magnetic guidance of labeled cells in the body, possibly in a manner similar to magnetic guidance of intravascular catheters (45) and magnetic beads (46). This approach could augment localization and simultaneous monitoring of cells in other organ systems such as the heart or the brain and may prove complementary to the systemic injection of cell therapies, thus expanding the horizon of cardiovascular interventions and the future of stem cell therapeutic strategies.
For more details regarding the methods and calculations used in this study, please see the online version of this article.
Supported by the UCL Institute of Child Health (Child Health Research Appeal Trust), the British Heart Foundation, the Alexander S. Onassis Public Benefit Foundation, and the Biotechnology and Biological Sciences Research Council. Panagiotis G. Kyrtatos and Dr. Lehtolainen contributed equally to this manuscript.
- Abbreviations and Acronyms
- common carotid artery
- endothelial progenitor cell
- Food and Drug Administration
- mononuclear cell
- magnetic resonance imaging/images
- superparamagnetic iron oxide
- Received May 4, 2009.
- Accepted May 20, 2009.
- American College of Cardiology Foundation
- Bartunek J.,
- Dimmeler S.,
- Drexler H.,
- et al.,
- Task Force of the European Society of Cardiology
- Pislaru S.V.,
- Harbuzariu A.,
- Gulati R.,
- et al.
- Polyak B.,
- Fishbein I.,
- Chorny M.,
- et al.
- Peichev M.,
- Naiyer A.J.,
- Pereira D.,
- et al.
- Gulati R.,
- Jevremovic D.,
- Peterson T.E.,
- et al.
- Korbling M.,
- Reuben J.M.,
- Gao H.,
- et al.
- Berridge M.V.,
- Tan A.S.
- Santiago F.S.,
- Ishii H.,
- Shafi S.,
- et al.
- Miyashiro J.K.,
- Poppa V.,
- Berk B.C.
- Losordo D.W.,
- Schatz R.A.,
- White C.J.,
- et al.
- Erbs S.,
- Linke A.,
- Adams V.,
- et al.
- Schachinger V.,
- Assmus B.,
- Britten M.B.,
- et al.
- Hill J.M.,
- Dick A.J.,
- Raman V.K.,
- et al.
- Hinds K.A.,
- Hill J.M.,
- Shapiro E.M.,
- et al.
- Faddis M.N.,
- Blume W.,
- Finney J.,
- et al.