Transmural capillary ingrowth is essential for confluent vascular graft healing

Accepted Manuscript Transmural capillary ingrowth is essential for confluent vascular graft healing Timothy Pennel, Deon Bezuidenhout, Josepha Koehne, Neil Davies, Peter Zilla PII: DOI: Reference:

S1742-7061(17)30666-9 https://doi.org/10.1016/j.actbio.2017.10.038 ACTBIO 5144

To appear in:

Acta Biomaterialia

Received Date: Revised Date: Accepted Date:

6 July 2017 24 October 2017 26 October 2017

Please cite this article as: Pennel, T., Bezuidenhout, D., Koehne, J., Davies, N., Zilla, P., Transmural capillary ingrowth is essential for confluent vascular graft healing, Acta Biomaterialia (2017), doi: https://doi.org/10.1016/ j.actbio.2017.10.038

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Transmural capillary ingrowth is essential for confluent vascular graft healing Timothy Pennel1,2, Deon Bezuidenhout1,2,3, Josepha Koehne1, Neil Davies 1 and Peter Zilla 1 1

Cardiovascular Research Unit and MRC IUCHRG, Christiaan Barnard Division of

Cardiothoracic Surgery, University of Cape Town, 203 Cape Heart Centre, Anzio Road, Observatory, 7925, Cape Town, South Africa

2

Authors contributed equally

3

Corresponding Author

Deon Bezuidenhout Cardiovascular Research Unit, 203 Cape Heart Centre Faculty of Health Sciences, University of Cape Town Anzio Road, Observatory, 7925, SOUTH AFRICA Tel: +27 21 406 6349 Cell: +27 82 894 5405 Fax: +27 21 448 5935 Email: [email protected].

KEYWORDS: Vascular Graft, Angiogenesis, Endothelialization, Growth Factor, Heparin

1

Abstract

Spontaneous endothelialization of synthetic vascular grafts may occur via three independent or concurrent modalities: transanastomotic (TA) outgrowth, transmural (TM) ingrowth or fallout (FO) from the blood. The limited TA and FO endothelialization, which occurs in humans, results in poor long-term patency in the small diameter position, where TM ingrowth may offer a clinically relevant alternative. To achieve sequential analysis of each mode of healing, loop grafts comprising anastomotically isolated angiopermissive polyurethane control grafts were abluminally sealed using either ePTFE wraps or solid polyurethane skins and implanted in the rat infrerenal aortic loop model for twelve weeks. Positive control grafts showed improved endothelialization and patency compared to the abluminally isolated midgrafts. Furthermore, the mid-graft healing was accelerated with surface heparin and heparingrowth factor (VEGF, PDGF) modification in a three-week sub-study. We are thus able to distinguish between the three vascular graft endothelialization modes, and conclude that fallout plays a secondary role to TM healing. The increased endothelialisation for growth factor presenting grafts indicates the promise of this simple approach but further optimization is required.

2

1. Introduction

Despite considerable effort to develop a synthetic alternative, autologous vessels remain the preferred conduit for surgeons performing coronary artery and peripheral vascular bypass grafting [1]. The outcome for small diameter synthetic grafts remain unsatisfactory, with less than 50% patency at five years in the infra-inguinal position [2]. Unfortunately, clinically implanted synthetic conduits have progressed little beyond the first reports of expanded polytetrafluoroethylene (ePTFE) and polyester fabric more than four decades ago [3]. Diverse pre-clinical design strategies, which include biodegradable polymers [4], decellularization of xenografts [5], non-degradable surface modifications [6] and the completely tissue engineered blood vessel (TEBV) [7] have failed to make clinical impact.

Regardless of the lack of consensus for the ideal graft construction, it is generally accepted that the absence of an endothelial layer accounts for the unacceptably high occlusion in the low flow, high resistance environment of peripherally implanted small caliber grafts [8,9]. This monolayer not only provides a physical interface between blood and surrounding tissues, but is regarded as the ‘master regulator’ of hemostatic equilibrium and prevents intimal hyperplasia [10], providing the optimal luminal surface coverage for synthetic vascular grafts [11]. A confluent endothelium has been shown to increase small diameter ePTFE patency in the infra-inguinal position, similar to that of veins through ex vivo transplantation of endothelial cells [12]. However, the complexity and cost of this two- stage endothelial seeding procedure has prevented its widespread therapeutic use.

The solution therefore lies in enhancing spontaneous in vivo healing of implanted grafts. Transanastomotic, fallout and transmural endothelialization have been well described as three potential sources [3] (Figure 1a), and novel animal model designs have documented the existence of all three phenomena, but never independently of one another.

3

Transanastomotic outgrowth (pannus), which originates from the adjacent vessel anastomosis can be rapid and complete in animals, but has been shown to be limited to only a few millimeters from the anastomosis in humans [13,14]. Auto-seeding of endothelial cells on polyethylene terephthalate (PET, Dacron®), suspended in the descending aorta of pigs was the first evidence of fallout endothelialization [15], and series of studies by Sauvage’s group now leaves little doubt of the existence of this form of healing [16-18]. Despite animal experimental models implementing optimization techniques by surface modification to enhance fallout endothelialization, [19,20], these methods have not yet translated to clinical use.

For these reasons, vascular grafts remain devoid of an endothelial surface for decades after implant in humans [21]. Transmural endothelialization on the other hand has the potential to heal synthetic grafts to confluence, provided that there is adequate porosity with sufficient pore-interconnectivity to allow capillary ingrowth. Clowes et al was able to demonstrate histological and corrosion cast evidence of capillaries penetrating the graft surface with increased endothelialization in baboons [11]. Transmural endothelialization arising from the capillary ingrowth of perivascular tissue, although generally accepted [22-24], has been shown by our group to occur independently from the transanastomotic endothelialization by employing anastomotic isolation [25]. We now describe the source of this mid-graft endothelium by additionally implementing abluminal isolation (wrap model) to determine whether the endothelial cells were predominantly of blood-born or perivascular capillary origin. Optimization of spontaneous endothelialization of these grafts through delivery of growth factors via heparin surface modification [26] was further investigated to apply results from subcutaneous disc implants to the circulatory system [27,28].

4

2. Materials and methods

2.1. Study Design Nine-centimeter-long looped conduits with a one centimeter isolated high-porosity mid-graft segment were implanted in the infra-renal aorta of Wistar rats weighing ≥ 350g for twelve weeks. Unmodified controls grafts [CG12] allowed for mid-graft transmural tissue ingrowth. Transmural tissue ingrowth was prevented with the addition of an ePTFE to form a wrap graft [WG12], while allowing for normal subcellular ‘communication’ through the external barrier. External coating of the mid-graft with polyurethane sealant to form a sealed graft [SG12] prevented all forms of transmural connection.

Furthermore, grafts were implanted for three weeks to determine the rate of transmural endothelialization through surface modification. Growth factor/heparin surface modification of the mid-graft segement[GF3] were implanted to optimize transmural in-growth rate and heparinized graft [HG3] segments without GF were also implanted to exclude heparin as a confounding agent in graft modification (Table 1). Control grafts implanted for three weeks are represented as CG3.

5

2.2. Porous graft production The porous polyurethane mid-graft segments (1.7mm [ID], 2.7mm [OD]) were produced to have 147±17.2µm pores with 67±17µm interconnections (82% porosity), as previously described by the authors(Figure 1d&e, 3b) [29].The manufacture involved tightly packing spherical gelatin beads (Thies Technologies, St Louis, 150-180µm sieve size) around a centrally aligned stainless steel mandrel (1.7 x 175mm) within a glass tube mold (2.7 x 8.0 x 100mm: ID x OD x Length). The voids were infiltrated with a proprietary polyurethane (PU, M48, Medtronic Inc, Shore hardness 80A) solution [30] (20% PU in N-methyl pyrrolidone [NMP], Sigma-Aldrich, RSA) at a pressure of 750kPa, with simultaneous application of a vacuum to the lower manifold at 100kPa to drive the PU solution through the packed column. After demolding the rod/graft from the glass tube, the polymer was precipitated by phase inversion (24 hours, room temperature, 96% ethanol), the porogen beads and remnant solvent extracted by extensive washing (5 days at 60°C in water) and graft segments were dried, inspected and cut to length.

2.3. Composite graft construction A 30cm long nylon cord (1.5mm [OD]) was wound tightly around a 13mm [OD] PTFE cylindrical mandrel and secured with cable ties (Figure 2a). This was heat set in boiling water for five minutes and rapidly cooled in room temperature (RT) water for one minute before the cable ties were removed allowing the cord to unravel (Figure 2b). The cord was then cut to lengths of approximately 110mm (1 ½ spirals), for use as individual spiraled mandrel. The PU mid-graft segment (described in 2.2) was cut to 10mm and beveled to accommodate the curvature of the loop graft and threaded over the mandrel. Low-porosity ePTFE (Zeus Industrial Products Inc., Orangeburg, SC; Internal diameter [ID]: 1.77 ± 0.05 mm; outer diameter [OD]: 2.48 ± 0.05 mm; internodal distance [IND]: 15-25µm, 40mm length, Figure 1b&c, 3b) isolation segments were glued with 10% PU in chloroform to each

6

side of the PU mid-graft (under x10 magnification) to complete the loop on the heat-set nylon thread (Figure 2 c&d). The completed loop-graft construct was further heat set for five minutes at 75oC to mold the PU to create a 3 to 4cm adjustable loop before and stored in a dry sterile container following 24hrs of 70% ethanol treatment.

2.4. Externally Isolated loop graft build [WG] The loop-grafts were built as above with an additional sheath of a 30mm long ePTFE (W. L. Gore, Flagstaff, AZ; USA, Internal diameter [ID]: 3.5 mm; outer diameter [OD]: 4.1mm; internodal distance [IND]: 30µm) around the mid-graft (Figure 2e&4a). Due to the size mismatch between the 3.5mm [ID] of the sheathed graft and 2.4mm[OD] of the isolation segment, the edges of the graft were then crimped and sealed with 10% PU in chloroform glue prevent any growth between the wrap and the loop graft (Figure 2f).

2.5. Isolated sealed loop graft build [SG] In place of an ePTFE wrap sheath as described above, a 10% solution of PU/chloroform was applied the abluminal surface of the high-porosity PU segment to form an impermeable 50µm layer/skin of PU (Figure 2g&4a). Specific care was taken not to allow the PU to wick into the pores and this was confirmed with stereomicroscopy as well as on explant histology. Furthermore, every 10th graft was excluded from the study for built quality control and examined under the SEM.

2.6. Isolated heparin treated loop graft build [HG] Details of the covalent heparinization of the polyurethane porous discs have been previously described [31]. In brief, PU graft segments were cleaned in isopropanol (IPA, 10min, RT), rinsed with deionized water (DI), and immersed in the grafting solution containing acrylic acid (AAc, 4.2 M, Aldrich, USA), acrylamide (AAm, 0.8 M, Aldrich), Cu(NO3)2 (0.1 M, Saarchem Holpro, RSA) and cerium ammonium nitrate (CAN, 0.006M, Saarchem) after

7

exclusion of oxygen by bubbling with Argon gas. Graft polymerization was performed for 35 min. at RT, after which the samples were rinsed in DI water, washed in PBS (pH = 7.4, RT, 24 h), rinsed again with DI, and dried in air. The PAAc-co-PAAm grafted samples were aminated by exposure to a 0.5M EDA solution in 0.5 M MES buffer containing 0.05 M EDC (pH 5.0, 2 h, RT), and subsequently rinsed in DI. Heparinization was achieved by exposing (2 h at 50oC) the aminated samples to a solution of nitrous acid de-aminated (NAD) heparin (2mg/mL, Celsus, USA) in an acetate buffer (0.4M, pH1⁄44.6) to which 0.01M sodium cyanoborohydrate (NaCNBH3, Saarchem) was added. The samples were consecutively rinsed in DI, PBS, DI, and 70% ethanol, and subsequent dried in air. Toluidine Blue and Ponceau S staining techniques were employed in order to follow and confirm each of the surface modification steps outlined above, and heparin quantification by 3-methyl- 2benzothiazolinehydrazone-hydrochloride (MBTH) [32] assay showed significant levels of heparin (11.05±1.91µg/mg vs 0.14±0.05 control, p<0.001). Pore (150±26µm) and window (69±14µm) sizes were not influenced by heparinization (p>0.4) [28].

2.7. Isolated heparin + growth factor treated loop graft build [GF] Growth factor concentrations and elution profiles have been previously been described by the authors [27]. In short, growth factors were adsorbed onto HG grafts under sterile conditions in a laminar flow hood by repeated compression/decompression of the grafts in the growth factor solution (27µl solution comprising 7.5µl VEGF @ 1.6µg/µl and 1.35µl PDGF-BB @ 1.3µg/µl with 18.15µl PBS) [27] and grafts were placed in a sterile petri dish with 1ml of saline for humidification (Table 1: 2C).

2.8. In vivo implantation Sixty male Wistar rats were operated under a 2% Isoflurane mask anesthesia. A midline laparotomy was performed and the infrarenal abdominal aorta was mobilized to expose the renal arteries to the iliac bifurcation. Once all the iliolumbar branches were ligated, a single

8

dose of heparin (1mg/kg) was administered in the iliolumbar vein, with no further anticoagulation or anti-platelet drugs. Following clamping, the infrarenal aorta was divided in a beveled fashion without excising the aorta, which allowed for manipulation of the loopgraft. Eight to twelve interrupted 9-0 nylon (Ethilon; Johnson & Johnson, New Brunswick, NJ) sutures were used to anastomose the proximal and distal aorta respectively. A visual patency test was performed before the abdominal cavity was closed in two layers (2-0 Ethibond; Johnson & Johnson, New Brunswick, NJ) (Figure 4b&c). Rats recovered from anesthesia and were kept in separate cages for with food and water ad libitum, following which they were paired for social interaction. All animal experiments were approved by the Animal Research and Ethics Committee of the University of Cape Town (Protocol 011/45, 012/51) and were in compliance with the Guide for the Care and Use of Laboratory Animals, Institute of Laboratory Animal Resources, Commission on Life Sciences, National Research Council and ARRIVE guidelines.

2.9. Histological processing Samples for histology were post-fixed in zinc solution (24 hours; 4oC) and embedded in paraffin. Specimens were cut into 3µm sections and dewaxed with 2,2,4-trimethylpentane. Following which, they were stained with haematoxylin (Merck, Damstadt, Germany) and eosin (BDH; WWR International, Poole, England) (H&E), Miller and Masson elastin trichrome stain, anti- AC133 (Abnova Taipei, Taiwan, PAB12663) anti-CD31 (Fitzgerald International, North Acton, Mass 10R-CD31gRT), anti-factor VIII von Willebrand (Dako, AS, Glostrup, Denmark A0082) antibodies for immunohistochemistry to label endothelial cells, and anti-alpha actin (Abcam, Cambridge, UK AB 5694-100) antibodies to label smooth muscle cells.

Slides were viewed and photographed using a Nikon eclipse 90i microscope (Nikon, Tokyo, Japan). The histologic presence of surface endothelium was cross-referenced against the

9

SEM analysis to confirm interpretation. Neo-intimal hyperplasia was defined as the tissue layer between the graft and the blood surface excluding thrombus appositions. The sample preparation was performed by technicians employed by the Cardiovascular Research Unit.

2.10. Cross-sectional vessel analysis The quantitative analysis for section 4.2 was performed with Visiopharm Integrated Systems (VIS) software package (Visiopharm A/S, Hørsholm, Denmark). CD31 positive structures (in a complete cross section) were detected following pre-training of the software; all analyses were assessed by a blinded observer for accuracy and, if necessary, manually corrected. Vessel count and area were measured and the vascular density obtained by dividing vessel number or area by cross-sectional area of the graft, which were reported as vessels/mm2 or area percentage (%), respectively.

2.11. Scanning electronmicroscopic processing (SEM) Explanted SEM specimens were post-fixed in glutaraldehyde (2.5%; 0.1 M phosphate buffer; 24 hours; 4oC), dehydrated in graded ethanol, critical point dried (Polaron, Evanston, Ill), sputter coated with gold, and analyzed in a Jeol JSM 5200. All samples were captured at x15 magnification and digitally ‘stitched’ in Photoshop (CS6; Adobe Systems, San Jose, CA) to create a single image of the whole graft. The samples were subsequently scanned at x75 to x300 magnification for evidence of endothelium, which was cross-referenced against the stitched image. Measurement of endothelium was then performed on Photoshop CS6 to capture area as well as minimum and maximum outgrowth length. All samples were crossreferenced against corresponding immunohistochemistry to confirm the presence of endothelium. During the SEM analysis of endothelial graft coverage, ostia on the surface of grafts were counted for the relevant group as categorical data.

10

2.12. Statistical analysis Results were expressed as mean±SD for continuous variables. Graphs are represented as means, and error bars represent standard error of the mean. All continuous data was tested for normality with a Shapiro-Wilk test and equivalence of variance was confirmed with a Bartlett’s test. Where appropriate, continuous data was analyzed with a student t-test and multi-group analysis by analysis of variance (ANOVA). The significant level for pairwise testing between categories was controlled by the Tukey’s HSD test. Non- parametric data analysis was performed with a Wilcoxon rank-sum test and multi-group analysis by KruskalWallis. Fisher’s exact test was used for categorical analysis. An ߙ of <0.05 was used for statistical significance.

11

3. Results Overall patency, presence of luminal surface endothelium presence and number of visible luminal ostia are given in Table 2. The number of implants in the SG12 groups was limited to 6 (compared to the 12 used in the CG12 and WG12 groups) due to 100% occlusion rates observed.

3.1. Transmural Isolation Model

General Findings During implantation, the CG12 oozed blood from the pores of the PU mid-section and required multiple clamping and unclamping to fully coagulate the graft wall prior to positioning it deep to the retroperitoneum (Figure 4b). Neither the SG12 nor the WG12 showed any permeability of blood through the barrier and hemostasis was complete immediately after clamp release. It was noted however that the space between the PU graft and the ePTFE wrap in the WG12, filled with blood and was under pressure (Figure 4c).

Despite 100% survival, 11/30 (36%) of the grafts were occluded at explant, without clinical evidence of hind limb ischemia. Graft occlusion was statistically higher in the SG12 6 (100%), when compared to the CG12 3(25%) and WG12 2(17%) groups, (p=0.0015). There was no statistical difference between the CG12 and WG12 with regards to patency.

Macroscopic analysis There was macroscopic evidence of collateral flow in the distal aorta in all the occluded grafts at the time of explant, which accounted for the lack of clinical hind limb malperfusion.

12

All WG12 were uniquely surrounded in a dense, thick capsule of fibrous tissue, which was initially mistaken for the ePTFE wrap due to its well-demarcated nature, as seen in (Figure 5k) marked with an *. On sectioning, fresh clot was noted between the ePTFE wrap and the high-porosity PU, with red discoloration of the PU graft wall absent of clot.

All the CG12 were well imbedded in the retroperitoneal tissue wall with no plane between the graft and the surrounding tissue. The patent CG12 had confluent glistening luminal surfaces and had macroscopic evidence of healing (Figure 5j). The SG12 showed no evidence of perivascular graft infiltration (Figure 5l) of the graft wall, which was only loosely adhered to the retroperitoneum. Macroscopically the SG12 graft looked similar to the pre-implant, with visibly pulsatile iliac arteries. All six SG12 lumina were completely occluded with a macroscopic appearance of clot once transected (Figure 5l).

Histological analysis Figure 5a,b,d,e shows the ingrowth of perigraft tissue within the prosthetic wall of the CG12, which is absent in both SG12 and WG12. The cyan staining, with typical wavy pattern on trichrome stained samples indicated the presence of collagen (Figure 5b). This collagenous tissue was absent in both SG12 and WG12, which were rich in erythrocytes and fibrin with a few scattered macrophages (Figure 5 d&e). Preparation of these small samples in crosssection does result in some sloughing of endothelial layer from the luminal surface when compared to SEM but endothelial cells were always documented on the luminal surface of the CG12 grafts. The three occluded CG12 grafts had mural vessels, which extended into the organized clot within the lumen (Figure 5g&h). This was not present in the other two groups with occlusions, where only clot was noted with minimal cellularity. No endothelium was noted on the clotted SG12 grafts, due to the 100% occlusion.

13

Flat nucleated cells sporadically lined the luminal surface of the WG12, which stained CD31+ although not with the same intensity as the CG12 group (Figure 5c). Macrophages cells were seen scattered throughout the wall of the WG, between the PU and the ePTFE. There were typical FBGC in the capsule surrounding the ePTFE wrap. Notably fewer macrophages cells were present in the CG12 than the WG12, but when present were larger multinucleate giant cells. The presence of PU mural vascularity was significantly higher in the CG12 (12/12), when compared to the WG12 (0/12) or the SG12 (0/12) (p=0.0001). None of the cells in any of the grafts stained positive for AC133 (Figure 5f).

Scanning electron microscopy An endothelial free zone between the anastomosis and the mid-graft test segment was confirmed in all samples with outgrowth rates comparable to those previously described by our group [25]. All patent CG12 grafts (9/9) had endothelium on the surface with coverage of 80±20%, whilst only three of the ten patent grafts showed any evidence of endothelium in the WG12, coverage 16±30% (p=0.0015) (Figure 5i). The categorical presence or absence of endothelium was also significantly higher in the CG12 compared the WG12 (p=0.003).

3.2. Optimizing Transmural Endothelialization

General Findings All 30/30 (100%) grafts were well embedded within the retroperitoneum and patent at explant. Patency could be confirmed by a pulsatile proximal and distal aorta, as well as by an unobstructed lumen on sectioning. There were no macroscopic differences from any of the three explanted groups, all of which appeared to be devoid of a confluent mid-graft endothelium to the naked eye.

14

Quantitative vessel analysis indexed over vessel area did not show a significant difference in the vessel number 33.0±19.8, 49.2±46.7 and 58.6±21.5 vessels/mm2 or area 1.56±0.86, 1.71±0.68, 2.00±0.84 %, for CG3, HG3 GF3 respectively [NS]. (Figure 6a).

Endothelial coverage The mid-graft endothelial coverage was statistically different between the three groups [p=0.0087]. On further analysis by comparison of means, only GF3 and HG3 were statistically different (15±20.5% vs. 0%, [p=0.0163]) (Figure 6b).

Categorical analysis for the presence of endothelium was also statistically different for the three groups [p=0.0020], which was driven by the presence of GF3 compared to HG3 [p=0.003]. Comparison between the other groups was not statistically significant.

3.3. Quantification of surface capillary ostia

At the completion of the experiment, only four capillary ostia were documented in all of the observed samples. Three of them in GF3 three-week implant and one in CG12 12-week implant [NS]. Capillary ostia could not be correlated with the presence, or the amount of endothelium present on the graft surface, but the endothelial outgrowth on the PU surface appeared to extend in a downstream direction from the capillaries in all observed cases (Figure 7).

4. Discussion

We have previously shown that confluent mid-graft endothelialization occurs spontaneously and independently of transanastomotic outgrowth [33]. However, it was not clear whether this complete in vivo healing was due to fallout endothelializaton through circulating EPCs or transmurally as we had proposed. By abluminally isolating the mid-graft from the

15

surrounding tissue, we could independently investigate these two forms of healing and alter its rate through growth factor surface modification. It was further confirmed that the looped isolation model is a reliable and reproducible implant procedure, which resulted in a 100% survival to explant. Although technically more challenging, once mastered the operative outcome of the loop-graft equals that of the standard straight implant. These results further validate the procedure as a high throughput model for the testing of transanastomotic independent healing of the mid-graft.

The high occlusion rate in the SG12 is evidence that some form of transmural communication is required for the patency of vascular conduits irrespective of mid-graft healing. Although this is not a new finding [34,35] it is an early warning sign to investigators pursuing ultra-low porosity grafts and reinforcement films. Subcellular transmural communication has been shown to prevent thrombus formation in PET grafts and also thought to play a critical role in fallout endothelialization [36]. There was complete lack of clinical signs of hind-limb ischaemia and the prominent collateral arterial supply found in all the occluded grafts, despite the ligation of all aortic branches, distal to the left renal artery to the aortic-iliac bifurcation at the time of surgery (The inferior mesenteric artery was always preserved). This would suggest that the occlusion progressed slowly allowing for capillary ingrowth within the organized clot to sufficiently supply the hind limbs [37].

Current clinically implanted ePTFE conduits have a reported IND of 30µm. Although there are advocates of porosity in moderation (30µm), for improved transmural growth [38], it is important to note that the IND of ePTFE does not equate to fixed diameter porosity of polyurethane sponges. ePTFE porosity is described by the IND but in reality, determined by the interconnecting fibril density, which is also heavily influenced by the wall thickness due to the random of fibril alignment. Thus, clinically implanted ePTFE grafts with wall thickness of 300µm are resistant to any transmural cellular ingrowth [3]. The external isolation segments of ePTFE used in this experimental model in the WG12 group, is also evidence of this.

16

The presence of abluminal surface clot between the PU and ePTFE wrap (WG12), was an anticipated finding due to the space between the two grafts at construction, but it was comforting to note that patency was not influenced. Despite the high concentration of growth factors that would exist in this platelet rich plasma collection [3], it had little effect on local endotheliazation as the mid-segment was isolated from all adjacent tissue. One can only speculate in the influence these circulating growth factors would have on systemic circulation, but there was no difference noted in the transanastomotic healing in this group.

High porosity allows for mural tissue ingrowth, which paradoxically prevents seepage of serous fluid, preventing seroma formation, which is a proposed complication of unwrapped ePTFE grafts. The high collagen content with FBG cells within the pores suggests that the foreign body capsule becomes integrated within the graft wall rather than surrounding it. Although this in-graft scar formation is not desirable, it may be solved in the future with the optimization of degradable polymers with similar porous macro-structure [4].

Patency between the CG12 and WG12 were similar despite significantly more endothelium on the CG12. Although this may appear counter-intuitive, the explanation lies in the exposed surface of the whole graft rather than the interposition segment. Every graft comprises of 90% ePTFE as isolation segments and only 10% of the test segment of polyurethane. Despite 80% endothelial coverage of the PU, this only comprises of 8% of the total graft coverage. This contact area is hardly enough to differentiate between endothelial covered and non-endothelialized grafts. To investigate the contribution of endothelium to patency, a loop comprised completely of PU would need to be compared as wrapped and non-wrapped groups. This would however be subject to TA endothelial ingrowth.

Whether in the form of capillary ingrowth, or merely by subcellular communication, optimization of endothelialization by transmural means has previously been documented with perigraft tissue coverage. External coverage with autologous vein [39] or meshed aorta

17

[40] has resulted improved healing, irrespective of the orientation of the vein wrap (Intimal vs. Adventitia) [17]. Covered stented grafts have also been shown to heal faster than surgically interposed grafts [41]. While omental wraps show improved synthetic graft endothelial coverage, an advantage is lost by devascularizing the omentum [42].

Despite these data, high-porosity grafts allowing for transmural ingrowth have not penetrated the clinical market. A single attempt to implant high porosity ePTFE (60µm) in humans [43] did not yield the same results seen in the baboon aorto-iliac model [22]. Few authors however, mention that the human study included an additional wrap on the porous graft simulating the WG12 in our current study and thus accounting for the discord between the baboon and human outcomes.

111

In-labelled platelet imaging was used as a surrogate for

endothelial surface coverage, and only excised segments of thrombosed grafts were analyzed by histology. Thus, the Kohler et al study is almost certainly the most misquoted paper on this subject of transmural healing, leading to numerous claims that TM growth will not occur in humans. It is thus our recommendation that the Kohler study should not be cited with reference to the lack of TM endothelialization in humans but rather that the prevention of TM growth by a wrap will result in less healing. Furthermore, it is essential that clinical studies with high porosity grafts be used to continue the research that lost its way more than 20 years ago.

The combination of significantly more mural vascularity together with increased surface endothelium in the CG12 suggests an association between the two, but does not give us any further insight into the mechanism of endothelialisation of the CG12 group. Previous authors have reported transmural luminal surface vessel breakthrough [22,44,45], which have confirmed continuity with the ablumen by latex microinfusion [22] as well as sequential histological cross-sectioning [22,46].

18

We describe four capillary ostia, with only one noted in the twelve-week group, which would suggest that ostia regress once endothelialization is complete. Although we were unable to correlate the presence on endothelial coverage with ostial breakthrough count, we do not agree with previous conclusions that this confirms lack of cause and effect [23]. The endothelium associated with ostia had a downstream pattern that would refute the proposal that the ostial origins are derived from fallout EPCs.

Despite intensive investigation of EPCs, the source of fallout endothelium remains elusive, and the lineage as well as the exact phenotype of cells responsible for fallout endothelialization is not yet known [47]. In vivo incorporation of endothelial-like cells differentiated from mononuclear blood cells extracted from healthy individuals was first described by Ashara et al [48] and has been subsequently confirmed [49]. In addition to CD34, early EPCs express AC133 (CD133) [50], and are generally characterized as AC133+/CD34+/VEGFR-2+/VE-cadherin+ [51]. CD34+ endothelial cells have been shown to be present in single stage bone marrow seeding [52], but have also been used to describe mature endothelium, which does not allow one to differentiate the source from transmural ingrowth. Furthermore, the expression of CD34 does not appear to be an absolute requirement for EPC identification and endothelial cell populations may also transdifferentiate from circulating monocytes [53].

To complicate matters further, bone marrow derived cells no longer express AC133 after differentiation [51]. This loss of AC133 on differentiation may be an explanation for the lack of AC133 expressing cells in any of our grafts. Despite this finding, AC133 has been documented on the surfaces of ventricular assist devices after many months of implantation in humans [54,55], with the only possible source being circulating cells. It is for this reason that we targeted this the immature hematopoietic stem cell marker AC133 as it appears to be a more reliable marker of EPCs [56]. The presence of endothelial islands on the WG12 mid-graft could only plausibly originate from circulating endothelial cells, but we were unable

19

to differentiate this form of healing through immunohistochemistry. However, the use of an external wrap is clear evidence that fallout endothelialization does occur, but in a clinically irrelevant amount.

Surface treatment with growth factors was anticipated to significantly increase the transmural capillary ingrowth with improved mid-graft endothelialization. This was based on previous data with equivalent porosity PU scaffold in the subcutaneous position [31]. The challenge in model design was to determine an implant timeframe that was short enough for transmural endothelialization to be absent on untreated samples, but sufficiently long for surface treatment to have an effect. Three weeks was chosen based on previously published data.[25] We have previously documented the binding of VEGF and PDGF to our heparinised porous polyurethane scaffolds with resultant increased neovascuarisation in a subcutaneous angiogenesis model, thus an intermediate group of heparin only treatment was implanted to determine the effect of heparin treatment alone.

The significant increase in mid-graft endothelialization was evident for coverage area as well as the categorical analysis for the presence of endothelium. Although no correlation between vessel ingrowth and endothelial overage was achieved, it appears logical that the apparent increase in mural vessels contributed to the surface endothelialization. Previous data has confirmed the positive influence of these growth factors on the mural angiogenesis in subcutaneous implants[27].

The ideal animal model to assess vascular graft healing does not exist [57] and cellular healing across species may mislead translational outcome. A further consideration is the age of the animal as this further confounds healing characteristics [3]. Small animal models are notoriously forgiving in both healing and patency, but their lower cost and higherthroughput ensure that they will remain the first line in vascular graft development. It is essential that one not only implement animal model appropriately for optimum outcome, but

20

are also able to interpret the findings of fellow researchers in the field. Adequate, clinically relevant lengths are essential when implemented in both small and large animal models and the application of isolation models may well have a place in large animals. Whilst the rat remains the dominant implant model for vascular graft we advocate a systematic implantation of experimental grafts in a straight and isolated loop model to assess mid-graft healing independent of the anastomosis, with implantation in a large animal model thereafter. We also caution against claims of mid-graft healing without transanastomotic isolation in grafts shorter than clinically relevant lengths.

5. Conclusions It is clear that transmural endothelialization is an independent form of healing that exists in the rat. The lack of adequate length and model design had blurred the lines between transanastomotic and other forms of spontaneous healing and limited progress in half a century of graft development.

The majority of publications describe experimental grafts of inadequate length to isolate the mid-graft from the anastomosis. By looping the graft, one can extend the low-porosity isolation segments to maintain broad endothelial-free zones for up to half a year. Transmural endothelialization can occur by two weeks, is consistently achieved at six weeks, and can be sped up with the addition of a heparin-growth factor surface treatment. By preventing perivascular ingrowth through a highly porous graft by external wrapping with low-porosity ePTFE, the mid-graft endothelialization is significantly reduced. Intimal hyperplasia does not increase in the mid-graft where transmural endothelialization occurs.

21

6. Acknowledgements The authors wish to thank Ms Helen Ilsley for histological processing and Mr Rodney Lucas of the UCT Faculty of Health Science’s Research Animal Facility for assistance with the animal study. Funding for this work was provided through the Competitive Program for Rated Researchers (CPRR) of the South African National Research Foundation (NRF), the South African Medical Research Council Inter-University Cape Heart Research Group (MRC-IUCHRG), and the German Heart Foundation. Opinions expressed and conclusions arrived at are those of the authors and are not necessarily to be attributed to the NRF.

7. Conflicts of interest None to Declare

8. REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12]

[13] [14] [15]

Y. Tomizawa, Vascular prostheses for aortocoronary bypass grafting: a review, Artif Organs. 19 (1995) 39–45. S. Tara, K.A. Rocco, N. Hibino, T. Sugiura, H. Kurobe, C.K. Breuer, T. Shinoka, Vessel Bioengineering, Circ J. (2014) 12–19. doi:10.1253/circj.CJ-13-1440. P. Zilla, D. Bezuidenhout, P. Human, Prosthetic vascular grafts: Wrong models, wrong questions and no healing, Biomaterials. 28 (2007) 5009–5027. doi:10.1016/j.biomaterials.2007.07.017. W. Wu, R.A. Allen, Y. Wang, Fast-degrading elastomer enables rapid remodeling of a cell-free synthetic graft into a neoartery, Nature Medicine. 18 (2012) 1148–1153. T. Pennel, G. Fercana, D. Bezuidenhout, A. Simionescu, T.-H. Chuang, P. Zilla, D. Simionescu, The performance of cross-linked acellular arterial scaffolds as vascular grafts; pre-clinical testing in direct and isolation loop circulatory models, Biomaterials. (2014). doi:10.1016/j.biomaterials.2014.04.062. X. Ren, Y. Feng, J. Guo, H. Wang, Q. Li, J. Yang, X. Hao, J. Lv, N. Ma, W. Li, Surface modification and endothelialization of biomaterials as potential scaffolds for vascular tissue engineering applications, Chem Soc Rev. 44 (2015) 5680–5742. doi:10.1039/c4cs00483c. N. L'Heureux, N. Dusserre, G. Konig, B. Victor, P. Keire, Human tissue-engineered blood vessels for adult arterial revascularization - Nature Medicine, Nature Medicine. (2006). K.R. Kidd, V.B. Patula, S.K. Williams, Accelerated endothelialization of interpositional 1-mm vascular grafts, J Surg Res. 113 (2003) 234–242. doi:10.1016/S0022-4804(03)00210-5. P. Verhamme, M.F. Hoylaerts, The pivotal role of the endothelium in haemostasis and thrombosis, Acta Clin Belg. 61 (2006) 213–219. doi:10.1179/acb.2006.036. A. de Mel, G. Jell, M.M. Stevens, A.M. Seifalian, Biofunctionalization of Biomaterials for Accelerated in Situ Endothelialization: A Review, 9 (2008) 2969–2979. doi:10.1021/bm800681k. A. Clowes, A. Gown, S. Hanson, M. Reidy, Mechanisms of arterial graft failure. 1. Role of cellular proliferation in early healing of PTFE prostheses, Am. J. Pathol. 118 (1985) 43. M. Deutsch, J. Meinhart, P. Zilla, N. Howanietz, M. Gorlitzer, A. Froeschl, A. Stuempflen, D. Bezuidenhout, M. Grabenwoeger, Long-term experience in autologous in vitro endothelialization of infrainguinal ePTFE grafts, Journal of Vascular Surgery. 49 (2009) 352–362. doi:10.1016/j.jvs.2008.08.101. K. Berger, L.R. Sauvage, A.M. Rao, S.J. Wood, Healing of arterial prostheses in man: its incompleteness, Ann. Surg. 175 (1972) 118. P. Zilla, D. Bezuidenhout, P. Human, Prosthetic vascular grafts: wrong models, wrong questions and no healing, Biomaterials. 28 (2007) 5009–5027. doi:10.1016/j.biomaterials.2007.07.017. M. Stump, G. Jordan, M. De Bakey, B. Halpert, Endothelium Grown from Circulating Blood on Isolated

22

[16] [17] [18] [19]

[20]

[21] [22] [23] [24] [25] [26] [27] [28]

[29] [30] [31] [32]

[33] [34] [35] [36] [37] [38] [39]

Intravascular Dacron Hub, Am. J. Pathol. 43 (1963) 361. Q. Shi, V. Bhattacharya, M. Hong-De Wu, L.R. Sauvage, Utilizing granulocyte colony-stimulating factor to enhance vascular graft endothelialization from circulating blood cells, Ann Vasc Surg. 16 (2002) 314–320. doi:10.1007/s10016-001-0238-x. Y. Onuki, N. Hayashida, M. Hong-De Wu, Q. Shi, Y. Kouchi, R. Ghali, L.R. Sauvage, Accelerated endothelialization model for the study of Dacron graft healing, Ann Vasc Surg. 11 (1997) 141–148. doi:10.1007/s100169900024. Y. Onuki, Y. Kouchi, H. Yoshida, M. Hong-De Wu, Q. Shi, L.R. Sauvage, Early Presence of Endothelial-like Cells on the Flow Surface of Porous Arterial Prostheses Implanted in the Descending Thoracic Aorta of the Dog, Ann Vasc Surg. 11 (1997) 604–611. doi:10.1007/s100169900098. J.I. Rotmans, J.M.M. Heyligers, H.J.M. Verhagen, E. Velema, M.M. Nagtegaal, D.P.V. de Kleijn, F.G. de Groot, E.S.G. Stroes, G. Pasterkamp, In Vivo Cell Seeding With Anti-CD34 Antibodies Successfully Accelerates Endothelialization but Stimulates Intimal Hyperplasia in Porcine Arteriovenous Expanded Polytetrafluoroethylene Grafts, Circulation. (2005). T.-Y. Kang, J.H. Lee, B.J. Kim, J.-A. Kang, J.M. Hong, B.S. Kim, H.J. Cha, J.-W. Rhie, D.-W. Cho, In vivo endothelization of tubular vascular grafts through in situ recruitment of endothelial and endothelial progenitor cells by RGD- fused mussel adhesive proteins, Biofabrication. 7 (2015) 1–11. doi:10.1088/1758-5090/7/1/015007. K. Berger, L.R. Sauvage, A.M. Rao, S.J. Wood, Healing of arterial prostheses in man: its incompleteness, Ann. Surg. 175 (1972) 118. A.W. Clowes, T.R. Kirkman, M.A. Reidy, Mechanisms of arterial graft healing. Rapid transmural capillary ingrowth provides a source of intimal endothelium and smooth muscle in porous PTFE prostheses, Am J Pathol. 123 (1986) 220–230. H. Tsuchida, S.E. Wilson, S. Ishimaru, Healing mechanisms of high-porosity PTFE grafts: significance of transmural structure, J Surg Res. 71 (1997) 187–195. H. Florey, S. Greer, J. Kiser, J. Poole, R. Telander, N. Werthessen, The development of the pseudointima lining fabric grafts of the aorta, Br J Exp Pathol. 43 (1962) 655–660. T. Pennel, P. Zilla, D. Bezuidenhout, Differentiating transmural from transanastomotic prosthetic graft endothelialization through an isolation loop-graft model, Journal of Vascular Surgery. 58 (2013) 1053– 1061. doi:10.1016/j.jvs.2012.11.093. S.S. Said, J.G. Pickering, K. Mequanint, Advances in growth factor delivery for therapeutic angiogenesis, Angiologica. 50 (2013) 35–51. doi:10.1159/000345108. N.H. Davies, C. Schmidt, D. Bezuidenhout, P. Zilla, Sustaining neovascularization of a scaffold through staged release of vascular endothelial growth factor-A and platelet-derived growth factor-BB, 18 (2012) 26–34. doi:10.1089/ten.tea.2011.0192. M. McLuckie, C.A. Schmidt, A. Oosthuysen, N. Sanchez-Macedo, H. Merker, D. Bezuidenhout, S.P. Hoerstrup, N. Lindenblatt, High heparin content surface modified polyurethane discs promote rapid and stable angiogenesis in full thickness skin defects through VEGF immobilization, (2017) 1–29. doi:10.1002/jbm.a.36108. D. Bezuidenhout, N. Davies, P. Zilla, Effect of well defined dodecahedral porosity on inflammation and angiogenesis, Asaio J. 48 (2002) 465–471. Medtronic, Inc., Biostable, segmented aliphatic polyurethanes and processes for their preparation, EP0361950B1, 1989. D. Bezuidenhout, N. Davies, M. Black, C. Schmidt, A. Oosthuysen, P. Zilla, Covalent surface heparinization potentiates porous polyurethane scaffold vascularization, Journal of Biomaterials Applications. 24 (2010) 401–418. doi:10.1177/0885328208097565. W.S. Choi, Y.K. Joung, Y. Lee, J.W. Bae, H.K. Park, Y.H. Park, J.-C. Park, K.D. Park, Enhanced Patency and Endothelialization of Small-Caliber Vascular Grafts Fabricated by Coimmobilization of Heparin and Cell-Adhesive Peptides, ACS Appl. Mater. Interfaces. 8 (2016) 4336–4346. doi:10.1021/acsami.5b12052. T. Pennel, P. Zilla, D. Bezuidenhout, Differentiating transmural from transanastomotic prosthetic graft endothelialization through an isolation loop-graft model, Journal of Vascular Surgery. 58 (2013) 1053– 1061. doi:10.1016/j.jvs.2012.11.093. R.A. White, The effect of porosity and biomaterial on the healing and long-term mechanical properties of vascular prostheses, 34 (1988) 95–100. S. Wesolowski, C. Fries, K. Karlson, M. Debakey, P. Sawyer, Porosity: Primary determinant of ultimate fate of synthetic vascular grafts, Surgery. 50 (1961) 91–96. Z. Zhang, S. Briana, Y. Douville, H. Zhao, N. Gilbert, Transmural communication at a subcellular level may play a critical role in the fallout based-endothelialization of dacron vascular prostheses in canine, J Biomed Mater Res A. 81A (2007) 877–887. doi:10.1002/jbm.a.31124. T. Okoshi, G. Soldani, M. Goddard, P. Galletti, Very small-diameter polyurethane vascular prostheses with rapid endothelialization for coronary artery bypass grafting, J. Thorac. Cardiovasc. Surg. 105 (1993) 791. A.J. Marshall, C.A. Irvin, T. Barker, E.H. Sage, K.D. Hauch, B.D. Ratner, Biomaterials with tightly controlled pore size that promote vascular in-growth, Polymer Preprints, American Chemical Society, Division of Polymer Chemistry. 45 (2004) 100–101. D.A. Bull, G.C. Hunter, H. Holubec, M.L. Aguirre, W.D. Rappaport, C.W. Putnam, Cellular origin and

23

[40] [41]

[42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52]

[53] [54] [55] [56] [57]

rate of endothelial cell coverage of PTFE grafts, J Surg Res. 58 (1995) 58–68. doi:10.1006/jsre.1995.1010. K.M. Sedlarik, P.B. van Wachem, H. Bartels, J.M. Schakenraad, Rapid endothelialization of microporous vascular prostheses covered with meshed vascular tissue: a preliminary report, Biomaterials. 11 (1990) 4–8. M.P. Ombrellaro, S.L. Stevens, J. Sciarrotta, D. Schaeffer, M.B. Freeman, M.H. Goldman, Effect of intra-arterial environment on endothelialization and basement membrane organization in polytetrafluoroethylene grafts, The American Journal of Surgery. 174 (1997) 29–32. doi:10.1016/S0002-9610(97)00049-4. A.V. Sterpetti, W.J. Hunter, R.D. Schultz, C. Farina, Healing of high-porosity polytetrafluoroethylene arterial grafts is influenced by the nature of the surrounding tissue, Surgery. 111 (1992) 677–682. T.R. Kohler, J.R. Stratton, T.R. Kirkman, K.H. Johansen, B.K. Zierler, A.W. Clowes, Conventional versus high-porosity polytetrafluoroethylene grafts: clinical evaluation, Surgery. 112 (1992) 901–907. R.K. Zacharias, T.R. Kirkman, A.W. Clowes, Mechanisms of healing in synthetic grafts, Journal of Vascular Surgery. 6 (1987) 429–436. doi:10.1016/0741-5214(87)90300-4. M.A. Golden, S.R. Hanson, T.R. Kirkman, P.A. Schneider, A.W. Clowes, Healing of polytetrafluoroethylene arterial grafts is influenced by graft porosity, Ymva. 11 (1990) 838–44– discussion 845. M.H. Wu, Q. Shi, Y. Onuki, Y. Kouchi, L.R. Sauvage, Histologic observation of continuity of transmural microvessels between the perigraft vessels and flow surface microostia in a porous vascular prosthesis, Ann Vasc Surg. 10 (1996) 11–15. C. Urbich, S. Dimmeler, Endothelial progenitor cells: characterization and role in vascular biology, Circulation Research. 95 (2004) 343–353. doi:10.1161/01.RES.0000137877.89448.78. T. Asahara, T. Murohara, A. Sullivan, M. Silver, R. van der Zee, T. Li, B. Witzenbichler, G. Schatteman, J.M. Isner, Isolation of putative progenitor endothelial cells for angiogenesis, Science. 275 (1997) 964– 967. V. Bhattacharya, P.A. McSweeney, Q. Shi, B. Bruno, A. Ishida, R. Nash, R.F. Storb, L.R. Sauvage, W.P. Hammond, M.H.-D. Wu, Enhanced endothelialization and microvessel formation in polyester grafts seeded with CD34(+) bone marrow cells, Blood. 95 (2000) 581–585. A.H. Yin, S. Miraglia, E.D. Zanjani, G. Almeida-Porada, M. Ogawa, A.G. Leary, J. Olweus, J. Kearney, D.W. Buck, AC133, a Novel Marker for Human Hematopoietic Stem and Progenitor Cells, …. (1997). R. Handgretinger, S. Kuçi, CD133-Positive Hematopoietic Stem Cells: From Biology to Medicine, in: D. Corbeil (Ed.), Prominin-1 (CD133): New Insights on Stem & Cancer Stem Cell Biology, Springer New York, New York, NY, 2012: pp. 99–111. doi:10.1007/978-1-4614-5894-4_7. T. Shinoka, G. Matsumura, N. Hibino, Y. Naito, M. Watanabe, T. Konuma, T. Sakamoto, M. Nagatsu, H. Kurosawa, Midterm clinical result of tissue-engineered vascular autografts seeded with autologous bone marrow cells, J. Thorac. Cardiovasc. Surg. 129 (2005) 1330–1338. doi:10.1016/j.jtcvs.2004.12.047. S.M. Scott, M.G. Barth, L.R. Gaddy, E.T. Ahl, F.T.D.O. Surgery, Pathology, Laboratory Medicine, Veterans Administration Medical Center, Asheville, Duke University Medical Center, Durham., The role of circulating cells in the healing of vascular prostheses, Ymva. 19 (1994) 585–593. O.H. Frazier, R.T. Baldwin, S.G. Eskin, J.M. Duncan, Immunochemical identification of human endothelial cells on the lining of a ventricular assist device, Tex Heart Inst J. 20 (1993) 78–82. S. Rafii, M.C. Oz, J.A. Seldomridge, B. Ferris, A.S. Asch, R.L. Nachman, F. Shapiro, E.A. Rose, H.R. Levin, Characterization of hematopoietic cells arising on the textured surface of left ventricular assist devices, Ats. 60 (1995) 1627–1632. doi:10.1016/0003-4975(95)00807-1. M. Peichev, A.J. Naiyer, D. Pereira, Z. Zhu, W.J. Lane, M. Williams, M.C. Oz, D.J. Hicklin, L. Witte, M.A. Moore, S. Rafii, Expression of VEGFR-2 and AC133 by circulating human CD34(+) cells identifies a population of functional endothelial precursors, Blood. 95 (2000) 952–958. M.J. Byrom, P.G. Bannon, G.H. White, M.K.C. Ng, Animal models for the assessment of novel vascular conduits, Ymva. 52 (2010) 176–195. doi:10.1016/j.jvs.2009.10.080.

24

List of Figures Figure 1 (A) Schematic representation of the three forms of vascular graft healing with high porosity conduit (purple arrow) with pore interconnectivity (orange arrow). (B,C) SEM of isolation segment (Zeus® 15-25µm IND) at (B) x100 and (C) x750. (D) x50 and (E) x150 magnification surface of high porosity PU (156±2µm) with pore interconnectivity of 70µm . Scale bars Black = 100µm, White= 500µm

Figure 2 Construction of loop graft: (A) The nylon cord is tightly wound around the PTFE mandrel and secured with cable ties. (B) Following the heat-setting water bath the nylon retains memory of the spiral form cut to lengths of ±110mm. (C) The mid-graft and isolation segments are aligned on the cord, glued and secured with a single tie. (D) Finally the mandrill is removed prior to ethanol submersion. Construction of the externally wrapped grafts. (E) ePTFE sheath is aligned over the PU mid-graft. (F) The edges are glued and crimped to prevent ingrowth from the side. The dashed line represents the PU mid- graft. (G) SEM x15 of the internal and external surface of the SG graft. Black Scale bar 10mm, White Scale bar 1mm Figure 3 (A) Schematic reaction scheme for the Ce(IV) initiated graft polymerization of acrylic acid/acrylamide (Aac/Aam) copolymer onto the surface of porous polyurethane grafts, followed by amination with 1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide activated coupling of ethylenediamine (EDA), and end-point attachment of nitrous acid degraded heparin (containing terminal aldehyde). (B and C) show micrographs of the unmodified (B) and heparinized (C) scaffolds respectively (350x orig. mag. Scale bar = 50µm). The presence of the undulating/cracked heparinized layer (caused by dehydration of the grafted hydrogel) is noted in (C).

Figure 4 (A) Construction of the mid-graft barrier wraps and skins to prevent abluminal perigraft tissue ingrowth. The left is a schematic representation of the implant design. The middle column is a low magnification SEM of the mid-graft with external treatment. The right column depicts a high magnification SEM of the external surface of the mid-graft and is representative of the degree of porosity. Scale bars: White = 500µm, Black = 50µm Implant of (B) pCG and (C) WG (D) SG in the Infra renal position. Figure 5 (A-H) Histological cross-section of the pPU mid-graft (A-C,G,H) CG (D-F) WG, (G,H) CD31 stain of occluded pCG graft with collateralization of mid-graft. SG excluded due

25

to 100% occlusion. (C,F) The immunohistochemistry is representative of the documented graft coverage. (C) CD31 (red), Dapi(blue), Actin(green) (F) AC133 (red), Dapi (blue). Scale bar= 200µm (I) Mid-graft endothelial coverage with graft patency overlay (red) (⌘) Denotes statistical significance for endothelial coverage, (⦿)denotes statistically significance for patency Macrophotograph with digital subtraction of the background of longitudinal section and cross section of (J) CG (K) WG (L) SG. (✽) Denotes the thick fibrous capsule around the mid-graft of WG Figure 6 (A) Vessel Number(black) and area(white), (B) vessel number indexed as a percentage off the total graft area for the three implant groups as well as the sub groups analysis. (C-E) Representative CD31 light stains of the three treatment groups. Scale bar= 200µm Figure 7 SEM of mid-graft surface. (A) Blue arrows indicate indentations on the surface, which may represent previous capillary openings. (B & C) Red arrows are confirmed capillary openings and also represent the direction of blood flow. (D) CD31 (Dapi) immunefluorescence of corresponding mid-graft. (E) SEM representing a capillary sprout through the graft wall onto the luminal surface with downstream endothelial outgrowth, Green dashed arrow represents the direction of blood flow and outgrowth of endothelium from the ostium. Scale bar = 100µm List of Tables

Table 1: Listing of graft codes, implant periods and number of replicates Table 2: Summary of implant groups: (1A-C) 12-week study to determine the effect of abluminal isolation on transmural endothelialization, and (2A-C) 3-week study to evaluate the potential of growth factor to accelerate and enhance vascularization and endothelialization.

26

27

28

29

30

31

32

33

Table 1: Listing of graft codes, implant periods and number of replicates

Label

Midgraft Modification

Implant time

N

(Weeks)

1A

CG12

None

12

12

1B

WG12

3.5mm ID ePTFE wrap

12

12

1C

SG12

PU external seal

12

6

2A

CG3

None

3

10

2B

HG3

Heparin Surface Modification

3

10

2C

GF3

Heparin+Growth Factor* Surface Modification

3

10

* Growth Factor (VEGF + PDGF)

34

Table 2: Summary of implant groups: (1A-C) 12-week study to determine the effect of abluminal isolation on transmural endothelialization, and (2A-C) 3-week study to evaluate the potential of growth factor to accelerate and enhance vascularization and endothelialization.

Label

Implant

Patency

time

Endothelial Endothelial

Surface

coverage

Present

Ostia

(Weeks)

1A

Control Graft [CG12]*

12

9/12 (75%)

80±20%

9/9(100%)

1

1B

Wrapped graft [WG12]

12

10/12 (83%)

16±30%

3/10(30%)

0

1C

Sealed graft [SG12]

12

0/6 (0%)

0%

NA

NA

2A

Control Graft [CG3]*

3

10/10 (100%)

2.5±6%

2/10(20%)

1

2B

Heparinized Graft [HG 3]

3

10/10 (100%)

0%

0/10(0%)

0

2C

Growth Factor graft [GF3]

3

10/10 (100%)

13.5±20%

3/10(30%)

3

* identical control grafts (CG) used as positive control in abluminal sealing study and negative control in improved angiogenesis section.

35

Statement of Significance In addition to the full elucidation of, and differentiation between, the three healing/endothelialisation modes of vascular grafts, the significance of the work relates to the near-complete lack of endothelialisation of small diameter vascular grafts in humans (12cm transanastomotic outgrowth on a graft that may be 60cm long) even after decades of implantation. The concomitant retained midgraft thrombogenicity leads, together with anastomotic hyperplastic responses, to poor long-term outcomes. The large impact of successful translation of the current research to the achievement of full endothelialisation of long peripheral grafts in humans via transmural ingrowth (half a millimetre distance; thickness of the graft wall), is evident, and supported by the large improvements in clinical patencies achievable in by pre-seeding of ePTFE grafts with confluent endothelia.

36