Article Text

Download PDFPDF

Original research
Autologous adipose-derived mesenchymal stem cells improve healing of coiled experimental saccular aneurysms: an angiographic and histopathological study
  1. Aymeric Rouchaud1,2,
  2. Waleed Brinjikji3,
  3. Daying Dai1,
  4. Yong-Hong Ding1,
  5. Tina Gunderson4,
  6. Dana Schroeder1,
  7. Laurent Spelle2,
  8. David F Kallmes1,
  9. Ramanathan Kadirvel1
  1. 1Applied Neuroradiology Laboratory, Mayo Clinic, Rochester, Minnesota, USA
  2. 2Department of Interventional Neuroradiology, NEURI Center, Le Kremlin-Bicetre, France
  3. 3Department of Radiology, Mayo Clinic, Rochester, Minnesota, USA
  4. 4Health Sciences Research, Division of Biomedical Statistics and Informatics, Mayo Clinic, Rochester, USA
  1. Correspondence to Dr Aymeric Rouchaud, Applied Neuroradiology Laboratory, Mayo Clinic, 200 First St SW, Rochester, MN 55905, USA; aymeric.rouchaud{at}gmail.com

Abstract

Purpose Long-term occlusion of coiled aneurysms frequently fails, probably because of poor intrasaccular healing and inadequate endothelialization across the aneurysm neck. The purpose of this study was to determine if attachment of autologous mesenchymal stem cells (MSCs) to platinum coils would improve the healing response in an elastase-induced aneurysm model in rabbits.

Materials and methods With approval from the institutional animal care and use committee, aneurysms were created in rabbits and embolized with control platinum coils (Axium; Medtronic) (n=6) or coils seeded ex vivo with autologous adipose-tissue MSCs (n=7). Aneurysmal occlusion after embolization was evaluated at 1 month with angiography. Histological samples were analyzed by gross imaging and graded on the basis of neck and dome healing on H&E staining. Fibrosis was evaluated using a ratio of the total area presenting collagen. Endothelialization of the neck was quantitatively analyzed using CD31 immunohistochemistry. χ2 and Student's t-test were used to compare groups.

Results Healing score (11.5 vs 8.0, p=0.019), fibrosis ratio (10.3 vs 0.13, p=0.006) and endothelialization (902 262 μm2 vs 31 810 μm2, p=0.041) were significantly greater in the MSC group. The MSC group showed marked cellular proliferation and thrombus organization, with a continuous membrane bridging the neck of the aneurysm. Angiographic stable or progressive occlusion rate was significantly lower in the MSC group (0.00, 95% CI 0.00 to 0.41) compared with controls (0.67, 95% CI 0.22 to 0.96) (p=0.02).

Conclusions Autologous MSCs attached to platinum coils significantly improve histological healing, as they result in improved neck endothelialization and collagen matrix formation within the aneurysm sac.

  • Aneurysm
  • Bioactive
  • Technique

Statistics from Altmetric.com

Introduction

Endovascular therapy of intracranial aneurysms is the gold standard and has supplanted the surgical approach for many types of aneurysms.1 However, numerous clinical studies indicate that up to 30% of aneurysms recur within 1 year after coiling,2 ,3 leading to frequent retreatment and associated perioperative risks and risk of hemorrhage if untreated.4–6 Thus, complete and durable aneurysm occlusion following coiling remains an important unmet clinical need. The recanalization phenomenon attributable to coil compaction is particularly observed in wide-necked or large aneurysms.7–13 The recurrence rate of coiled aneurysms is most likely due to the inert nature of platinum coils,14 ,15 which fail to induce attraction and proliferation of either circulating or resident cells that might reconstruct the deficient segment of the vessel wall. Aneurysm recurrence after endovascular treatment is due to coil compaction, lack of endothelial cell growth across the neck of the aneurysm, and deficient deposition of extracellular matrix proteins in the aneurysm cavity.9 ,13 ,16–18 Thus, the aneurysm cavity remains unable to withstand the ongoing mechanical stress from continued arterial pulsation, leading to aneurysm recurrence and possible rupture.

Many different strategies have been proposed to improve aneurysm occlusion rates, including modification of coils by adding polymers, growth factors, β emitters or cells.14 ,19–28 To date, clinical trials have been disappointing regarding the efficacy of these second-generation, modified-coil technologies.2 ,29–40 Other studies have used cell therapy with mature and differentiated cells to promote aneurysm healing.41–52 However, terminally differentiated, implanted cells do not differentiate into the other cell types normally present in the blood vessel wall. Ideally, the cells lining the implanted coils would differentiate into arterial endothelial cells, forming a confluent layer of endothelial cells across the neck of the aneurysm, and those residing deep in the endothelial lining would become medial smooth muscle cells with the formation of collagen.13 ,53 ,54 From a biological perspective, mesenchymal stem cells (MSCs) represent an ideal cell type with which to populate saccular aneurysms.26 ,41 ,53 ,55–58

To study the effect of autologous MSCs on aneurysm healing, we conducted a case–control study comparing the healing effects of MSCs on coils with bare platinum coils in the treatment of an elastase-induced aneurysm model in rabbits. The purpose of this study was to determine if attaching autologous MSCs to platinum coils would improve the healing response of experimental aneurysms.

Materials and methods

The animal care and use committee at our institution approved the animal procedures.

Aneurysm creation

Elastase-induced saccular aneurysms were created in 13 rabbits as previously described.59 ,60 Briefly, under general anesthesia, the right common carotid artery (RCCA) was exposed and ligated distally. A 5 F vascular sheath was advanced retrogradely into the RCCA, and a 3 F Fogarty balloon inflated with iodinated contrast medium was advanced through the sheath to the level of the origin of the RCCA under fluoroscopic guidance. Porcine elastase was incubated within the lumen of the RCCA above the inflated balloon for 20 min, after which the catheter, balloon and sheath were removed. The RCCA was then ligated below the arteriotomy. No morbidity or mortality was observed with aneurysm creation.

Autologous MSC isolation and culture

Adipose tissue was harvested during the aneurysm creation procedure from the neck of the rabbit. Autologous MSCs were isolated from adipose tissue extracted from the fat of the rabbit neck. Briefly, adipose tissue was digested with collagenase, then passed through a 70 μm cell strainer (BD Biosciences, San Diego, California, USA) and incubated in red blood cell lysis buffer (Sigma-Aldrich). Harvested cells were cultured in a medium consisting of Dulbecco's Modified Eagle Medium with 10% fetal bovine serum and 2 mM l-glutamine. Adherent MSCs were cultured until 5×106 cells had been obtained from each rabbit with about four cell passages for each rabbit. MSC membranes were labeled with fluorescent Dil (Sigma/Aldrich Chemical Co).

Coil preparation

Axium detachable coils (Medtronic, Irvine, California, USA) were placed in 15 mL Falcon tubes. An MSC suspension (1×106 cells/mL) was poured into the Falcon tubes containing the coils. The tubes were agitated at 300 rpm for 1 hour in an orbital shaker, and then the coils were cultured in the same culture medium as previously described and stored in the cell incubator for 24 hours at 37°C. Cell-seeded coils were then thoroughly washed with phosphate-buffered saline to exclude non-adherent cells, and resheathed in the delivery sheath. Axium coils incubated with culture medium without MSCs served as controls. Figure 1 illustrates MSCs growing on the coil 24 hours after seeding viewed by scanning electronic microscopy (figure 1A) and by fluorescence microscopy after nuclear staining (Syto16; Thermo Fisher Scientific, Waltham, Massachusetts, USA) (figure 1B).

Figure 1

Platinum coils seeded with adipose-derived autologous mesenchymal stem cells (MSCs). (A) Scanning electron microscopic image showing coil seeded with MSCs. (B) Confocal microscopic image of MSCs seeded on the coil surface, stained with nuclear stain Syto16 (green).

To confirm that the resheathing and delivery of the coil through the microcatheter had not stripped a large proportion of cells off the coils, we pushed a cell-seeded coil through an Excel (Target Therapeutics) delivery microcatheter ex vivo and then flushed the catheter with saline. We measured the number of non-adherent cells afterwards, which was less than 5% of the initial number of cells seeded on the coil. The non-adherent cells were cultured in the culture medium to confirm that the cells were alive after the resheathing and delivery steps.

Coiling of the experimental aneurysms

Aneurysms were treated 3 weeks after their creation.61 Under general anesthesia, a 5 F catheter was advanced from the right common femoral artery into the brachiocephalic artery. Using the coaxial technique, an Excel microcatheter was advanced into the aneurysm. Aneurysms were embolized with Axium coils without MSCs (control group, n=6) or Axium coils carrying autologous MSCs (test group, n=7). After embolization, a final control DSA was performed. A blinded reader with 20 years of experience estimated aneurysm occlusion and packing density with the Angiocalc tool (http://www.angiocalc.com).

Follow-up

Control DSA was performed 4 weeks after coiling via the left auricularis rostralis artery under general anesthesia. All angiograms at death were compared with immediate post-treatment angiograms and assessed for any increase or decrease in contrast filling of the aneurysms. Aneurysms were assigned to one of three result groups: stable occlusion (no modification between immediate post-treatment and 4-week angiograms), progressive occlusion (decrease in aneurysm opacification between immediate post-treatment and 4-week angiograms), or coil compaction (increase in aneurysm opacification between immediate post-treatment and 4-week angiograms). Angiography results are expressed as the proportion of aneurysms demonstrating stable or progressive occlusion comparing immediate post-coiling and follow-up results. For this evaluation, the reader was blinded to the treatment group.

Tissue harvest and processing

After follow-up DSA, animals were killed by lethal injection of pentobarbital. Median sternotomy and pericardiotomy were performed. The animals were then perfusion-fixed with 10% buffered formalin for 10 min followed by flushing with heparinized saline for 5 min. The coiled aneurysm was then harvested and immersed in Tris-buffered saline (TBS). Under a dissection microscope (Leica MZ 125), the parent artery was cut longitudinally to expose the neck orifice for gross inspection to evaluate tissue growth at the neck; the sample was then photographed using the MicroPublisher 5.0 RTV camera attached to the dissection microscope. After photography, the sample was fixed in 10% formalin for 2 hours for further whole tissue mount staining.

Whole-mount en face immunostaining

Details of the procedure are described in a previous study.62 Briefly, after the macrophotographs were taken, the aneurysms were fixed in 10% neutral buffered formalin and then washed with TBS; they were then incubated with 5% normal donkey serum in 0.3% Tween in TBS. The samples were then incubated with primary antibodies against CD31 (1:30; Dako, Carpentaria, California, USA) or smooth muscle actin (1:200; Dako) at 4°C. Specific binding was visualized using a secondary antibody Cy3 or fluorescein isothiocyanate-conjugated IgG (1:200; Jackson Immuno Research, West Grove, Pennsylvania, USA). Sytox green (1:1000; Life Technologies, Grand Island, New York, New York, USA) served as a nuclear counterstain to identify inflammatory cells.

Histological analysis

Tissues were embedded in paraffin and then sectioned at 1000 μm. Metallic coil fragments were gently removed with dissection forceps under a dissecting microscope. Paraffin sections were then re-embedded in paraffin blocks and sectioned at 4 μm intervals. Mounted sections were stained with either H&E for conventional histopathological evaluation or Masson Trichrome stain for collagen deposition. Sections were viewed by two blinded and independent reviewers paying particular attention to the thickness of the cell layer across the neck of the aneurysms and the cellularity within the aneurysm dome. To evaluate histological healing, we used an ordinal grading classification, described in a previous report.63 Briefly, neck healing score was calculated on the basis of tissue coverage, the score for coil microcompaction at the neck was based on the shape of the coil holes across the neck, and the healing characteristic score in the dome was categorized on the basis of the density of cellular infiltration and area of organized tissue, by using an image analysis technique based on Photoshop software (Adobe Systems, San Jose, California, USA). These scores used for the ordinal grading classification (neck average, microcompaction and healing) were combined to obtain a total score representative of the aneurysm's healing. The degree of inflammation was defined and scored as: 0, no inflammatory cell infiltration; 1, mild, scant, scattered inflammatory cell infiltration; 2, moderate, patchy inflammatory cell infiltration; 3, marked, attenuated, diffuse inflammatory cell infiltration.47

Neck endothelialization at the level of the aneurysm

For endothelialization analysis, we used whole-mount en face immunostaining with primary antibodies against CD31. We quantitatively analyzed the endothelialization of the neck area with measurement of the CD31-positive field viewed from the neck orifice.

Analysis of collagen deposition

We quantitatively analyzed collagen deposition or fibrosis identified with Masson Trichrome staining using an image analysis technique based on Photoshop software.47 A fibrosis ratio—the total area of fibrosis within the aneurysmal cavity and neck divided by the total area of the aneurysmal cavity and neck—was calculated for each aneurysm.

Immunohistochemistry

Sections were stained immunohistochemically64 for immunostained RAM-11 (1:50, for rabbit; Dako, Carpentaria, California, USA). RAM-11 is a macrophage-specific monoclonal antibody65 used to detect inflammation in the aneurysm dome.

Statistical analysis

Wilcoxon rank-sum tests were performed to compare geometries and packing density between groups. Differences in histological findings between groups were assessed using the between-group Wilcoxon rank-sum test. Results are expressed as median and interquartiles (25th and 75th). For DSA outcomes, groups were compared using Fisher’s exact test and reported with exact Klopper-Pearson 95% CIs. Statistical significance was set at p<0.05.

Results

Seeding of autologous MSCs on Axium coils

Diffuse, confluent cell growth on the coil’s surface and between coil loops was obtained after 24 hours of seeding, which was confirmed by scanning electron microscopy and fluorescence microscopy (figure 1).

Angiographic findings

There were no significant differences in aneurysm neck size (p=0.69), aneurysm width (p=0.37), and aneurysm height (p=0.64) between groups. Packing density was significantly different between groups, with median (IQR) of 22.5% (18.8–24.8%) for the control group and 30.0% (25.0–34.0%) for the MSC group (p=0.01).

All the aneurysms in the MSC group had coil compaction at follow-up with neck recurrences, whereas there were two cases of coil compaction, three stable total occlusions and one progressive total occlusion in the control group. The stable or progressive occlusion rate was significantly lower in the MSC group (0.00, 95% CI 0.00 to 0.41) than in the control group (0.67, 95% CI 0.22 to 0.96) (p=0.02).

Qualitative histology

Gross microscopy showed the coil loops at the neck orifice were bare with no membranous tissue covering (figure 2B) in the control group; no endothelialization on the coil surface was observed with CD31 immunostaining (figure 2C). In the MSC group, the coil loops that crossed the neck orifice were covered with transparent and translucent membranous tissue (figure 3B); this membranous tissue was lined with a layer of CD31-positive cells (figure 3C), indicating endothelialization on the coil surface.

Figure 2

A set of representative images of a single aneurysm packed with Axium coils without mesenchymal stem cells (MSCs) (control group). (A) Angiograms show stability of the aneurysm occlusion between the immediate post-treatment and 1-month follow-up angiograms. (B) Macrophotograph of the neck orifice shows that the coil loops crossing the neck orifice are bare (original magnification 3.2×). (C) Whole tissue mount staining (antibody for CD31) shows there are no CD31+ cells on the coil surface at the neck orifice (water lens, original magnification 20×). (D) Microphotograph shows poorly organized thrombus at the neck interface; loose connective tissue with thin extracellular matrix in the dome (H&E, original magnification 12.5×). Higher magnification (40×) is provided for the blue square area. (E) Masson Trichrome staining shows that the aneurysm barely has any collagen deposit within the cavity (Masson Trichrome, original magnification, 12.5×) (F) RAM-11 immunostaining shows a few scattered macrophages within the dome tissue (immunohistochemistry, original magnification 25×).

Figure 3

A set of representative images of a single aneurysm packed with Axium coils with seeded autologous mesenchymal stem cells (MSCs) (MSC group). (A) Angiograms show stability of the aneurysm occlusion between the immediate post-treatment and 1-month follow-up angiograms. (B) Macrophotograph of the neck orifice shows that the coil loops at the neck orifice are almost completely covered with white, translucent and transparent membranous tissue (original magnification 6.3×). (C) Whole tissue mount staining (antibody for CD31) shows a single layer of confluent CD31+ cells on the coil loop at the neck orifice (water lens, original magnification 20×). (D) Microphotograph shows that the dome is completely filled with dense connective tissue; endothelium lined thin and thick neointima completely transverses the entire neck interface. Local, dense inflammation is also shown at the top of the dome (H&E, original magnification 12.5×). Higher magnification (40×) is provided for the blue square area. (E) Masson Trichrome staining shows dense packing of collagen deposition in an organized fashion in the aneurysm sac (Masson Trichrome, original magnification 100×). (F) RAM-11 immunostaining shows local and diffuse dense macrophages within the dome surrounding the coil struts (immunohistochemistry, original magnification 25×).

Microscopy showed six of six aneurysms in the control group with a large neck remnant and poorly organized thrombus crossing the neck interface (figure 2D), loose connective tissue associated with scattered macrophages and thin extracellular matrix filling the aneurysm dome, and a small amount of collagen deposition (figure 2D–F). In contrast, in the MSC group, microscopy showed endothelium-lined, thin and thick neointima completely covering the neck interface (figure 3D). Denser connective tissue associated with diffuse macrophage infiltration, thick extracellular matrix and collagen deposit in the aneurysmal cavity was also observed in the MSC group (figure 3D–F).

Histological scoring

The MSC group had a significantly higher inflammation score than the control group (mean (range) 3.0 (3.0–3.0) vs 2.0 (2.0–2.0), p<0.01) indicating a larger number of multinuclear giant cells and macrophages. Overall, the total histological healing score was significantly higher in the MSC group (11.5 (11.0 to 12.0)) than the control group (8.0 (6.25 to 9.0)) (p=0.019).

Quantitative analysis of aneurysm fibrosis

Quantitative analysis of aneurysm fibrosis based on evaluation of collagen66 in the aneurysm cavity demonstrated a significantly higher fibrosis ratio in the MSC group (10.3 (IQR 7.98–28.3)) than in the control group (0.13 (IQR 0.055–0.272)) (p=0.006).

Quantitative analysis of coil endothelialization at the level of the aneurysm neck

There was significantly greater endothelialization of the aneurysm neck in the MSC group (902 262 μm2 (IQR 608 991–1 799 849)) than in the controls (31 810 μm2 (IQR 14 106–397 315)) (p=0.041).

Discussion

This study demonstrated that autologous MSCs attached to platinum coils improved endothelial cell growth across the neck of the aneurysm and increased fibrosis within the aneurysm sac. Thus, ex vivo loading of standard coils can achieve improved biological healing of aneurysm cavities.

Notably, however, angiographic outcomes were better in controls than in the MSC group. It is possible that the induced fibrosis from the implanted cells caused retraction of the coils, favoring coil compaction. Prior to clinical translation therefore substantial additional preclinical evaluation of ex vivo autologous MSC implantation is warranted.

Prior studies used therapy with differentiated cells to promote aneurysm healing.41–52 However, none of these studies demonstrated the improvements in healing responses in both aneurysm neck and aneurysm sac seen in our study. Some authors demonstrated that fibroblasts improve intrasaccular fibrosis but have no impact on endothelialization and neointima formation.47 ,48 ,50 ,51 Furthermore, fibroblasts have important procoagulant activity due to intense expression of tissue factor (TF).53 Unlike fibroblasts, MSCs, whether from adipose tissue or bone marrow, express TF poorly and are devoid of procoagulant activity, limiting thrombus formation.53 Other groups used smooth muscle cells delivered into the aneurysm. Raymond et al45 and Marbacher et al42 observed thicker neointima with smooth muscle cell implantation, but observed no difference in fibrosis and endothelialization with this technique. Meanwhile, studies reporting the use of endothelial progenitor cells in preclinical aneurysm models demonstrated confluent monolayers of endothelial cells with underlying neointima but without effect on intrasaccular healing.42 ,49

These previous studies confirm that differentiated implanted cells do not differentiate into the other cell types normally present in the blood vessel wall. Ideally, the cells lining the implanted coils at the neck interface would differentiate into arterial endothelial cells, and those residing deeper in the aneurysm sac would become medial smooth muscle cells.13 ,53 ,54 From a biological perspective, MSCs represent an ideal cell type with which to populate saccular aneurysms because of their potential to differentiate into various cell types based on their specific environment.26 ,41 ,53 ,55–57 While our current methodology did not allow specific histopathological identification of implanted versus recruited cells, previous studies have demonstrated that transplanted MSCs can differentiate into specific phenotypes of damaged cells in vivo under the influence of local host factors.67–69 Under conditions of laminar flow, shear stress induces MSCs to differentiate into endothelial cells, so the implanted MSCs lining the neck of the aneurysm preferentially differentiate into cells lining the arterial wall.70 ,71 This was also observed in the present study using autologous stem cells with continuous endothelium at the aneurysm necks and thick neointima, assumed to originate from the implanted autologous MSCs. Furthermore, the embolized aneurysm cavity is rich in platelets stimulating MSCs to differentiate into smooth muscle cell phenotypes;72 this was observed in our study with smooth muscle cells assumed to originate from the implanted autologous MSCs. The smooth muscle cells present in the aneurysm sacs are then responsible for production of the collagen, which is responsible for fibrosis and wound healing.

Recently published studies demonstrate the benefits of MSCs in promoting intracranial aneurysm healing.53 ,58 ,73 In the same elastase model in rabbits, Rouchaud et al53 used autologous MSCs from bone marrow injected directly into the aneurysm by an endovascular route through a catheter placed in the aneurysm with no scaffold. Satisfactory results regarding reduction in aneurysm size and histological healing were obtained in this study, but the number of cells that effectively attached themselves to the aneurysm sac was low because of the absence of a scaffold to hold them inside the aneurysm. This diminished the beneficial effect of the MSC therapy and also increased the risk of cells migrating into the intracranial circulation and causing an ischemic stroke. Recently, Adibi et al58 published a study of three rabbits treated with combined endovascular coiling and intra-aneurysmal allogeneic bone-marrow MSC injection into the aneurysm immediately after placement of the framing coil. They observed a significant improvement in histological scores.58 The authors used allogeneic MSCs with no deleterious impact in rabbits, but this technique is likely to induce an immune response against transplanted cells in other animals and humans.74 Another limitation of this study is the fact that the authors relied on the framing coil to promote intra-aneurysmal stasis, thus containing the implanted cells in the aneurysm sac and avoiding cell migration. Further analysis of this technique is needed to assess the risk of cell migration and to evaluate the risk of ischemic stroke.

Limitations

Our study is limited by its small sample size. Another limitation is that the rabbits were killed at only one time point, making it impossible to evaluate if the effects were permanent; also it is possible that the beneficial effects on histological healing observed with MSCs could appear earlier than 1 month. Furthermore, only a dose of 5×106 cells was evaluated, which might not be the appropriate dose. Also, MSCs were not labeled for in vivo tracking, therefore we cannot determine if the implanted cells were viable at the time of death or whether the cells were responsible for the tissue changes noted. Furthermore, the packing density was significantly different between the groups, being higher in the MSC group, which may have biased some of the histological results.

To the best of our knowledge, this study is the largest to use autologous stem cells for the treatment of intracranial aneurysms in an experimental study. Furthermore, despite the small sample size, the effect was statistically significant in favor of MSCs for histological healing, encouraging larger experiments to evaluate safety and long-term outcomes. Further studies are required to evaluate outcomes at different time points to determine the effect of MSCs over time and evaluate the stability or progression of neck remnants. Also, an autologous approach with MSCs does not seem to be appropriate for acute treatment of ruptured aneurysms, since time is needed to produce sufficient cells for each patient. We used a cell dose of 5×106 cells, which was used in the previous studies by Adibi et al58 and Rouchaud et al,53 but further studies with different amounts of cells are needed to determine the appropriate dose–response efficiency.

Conclusion

Autologous MSCs attached to platinum coils significantly improve histological healing compared with standard coils, with more endothelial cells across the neck of the aneurysm and collagen matrix within the aneurysm sac. However, these findings did not correlate with the angiographic findings.

Acknowledgments

We thank Medtronic (Irvine, California, USA) for generously providing the Axium coils used for this study.

References

Footnotes

  • Contributors AR, DFK, and RK participated in the conception and design of the study. AR, DD and WB participated in drafting the article. AR, WB, LS, DFK, and RK participated in revising it critically for important intellectual content. AR, WB, DD, YHD, TG, LS, DFK, and RK made substantial contributions to conception and design, acquisition of the data, and analysis and interpretation of the data. All authors provided final approval of the version to be published.

  • Funding This work was supported by research grant NS0767491 from the National Institute of Health and Medtronics. This work was partially funded by the SNIS Foundation Fellow Research Grant. AR was supported by research grants from the French Society of Radiology and Therese Planiol Foundation.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement The data discussed in this article are taken from our institution. Unpublished data may be available on request to the corresponding author.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.