Host response to autologous endothelial seeding☆☆☆★
Article Outline
Abstract
Dacron and expanded polytetrafluoroethylene grafts, seeded with autologous venous endothelial cells at the time of implantation, subsequently develop endothelial linings. However, it has not been shown whether the endothelial cells in these linings are derived from the seeded cells, or whether the seeding process itself stimulates host endothelial cells to proliferate and cover the grafts. As a first step in testing this hypothesis, bilateral end-to-side aortoiliac expanded polytetrafluoroethylene grafts with an internal diameter of 6 mm, an internodal distance of 22 μm, and an average length of 9.2 cm were placed in 10 adult mongrel dogs weighing 20 to 25 kg; the aorta was ligated just distal to the origin of the grafts. The graft on one side, chosen at random, was seeded with autologous endothelium that was harvested by enzyme single-stage technique from external jugular veins; the other side was not seeded. After 4 weeks the animals were anesthetized and heparinized, and the grafts were fixed by perfusion in vivo with 2.5% glutaraldehyde solution before they were removed. Both grafts occluded in two animals, and both grafts were patent in five animals. In two animals the seeded grafts were open and the unseeded grafts were occluded. In one the seeded graft was occluded and the unseeded graft was patent. There was no significant difference in clot-free surface area between seeded (29% ± 18%) and unseeded (31% ± 11%) grafts. Scanning electron microscopy showed the presence of an endothelial monolayer that averaged 39% ± 20% and 36% ± 26% coverage, respectively, in the clot-free midgraft portions of all seeded and unseeded patent grafts. In addition, transmission electron microscopy revealed endothelial cells under a thin adherent layer of fibrin and red cells, seen by light microscopy over an average of 35% of the surface area. The lack of differences between seeded and unseeded grafts in patency and clot-free surface, and the extent and distribution of the endothelial coverage in unseeded grafts after only 4 weeks implantation, provide supportive, although inconclusive, evidence that the seeding process may be more important as a stimulus for growth of host endothelium than as a vehicle providing cells to line grafts. (J Vasc Surg 1989:9:656–64.)
The demonstration that endothelial linings of synthetic vascular grafts could be produced in dogs by “seeding” autologous cells onto the lumen of the grafts immediately before implantation stimulated a variety of studies related to this subject. In addition to those studies about the makeup and stability of the lining, many studies concern the resistance to thrombosis, which is presumed to result from such a lining; and under rigid experimental conditions, certain benefits have been shown. However, one issue that has had little attention is whether the seeding process provides the cells that ultimately compose the lining of such grafts, or whether it acts as a stimulus to host endothelium to proliferate and cover the grafts. This study was done as an attempt to provide initial data concerning this subject.
Material and methods
The dog was chosen as the experimental animal because of the similarities of endothelial response to grafting between dog and man, and also because many previous “seeding” studies, including the original ones, were in dogs. Clinical expanded polytetrafluoroethylene (ePTFE) was chosen as the experimental graft because it has the least spontaneous coverage with endothelium, because there is minimal transmural migration of multipotential cells, and because the linings that develop after seeding do not undergo progressive thickening, as they do in Dacron and porous PTFE. Antiplatelet agents were not used because we wanted to minimize pharmacologic manipulations and because we expected adequate numbers of patent grafts in the aortoiliac system.
Animal care and handling complied with all aspects of the “Principles of Laboratory Animal Care” and the “Guide for the Care and Use of Laboratory Animals” (NIH Publication No. 80-23, revised 1985). Ten adult mongrel dogs weighing 20 to 25 kg were anesthetized with pentobarbital sodium (25 mg/kg), were intubated, and were mechanically ventilated by means of a constant-volume ventilator. Both external jugular veins in each dog were harvested through bilateral longitudinal neck incision by use of sterile technique. After primary closure of these incisions, the dogs were systemically heparinized (100 U/kg), and bilateral ePTFE grafts (W. L. Gore and Associates, Elkton, Md.) were placed from the sides of the infrarenal aorta to the sides of the external iliac arteries (Fig. 1).
Fig. 1.
Diagram of animal model used shows bilateral end-to-side grafts of aorta to external iliac artery. Graft material was ePTFE with 6 mm internal diameter and internodal distances of 22 μm. Average graft length was 9.2 cm.
Each pair of ePTFE grafts was obtained from a single length of graft material that had been preclotted with 10 ml of unheparinized blood drawn from the cephalad portion of an external jugular vein. This was accomplished by forcing the blood under pressure into the graft, with one end occluded until the blood leaked through the interstices, and then removing excess clot with a Fogarty embolectomy catheter. The original graft was then divided into equal lengths; and one side, chosen at random, was inoculated with a suspension of endothelial cells. The inoculum was introduced into the graft in four aliquots, as described by Kesler et al.1 Before each injection the graft was rotated through 90 degrees in a clockwise direction, and each aliquot was left in the graft for 10 minutes. The control graft was simultaneously inoculated with tissue culture medium 199 (M-199) without endothelium.
The suspension of endothelial cells was obtained from both external jugular veins by an enzymatic single-stage technique,2 which involved everting the veins on a 16-gauge catheter and exposing the intimal surface to 0.2% Worthington type II collagenase in phosphate-buffer saline for 20 minutes. These veins were then irrigated with a steady stream of M-199. The loosened cells were collected and concentrated by centrifugation and then resuspended in 6 ml of M-199, 4 ml of which was used to inoculate the experimental graft. The remaining endothelial cell suspension from each animal was placed in 35 mm tissue culture plates. After cells had grown to confluence, the purity of the endothelial culture was confirmed by staining for factor VIII-related antigen by means of the peroxidase-antiperoxidase method adapted from Baughman et al.3
No special measures were taken to prevent migration of the endothelial cells from the seeded to the unseeded graft. The unseeded grafts, after inoculation with cell culture medium containing no endothelial cells, were installed first in all cases; no flushing was done at the completion of the anastomoses. The seeded grafts were placed as soon as the anastomoses for unseeded grafts were complete; the inoculation was done during installation of unseeded grafts so that no contamination with cells and no mixup of solutions were possible. No prograde or retrograde arterial flow was allowed until both grafts were in place. At that time the aortic clamp was removed, and blood flowed simultaneously into the proximal anastomoses of both grafts. Although the aorta was ligated just below the origin of both grafts, no ligatures were placed on the iliac arteries to prevent washout of cells from the seeded side to the unseeded side.
Four weeks after implantation (26 to 37 days), the animals were anesthetized with pentobarbital sodium and heparinized (100 U/kg), and the grafts were fixed by perfusion in situ with 2.5% glutaraldehyde solution in cacodylate buffer. The aorta was occluded cephalad to the proximal graft anastomoses and cannulated with a 20-gauge Angiocath (Critikon, Inc., Tampa, Fla.) at a point between the aortic cross-clamp and the anastomoses. Small transverse incisions were then made in both external iliac arteries beyond the distal graft anastomoses, and these arteries were occluded immediately proximal to these anastomoses. The glutaraldehyde solution was introduced under pressure through the Angiocath, and thus through both graft lumina. Flow was continued until the effluent from the iliac arteriotomies was clear; the endothelial linings were fixed on the prosthetic surfaces in situ. Next, the grafts were excised and opened longitudinally. The luminal surfaces were photographed, and the grafts were then immersed in 2.5% glutaraldehyde solution for subsequent histologic studies.
The percentage of clot-free surface area was determined by planimetric measurement of each graft from projections of the color transparencies on quadrille paper. For histologic studies the grafts were bisected longitudinally; half were used for light microscopy and half for electron microscopy. Samples designated for light microscopy were removed from the fixative, dehydrated in ethanol, and embedded in JB-4 plastic (Polysciences Inc., Warrington, Pa.). After they were hardened and desiccated, sections were cut with a Sorvall (DuPont Co., Wilmington, Del.) microtome, placed on a slide, fixed with heat, and stained with filtered toluidine blue.
The half of each graft designated for electron microscopy study was dried to the critical point, coated with gold-palladium, and examined by scanning electron microscopy to determine what percentage was covered by endothelium. Samples for both light and electron microscopy were taken of all grafts near each anastomosis and in the midportion of the graft to provide representative sections of the entire graft for analysis. To determine if endothelium was present beneath the areas covered by adherent red blood cells, platelets, and fibrin, selected samples were embedded in Epon (Ladd Research, Burlington, Vt.) for transmission electron microscopy.
Results
Two animals had both grafts occluded, and five had both grafts patent. Of the remaining three animals, two had the seeded side patent and the unseeded side occluded. The other had a patent unseeded graft and an occluded seeded one. Cumulative patency rates were 70% for the seeded grafts and 60% for the unseeded grafts. This difference was not statistically significant.
Three distinct patterns of surface healing were noted on gross examination of the grafts: (1) clear thrombus-free areas, (2) areas with gross thrombus, and (3) areas with adherent fibrin, platelets, and red blood cells (Fig. 2).
Fig. 2.
Specimen photograph shows three patterns of graft surface healing: A, gross thrombus; B, region of adherent fibrin, platelets, and red blood cells; C, a clear thrombus-free area.
Light microscopic examination of the anastomotic and midgraft regions did not show a significant difference between seeded and unseeded grafts in terms of the presence of an endothelial monolayer, adherent red cells, or intimal hyperplasia (Tables I and II).
Table I. Graft characteristics by light microscopy in the region of the graft anastomoses
| Seeded (n = 14) | Unseeded (n = 12) | |
|---|---|---|
| Endothelial cell monolayer | 11 (79%) | 11 (92%) |
| Adherent red blood cells | 5 (35%) | 6 (50%) |
| Hyperplasia | 4 (29%) | 4 (33%) |
Table II. Graft characteristics by light microscopy in the midportion of each graft
| Seeded (n = 7) | Unseeded (n = 6) | |
|---|---|---|
| Endothelial cell monolayer | 5 (71%) | 5 (83%) |
| Adherent red blood cells | 5 (71%) | 6 (100%) |
| Hyperplasia | 0 (0%) | 0 (0%) |
Scanning electron microscopic examination of the graft surfaces confirmed the presence of a monolayer of endothelial cells, with the characteristic cobblestone pattern, nuclear bulge, and well-defined intercellular borders present in both seeded and unseeded grafts. Some areas of both experimental and control grafts were lined by a layer of fibrin, adherent platelets, and red cells, whereas a few areas showed the nodes and fibrils of the ePTFE surface. The percentage of experimental and control graft surfaces covered by a monolayer of endothelium determined by scanning electron microscope measurement was 39% ± 20% and 36% ± 26%, respectively, and this difference was not significant (Figs. 3 and 4).
Fig. 3.
Scanning electron micrograph from the midportion of a patent seeded graft shows a confluent monolayer of endothelial cells. (Original magnification × 1200.)
Fig. 4.
Scanning electron micrograph from the midportion of a patent unseeded control graft shows a confluent monolayer of endothelial cells, essentially identical to that seen in Fig. 3. (Original magnification × 1200.)
Transmission electron micrographs of the areas covered with the fibrin and red cells showed that beneath this layer there was a monolayer of endothelium lining the surface of the graft. These cells exhibited the characteristic nuclear, cytoplasmic, and junctional features of endothelium (Figs. 5 and 6).
Fig. 5.
Transmission electron micrograph from the midportion of a patent seeded graft. Nucleus of an endothelial cell is visible under the overlying fibrin and red blood cells. (Original magnification × 7700.)
Fig. 6.
Transmission electron micrograph from the midportion of a patent seeded graft shows characteristic endothelial cell junction under overlying fibrin and red blood cells. (Original magnification × 5675.)
Discussion
The lack of differences in patency between seeded and unseeded grafts 6 mm in internal diameter was not surprising in the high-flow, high-pressure aortoiliac system. Studies showing differences in patency have nearly all been done in smaller grafts, usually 4 mm internal diameter, and have involved the use of antiplatelet agents started at or before operation.
The amount of endothelial coverage in seeded grafts at 4 weeks was greater than anticipated because endothelial cells were identified in all areas of the grafts: in clot-free surface, beneath a thin layer of adherent fibrin, and even beneath gross thrombus (Fig. 7).
Fig. 7.
Scanning electron micrograph from the midportion of a patent seeded graft. Characteristic cobblestone pattern and nuclear bulge of the endothelial cell monolayer is visible, even under a thin layer of fibrin. (Original magnification × 600.)
However, in contrast to all previous work in dogs, this study showed no significant differences in endothelial coverage between seeded and unseeded grafts. For example, the original article by Herring et al.6 found an average of 44% more clot-free surface (76.4% vs 32.4%) in 6 cm lengths of knitted Dacron grafts, 6 mm in internal diameter.6 They used one graft per animal interposed end to end between the infrarenal and the terminal aorta. Using a much longer segment of graft, 6 mm knitted Dacron in the thoracoabdominal position, Graham et al.7 found an average of 48.4% more clot-free surface in seeded grafts between 14 and 28 days. Using the same preparations, but with ePTFE grafts, they subsequently found dramatic differences in clot-free surface between seeded and unseeded grafts: 64% to 91% at 2 and 4 weeks, respectively, in seeded grafts versus less than 10% at both times in unseeded grafts. Some grafts in this latter study were 6 mm in internal diameter, and others were 10 mm.4
In a longitudinal study of the cellular events of healing, Herring et al.5 compared ePTFE with knitted Dacron using short, straight grafts in the infrarenal position. Small numbers of grafts were studied at a variety of intervals, up to 7 months after implantation; consequently, it was difficult to make many statistical comparisons. However, it was clear that seeded grafts provided significantly greater coverage of the surface with endothelium than did unseeded ones. In addition, there was minimal endothelial coverage of unseeded ePTFE at 30 days; in fact, only one of three grafts had any identifiable endothelial cells except immediately adjacent to the anastomoses. This finding was consistent with the studies by Graham et al.4 who found endothelial coverage limited to pannus ingrowth and “never exceeding 10% of the graft surface.” Furthermore, these studies also had remarkable agreement concerning the amount of pannus ingrowth from anastomoses in ePTFE; there was less than 5 mm ingrowth by 30 days.
Allen et al.8 reported an experimental study of patency of Dacron grafts using seeded and unseeded grafts in the same animals. Four grafts were placed in each animal: One seeded and one unseeded graft were placed in the carotid positions, and one seeded and one unseeded graft were placed in the femoral positions (all as interposition grafts). Unseeded grafts explanted at 1 month showed pannus ingrowth only, with no endothelial coverage of the midgraft segments, whereas seeded grafts showed confluent endothelial coverage. Kempczinski et al.7 also studied seeded and unseeded carotid interposition grafts in the same animals; they used an experimental, porous ePTFE graft. Except for pannus ingrowth, endothelial cells were not seen on the luminal surface of control grafts, even in thrombus-free areas.
In the present study the average endothelial coverage of unseeded grafts was 36% at 1 month after implantation; and endothelial cells could be identified by scanning electron microscopy at the midgraft in all patent unseeded grafts. These findings, which are in marked contrast to those in unseeded grafts in all previous studies, were obviously surprising to us, and there are several possible explanations for this difference. One explanation is technical differences in examination or interpretation of histologic data, but reexamination of the midportion of ePTFE grafts from the previous study by Herring et al.5 of the events of graft healing confirmed the lack of endothelial cells at the midportions of grafts harvested between 15 and 45 days after implantation.*Another possible explanation is that cells from the seeded graft spilled over into the unseeded graft and attached; thus they produced an endothelial lining. Whereas one cannot rule out such an explanation, it seems like an amazing feat when one considers that fewer than 5% of the harvested cells remain on the seeded graft after 24 hours. Furthermore, it seems that the other studies of seeded and unseeded grafts in the same animals would have found sporadic midgraft endothelialization, something that has not been reported to date in ePTFE grafts. Consequently, it seems more likely that some facet of the experimental model is responsible.
One difference in the model is that no antiplatelet agents were used. However, it has been shown that these agents in the usual doses do not adversely affect endothelial seeding.10 It seems more likely that some aspect of the seeding process is mitogenic. To expect 5% of cells that have been harvested by damaging enzymes to set up housekeeping on a hydrophobic surface, to reproduce enough to cover the surface, and then to resume physiologic function and prevent clotting seems a little far-reaching. However, it is clear that seeding is followed by endothelialization; whether the seeded cells are the same as those found on the luminal surface of the graft a month later is still not known. Whereas studies of cells with genetic markings are being carried out, none has been reported at the time of this writing.
An obvious difference in the model used in the study of Burkel et al.10 is the hemodynamics of end-to-side flow in the high-pressure and high-flow aortoiliac system. That such forces affect endothelial linings is shown by the greater association of intimal hyperplasia with end-to-side anastomoses compared to end-to-end anastomoses. Furthermore, endothelium is known to produce a humoral agent, endothelium-derived relaxing factor, which induces relaxation of underlying smooth muscle in response to increases in shear stress caused by sudden increases in blood velocity.11 This agent is probably only one of several yet-to-be-described factors that influence endothelial cell activity. Of course it is purely speculative to suggest that such agents responding to shear forces may stimulate host endothelial cells to line grafts; but it is probably as plausible as suggesting that a sparse seeding of cells, after being subjected to enzymatic harvest, centrifugation, resuspension, and sudden high shear stress resulting from resumption of arterial flow, produces confluent endothelial coverage. The only other experiment with hemodynamics similar to ours was that of Graham et al.,12 but seeded and unseeded grafts were not placed in the same animal. Furthermore, the grafts were 4 mm in internal diameter; and six of eight unseeded grafts were occluded at 4 weeks, with the others being clotted at 16 weeks.
Since humans appear to be the only mammals that do not spontaneously develop an endothelial surface for vascular grafts, one can speculate that something about the process of grafting is inhibiting them from responding normally. Dogs appear to have only a partial inhibition of this response; and it can be overcome by endothelial seeding. We believe that these studies provide preliminary evidence that the seeding process acts in this manner, rather than acting as an autologous carpet to cover the bare graft surface.
Acknowledgements
The authors acknowledge the support of W. L. Gore and Associates, Elkton, Md., and the assistance of Beth Raper in performing histologic analysis.
Discussion
Dr. James C. Stanley (Ann Arbor,Mich.). I am surprised that the patency rates were 60% and 70% for a 6 mm conduit that was less than an average of 10 cm long. Although the difference may be only one or two animals in the small group, I would have expected patency rates of 80% or 90%. However, I am not surprised that you saw endothelialized areas in your unseeded grafts with that particular model. This is not an unprecedented finding, and we have three Dacron iliofemoral bypass grafts that developed complete endothelialization in our own laboratory despite the fact that they were unseeded. We initially felt that this phenomenon might be normal in dogs, if not in humans, but there are other ways to explain it without trying to evoke some peculiar endothelial-derived growth factor.
For example, there may be spillover from one iliac graft to the other at the time the clamps are removed. Detached cells may recirculate and, like labeled platelets in so many other studies, drop down on that thrombogenic surface inside the contralateral unseeded graft. It is difficult for me to believe that endothelial-derived growth factor will really influence a graft in another area when it takes a while even to have a significant effect on the growth of cells in culture media. There are a number of examples in the literature of seeded grafts on one side and unseeded ones on the other, but none of these studies has reported crossover endothelialization or patency rates approaching those described in your report. They all have had a “black-and-white” experience: Unseeded grafts had nothing except pannus ingrowth, whereas seeded grafts were close to 90% covered with endothelium in the same animal.
This argument may become moot very quickly. Using common DNA technology one may tag endothelial cells with biologic markers before they are seeded onto a graft. I am aware of at least one laboratory in which recent work has shown complete coverage of implanted grafts with cells having the same genetic tag as those that were used to seed it. This is clear evidence that progeny of the seeded cells are the ones responsible for endothelialization, not host cells that somehow happened to grow in. I was interested in the results presented today, but I am concerned about the bilateral iliofemoral or aortoiliac model because of the possibility of spillover or the adherence of circulating seeded cells to the other side.
Dr. Glover. Dr. Stanley indicated that 80% to 90% patency would be expected with this seeded model. In one sense it was 80% in terms of the whole animal because eight of the 10 dogs had patent grafts. One may argue that 80% to 90% of the grafts on each side ought to be patent, but I would not agree with that. Nevertheless, it should be emphasized that our unprecedented observations were made in ePTFE, not in Dacron. There probably is a significant component of transmural migration of multipotential cells that contributes to the formation of endothelium in Dacron, but not in ePTFE, and that is why we chose this model.
Spillover may be a factor, and if it is, I want to know about it. Biologic tagging may be one approach to clarifying this issue, but in the meantime I am not uncomfortable invoking some unusual factors. Humans are unusual in this situation because they do not endothelialize grafts, whereas so many other animals seem to do just that.
Dr. Malcolm B. Herring (Indianapolis, Ind.). I think that Dr. Stanley raised several pertinent questions. We know that growth factors are elaborated by endothelial cells, but even if the replicating cells on the seeded side would enhance cellular coverage on the unseeded side, those cells have to come from someplace. We also know from Richard Kempczinski's work in Cincinnati that nearly 97% of seeded cells are flushed into the distal circulation from the graft surface within the first 24 hours after implantation. What happens to those cells? Do they survive circulation through two capillary beds? I do not know, but considering some reports of circulating endothelium that can be isolated from the blood after myocardial infarction, it would not surprise me that circulating cells in our models are capable of attaching to another unseeded surface.
I suspect that this is the explanation for the endothelium discovered in the unseeded grafts in this study. In fact, the only unseeded graft that did not contain endothelium was paired with a seeded side that had occluded. Since it is possible that this particular graft occluded soon after implantation, the cells with which it was seeded may not have had the opportunity to rinse into the systemic circulation. Perhaps one could approach this problem by constructing another similar series of grafts and ligating half of the seeded sides immediately after implantation to provide a control group. We currently are planning to use tagged endothelial cells, and I think that approach may answer the question as well.
Dr. Glover (closing). I agree that this would be a reasonable way to determine what is going on in this preparation.
References
- . Sequential inoculation for optimal cell distribution on tubular grafts. In: Herring MB, Glover JL editor. Endothelial seeding in vascular surgery. New York: Grune & Stratton; 1987;p. 103–118
- . A method of immediate endothelial seeding suitable for clinical application. In: Herring MB, Glover JL editor. Endothelial seeding in vascular surgery. New York: Grune & Stratton; 1987;p. 165–170
- . The peroxidase-antiperoxidase staining of factor VIII-related antigen on cultured endothelial cells. J Biomed Mater Res. 1984;18:561–567
- . Expanded polytetrafluoroethylene vascular prostheses seeded with enzymatically derived and cultured canine endothelial cells. Surgery. 1982;91:550–559
- Endothelial seeding of Dacron and polytetrafluoroethylene grafts: the cellular events of healing. Surgery. 1984;96:745–754
- . A single-staged technique for seeding vascular grafts with autogenous endothelium. Surgery. 1978;84:498–504
- . Immediate seeding of enzymatically derived endothelium in Dacron vascular grafts: early experimental studies with autologous canine cells. Arch Surg. 1980;115:1289–1294
- . Influence of endothelial cell seeding on platelet deposition and patency in small-diameter Dacron arterial grafts. J Vasc Surg. 1984;1:224–233
- . Endothelial cell seeding of a new PTFE vascular prosthesis. J Vasc Surg. 1985;2:424–429
- . Endothelial seeding of enzymatically derived and cultured cells on prosthetic grafts. In: Stanley JC, Burkel WE, Lindenauer SM, et al. editor. Biologic & Synthetic Vascular Prostheses. New York: Grune & Stratton; 1982;p. 631–651
- . The endothelium-modulator of vascular smooth-muscle tone. N Engl J Med. 1988;319:512–513
- . Improved patency of endothelial-cell-seeded, long, knitted Dacron and ePTFE vascular prostheses. ASAIO Journal. 1985;8:65–73
- * *Personal communication from Beth Raper, Indiana University School of Medicine, Department of Anatomy, Sept. 21, 1988.
☆ Supported by the William Beaumont Hospital Research Institute, No. RI-87-11.
☆☆ Reprint requests: John L. Glover, MD, Chairman, Department of Surgery, William Beaumont Hospital, 3601 W. Thirteen Mile Rd., Royal Oak, MI 48072.
★ J Vasc Surg 1989:9:656–64
PII: S0741-5214(89)70036-7
© 1989 Society for Vascular Surgery and International Society for Cardiovascular Surgery, North American Chapter. Published by Elsevier Inc. All rights reserved.
