Optimal Seeding Conditions for Human Endothelial Cells K. Craig Kent, MD*, Akira Oshima, MD ~, Anthony D. Whittemore, MD "~, Boston, Massachusetts
An in vitro model of endothelial cell seeding has been developed to individually evaluate the steps required for seeding arterial prostheses. Human saphenous vein endothelial cells are radiolabeled with tritiated thymidine and seeded onto 4 mm polytetrafluoroethylene grafts. Grafts are then placed into a perfusion circuit for determination of cellular retention. Using this model, the following variables were studied: (1) graft coating (fibronectin versus serum versus plasma); (2) time of incubation of cells with graft (0, 20, 90 minutes); (3) density of the initial seeding solution (4 x 103-6 x 10 s cells/cm2). The data suggest that incubation of a graft with plasma provides an adhesive surface that is as effective as fibronectin for enhancing cell retention. With this particular model, seeding densities between 1 and 2 x 10 s cells/cm = produce a confluent monolayer with optimal utilization of cells. A shorter 20 minute incubation period resulted in the retention of only half of the seeded cells, while postperfusion attachment increased significantly with a 90 minute incubation period. Data derived from this system can be used to construct a protocol that may be useful for clinical in vivo seeding trials. (Ann Vasc Surg 1992;6:258-264). KEY WORDS: prostheses.
Endothelial cells; seeding; polytetrafluoroethylene (PTFE) grafts;
The diminished patency rate associated with currently available small diameter prosthetic vascular grafts is in part related to the thrombogenic nature of the graft surfaces. Graft occlusion may be the direct result of thrombosis. Occlusion can also be secondary to intimal hyperplasia (which may develop in response to smooth muscle cell mitogens released from aggregating platelets). Lining prosthetic vascular grafts with endothelium may increase their patency by reducing platelet aggregation and providing a surface that is resistant to thrombosis. From the Department of Surgery, Beth Israel Hospital*, the Department of Surgery, Brigham & Women's Hospital§, and the Harvard Medical School, Boston, Massachusetts. Reprint requests: Anthony D. Whittemore, MD, Department of Surgery, Brigham & Women's Hospital, 75 Francis" Street, Boston, Massachusetts 02115.
Although a variety of studies have analyzed the kinetics of endothelial cell seeding using cells derived from animals or human umbilical vein, few protocols have looked at attachment and retention of cells derived from human saphenous vein [I-5]. As more is learned about large vessel human endothelium, it is clear that these cells have characteristics distinct from cells derived from other sources [6]. Knowledge of the behavior of large vessel human endothelium is necessary before a technique of seeding in humans can be expected to be successful. Although seeded endothelial cells may initially adhere to a graft surface, it is essential that they remain attached when the graft is exposed to shear stress associated with physiologic circulation. Many previous studies have evaluated cellular attachment under static conditions, but seeding methods must be optimized so that cellular retention occurs following the introduction of flow.
258
VOLUME 6 N o 3 - 1992
SEEDING OF H U M A N ENDOTHELIAL CELLS
An in vitro system was created to evaluate the attachment of human saphenous vein endothelial cells to a polytetrafluoroethylene (PTFE) graft under physiologic flow conditions. Using this model, we evaluated several factors which are important in establishing an endothelial monolayer that survives the shear stresses provided by flow. Specifically, we varied the type of graft coating, the cellular density of the seeding suspension, and the time of incubation of the cells with the graft prior to subjecting them to flow. Our goal was to learn more about which conditions are most conducive for retention of cells by a graft surface.
MATERIALS AND METHODS
Port for medium change
259
i 2j ....
:" /
4 MM
Cell harvest, culture, and labeling
Endothelial cells were harvested from excess saphenous vein retrieved from patients undergoing coronary artery bypass surgery. Following their removal, veins were flushed with and stored in chilled plasmalyte solution. When the vascular reconstruction was complete (usually two hours following vein removal), the unused vein segments were released and transferred to the laboratory in chilled heparinized blood. Cell harvest was then immediately performed. Our techniques of harvest and cell passage have been previously described [7]. Using standard tissue culture techniques and culture media (CM) consisting of M-199, 50 p.g/ml penicillin and 50 p,g/ml streptomycin*, 20% Fetal Bovine Serum (FBS), 100/~g/ml of sodium heparin*, 100 /~g/ml L-glutamine and 100 /zg/ml of pituitary derived growth factor s, six T-75 flasks of confluent endothelial cells in their second passage were produced from each vein segment. Confirmation of these cells as endothelium was made with lowdensity lipoprotein (LDL) staining. Cells were labeled by incubating confluent cultures for 48 hours with CM containing 5 /zCi/cc of titrated thymidine (specific activity 6.7 p,Ci/ retool)**. Cells were removed from the T-75 flask with 0.05% trypsin and then washed twice with Dulbecco's phosphate buffered saline (PBS) before resuspension in culture media. Loss of radiolabel from endothelial cells was determined by aliquoting 100 /A samples of the labeled cell suspension into Eppindorf tubes. These tubes were incubated at 37°C for 15, 30, 60 and 120 minutes. After centrifugation, the cell pellet was lysed with .3 mol/L NaOH and the counts per *Gibco, Grand Island, New York. *Sigma Chemical Co., St. Louis, Missouri. ~Biomedical Technologies. Stoughton, Massachusetts. **New England Nuclear, Boston, Massachusetts.
PTFE Groft
2,!.".,
:f
.'
/
q
Ties for ottochment onto gloss tubing , i
"Y
:
Fig. 1. Diagram of the apparatus used for precoating and seeding a 4 mm polytetrafluoroethylene graft. Graft was interposed between two glass tubes; one acted as the inflow and the other as outflow for coating and seeding solutions.
minutes (CPM) in the pellet and supernatant were determined with a beta scintillation counter. Percent isotope loss was calculated as: cpm (supernatant) x 100 cpm (supernatant) + cpm (pellet) Precoating and seeding of grafts
The apparatus used for precoating and seeding of grafts is diagrammed in Figure 1. A 14 cm segment of 4 mm ID PTFE vascular graft tt was interposed between two segments of glass tubing, which acted as inflow and outflow ports for introducing the coating and seeding suspensions. There was a 1 cm overlap of graft and tubing on each end with the final usable graft length being 12 cm. The graft and tubing were then inserted into a Kimax culture tube **W.L. Gore, Flagstaff, Arizona.
260
SEEDING OF H U M A N ENDOTHELIAL CELLS
with the inflow and outflow ports exiting through a rubber stopper. Three different graft coatings were used (plasma, serum and a solution of purified fibronectin). Fibronectin ~'~ was prepared in M-199 at a concentration calculated to allow 10 p,g of solubilized fibronectin per 1 cm 2 of graft surface. Human serum and plasma were obtained from healthy volunteers. Serum was prepared by centrifuging clotted blood for l0 minutes at 1000 g. Plasma was similarly prepared from specimens collected in 3.8% sodium citrate. Grafts were precoated by injecting 1.76 cc of coating solution into the inflow port (1.76 cc is the volume required to fill the lumen o f a 4 mm graft 12 cm in length). Subsequent injection of air allowed centering of the fluid column so that it remained within the confines of the graft. The culture tube was then placed in the incubator on its side and rotated one-quarter turn every five minutes for the first 20 minutes. After one hour, the coating solution was gently flushed from the graft with 20 cc of M-199. Endothelial cells used for seeding were trypsinized and resuspended in culture media. The cellular density of the resultant suspension was determined by counting with a hemocytometer. The grafts were seeded with suspensions of labeled endothelial cells using the same process as was used for precoating. The seeding suspension was incubated with the graft for varying time periods ranging from 0 to 90 minutes. Graft perfusion
Following the seeding process, the graft was bisected. One half of the graft was removed for counting, and the second half was placed in a perfusion circuit driven by a peristaltic roller pump (Fig. 2). The graft was perfused for one hour at a pressure of 100 mmHg and a flow rate of 220 cc/min with a solution of M-199 with 20% fetal bovine serum. After perfusion, this segment of the graft was removed for counting.
ANNALSOF
VASCULAR SURGERY
Culfure
Fig. 2. Diagram of circuit used for perfusion. Polytetrafluoroethylene graft was placed in a flow system driven by a peristaltic roller pump.
incubated for two hours with ! cc of ProtosaF~***. In earlier experiments we compared Protosal ® and NaOH and found Protosal ® to be much more effective in recovering tritium from cells attached to a prosthetic graft surface, All samples were then counted in a Beckman scintillation counter for one minute. The percentage of seeded cells that remained attached to the graft surface was calculated by the following formula: % attachment = [avg. cpm (1 cm graft segment) x 12] cpm (initial seeding suspension)
x 100
All statistical comparisons of attachment rates were made using an unpaired Student T-test. To evaluate whether free tritiated thymidine adheres to the PTFE grafts (producing falsely high cellular attachment rates), six grafts (two with each of the three different precoating solutions) were placed in the seeding apparatus and incubated for one hour with 5 ~Ci/ml tritium in M-199. The grafts were then flushed, divided into 1 cm segments and counted.
Calculation of cellular attachment
Grafts were divided into five 1 cm segments, each of which was counted separately. Graft segments, samples of the initial seeding suspension, the effluent removed following seeding of the graft, and samples of media used to perfuse the grafts, were all counted. Cells must be lysed so that intracellular tritium can be released and appropriately counted. Graft segments, pellets, and supernatants were all
Isotope retrieval
The retrieval of the tritiated thymidine label for each experiment was determined using the following formula: % retrieval = [cpm (total graft) + cpm (seeding effluent & graft perfusate)] cpm (initial seeding suspension)
~~Boehringer Mannheim Biochemicals, Indianapolis, Indiana.
x 100
***New England Nuclear, Boston, Massachusetts.
VOLUME 6
SEEDING OF HUMAN ENDOTHELIAL CELLS
N o 3 - 1992
261
TABLE I.mPercent of endothelial attachment as a function of graft coating Preperfusion Postperfusion
Control 10 -+ 7 (n = 7) 4+--3(n =6)
Fibronectin 69 +- 9 (n = 10) 62-+ 15(n = 11)
Morphology of cell monolayers
Samples of seeded grafts were stained with hematoxylin and observed under an inverted microscope. This allowed identification of endothelial cell nuclei, and the density of the cell layer could be observed as well as the presence or absence of layering of cells. We were particularly interested in the morphologic behavior of endothelial cells seeded at high density, and we used a tissue culture system to further study the effect of seeding at a density greater than confluence. Two cm 2 gelatin-coated tissue culture wells were seeded with human saphenous vein endothelial cells at a density of 2 x 108 cells/cc. After 90 minutes of incubation, the suspension was removed and the wells washed vigorously three times with PBS solution and observed under the inverted microscope. Culture media was replaced and cells were then maintained at 37°C. At 24 hour intervals, the wells were again washed vigorously three times with PBS, the culture media changed, and the morphology of the endothelial layer again observed under the inverted microscope. This process was repeated until endothelial cells no longer existed in layers and only a confluent monolayer remained. Similar experiments were performed using endothelial cells derived from human umbilical veins. RESULTS We found minimal loss of the radiolabel from endothelial cells during periods of incubation up to 120 minutes. The leakage rate was less than 1% for all time periods studied. Free tritiated thymidine did not attach to the grafts. When coated grafts were perfused with solutions containing high concentrations of tritiated thymidine, there was minimal ( .25). As expected, there was minimal retention of cells (ANOVA) on an uncoated graft (Table I), In contrast, 69% of seeded cells initially attached to a fibronectin coated graft, and the percentage of attached cells diminished insignificantly to 62% following one hour of perfusion. Precoating the grafts with serum produced initial attachment rates that were equal to that of fibronectin (69%, p = .375). Initial attachment rates with plasma were significantly greater than with fibronectin (81%, p < .025). After one hour of perfusion, there was attrition of cells from both serum and plasma coated grafts. Postperfusion attachment in the plasma treated grafts remained equivalent to the attachment found in grafts treated with fibronectin (plasma 69%, p > .375). There was significantly less postperfusion attachment, however, in the serum treated grafts when they were compared to grafts treated with fibronectin (serum 45%, p < .025). Incubation time: The time of incubation of the endothelial cells with the grafts was varied from 0, 20 and 90 minutes. Polytetrafluoroethylene grafts were pretreated with fibronectin, and seeding densities were always less than confluence. The following are the average seeding densities (cells/cm z) for each time period used: 0 minutes--6.7 × 104, 20 minutes--8.9 × 10 4, 90 minutes--8.2 x 10 4 (p > .25 by ANOVA). There was minimal attachment when cells were incubated with the graft for only a few seconds. After 20 minutes of incubation, 54% of cells initially attached to the graft and a statistically equivalent number of cells (50%) remained attached following perfusion. The rate of attachment of cells to the graft was significantly greater if the incubation TABLE ll.--Percent of endothelial attachment as a function of incubation time
Attachment studies
Graft coatings: PTFE grafts either received no treatment or they were exposed to fibronectin, serum, or plasma as described. For these studies, the incubation time was fixed at 90 minutes. Cells
Serum 69 -+ 10 (n = 10) 45±20(n =7)
Preperfusion Postperfusion
Incubation time (minutes) 0 20 90 (n = 2) (n = 10) (n = 11) 2 -+ 1 54 -+ 13 69 -+ 9 1 -+ 1 50 -+ 12 62 -+ 6
SEEDING OF H U M A N ENDOTHELIAL CELLS
262
period was increased to 90 minutes (69%, p < .05 ). The postperfusion attachment was also greater (62%) when cells were incubated 90 versus 20 minutes (p < .025). Seeding density: With fibronectin as the coating and 90 minutes as the incubation time, cells were seeded at densities ranging from 4 x 103 to 6 x 105 cell/cm 2. Figure 3 shows the relationship between density of the seeding suspension and the percentage of cells that remain attached to the graft after perfusion. As the density of cells in the seeding suspension increases, the percentage of seeded cells that remain attached to the graft diminishes. Each of these grafts was microscopically inspected. In all grafts, cells were evenly distributed. At seeding densities below 105, a homogenous cellular layer was found, but cells were not confluent. At densities between I x 105 and 2 x 105, aconfluent layer was present with some stacking of the cells. At densities greater than 2 x 105, a confluent layer was present, but endothelial cells were stacked in multiple layers. This layering of cells was present before and after one hour of peffusion. Since endothelial cells prefer to exist in monolayers, it is expected that with continued perfusion, these layers of cells would eventually be washed into the perfusate leaving behind a confluent cellular lining on the graft. In tissue culture, this same layering was observed when cells were seeded at high density. It would take an average of four days of vigorous washing before all layered cells could be removed and the underlying monolayer visualized.
DISCUSSION Lining a prosthetic graft with a confluent layer of endothelial cells offers two potential advantages. An intact, quiescent layer of endothelium might increase early patency in grafts whose surfaces are inherently thrombogenic. Also, a diminution in platelet aggregation may reduce the inevitable formation of myointimal hyperplasia. In 1979 Herring successfully demonstrated improved patency of endothelial seeded Dacron grafts placed in the carotid position in dogs [2]. Over the subsequent 12 years, the technique has been applied in a variety of animal species and more recently humans. In animal models, grafts seeded with endothelium exhibit diminished platelet activation, resistance to infection and most importantly increased patency. Unfortunately, success in human clinical trials has not been so readily achieved [1,8-12]. It is probable that all endothelial cells are not created equally. Differences in morphology, attachment and potential for growth are found when venous endothelium is compared to arterial endothelium, when microvascular endothelium is compared to that of large arteries, and when endothelium from humans is corn-
ANNALS OF VASCULAR SURGERY
pared to that of animals [6,13]. The requirements for growth and maintenance of human large vessel endothelium are different than the requirements for growth of endothelium derived from most other sources [6]. It would follow that seeding techniques that have been successfully employed in animal models might need to be modified in order to produce equivalent results in humans. In the majority of the animal studies, the technique of "immediate seeding" has been used. Cells which have been freshly harvested are seeded onto a graft at subconfluent density just prior to its implantation. Unfortunately, these cells have been recently exposed to the detrimental effects of collagenase and have little time to attach to the graft surface before flow is reestablished [14]. The initial subconfluent layer of cells which lines the graft sustains further cell loss when exposed to the arterial shear stress. An alternative would be to raise large cell numbers in tissue culture, seed at confluent density, and allow an adequate incubation period so that cells are firmly adherent to the graft prior to implantation [1]. With this method, a confluent layer of endothelium would provide an immediate nonthrombogenic surface when the graft is exposed to the circulation. The model used in this study seeks to exploit this method. Seeding studies performed with Indium-Ill as the endothelial cell marker are subject to considerable error [15]. We used tritiated thymidine as the cell label in that it is relatively nontoxic to cells [16], and has previously been shown to be an excellent label for endothelial cells. The rate of loss of radiolabel from the cells is low, We found that less than 1% of the radiolabel was released from viable cells over a two hour time period. Hasson, using labeling techniques very similar to ours, showed a linear relationship between radioactive counts and cell number. Thus, there appears to be even labeling of cells with this technique. In each experiment, we calculated the recovery of the radiolabel. The average retrieval of tritiated thymidine from our system was 101 -+ 10% (n = 20). The ability of an endothelial cell to attach to a prosthetic graft surface is dependent upon the affinity of the cell for that surface. In vivo this surface is the basement membrane which consists of collagen, laminin and glycoproteins. Existing prosthetic graft materials including Dacron and PTFE will retain very few cells unless a coating is used. Our experience with uncoated PTFE concurs, in that only 4% of seeded cells were retained after one hour of flow. There has been an extensive search for the ideal graft substrate. Fibronectin is a surface glycoprorein that is found in plasma and is also secreted by endothelial cells into adjacent basement membrane [ 17]. Our results show excellent adherence of endothelial cells to a PTFE graft coated with exogenous fibronectin, with 62% of seeded cells remaining
VOLUME 6 N o 3 - 1992
S E E D I N G OF H U M A N E N D O T H E L I A L C E L L S
after one hour of flow. The ability of fibronectin to enhance human endothelial cell attachment has been previously documented, and this glycoprotein has been used in various seeding protocols. Although a variety of other substances, including collagen, laminin, amniotic membrane and Matrigel, have been evaluated, none have produced attachment rates that are superior to that of fibronectin [18]. Purified fibronectin is derived from pooled plasma, and thus its administration presents potential risks, including the transfer of viral infection. Plasma and serum also contain fibronectin in a soluble form which will precipitate when either blood product is incubated with a graft surface [!9]. With this in mind, we evaluated the initial attachment and retention of endothelial cells to a PTFE graft that had been pretreated for one hour with autogenous plasma or serum. Following an hour of perfusion, the retention of endothelial cells by grafts treated with plasma was equivalent to the retention found in fibronectin-coated grafts. In serum treated grafts, the retention of cells was significantly less. It appears that incubation of a grail with plasma provides an adhesive surface that is as effective as fibronectin in retaining cells. Use of autogenous plasma is convenient and simple, and avoids the inherent risks associated with the use of exogenous fibronectin. A limiting factor in any seeding protocol is the availability of a sufficient number of endothelial cells to completely cover a graft surface. Choosing the proper seeding density is important so that cells are not wasted. Since the density of saphenous vein endothelial cells at confluence is approximately 1.2 × 105 [6] establishment of an immediate confluent lining on a prosthetic graft is likely to require a similar minimal density of cells. In many of the immediate seeding protocols, lesser cell densities have been used, thus producing a subconfluent layer that requires days to weeks to become confluent in vivo. Others have tried to produce immediate confluence by seeding at cellular concentrations much greater than 1.2 × 105 [20]. Using optimum ex vivo conditions, we wished to determine the seeding density that would result in the greatest effective utilization of cells. Our goal was to produce a confluent layer of endothelium that persisted after institution of flow. Standardizing graft coating and incubation time, we found that, as the density of the seeding suspension is increased, the "percentage" of seeded cells that remain attached to the graft diminishes (Fig. 3). Therefore, seeding at lower densities results in efficient use of cells and seeding at high densities results in significant wastage of cells. If cells were seeded at densities less than 105 cells/cm2, the rate of attachment was high, but histologically a confluent monolayer was not pro-
263
100 ,
o
60 O 0
O
•
40
20 100
200 300 400 500 Seeding Density (cells x 103 )
600
700
Fig. 3. Relationship between density of seeding suspension and percentage of cells that remain attached to the graft following perfusion (11 nonconfluent monoiayer, ~ confluent monolayer, minimal stacking of cells, ° confluent monolayer, multiple layers of cells).
duced. When cells were seeded at concentrations between 1 and 2 × 105 cells/cm 2, the rate of cellular attachment diminished somewhat (50-60%), but grafts were covered with a fairly uniform monolayer of cells (Fig. 4). Seeding at densities greater than 2 × 105 resulted in a diminished rate of attachment. Also, high density seeding produced layering of cells on the graft surface. Presumably, with longer periods of perfusion, these extra layers of cells would detach and be lost into the circulation. Seeding at high densities in tissue culture produced this same phenomenon. Cells would stack on top of each other, and it would take several days of washing before the layered cells departed and only a monolayer remained. With this particular model,
Fig. 4. Polytetrafluoroethylene graft seeded with human endothelial cells at a density of 2 x 10 = cells/cm =. Following one hour of perfusion, homogenous monolayer of cells remains (graft, stained with hematoxylin, observed under inverted microscope l Ox).
264
SEEDING OF H U M A N ENDOTHELIAL CELLS
it would appear that seeding densities between 1 and 2 × 10~cells/cm 2 produce a confluent monolayer with optimal utilization of cells. The time of incubation of the cells with the graft is the third critical determinant in a seeding protocol. The longer that endothelial cells are incubated with the graft surface prior to institution of flow, the more adherent they may become. Electron microscopy has demonstrated that seeded cells are initially spherical and barely adhere to the recipient surface. With time, these cells will spread on the graft surface, and a larger portion of the endothelial membrane becomes attached. The following incubation times were used (0, 20 minutes, 90 minutes). A 20 minute incubation period resulted in fairly good cellular attachment (54%), and perfusion of these grafts for one hour did not produce significant attrition of cells (50%). Both the pre- and postperfusion attachment increased significantly with a 90 minute incubation period.
CONCLUSION Endothelial cell seeding may improve the patency of small diameter vascular grafts. Seeding techniques must be refined so that the least number of cells can be used to produce an endothelial monolayer that is available at the time of graft implantation. We have used tritiated thymidine as a radiolabel to study the retention of human saphenous vein endothelial cells by a FI'FE graft. We conclude that plasma is as effective as fibronectin in promoting cellular attachment. In this particular model, optimal usage of cells was achieved when the density of the seeding suspension varied from 1 to 2 × 105 cells/cm z. Although an incubation time of 20 minutes provided adequate attachment, longer incubation times significantly improve the rate of attachment. Only in vivo human trials will allow determination of the optimal method for seeding endothelial cells onto a prosthetic vascular graft. Hopefully, information derived from in vitro systems such as ours can be used to design potentially effective clinical protocols.
REFERENCES 1. SHINDO S, TAKAGI A, WHITTEMORE AD. Improved patency of collagen-impregnated grafts after a vitro autogenous endothelial cell seeding. J Vase Surg 1987:6:325-332. 2. HERRING MB, et al, Seeding endothelium onto canine arterial prostheses. Arch Surg 1979;114:679-682. 3. SHEPARD AD, et al. Endothelial cell seeding of small-caliber synthetic grafts in the baboon. Surge~. 1986;99:318--325. 4. HOLLIER LH, et al. Are seeded endothelial cells the origin of neointima on prosthetic vascular grafts? J Vase Surg 1986;3:65-73. n u n
ANNALS OF VASCULAR SURGERY
5. SCHNEIDER PA, HANSON SR, PRICE TM, et al. Durability of confluent endothelial cell monolayers on small caliber vascular prosthesis in vitro. Sttrgeo' 1989;103(4):456--472, 6, KENT KC, et al. Species variation and the success of endothelial cell seeding. J Vase Surg 1989;9:271-276. 7. KENT KC, OSHIMA A, IKEMOTO T, et al. An in vitro model for human endothelial cell seeding of a small diameter vascular graft. Trans Am Soe Artif lntern Organs 1988;34: 578-580. 8. HERRING MB, et al. Seeding human arterial prostheses with mechanically derived endothelium. J Vase Surg 1984; 1:279-289. 9. ORTENWALL P, et al. Reduction in deposition of indium 11 I-labeled platelets after autotogous endothelial cell seeding of Dacron aortic bifurcation grafts in humans: a preliminary report. J Vase Sttrg 1986:6:17-25. 10. ZILLA P. et al. Endothelial cell seeding of polytetrafluoroethylene vascular grafts in humans: a preliminary report. J Vase Surg 1987:6:535-541, I 1. ROLAND F, et al. Human endothelial cell seeding: evaluation of its effectiveness by platelet parameters after one year. J Vase Sttrg 1989:9:432--436. 12. DEUTSCH M, F1SCHLEIN T, EBERI T, et al. Human in vitro endothe[ialization of ePTFE vascular graft: an initial clinical report (Abstract), Program of the Int Soc Cardiovasc Surg, 38th Annual Meeting. 1990, p. 30. 13. WAGNER WH. HENDERSON RM, HICKS HE, et al. Differences in morphology, growth rate, and protein synthesis between cultured arterial and venous endothelial cells. J Vase Sur L, 1988:8:509-519. 14. SHAREFKIN JB, VAN WART HE. CRUESS DF, et al, Adult human endothelial cell enzymatic harvesting: estimates of efficiency and comparison of crude and partially purified bacterial collagenase presentations by replicate microwell culture and fibronectin degradation measured by enzyme-linked immunosorbent assay. J Vase Surg 1986:4:566-577. 15. PATTERSON RB, MAYFIELD G, SILBERSTEIN EB, et al. The potential unreliability of indium-Ill oxine labeling in studies of endothelial cell kinetics (Abstract). Program of the Soc Vasc Surg, 43rd Annual Meeting, 1989, p. 74. 16. HASSON JE, WIEBE DH, SHAREFKIN JB, et al, Use of tritiated thymidine as a marker to compare the effects of matrix proteins on adult human vascular endothelial cell attachment: implications for seeding of vascular prostheses. Surgery 1986:100(5):884-891. 17. EFFRON MK, HARRISON DC. Fibronectin: cardiovascular aspects of a ubiquitous glycoprotein. Am J Cardiol 1983:32:206-208. 18. PATTERSON RB, KELLER JD, SILBERSTEIN EB, et al. A comparison between fibronectin and Matrigel pretreated ePTFE vascular grafts. Ann Vase Surg 1989;3:t60-166. 19. VAN WACHEM PB, VRERIKS CM, BEUGEL1NG T, et al. The influence of protein adsorption on interactions of cultured human endothelial cells with polymers. J Biomed Mat Res 1987;2I(6):701-718. 20. T A N N E N B A U M G. AHLBORN T, BENVENISTY A, et al. High-density seeding of cultured endothelial cells leads to rapid coverage of polytetrafluoroethylene grafts. Curr Prob Surg 1987:44(4):318-321. 21. PRATT KJ, JARRELL BE, WILLIAMS SK, et al. Kinetics of endothelial cell surface attachment forces. J Vase Surg 1988:7:591-599. 22. JARRELL BE, WILLIAMS SK, SOLOMON L, et al. Use of an endothelial monolayer on a vascular graft prior to implantation. Ann Surg 1986;203(6):671-678.
An in vitro model of endothelial cell seeding has been developed to individually evaluate the steps required for seeding arterial prostheses. Human saphenous vein endothelial cells are radiolabeled with tritiated thymidine and seeded onto 4 mm polytetrafluoroethylene grafts. Grafts are then placed into a perfusion circuit for determination of cellular retention. Using this model, the following variables were studied: (1) graft coating (fibronectin versus serum versus plasma); (2) time of incubation of cells with graft (0, 20, 90 minutes); (3) density of the initial seeding solution (4 x 103-6 x 10 s cells/cm2). The data suggest that incubation of a graft with plasma provides an adhesive surface that is as effective as fibronectin for enhancing cell retention. With this particular model, seeding densities between 1 and 2 x 10 s cells/cm = produce a confluent monolayer with optimal utilization of cells. A shorter 20 minute incubation period resulted in the retention of only half of the seeded cells, while postperfusion attachment increased significantly with a 90 minute incubation period. Data derived from this system can be used to construct a protocol that may be useful for clinical in vivo seeding trials. (Ann Vasc Surg 1992;6:258-264). KEY WORDS: prostheses.
Endothelial cells; seeding; polytetrafluoroethylene (PTFE) grafts;
The diminished patency rate associated with currently available small diameter prosthetic vascular grafts is in part related to the thrombogenic nature of the graft surfaces. Graft occlusion may be the direct result of thrombosis. Occlusion can also be secondary to intimal hyperplasia (which may develop in response to smooth muscle cell mitogens released from aggregating platelets). Lining prosthetic vascular grafts with endothelium may increase their patency by reducing platelet aggregation and providing a surface that is resistant to thrombosis. From the Department of Surgery, Beth Israel Hospital*, the Department of Surgery, Brigham & Women's Hospital§, and the Harvard Medical School, Boston, Massachusetts. Reprint requests: Anthony D. Whittemore, MD, Department of Surgery, Brigham & Women's Hospital, 75 Francis" Street, Boston, Massachusetts 02115.
Although a variety of studies have analyzed the kinetics of endothelial cell seeding using cells derived from animals or human umbilical vein, few protocols have looked at attachment and retention of cells derived from human saphenous vein [I-5]. As more is learned about large vessel human endothelium, it is clear that these cells have characteristics distinct from cells derived from other sources [6]. Knowledge of the behavior of large vessel human endothelium is necessary before a technique of seeding in humans can be expected to be successful. Although seeded endothelial cells may initially adhere to a graft surface, it is essential that they remain attached when the graft is exposed to shear stress associated with physiologic circulation. Many previous studies have evaluated cellular attachment under static conditions, but seeding methods must be optimized so that cellular retention occurs following the introduction of flow.
258
VOLUME 6 N o 3 - 1992
SEEDING OF H U M A N ENDOTHELIAL CELLS
An in vitro system was created to evaluate the attachment of human saphenous vein endothelial cells to a polytetrafluoroethylene (PTFE) graft under physiologic flow conditions. Using this model, we evaluated several factors which are important in establishing an endothelial monolayer that survives the shear stresses provided by flow. Specifically, we varied the type of graft coating, the cellular density of the seeding suspension, and the time of incubation of the cells with the graft prior to subjecting them to flow. Our goal was to learn more about which conditions are most conducive for retention of cells by a graft surface.
MATERIALS AND METHODS
Port for medium change
259
i 2j ....
:" /
4 MM
Cell harvest, culture, and labeling
Endothelial cells were harvested from excess saphenous vein retrieved from patients undergoing coronary artery bypass surgery. Following their removal, veins were flushed with and stored in chilled plasmalyte solution. When the vascular reconstruction was complete (usually two hours following vein removal), the unused vein segments were released and transferred to the laboratory in chilled heparinized blood. Cell harvest was then immediately performed. Our techniques of harvest and cell passage have been previously described [7]. Using standard tissue culture techniques and culture media (CM) consisting of M-199, 50 p.g/ml penicillin and 50 p,g/ml streptomycin*, 20% Fetal Bovine Serum (FBS), 100/~g/ml of sodium heparin*, 100 /~g/ml L-glutamine and 100 /zg/ml of pituitary derived growth factor s, six T-75 flasks of confluent endothelial cells in their second passage were produced from each vein segment. Confirmation of these cells as endothelium was made with lowdensity lipoprotein (LDL) staining. Cells were labeled by incubating confluent cultures for 48 hours with CM containing 5 /zCi/cc of titrated thymidine (specific activity 6.7 p,Ci/ retool)**. Cells were removed from the T-75 flask with 0.05% trypsin and then washed twice with Dulbecco's phosphate buffered saline (PBS) before resuspension in culture media. Loss of radiolabel from endothelial cells was determined by aliquoting 100 /A samples of the labeled cell suspension into Eppindorf tubes. These tubes were incubated at 37°C for 15, 30, 60 and 120 minutes. After centrifugation, the cell pellet was lysed with .3 mol/L NaOH and the counts per *Gibco, Grand Island, New York. *Sigma Chemical Co., St. Louis, Missouri. ~Biomedical Technologies. Stoughton, Massachusetts. **New England Nuclear, Boston, Massachusetts.
PTFE Groft
2,!.".,
:f
.'
/
q
Ties for ottochment onto gloss tubing , i
"Y
:
Fig. 1. Diagram of the apparatus used for precoating and seeding a 4 mm polytetrafluoroethylene graft. Graft was interposed between two glass tubes; one acted as the inflow and the other as outflow for coating and seeding solutions.
minutes (CPM) in the pellet and supernatant were determined with a beta scintillation counter. Percent isotope loss was calculated as: cpm (supernatant) x 100 cpm (supernatant) + cpm (pellet) Precoating and seeding of grafts
The apparatus used for precoating and seeding of grafts is diagrammed in Figure 1. A 14 cm segment of 4 mm ID PTFE vascular graft tt was interposed between two segments of glass tubing, which acted as inflow and outflow ports for introducing the coating and seeding suspensions. There was a 1 cm overlap of graft and tubing on each end with the final usable graft length being 12 cm. The graft and tubing were then inserted into a Kimax culture tube **W.L. Gore, Flagstaff, Arizona.
260
SEEDING OF H U M A N ENDOTHELIAL CELLS
with the inflow and outflow ports exiting through a rubber stopper. Three different graft coatings were used (plasma, serum and a solution of purified fibronectin). Fibronectin ~'~ was prepared in M-199 at a concentration calculated to allow 10 p,g of solubilized fibronectin per 1 cm 2 of graft surface. Human serum and plasma were obtained from healthy volunteers. Serum was prepared by centrifuging clotted blood for l0 minutes at 1000 g. Plasma was similarly prepared from specimens collected in 3.8% sodium citrate. Grafts were precoated by injecting 1.76 cc of coating solution into the inflow port (1.76 cc is the volume required to fill the lumen o f a 4 mm graft 12 cm in length). Subsequent injection of air allowed centering of the fluid column so that it remained within the confines of the graft. The culture tube was then placed in the incubator on its side and rotated one-quarter turn every five minutes for the first 20 minutes. After one hour, the coating solution was gently flushed from the graft with 20 cc of M-199. Endothelial cells used for seeding were trypsinized and resuspended in culture media. The cellular density of the resultant suspension was determined by counting with a hemocytometer. The grafts were seeded with suspensions of labeled endothelial cells using the same process as was used for precoating. The seeding suspension was incubated with the graft for varying time periods ranging from 0 to 90 minutes. Graft perfusion
Following the seeding process, the graft was bisected. One half of the graft was removed for counting, and the second half was placed in a perfusion circuit driven by a peristaltic roller pump (Fig. 2). The graft was perfused for one hour at a pressure of 100 mmHg and a flow rate of 220 cc/min with a solution of M-199 with 20% fetal bovine serum. After perfusion, this segment of the graft was removed for counting.
ANNALSOF
VASCULAR SURGERY
Culfure
Fig. 2. Diagram of circuit used for perfusion. Polytetrafluoroethylene graft was placed in a flow system driven by a peristaltic roller pump.
incubated for two hours with ! cc of ProtosaF~***. In earlier experiments we compared Protosal ® and NaOH and found Protosal ® to be much more effective in recovering tritium from cells attached to a prosthetic graft surface, All samples were then counted in a Beckman scintillation counter for one minute. The percentage of seeded cells that remained attached to the graft surface was calculated by the following formula: % attachment = [avg. cpm (1 cm graft segment) x 12] cpm (initial seeding suspension)
x 100
All statistical comparisons of attachment rates were made using an unpaired Student T-test. To evaluate whether free tritiated thymidine adheres to the PTFE grafts (producing falsely high cellular attachment rates), six grafts (two with each of the three different precoating solutions) were placed in the seeding apparatus and incubated for one hour with 5 ~Ci/ml tritium in M-199. The grafts were then flushed, divided into 1 cm segments and counted.
Calculation of cellular attachment
Grafts were divided into five 1 cm segments, each of which was counted separately. Graft segments, samples of the initial seeding suspension, the effluent removed following seeding of the graft, and samples of media used to perfuse the grafts, were all counted. Cells must be lysed so that intracellular tritium can be released and appropriately counted. Graft segments, pellets, and supernatants were all
Isotope retrieval
The retrieval of the tritiated thymidine label for each experiment was determined using the following formula: % retrieval = [cpm (total graft) + cpm (seeding effluent & graft perfusate)] cpm (initial seeding suspension)
~~Boehringer Mannheim Biochemicals, Indianapolis, Indiana.
x 100
***New England Nuclear, Boston, Massachusetts.
VOLUME 6
SEEDING OF HUMAN ENDOTHELIAL CELLS
N o 3 - 1992
261
TABLE I.mPercent of endothelial attachment as a function of graft coating Preperfusion Postperfusion
Control 10 -+ 7 (n = 7) 4+--3(n =6)
Fibronectin 69 +- 9 (n = 10) 62-+ 15(n = 11)
Morphology of cell monolayers
Samples of seeded grafts were stained with hematoxylin and observed under an inverted microscope. This allowed identification of endothelial cell nuclei, and the density of the cell layer could be observed as well as the presence or absence of layering of cells. We were particularly interested in the morphologic behavior of endothelial cells seeded at high density, and we used a tissue culture system to further study the effect of seeding at a density greater than confluence. Two cm 2 gelatin-coated tissue culture wells were seeded with human saphenous vein endothelial cells at a density of 2 x 108 cells/cc. After 90 minutes of incubation, the suspension was removed and the wells washed vigorously three times with PBS solution and observed under the inverted microscope. Culture media was replaced and cells were then maintained at 37°C. At 24 hour intervals, the wells were again washed vigorously three times with PBS, the culture media changed, and the morphology of the endothelial layer again observed under the inverted microscope. This process was repeated until endothelial cells no longer existed in layers and only a confluent monolayer remained. Similar experiments were performed using endothelial cells derived from human umbilical veins. RESULTS We found minimal loss of the radiolabel from endothelial cells during periods of incubation up to 120 minutes. The leakage rate was less than 1% for all time periods studied. Free tritiated thymidine did not attach to the grafts. When coated grafts were perfused with solutions containing high concentrations of tritiated thymidine, there was minimal ( .25). As expected, there was minimal retention of cells (ANOVA) on an uncoated graft (Table I), In contrast, 69% of seeded cells initially attached to a fibronectin coated graft, and the percentage of attached cells diminished insignificantly to 62% following one hour of perfusion. Precoating the grafts with serum produced initial attachment rates that were equal to that of fibronectin (69%, p = .375). Initial attachment rates with plasma were significantly greater than with fibronectin (81%, p < .025). After one hour of perfusion, there was attrition of cells from both serum and plasma coated grafts. Postperfusion attachment in the plasma treated grafts remained equivalent to the attachment found in grafts treated with fibronectin (plasma 69%, p > .375). There was significantly less postperfusion attachment, however, in the serum treated grafts when they were compared to grafts treated with fibronectin (serum 45%, p < .025). Incubation time: The time of incubation of the endothelial cells with the grafts was varied from 0, 20 and 90 minutes. Polytetrafluoroethylene grafts were pretreated with fibronectin, and seeding densities were always less than confluence. The following are the average seeding densities (cells/cm z) for each time period used: 0 minutes--6.7 × 104, 20 minutes--8.9 × 10 4, 90 minutes--8.2 x 10 4 (p > .25 by ANOVA). There was minimal attachment when cells were incubated with the graft for only a few seconds. After 20 minutes of incubation, 54% of cells initially attached to the graft and a statistically equivalent number of cells (50%) remained attached following perfusion. The rate of attachment of cells to the graft was significantly greater if the incubation TABLE ll.--Percent of endothelial attachment as a function of incubation time
Attachment studies
Graft coatings: PTFE grafts either received no treatment or they were exposed to fibronectin, serum, or plasma as described. For these studies, the incubation time was fixed at 90 minutes. Cells
Serum 69 -+ 10 (n = 10) 45±20(n =7)
Preperfusion Postperfusion
Incubation time (minutes) 0 20 90 (n = 2) (n = 10) (n = 11) 2 -+ 1 54 -+ 13 69 -+ 9 1 -+ 1 50 -+ 12 62 -+ 6
SEEDING OF H U M A N ENDOTHELIAL CELLS
262
period was increased to 90 minutes (69%, p < .05 ). The postperfusion attachment was also greater (62%) when cells were incubated 90 versus 20 minutes (p < .025). Seeding density: With fibronectin as the coating and 90 minutes as the incubation time, cells were seeded at densities ranging from 4 x 103 to 6 x 105 cell/cm 2. Figure 3 shows the relationship between density of the seeding suspension and the percentage of cells that remain attached to the graft after perfusion. As the density of cells in the seeding suspension increases, the percentage of seeded cells that remain attached to the graft diminishes. Each of these grafts was microscopically inspected. In all grafts, cells were evenly distributed. At seeding densities below 105, a homogenous cellular layer was found, but cells were not confluent. At densities between I x 105 and 2 x 105, aconfluent layer was present with some stacking of the cells. At densities greater than 2 x 105, a confluent layer was present, but endothelial cells were stacked in multiple layers. This layering of cells was present before and after one hour of peffusion. Since endothelial cells prefer to exist in monolayers, it is expected that with continued perfusion, these layers of cells would eventually be washed into the perfusate leaving behind a confluent cellular lining on the graft. In tissue culture, this same layering was observed when cells were seeded at high density. It would take an average of four days of vigorous washing before all layered cells could be removed and the underlying monolayer visualized.
DISCUSSION Lining a prosthetic graft with a confluent layer of endothelial cells offers two potential advantages. An intact, quiescent layer of endothelium might increase early patency in grafts whose surfaces are inherently thrombogenic. Also, a diminution in platelet aggregation may reduce the inevitable formation of myointimal hyperplasia. In 1979 Herring successfully demonstrated improved patency of endothelial seeded Dacron grafts placed in the carotid position in dogs [2]. Over the subsequent 12 years, the technique has been applied in a variety of animal species and more recently humans. In animal models, grafts seeded with endothelium exhibit diminished platelet activation, resistance to infection and most importantly increased patency. Unfortunately, success in human clinical trials has not been so readily achieved [1,8-12]. It is probable that all endothelial cells are not created equally. Differences in morphology, attachment and potential for growth are found when venous endothelium is compared to arterial endothelium, when microvascular endothelium is compared to that of large arteries, and when endothelium from humans is corn-
ANNALS OF VASCULAR SURGERY
pared to that of animals [6,13]. The requirements for growth and maintenance of human large vessel endothelium are different than the requirements for growth of endothelium derived from most other sources [6]. It would follow that seeding techniques that have been successfully employed in animal models might need to be modified in order to produce equivalent results in humans. In the majority of the animal studies, the technique of "immediate seeding" has been used. Cells which have been freshly harvested are seeded onto a graft at subconfluent density just prior to its implantation. Unfortunately, these cells have been recently exposed to the detrimental effects of collagenase and have little time to attach to the graft surface before flow is reestablished [14]. The initial subconfluent layer of cells which lines the graft sustains further cell loss when exposed to the arterial shear stress. An alternative would be to raise large cell numbers in tissue culture, seed at confluent density, and allow an adequate incubation period so that cells are firmly adherent to the graft prior to implantation [1]. With this method, a confluent layer of endothelium would provide an immediate nonthrombogenic surface when the graft is exposed to the circulation. The model used in this study seeks to exploit this method. Seeding studies performed with Indium-Ill as the endothelial cell marker are subject to considerable error [15]. We used tritiated thymidine as the cell label in that it is relatively nontoxic to cells [16], and has previously been shown to be an excellent label for endothelial cells. The rate of loss of radiolabel from the cells is low, We found that less than 1% of the radiolabel was released from viable cells over a two hour time period. Hasson, using labeling techniques very similar to ours, showed a linear relationship between radioactive counts and cell number. Thus, there appears to be even labeling of cells with this technique. In each experiment, we calculated the recovery of the radiolabel. The average retrieval of tritiated thymidine from our system was 101 -+ 10% (n = 20). The ability of an endothelial cell to attach to a prosthetic graft surface is dependent upon the affinity of the cell for that surface. In vivo this surface is the basement membrane which consists of collagen, laminin and glycoproteins. Existing prosthetic graft materials including Dacron and PTFE will retain very few cells unless a coating is used. Our experience with uncoated PTFE concurs, in that only 4% of seeded cells were retained after one hour of flow. There has been an extensive search for the ideal graft substrate. Fibronectin is a surface glycoprorein that is found in plasma and is also secreted by endothelial cells into adjacent basement membrane [ 17]. Our results show excellent adherence of endothelial cells to a PTFE graft coated with exogenous fibronectin, with 62% of seeded cells remaining
VOLUME 6 N o 3 - 1992
S E E D I N G OF H U M A N E N D O T H E L I A L C E L L S
after one hour of flow. The ability of fibronectin to enhance human endothelial cell attachment has been previously documented, and this glycoprotein has been used in various seeding protocols. Although a variety of other substances, including collagen, laminin, amniotic membrane and Matrigel, have been evaluated, none have produced attachment rates that are superior to that of fibronectin [18]. Purified fibronectin is derived from pooled plasma, and thus its administration presents potential risks, including the transfer of viral infection. Plasma and serum also contain fibronectin in a soluble form which will precipitate when either blood product is incubated with a graft surface [!9]. With this in mind, we evaluated the initial attachment and retention of endothelial cells to a PTFE graft that had been pretreated for one hour with autogenous plasma or serum. Following an hour of perfusion, the retention of endothelial cells by grafts treated with plasma was equivalent to the retention found in fibronectin-coated grafts. In serum treated grafts, the retention of cells was significantly less. It appears that incubation of a grail with plasma provides an adhesive surface that is as effective as fibronectin in retaining cells. Use of autogenous plasma is convenient and simple, and avoids the inherent risks associated with the use of exogenous fibronectin. A limiting factor in any seeding protocol is the availability of a sufficient number of endothelial cells to completely cover a graft surface. Choosing the proper seeding density is important so that cells are not wasted. Since the density of saphenous vein endothelial cells at confluence is approximately 1.2 × 105 [6] establishment of an immediate confluent lining on a prosthetic graft is likely to require a similar minimal density of cells. In many of the immediate seeding protocols, lesser cell densities have been used, thus producing a subconfluent layer that requires days to weeks to become confluent in vivo. Others have tried to produce immediate confluence by seeding at cellular concentrations much greater than 1.2 × 105 [20]. Using optimum ex vivo conditions, we wished to determine the seeding density that would result in the greatest effective utilization of cells. Our goal was to produce a confluent layer of endothelium that persisted after institution of flow. Standardizing graft coating and incubation time, we found that, as the density of the seeding suspension is increased, the "percentage" of seeded cells that remain attached to the graft diminishes (Fig. 3). Therefore, seeding at lower densities results in efficient use of cells and seeding at high densities results in significant wastage of cells. If cells were seeded at densities less than 105 cells/cm2, the rate of attachment was high, but histologically a confluent monolayer was not pro-
263
100 ,
o
60 O 0
O
•
40
20 100
200 300 400 500 Seeding Density (cells x 103 )
600
700
Fig. 3. Relationship between density of seeding suspension and percentage of cells that remain attached to the graft following perfusion (11 nonconfluent monoiayer, ~ confluent monolayer, minimal stacking of cells, ° confluent monolayer, multiple layers of cells).
duced. When cells were seeded at concentrations between 1 and 2 × 105 cells/cm 2, the rate of cellular attachment diminished somewhat (50-60%), but grafts were covered with a fairly uniform monolayer of cells (Fig. 4). Seeding at densities greater than 2 × 105 resulted in a diminished rate of attachment. Also, high density seeding produced layering of cells on the graft surface. Presumably, with longer periods of perfusion, these extra layers of cells would detach and be lost into the circulation. Seeding at high densities in tissue culture produced this same phenomenon. Cells would stack on top of each other, and it would take several days of washing before the layered cells departed and only a monolayer remained. With this particular model,
Fig. 4. Polytetrafluoroethylene graft seeded with human endothelial cells at a density of 2 x 10 = cells/cm =. Following one hour of perfusion, homogenous monolayer of cells remains (graft, stained with hematoxylin, observed under inverted microscope l Ox).
264
SEEDING OF H U M A N ENDOTHELIAL CELLS
it would appear that seeding densities between 1 and 2 × 10~cells/cm 2 produce a confluent monolayer with optimal utilization of cells. The time of incubation of the cells with the graft is the third critical determinant in a seeding protocol. The longer that endothelial cells are incubated with the graft surface prior to institution of flow, the more adherent they may become. Electron microscopy has demonstrated that seeded cells are initially spherical and barely adhere to the recipient surface. With time, these cells will spread on the graft surface, and a larger portion of the endothelial membrane becomes attached. The following incubation times were used (0, 20 minutes, 90 minutes). A 20 minute incubation period resulted in fairly good cellular attachment (54%), and perfusion of these grafts for one hour did not produce significant attrition of cells (50%). Both the pre- and postperfusion attachment increased significantly with a 90 minute incubation period.
CONCLUSION Endothelial cell seeding may improve the patency of small diameter vascular grafts. Seeding techniques must be refined so that the least number of cells can be used to produce an endothelial monolayer that is available at the time of graft implantation. We have used tritiated thymidine as a radiolabel to study the retention of human saphenous vein endothelial cells by a FI'FE graft. We conclude that plasma is as effective as fibronectin in promoting cellular attachment. In this particular model, optimal usage of cells was achieved when the density of the seeding suspension varied from 1 to 2 × 105 cells/cm z. Although an incubation time of 20 minutes provided adequate attachment, longer incubation times significantly improve the rate of attachment. Only in vivo human trials will allow determination of the optimal method for seeding endothelial cells onto a prosthetic vascular graft. Hopefully, information derived from in vitro systems such as ours can be used to design potentially effective clinical protocols.
REFERENCES 1. SHINDO S, TAKAGI A, WHITTEMORE AD. Improved patency of collagen-impregnated grafts after a vitro autogenous endothelial cell seeding. J Vase Surg 1987:6:325-332. 2. HERRING MB, et al, Seeding endothelium onto canine arterial prostheses. Arch Surg 1979;114:679-682. 3. SHEPARD AD, et al. Endothelial cell seeding of small-caliber synthetic grafts in the baboon. Surge~. 1986;99:318--325. 4. HOLLIER LH, et al. Are seeded endothelial cells the origin of neointima on prosthetic vascular grafts? J Vase Surg 1986;3:65-73. n u n
ANNALS OF VASCULAR SURGERY
5. SCHNEIDER PA, HANSON SR, PRICE TM, et al. Durability of confluent endothelial cell monolayers on small caliber vascular prosthesis in vitro. Sttrgeo' 1989;103(4):456--472, 6, KENT KC, et al. Species variation and the success of endothelial cell seeding. J Vase Surg 1989;9:271-276. 7. KENT KC, OSHIMA A, IKEMOTO T, et al. An in vitro model for human endothelial cell seeding of a small diameter vascular graft. Trans Am Soe Artif lntern Organs 1988;34: 578-580. 8. HERRING MB, et al. Seeding human arterial prostheses with mechanically derived endothelium. J Vase Surg 1984; 1:279-289. 9. ORTENWALL P, et al. Reduction in deposition of indium 11 I-labeled platelets after autotogous endothelial cell seeding of Dacron aortic bifurcation grafts in humans: a preliminary report. J Vase Sttrg 1986:6:17-25. 10. ZILLA P. et al. Endothelial cell seeding of polytetrafluoroethylene vascular grafts in humans: a preliminary report. J Vase Surg 1987:6:535-541, I 1. ROLAND F, et al. Human endothelial cell seeding: evaluation of its effectiveness by platelet parameters after one year. J Vase Sttrg 1989:9:432--436. 12. DEUTSCH M, F1SCHLEIN T, EBERI T, et al. Human in vitro endothe[ialization of ePTFE vascular graft: an initial clinical report (Abstract), Program of the Int Soc Cardiovasc Surg, 38th Annual Meeting. 1990, p. 30. 13. WAGNER WH. HENDERSON RM, HICKS HE, et al. Differences in morphology, growth rate, and protein synthesis between cultured arterial and venous endothelial cells. J Vase Sur L, 1988:8:509-519. 14. SHAREFKIN JB, VAN WART HE. CRUESS DF, et al, Adult human endothelial cell enzymatic harvesting: estimates of efficiency and comparison of crude and partially purified bacterial collagenase presentations by replicate microwell culture and fibronectin degradation measured by enzyme-linked immunosorbent assay. J Vase Surg 1986:4:566-577. 15. PATTERSON RB, MAYFIELD G, SILBERSTEIN EB, et al. The potential unreliability of indium-Ill oxine labeling in studies of endothelial cell kinetics (Abstract). Program of the Soc Vasc Surg, 43rd Annual Meeting, 1989, p. 74. 16. HASSON JE, WIEBE DH, SHAREFKIN JB, et al, Use of tritiated thymidine as a marker to compare the effects of matrix proteins on adult human vascular endothelial cell attachment: implications for seeding of vascular prostheses. Surgery 1986:100(5):884-891. 17. EFFRON MK, HARRISON DC. Fibronectin: cardiovascular aspects of a ubiquitous glycoprotein. Am J Cardiol 1983:32:206-208. 18. PATTERSON RB, KELLER JD, SILBERSTEIN EB, et al. A comparison between fibronectin and Matrigel pretreated ePTFE vascular grafts. Ann Vase Surg 1989;3:t60-166. 19. VAN WACHEM PB, VRERIKS CM, BEUGEL1NG T, et al. The influence of protein adsorption on interactions of cultured human endothelial cells with polymers. J Biomed Mat Res 1987;2I(6):701-718. 20. T A N N E N B A U M G. AHLBORN T, BENVENISTY A, et al. High-density seeding of cultured endothelial cells leads to rapid coverage of polytetrafluoroethylene grafts. Curr Prob Surg 1987:44(4):318-321. 21. PRATT KJ, JARRELL BE, WILLIAMS SK, et al. Kinetics of endothelial cell surface attachment forces. J Vase Surg 1988:7:591-599. 22. JARRELL BE, WILLIAMS SK, SOLOMON L, et al. Use of an endothelial monolayer on a vascular graft prior to implantation. Ann Surg 1986;203(6):671-678.
