Magnetically Targeted Endothelial Cell Localization in Stented Vessels
Author + information
- Received May 3, 2006
- Revision received June 26, 2006
- Accepted June 26, 2006
- Published online November 7, 2006.
Author Information
- Sorin V. Pislaru, MD, PhD⁎,
- Adriana Harbuzariu, MD⁎,
- Rajiv Gulati, MD, PhD⁎,
- Tyra Witt⁎,
- Nicole P. Sandhu, MD, PhD†,
- Robert D. Simari, MD, FACC⁎ and
- Gurpreet S. Sandhu, MD, PhD, FACC⁎,⁎ (sandhu.gurpreet{at}mayo.edu)
- ↵⁎Reprint requests and correspondence:
Dr. Gurpreet S. Sandhu, Division of Cardiovascular Diseases, Mayo Clinic College of Medicine, 200 First Street SW, Rochester, Minnesota 55905.
Abstract
Objectives A novel method to magnetically localize endothelial cells at the site of a stented vessel wall was developed. The application of this strategy in a large animal model is described.
Background Local delivery of blood-derived endothelial cells has been shown to facilitate vascular healing in animal models. Therapeutic utilization has been limited by an inability to retain cells in the presence of blood flow. We hypothesized that a magnetized stent would facilitate local retention of superparamagnetically labeled cells.
Methods Cultured porcine endothelial cells were labeled with endocytosed superparamagnetic iron oxide microspheres. A 500:1 microsphere-to-cell ratio was selected for in vivo experiments based on bromo-deoxyuridine incorporation and terminal deoxynucleotidyl transferase mediated dUTP nick end labeling assays. Stents were magnetized and implanted in porcine coronary and femoral arteries using standard interventional equipment. Labeled endothelial cells were delivered locally during transient occlusion of blood flow.
Results The delivered cells were found attached to the stent struts and were also distributed within the adjacent denuded vessel wall at 24 h.
Conclusions Magnetic forces can be used to rapidly place endothelial cells at the site of a magnetized intravascular stent. The delivered cells are retained in the presence of blood flow and also spread to the adjacent injured vessel wall. Potential applications include delivering a cell-based therapeutic effect to the local vessel wall as well as downstream tissue.
The complete endothelialization of implanted cardiovascular devices such as stents, vascular grafts, and valves may take several months (1,2). Although drug-eluting stents have substantially reduced the incidence of restenosis (3), recent reports now describe late stent thrombosis (4–6) and an increase in late neointima formation (7). Sirolimus, while exerting beneficial anti–smooth-muscle cell effect, may also impair re-endothelialization via direct effect on circulating progenitor cells (8). The requirement for prolonged dual anti-platelet therapy leads to an increased risk for bleeding-related complications (9), and any premature discontinuation leaves individuals at risk for stent thrombosis, with consequent high morbidity and mortality (6). Previous methods of placing endothelial cells on intravascular stents have included precoating stents with cultured cells (10–13) and capturing cells with ligands or antibodies attached to the stents (14). The recruitment strategies are limited by the small number of progenitor cells in circulation and may result in non-specific capture of inflammatory cells.
Local delivery of cultured cells has also been described but is limited by the requirement for prolonged occlusion of blood flow. Among other intravascular devices, synthetic vascular grafts coated with autologous endothelial cells demonstrated improved patency rates in a report published in 1978 (15). Development of earlier strategies was hampered by an inability to generate an adequate number of endothelial cells from autologous sources. Endothelial cells can now be cultured from circulating progenitors, and these cells appear to be capable of facilitating re-endothelialization and can normalize vascular function in injured vessels (16–21). We have previously shown (22) that magnetic forces can be used to capture and retain superparamagnetically labeled endothelial cells on modified vascular grafts. This report describes the results of superparamagnetic cell labeling and the rapid attachment of these labeled cells to magnetized stents placed in porcine coronary and femoral vessels. This novel method of magnetic cell targeting may have multiple future clinical applications, including the ability to provide localized or regional cell-based therapies.
Materials and methods
Generation of endothelial outgrowth cells
Peripheral blood mononuclear cells were isolated from 50 to 200 ml porcine blood by Ficoll density gradient (Histopaque 1077, Amersham, Piscataway, New Jersey) as previously described (17). These were placed on fibronectin-coated plates and grown in endothelial basal culture media (EBM) supplemented with EGM-2 SingleQuotes (Cambrex, Baltimore, Maryland). The presence of highly proliferative cells with endothelial characteristics (endothelial outgrowth cells [EOCs]) was noted after 7 to 10 days. Our laboratory (17–19) and others (16,23) have previously described the functional and phenotypic characteristics of these cells. A total of 2 weeks in culture yielded large numbers of cells for in vitro and in vivo experiments.
Cell labeling
Superparamagnetic microspheres (SPM) (0.9 micrometer diameter, 63.4% iron oxide, pre-coated with a green fluorescent tag; Bangs Labs, Fishers, Indiana) were added to the EGM-2 culture media at serial concentrations for the in vitro studies. For in vivo experiments, a 500 SPM particles/cell co-incubation ratio was selected based upon the in vitro results as described later. Cells were co-incubated with SPM at 37°C for 16 h, washed 3 times in phosphate-buffered saline, trypsinized, and used in subsequent experiments. Endothelial outgrowth cells were labeled with a second fluorescent tag by incubation with the red carbocyanine tag CM-DiI (Molecular Probes Inc., Eugene, Oregon) at 37°C for 30 min.
Effects of iron loading on cell survival, proliferation, and function
Endothelial outgrowth cells were incubated with escalating doses of SPM particles (SPM-to-EOC ratio of 0 to 4,000/cell). Cell proliferation was assessed by spectrophotometric measurements of the bromo-deoxyuridine (BrdU) incorporation in replicating cellular deoxyribonucleic acid (Cell Proliferation ELISA, BrdU colorimetric, Roche, Nutley, New Jersey). The superparamagnetic particles within cells were associated with substantial backgrounds; therefore, cell-free supernatants were analyzed in 96-well plates using a standard plate reader (Spectra Max Plus Plate Reader, Molecular Devices, Sunnyvale, California). Endothelial outgrowth cell apoptosis after exposure to SPM was evaluated with the terminal deoxynucleotidyl transferase mediated dUTP nick end labeling (TUNEL) method (In Situ Cell Death Detection Kit, TMR red, Roche). All experiments were performed at least in triplicate.
To characterize the effects of SPM loading on EOC function, we plated EOCs on 6-well plates coated with human fibronectin and added SPM at a 500:1 ratio. Control EOCs were plated without SPM. To avoid contamination from EGM-2 SingleQuotes, EBM alone was used in these experiments. The conditioned medium was concentrated with a CentriCon tube and analyzed on the Human Cytokine Array V (RayBiotech, Norcross, Georgia). This detection system is designed for qualitative analysis of proteins present in conditioned media. Proteins specifically bound to spots of a membrane and were detected by a cocktail of biotin-conjugated antibodies (for 72 different cytokines). After incubation of the membrane with horseradish peroxidase–conjugated streptavidin and detection solution, the membrane was exposed to Biomax MR film (Kodak, Rochester, New York).
Development of magnetizable stents
A variety of stainless steel, chromium cobalt, and nitinol stents were exposed to a 5,000 G magnetic field generated by a neodymium-iron-boron supermagnet. These commercially available stents did not retain any significant magnetic charge capable of attracting SPM-loaded cells. Multiple grades of steel; nickel; and alloys such as Monel, Permalloy, Nitinol, and nickel chromium were obtained in a 0.005- to 0.010-inch diameter wire form (McMaster Inc., Atlanta, Georgia) so as to approximate the dimensions of stent struts. We magnetized 10-mm wire loops by a 5-min exposure to the neodymium supermagnet. An electromagnetic degaussing device was used to demagnetize the negative control wires. The ability to attract labeled cells was evaluated by live video recording. Magnetized nickel provided excellent cell capture but does not have the structural properties required for stent construction. We therefore combined the structural properties of steel with the magnetic properties of nickel by coating commercially available stainless-steel stents with a 10-μm-thick layer of nickel. Ferrofluid (10-nm magnetite particles dispersed in liquid base) was brought into contact with the stent to define the magnetic field generated by the magnetized struts. The magnetite particles in ferrofluid align with magnetic fields and can be visualized with light microscopy. This mapping demonstrated the presence of uniformly distributed small magnetic domains on the surface of the struts. Confocal and scanning electron microscopy showed an evenly distributed adherence of SPM-loaded EOCs on these plated stents. The ability to attract and retain iron-loaded cells in flow conditions was tested by exposing magnetized stents to a circulating suspension of iron-loaded EOCs at various flow rates. The stents were then inspected by standard fluorescent and confocal microscopy.
Porcine studies
All animal experiments were performed in accordance with the “Position of the American Heart Association on Research Animal Use” and were approved by the Institutional Animal Care and Utilization Committee of the Mayo Clinic. Domestic pigs weighing 30 to 40 kg were used. We obtained 200 ml blood under sedation 2 weeks before intervention and used it for generation of autologous EOCs. Superparamagnetic microsphere loading was performed by a 16-h co-incubation with EOCs at a 500:1 SPM-to-EOC ratio. The cells were then labeled with CM-DiI on the morning of the experiment. The animals were sedated with a combination of ketamine and xylazine and anesthetized with isoflurane. The carotid arteries were dissected free and 9-F arterial sheaths were placed under sterile conditions. Arterial segments devoid of visible side branches in the superficial femoral and coronary arteries were selected for delivery. Denudation of the targeted segment was performed by passing back and forth an oversized balloon (balloon-to-vessel ratio of 1.2 to 1.4) for a total of 5 times. Nickel-coated stainless steel stents were brought into contact with the 5,000 G neodymium magnet for 5 min and were deployed in the vessel with an over-the-wire balloon catheter. Control non-magnetized stainless steel stents were placed either immediately upstream or downstream of the nickel-coated stent. The balloon was then withdrawn until the tip was just proximal to the deployed stents and was re-inflated at low pressure. Complete vessel occlusion was confirmed by injecting a small amount of contrast material upstream from the balloon. The guidewire was withdrawn, and the central lumen of the balloon was used for cell delivery. Approximately 2.5 to 6 × 106autologous EOCs suspended in 4 ml PBS were injected over 2 min and blood flow was restored after a total occlusion time of 5 min in the coronary arteries. The femoral artery underwent a longer total occlusion time of 10 min. The adjacent magnetized and control stent deployment enabled us to expose both stents to similar cell loads and flow conditions in a single artery. All animals received oral aspirin (325 mg/day) and clopidogrel (75 mg/day) for 5 days before the procedure and on the day after procedure day. The animals were allowed to recover and were sacrificed 24 h after stent deployment. The targeted femoral and coronary artery segments were carefully dissected and immersed in cold phosphate-buffered saline. The vessels were then sectioned longitudinally into 2 halves and the stent struts carefully removed. The arterial segments and corresponding stent struts were examined by en face confocal or fluorescent microscopy. The presence of dual-labeled cells (green from the superparamagnetic iron microspheres and red from CM-DiI) was evaluated by visual inspection, and images were captured at low (40×) and high (200×) magnifications. The number of labeled cells was then counted on 10 random high-power fields selected from segments corresponding to the coated and control stents.
Statistical analysis
Statistical analysis was performed using SAS software (SAS, Cary, North Carolina). The normal distribution was tested with the Shapiro-Wilk statistic, and transformations were performed when appropriate. Data were analyzed with one- or two-way ANOVA as appropriate. Tukey’s ttest was used for multiple pairwise comparisons. Pearson’s correlation coefficient was used for evaluating the relationship between cell dose and corresponding stent counts. A p value of <0.05 was considered significant. Data are presented as mean values ± SD.
Results
In vitro experiments
The effects of SPM particle loading on BrdU incorporation are given in Figure 1.There was a minimal decrease in proliferation at a 500:1 SPM to EOC co-incubation dose. However, a statistically significant reduction in proliferation was noted at higher doses. Similar results were observed during TUNEL staining. A minimal increase in apoptosis was noted up to 500:1 SPM-to-EOC co-incubation ratio, with significant cell death occurring at higher doses (representative images shown in Figs. 2Aand 2B). Cytokine array studies (Figs. 2C and 2D) also demonstrated no change in the secretory profile of EOCs after SPM loading at the 500:1 ratio. On the basis of these results, and on our observation that cells could easily be attracted to magnetized metallic devices at a 500:1 SPM-to-EOC ratio, we selected this dose for subsequent in vivo experiments.
Escalating superparamagnetic microsphere dose inhibits cell proliferation. Exposure of endothelial outgrowth cell (EOC) to escalating doses of superparamagnetic microsphere (SPM) results in a small but significant decrease in bromo-deoxyuridine (BrdU) incorporation at high, but not at low doses. * p < 0.05 versus no SPM.
Apoptosis and secretory profiles are stable with low-dose SPM. (A and B)Terminal deoxynucleotidyl transferase mediated dUTP nick end labeling (TUNEL) assays: composite confocal images. Green fluorescence is given by SPM whereas red fluorescence represents TUNEL-positive apoptotic cells. Exposure to SPM at a 500:1 ratio results in minimal apoptosis (A), whereas significantly increased presence of apoptotic cells is evident at a higher SPM to EOC ratio (B, 4,000:1 ratio). (C and D)Cytokine array studies. EOC secretory profile was very similar for untreated cells (C)and cells exposed to SPM at a 500:1 ratio (D). The top four left dotsand bottom two right dotsare built-in positive controls of the array. All other dotsrepresent an individual cytokine (72 in total). With these studies, secretion of interleukins (IL)-8 and IL-10; growth factors EGF, HGF, and GDNF; and metalloproteinase inhibitors TIMP-1 and TIMP-2 and MIP-1b were identified. Other abbreviations as in Figure 1.
The ability of magnetized metallic devices to attract SPM-loaded EOCs is illustrated in Figure 3.Confocal microscopy showed the presence of large numbers of dual-labeled cells on the magnetized but not on the control devices with degaussed stainless steel wire loops (Figs. 3A and 3B). Nickel coating of commercially available stainless-steel stents resulted in improvement of magnetic properties. Ferrofluid mapping demonstrated the presence of small magnetic domains along the stent struts (Fig. 3C). These small magnetic domains resulted in a more uniform coverage of cells on the stent struts, as shown in the confocal microscopy image (Fig. 3D). These magnetic forces were strong enough to allow retention of cells on the stent struts in flow conditions as high as 200 ml/min. An increase in flow rates above this level resulted in progressive cell loss (data not shown). Transmission electronic microscopy (Fig. 3E) shows the endocytosed SPM particles within the cytoplasm of the EOCs. Scanning electronic microscopy at the time of capture shows spherical cells on the stent surface (Fig. 3F).
SPM-labeled EOCs: interactions with magnetic fields. (A and B)Stacked confocal images obtained along the z-axis (100× magnification). EOCs are labeled with CM-DiI (red)and SPM (green). Exposure of SPM-loaded EOCs to a degaussed stainless steel loop does not result in significant cell attraction (A). On the contrary, in the presence of a magnetized stainless steel loop, SPM-loaded cells are rapidly cleared from the suspension and accumulate predominantly on the loop bend (B). A green reflection from the stainless-steel loop is evident. (C and D)Nickel coating of a commercially available stent results in multiple, small, uniformly distributed magnetic domains, as shown in ferrofluid studies (C). Presence of numerous magnetic microdomains is evident (arrows). This translated into a more uniform coverage of the stent surface (D). Endothelial outgrowth cells are labeled with SPM (green). (E and F)Transmission electronic microscopy confirms uniform cytoplasmic capture of SPM microspheres (E). Scanning electron microscopy shows early accumulation of rounded cells to the metal surface (F). (G and H)Representative en face fluorescent microscopy images (200× magnification) from explanted coronary arteries. Local delivery of CM-DiI (red)–labeled EOCs to a non-magnetized stent containing arterial segment results in retention of cells in small numbers (G). Local delivery is greatly enhanced in the segment that received a nickel-coated, magnetized stent before EOC delivery. Note presence of CM-diI–labeled cells along a stent strut (H). Abbreviations as in Figure 1.
In vivo experiments
A total of 10 animals were used in this study, and a total of 24 stents were implanted: 11 in the coronary arteries (7 magnetized, 4 non-magnetized control animals) and 13 in femoral arteries (10 magnetized, 3 non-magnetized control animals). An average of 3.7 ± 1.3 × 106cells was delivered per artery (range 2.5 to 6 × 106cells/artery).
All stented arteries were widely patent 24 h after implantation. The delivery of SPM-loaded cells resulted in a significant retention of cells at the site of the magnetized stent (220 ± 156 cells/high-power field). Very few cells were seen in the region with non-magnetic stents and no cells were seen in areas immediately proximal or distal to the deployment site. The presence of the magnetized stent was associated with a statistically significant 6- to 30-fold increase in the number of retained cells in both the coronary and femoral positions (p < 0.01). Representative en face microscopic images are given in Figures 3G and 3H. Slightly better EOC retention was noted in the coronary vessels for both magnetized and control stents (Fig. 4A).There was no significant correlation between the EOC dose and the number of cells retained at the site of either magnetic or non-magnetic stents (r2= 0.003 and r2= 0.023 respectively; p = NS for both). Low-power images demonstrated localization of cells in the vessel wall along the stent struts as well as within the adjacent denuded areas (Fig. 4B).
Arteries retain more EOCs in the presence of magnetized stents. (A)Cell retention in vivo was significantly higher on magnetized (solid bars)versus non-magnetized control stents (open bars)for both femoral and coronary position. *p < 0.05 for magnetized versus non-magnetized stent. (B)Overlapping, composite, low-power microscopic images of the luminal surface of a porcine coronary artery showing distribution of labeled cells along stent tines and the adjacent denuded endothelium. Abbreviations as in Figure 1.
Discussion
The deployment of intravascular stents results in endothelial denudation over approximately 40% of the stented surface (24). Under normal conditions, endothelial cells gradually repopulate the denuded areas and cover the surface of implanted cardiovascular devices such as stents, valves, and vascular grafts over a period of weeks to months. The ability to rapidly “endothelialize” implanted devices could accelerate local healing and thereby reduce the risk of thrombosis. A method of localizing cells within the vasculature would also enhance the ability to develop novel local or regional cell-based therapies.
Several methods of coating stents with endothelial cells have been described, but none appear to have provided the efficacy, safety, and ease of use required for clinical application. Chemical and biologic modifications of stent surfaces, including the addition of adhesion molecules and antibodies, have been proposed (10,25–30). A common drawback to these approaches is that a peripherally or locally delivered cell must first come into contact with its putative binding site by random interaction and then overcome the shear forces generated by blood flow until additional cellular attachments are formed. Early reports with a CD34+ cell capturing stent show promise, with a restenosis rate of 13.3% at 6 months (28), but long-term results are not available. Because the exact nature and cell markers of the putative endothelial progenitor cell remain elusive, most reports have focused on coating devices with receptors for known/presumed markers. Although endothelial progenitor cells likely express such markers, their exclusivity is unknown. Recruiting circulating progenitor cells based on such an approach raises questions about the possibility of unpredicted/unwanted effects owing to the capture of inappropriate cells. A recent report showed that coverage of arterio-venous grafts with a CD34 receptor results in near-complete endothelialization; however, morphometric analysis of grafts 4 weeks after implantation showed significantly higher restenosis rates in the CD34-coated grafts when compared with control subjects (30). Our approach consists of using endothelial outgrowth cells, which during culture have achieved a more defined and homogenous endothelial cell phenotype (16,17).
Local cell delivery during prolonged occlusion of blood flow has been shown to result in cell attachment to the vessel wall but is impractical for clinical use (18). The concept of localizing iron-loaded cells by external magnetic fields was first described by Consigny et al. (31). In a rabbit model of femoral artery injury, increased adhesion of superparamagnetic-loaded endothelial cells was achieved by applying a magnet external to the injured area. However, extended dwell times were needed, and the cell adhesion was primarily on the vessel wall closest to the magnet. We have previously used superparamagnetic labeling to instantly coat the inner surface of magnetized Dacron grafts (22) with endothelial cells. These grafts were used to replace sections of porcine carotid arteries and demonstrated excellent cell retention after being subjected to high flow rates in vivo. A similar “pre-coating” of stents, although feasible, would result in significant cell loss during balloon inflation for stent deployment.
Magnetic forces may provide an elegant solution for cell capture. Cells could potentially be guided to a desired site and retained in place until they are able to form enduring biologic attachments. Literature describing the development of superparamagnetic tags used for magnetic resonance imaging (MRI)-based cell tracking supports the concept that endocytosed superparamagnetic particles are effectively inert in the concentrations commonly used (32,33). We were able to successfully identify a concentration that permitted magnetic capture while having little impact on the viability, proliferation, or cytokine secretion profile of the cultured EOCs. These labeled cells resisted dislodgement when subjected to flow chamber rates similar to physiologic conditions. Local cell delivery during brief cessation of flow was sufficient to allow cell capture in the porcine model. A 5-min cessation of flow would be undesirable for human coronary applications. This time could be reduced by increasing the magnetic charge on the stent or by concentrating cells further by providing simultaneous proximal and distal occlusion. A double-balloon system with a perfusion port to permit distal blood flow may provide additional safety.
The weak magnetic field generated by magnetized stents is likely to create localized imaging artifacts if an individual requires an MRI, but is unlikely to result in contraindication of future imaging. Iron oxide particles can be detected by MRI, but the limited quantity of particles used in our method, along with potential stent artifacts, make it unlikely that these could be imaged in vivo. Transiently magnetizable surfaces, as well as biodegradable magnetic components, can be developed to completely eliminate magnetic charges after the cells have formed stable biochemical bonds.
The bulk magnetization of stents is impractical for uniform cell capture because the magnetic poles localize at the ends of the device and at bends in the struts, resulting in uneven coverage. The nickel plating on our stents resulted in microscopic surface irregularities, thereby generating multiple small magnetic domains, as shown by the ferrofluid mapping. This enabled an even distribution of delivered cells. This magnetic interaction was sufficient to resist the shear forces of blood flow and allowed the cells to form stable biochemical bonds as shown by the substantial in vivo cell retention at 24 h. Not only did the cells line up along the stent struts, but they were also found to have spread to denuded areas between the stent struts, thereby indicating procedural efficacy extending beyond simple retention on the stent alone. Control non-magnetized stents showed only a few cells randomly distributed on the vessel wall.
Nickel can be allergenic in certain individuals and was used for our short-term studies to demonstrate proof of concept. Fully biocompatible stents are being developed for long-term studies. In a manner analogous to drug-eluting stents, magnetic particles may be dispersed in a biocompatible polymer, thereby preventing direct contact with the blood. The long-term effects of iron-containing EOCs and whether they facilitate endothelialization are unknown. These labeled EOCs, or iron particles released by cell turnover, could paradoxically lead to accelerated neointima formation by triggering local inflammation. The SPM particles contain polystyrene and are not biodegradable. Human studies will likely require particles similar to the clinically approved ferumoxide.
The procedure as described was carried out with cultured cells and cannot, in the present form, be used for individuals requiring emergent procedures. However, it would also be possible to place a magnetizable device during an acute procedure, followed by the delivery of labeled cells at a later time. A staged delivery of cells would necessitate a second invasive procedure. Future studies would also need to address the safety of re-exposure to iron oxide particles in the event that additional interventions are required. Future refinements of our strategy include the ex vivo concentration of progenitor cells followed by immediate labeling and redelivery to reduce the time frame to <1 day. Developments in progenitor cell biology may lead to identification of alternative cell populations that may be isolated rapidly, yet retain the safety of EOC delivery. Genetically modified cells, a combination of cells, or serial delivery of different cells designed to treat specific conditions can also be envisaged. Recent experiments in models of vascular injury with overexpression of eNOS, tissue plasminogen activator, and factor VIII appear promising for localized therapy (20,34,35). The clinical implications of our approach extend beyond the re-endothelialization of implantable cardiovascular devices. The ability to localize cells within the vasculature opens the possibility that local healing could be accelerated and ischemic downstream tissue treated by overexpression of various factors in modified endothelial cells.
Footnotes
This research was supported by the National Institutes of Health (HL75566 to Dr. Simari).
- Abbreviations and Acronyms
- BrdU
- bromo-deoxyuridine
- EOC
- endothelial outgrowth cell
- MRI
- magnetic resonance imaging
- SPM
- superparamagnetic microsphere
- TUNEL
- terminal deoxynucleotidyl transferase mediated dUTP nick end labeling
- Received May 3, 2006.
- Revision received June 26, 2006.
- Accepted June 26, 2006.
- American College of Cardiology Foundation
References
- ↵
- Van Belle E.,
- Tio F.O.,
- Couffinhal T.,
- Maillard L.,
- Passeri J.,
- Isner J.M.
- ↵
- Roiron C.,
- Sanchez P.,
- Bouzamondo A.,
- Lechat P.,
- Montalescot G.
- ↵
- Jeremias A.,
- Sylvia B.,
- Bridges J.,
- et al.
- ↵
- ↵
- Carter A.J.,
- Aggarwal M.,
- Kopia G.A.,
- et al.
- ↵
- Fukuda D.,
- Sata M.,
- Tanaka K.,
- Nagai R.
- ↵
- ↵
- Dichek D.,
- Neville R.,
- Zwiebel J.,
- Freeman S.,
- Leon M.,
- Anderson F.
- ↵
- Kipshidze N.,
- Dangas G.,
- Tsapenko M.,
- et al.
- ↵
- ↵
- ↵
- Gulati R.,
- Jevremovic D.,
- Peterson T.E.,
- et al.
- ↵
- Gulati R.,
- Jevremovic D.,
- Peterson T.E.,
- et al.
- Gulati R.,
- Jevremovic D.,
- Witt T.A.,
- et al.
- ↵
- Griese D.P.,
- Achatz S.,
- Batzlsperger C.A.,
- et al.
- Griese D.P.,
- Ehsan A.,
- Melo L.G.,
- et al.
- ↵
- Pislaru S.V.,
- Harbuzariu A.,
- Agarwal G.,
- et al.
- He T.,
- Smith L.A.,
- Harrington S.,
- Nath K.A.,
- Caplice N.M.,
- Katusic Z.S.
- ↵
- Rogers C.,
- Parikh S.,
- Seifert P.,
- Edelman E.R.
- Nugent H.M.,
- Rogers C.,
- Edelman E.R.
- ↵
- Aoki J.,
- Serruys P.W.,
- van Beusekom H.,
- et al.
- Bhattacharya V.,
- McSweeney P.A.,
- Shi Q.,
- et al.
- ↵
- Rotmans J.I.,
- Heyligers J.M.,
- Verhagen H.J.,
- et al.
- ↵
- ↵
- Hinds K.A.,
- Hill J.M.,
- Shapiro E.M.,
- et al.
- Hill J.M.,
- Dick A.J.,
- Raman V.K.,
- et al.
- Kong D.,
- Melo L.G.,
- Mangi A.A.,
- et al.
- Lin Y.,
- Chang L.,
- Solovey A.,
- Healey J.F.,
- Lollar P.,
- Hebbel R.P.