A new canine model for evaluating blood prosthetic arterial graft interactions Colleen M. Brophy,* Ralph K. Ito, William C. Quist, Michael S. Rosenblatt, Mauricio Contreras, Athanassios Tsoukas, and Frank W. LoGerfo New England Deaconess Hospital Harvard Surgical Service, Boston, Massachusetts 02215 Various models have been proposed to examine blood-prosthetic materials interactions in terms of the effect of the prosthetic material on platelet structure and function, blood coagulation and fibrinolysis, and tissue infiltrates (cellular or acellular). In addition, these models have been used to examine the change in the graft surface over time. Particular difficulties in examining graft-materials interactions include species differences, short residence time for blood-materials interactions with commonly employed short grafts, and length of study limitations with ex vivo shunts. In this paper we report a canine, carotid-aorta subcutaneous prosthetic graft model. The specific
advantages of this model are the length of the graft, which allows prolonged contact of blood with the prosthetic surface; the subcutaneous location of the graft, which allows repeated sampling of blood along the graft; and the healing characteristics of canine grafts. We selected the canine model because the healing characteristics are morphologically similar to those in humans in that endothelialization of the prosthetic surface is limited. Other models, such as the pig, are favored for use when examining blood coagulation, platelet, or fibrinolytic studies; however, these models can fully endothelialize prosthetic surfaces.
Significant advances in vascular surgical technique have resulted in excellent short-term patency of arterial bypass grafts. In spite of this, long-term patency remains limited, especially for prosthetic grafts. The major factor contributing to decreased long-term patency is anastomotic intimal hyperplasia.1,2Intimal hyperplasia has been shown to be greater at the downstream anastomosis, suggesting that blood interactions with the prosthetic materials may augment the hyperplastic res~onse.~ In particular, platelet activation may have a role in intimal hyperplasia in that platelet derived mitogens (PDGF) may induce smooth muscle cell pr~liferation.~ A model was developed to examine interaction between the flowing blood and the prosthetic conduit. The model is designed to maximize the contact activation of platelets by the artificial surface under arterial flow conditions and is similar in flow characteristics and length to axillofemoral grafts in hu*To whom correspondence should be addressed at New England Deaconess Hospital, 110 Francis Street, Suite 7C, Boston, MA 02215. Journal of Biomedical Materials Research, Vol. 25, 1031-1038 (1991) CCC 0021-9304/91/081031-08$4.00 0 1991 John Wiley & Sons, Inc.
BROPHY ET AL.
mam5 This model will allow correlation of blood prosthetic material interactions with morphologic events as the graft matures and the flow surface undergoes passivation. MODEL
Large (25-30 kg) mongrel dogs are kept NPO the night before surgery and are anesthetized with (20 mg/kg) thioamylal intravenously (IV) and anesthesia is maintained with endotracheally administered halothane. One gram of a cephalosporin is administered IV preoperatively, a repeat dose is administered 4 h later, and subsequently the dogs receive one gram of cephalosporin daily intramuscularly for 3 days. The animals are placed on the operating table in the right lateral decubitus position, shaved, prepped, and draped. The animals receive 1 liter of intravenous normal saline with 5% dextrose throughout the procedure . We use a retroperitoneal approach to the abdominal aorta. A curvilinear incision is made from the junction of the lowest rib and the anterior aspect of the spine to the left lateral abdomen just above the iliac crest. The external oblique is split, the internal oblique divided, and the fascia of the rectus abdominus is detached from its posterior insertion. The aorta is readily located anterior to the spine. Care must be taken in obtaining control of the aorta to avoid entering the cysterna chyle which lies directly on top of the aorta. If the cysterna chyle is inadvertently injured it makes the ensuing dissection more difficult but we have not noted any adverse clinical sequelae. We have found that starting the disection distally (just above the aortic bifurcation) and proceeding proximally will usually allow complete exposure of the aorta without damage to the cysterna. There are usually a pair of lumbar arteries located midway between the renal arteries and the bifurcation and once identified and dissected these branches are encircled with elastic vessel loops. There are rarely any other branches posteriorly. Small anterior branches can be ligated without sequelae. Doubly looped small red rubber vessel loops are used to obtain proximal and distal control of the aorta. The common carotid artery is then exposed, through an oblique incision in the neck, posterior to the sternocleidomastoid muscle. The carotid sheath is infiltrated with a balanced salt solution containing 60 mg/500 mL papaverine hydrochloride and 2000 U of heparin sodium prior to dissection to avoid spasm.6 Baseline arterial samples are collected from the aorta with a 19-gauge butterfly needle. Additionally, 20 cc of blood is obtained to preclot the graft. The 0.8 X 70 cm knitted Dacron graft (C. R. Bard Inc., Billerica, MA) is preclotted by infusing it with blood, stripping the graft of blood, and allowing it to set in blood until the blood clots. The graft is irrigated with heparinized saline prior to insertion. A subcutaneous tunnel from the carotid incision to the retroperitoneal abdominal incision is made using blunt dissection with a large vascular clamp. Two counter incisions are made, just above the scapula and at the midchest to facilitate tunneling. The graft is then pulled through the tunnel over the left shoulder to where it is passed between the 12th rib at
PROSTHETIC ARTERIAL GRAFT INTERACTIONS
Figure 1. An artist’s rendition of the canine carotid-aorta subcutaneous arterial graft. Arterial blood samples are obtained through direct graft puncture, at various time intervals, at sites W (at the proximal anastomosis), X ( 5 cm from the proximal anastomosis), Y (25 cm from the proximal anastomosis), and Z (50 cm from the proximal anastomosis).
its spinal insertion point. Approximately 8 cm of graft is thus in the retroperitoneal space and inaccessable to sampling. The animal is heparinized with 5000 U of heparin IV and another set of blood samples are obtained from the distal aorta. The vessel loops are cinched up to impede blood flow in the aorta and an arteriotomy is made along the anterolateral aspect of the aorta just above the aortic bifurcation with a knife. The arteriotomy is extended to a total length of approximately 2 cm with Potts scissors. The graft is cut obliquely and the distal anastomosis is constructed between the end of the graft and the side of the aorta with a running 6-0 prolene. The graft is flushed with blood and clamped just above the anastomosis, restoring blood flow through the aorta to the lower extremities. The proximal anastomosis is then constructed end-to-side onto the carotid artery with a running 6-0 prolene. Proximal control on the carotid artery is
BROPHY ET AL.
maintained with a small vascular clamp. The graft flushed prior to securing the suture. The distal carotid artery is then ligated and divided. The aorta, proximal to the graft is then ligated with a cotton umbilical tape. It is important to completely occlude inflow from the aorta to avoid graft occlusion secondary to competitive blood flow from the carotid and aorta. Flow rates are measured through the graft with an ultrasonic flow meter (Transonics Corp., Ithaca, NY). Hemostasis is obtained and the incisions closed with 2-0 and 3-0 interupted absorbable suture. Sham operations are performed in a similar manner except both anastomoses were omitted. In the sham operations, an arteriotomy is made in the distal aorta and closed with a running 6-0 prolene and the left carotid artery exposed and ligated. Blood samples may be obtained with a 19-gauge needle percutaneously, directly through the graft, at the distal portion of the graft, -50 cm from the proximal (Z), the midportion of the graft, -25 cm ( Y ) , 5 cm distal to the proximal anastomosis (X), and at the proximal anastomosis (W). In a similar fashion, samples may be obtained at various time intervals to assess changes in blood reactivity as the graft matures. The specimens are collected directly into syringes containing appropriate anticoagulants. Femoral artery samples are collected in sham dogs. All blood drawing is performed with the dogs under thioamyl sedation. The dogs are kept in cotton jackets until their wounds are healed. The cotton jackets were specifically constructed out of thick denim material with two holes for the forefeet and are secured over the back of the animal with a Velcro strap. At sacrifice, the dogs are anesthetized and the previous incisions reopened. Careful sharp and blunt dissection is used to obtain exposure of the carotid artery proximal to the graft and distally along the aorta at the aortic anastomosis. Heparin, 5000 U, is infused intravenously. A 16-gauge butterfly catheter is placed in the carotid artery proximally and the aorta distal to the anastomosis. The carotid is clamped proximal to the butterfly catheter and the aorta is clamped distal to the butterfly catheter. The dogs are sacrificed with an intravenous bolus to a saturated potassium chloride solution. The inflow butterfly catheter is infused with normal saline until the effluent is clear. The distal butterfly is left open to drain. The grafts are then perfused with 2% glutaraldehyde, 4% formaldehyde in phosphate buffered saline (pH 7.4) at 100 mm Hg pressure from a pressure bag for 15 min.7 The distal butterfly is clamped and the proximal infusion line is clamped so that the pressure in the graft remains at 100 mm Hg. The graft is exposed throughout its length and removed, along with the perigraft tissue. The tissue may then be processed for both routine histological and ultrastructural analysis. We have currently implanted 14 grafts with this canine model of a long subcutaneous carotid-aorta Dacron prosthetic graft and have performed seven sham operations. This model results in a flow rate of 125 f 35 mL/min through the graft. Of the 14 animals with grafts, 4 were sacrificed for graft infection or graft exposure secondary to self-inflicted wounds. These complications occurred
PROSTHETIC ARTERIAL GRAFT INTERACTIONS
prior to implementing cotton dog jackets and were effectively prevented by covering the wounds with the jackets. There was one early thrombosis and one graft occluded at 1.5 months. Three perioperative deaths occurred early in our experience with the model secondary to technical mishaps and anesthetic complications. The remaining grafts remained patent until the dogs were sacrificed at 8 months. Scanning electron micrographs of the midportion of the grafts revealed red blood cells enmeshed in fibrin with scattered adherent platelets but no endothelial lining (Fig. 2).
The implantation of prosthetic arterial bypass grafts results in prolonged exposure of blood to an artificial surface. Acutely, this results in platelet activation, adhesion, and release; activation of the intrinsic pathway; and the formation of thrombin.8 This results in modification of the graft surface with deposition of fibrinogen, immunoglobulins, and subsequently highmolecular-weight kininogens.' With time the graft surface becomes progressively less reactive, a process referred to as "passivation."l"Although dogs appear to lack sufficient levels of factor XI1 and prekallikrein to be detected
Figure 2. Scanning electron micrograph of the endoluminal Dacron graft surface that had been implanted for 8 months. Note the lack of an endothelial surface. There are numerous red blood cells enmeshed in a fibrin lattice. There are scattered adherent platelets present.
BROPHY ET AL.
by dextran sulfate/acetone activation, or kaolin," they represent a good model in that arterial bypass grafts in dogs do not form an intact endothelial surface throughout the length of the graft. The graft surface "passivates" in a fashion similar to that which occurs with human grafts." In this model the surface of the midportion of the graft did not endothelialize in any of the dogs over the course of 8 months. A pannus ingrowth of endothelium occurred at each anastomosis. This ingrowth extended only 1-2 cm into the graft. Other investigators, using immunohistochemical techniques, have identified the presence of a pseudointima containing fibronectin, fibrinogen/ fibrin, and platelet membranes in the luminal surface of dacron grafts that have been in vivo for up to 6 years in h ~ m a n s .Further '~ work with this model is needed to determine the exact matrix composition of the pseudointima as it changes over time in this dog model. The long-term effects of the graft surface, as it changes with time, on blood components, platelet reactivity, thrombosis, and fibrinolysis are not well known. Various models have been proposed to study in vivo blood-arterial prosthetic interactions. Inherent differences in platelet responses, coagulation parameters, and graft healing are present among different mammalian species (Table I).14,15 Clagett, using a canine thoracoabdominal Dacron graft, has shown persistent biochemical and functional alterations in platelets over time.I6In particular platelet counts diminished postoperatively and normalized 4-6 weeks postoperatively; platelet survival decreased, was maximally shortened during the first 6 months postoperatively, and gradually lengthened over a year; and platelet serotonin levels decreased coincident with platelet survival. Although this model has provided important information about platelet reactivity to in vivo grafts over time, the graft is placed in a deep location and is as such inaccessible for measuring changes within or across the graft. The canine carotid-aorta Dacron graft model described in this report represents an opportunity to examine graft materials interactions in long-term survival studies with repeated sampling. This allows analysis of changes in TABLE I A Comparison of the Different Mammalian Species Commonly Used to Evaluate Blood-Prosthetic Graft Interactions Human
Extrinsic clotting Intrinsic clotting Fibrinolysis PLT AGG-ADP Thrombin Collagen Dacron graft
++ ++ + -e
E or I E or I E or I
++ + + +e
E or D I E or D
I I I
E I E
+ + +
Key: E = equivalent, I = increased, D = decreased; + = aggregates, +/- = variable response; -e the implanted arterial prostheses are covered with a pseudointima and pannus ingrowth is present at the anastomosis but there is no endothelial surface at the midportion of the graft; +e the implanted arterial prostheses are covered with an endothelial surface.
PROSTHETIC ARTERIAL GRAFT INTERACTIONS
reactive components across the graft over time. Important considerations in using this model are a retroperitoneal approach to minimize fluid loss, the need for meticulous dissection since transfusion is not available, and the necessity of protecting the surgical wounds with a cotton jacket. The major advantages of our model to study blood-materials interaction and subsequent healing responses are: (a) the direct anastomosis between the native artery and the graft material eliminates blood contact with other biomaterials as occurs in ex vivo shunt models, (b) the subcutaneous positioning of the graft allows easy access for chronic sampling studies and (c) the length of the graft provides extended residence time for the blood-materials interactions and enhances the progression of bioreactive events for study, and (d) graft healing in dogs is similar to humans in that there is an absence of an endothelialized surface in the midportion of the graft. This work was supported by a grant from the NIH (RO1 HL21796).
References 1. J. A. DeWeese, ”Anastornotic intimal hyperplasia,” in Vascular Grafts, P. M. Sawyer and M. J. Kaplitt (eds.), Appleton-Century-Crofts, New York, 1978. 2. A.M. Imparato, A. Bracco, G. E. Kim, and R. Zeff, “Intimal and neointima1 fibrous proliferation causing failure of arterial reconstructions,” Surgery, 72, 1007-1017 (1972). 3. EW. LoGerfo, W.C. Quist, M. D. Nowak, H.M. Crawshaw, and C.C. Haudenschild, “Downstream anastomotic hyperplasia: A mechanism of failure in dacron arterial grafts,” Ann. Surg., 197, 479 (1983). 4. R. Ross, E.W. Raines, and D. F. Bowen-Pope, “The biology of plateletderived growth factor,” Cell, 46, 155-169 (1986). 5. F.W. LoGerfo, W.C. Johnson, J.D. Corson, et al., “A comparison of the late patency rates of axillobilateral femoral and axillounilateral femoral grafts,” Surgery, 81, 33-40 (1988). 6. F.W. LoGerfo, A. N. Sidawy, W.C. Quist, “A technique for prevention of spasm in in situ vein grafts,” Contemp. Surg., 28, 18-22 (1986). 7. V.S. Sottiurai, Sue S. Lim, M. K. Hsu, W. K. Mann, and R.C. Batson, ”Pseudointima formation in woven and knitted dacron Erafts: A comparitive ultrastructural analysis,” J. Cardiovasc. Surg., 30; 808-816 (1989). 8. R.W. Coleman, J. Hirsh, V. J. Marder, E.W. Salzman, (eds.), Hemostasis 9. 10.
and Thrombosis: Basic Principles and Clinical Practices, 2nd Ed., J. B. Lippincott, Philadelphia, 1987. L. Vroman and A. I. Adams, “Identification of rapid changes at plasmasolid interfaces,” J. Biomed. Mat. Res., 3/43-67 (1969). E.W. Salzman and E.W. Merrill, “Interactions of blood with artificial surfaces,“ in Hemostasis and Thrombosis: Basic Principles and Clinical Practice, 2nd Ed., J. B. Lippincott, Philadelphia, 1987, pp. 606-617. R. K. Ito and 6.E. Statland, “Centrifugal analysis for plasma kallikrein activity, with use of the chromogenic substrate S-2302,“ Clin. Chem., 27, 586-593 (1981). L.R. Sauvage, K. E. Berger, S. J. Wood, S.G. Yates, J.C. Smith, P. B. Mansfield, “Interspecies healing of porous arterial prostheses,” Arch. Surg., 109, 698-705 (1974).
BROPHY ET AL.
13. J. M. Anderson, M. E Abbuhl, T. Hering, and K. H. Johnston, "Immunohistochemical identification of components in the healing response of human vascular grafts," ASAIO, 8, 79-85 (1985). 14. Guidelines for Blood-Materials Interaction: Report of the National Heart, Lung, and Blood Institute Working Group. NIH Publication. #85-2185, September, 1985. 15. C. F. Scott, "Appropriate animal models for research in blood in contact with artificial surfaces," in Blood in Contact with Natural and Artificial Surfaces, New York Academy of Sciences Press, New York, 1987, pp. 636-637. 16. G. P. Clagett, M. Russo, and H. Hufnagel, "Platelet changes after placement of aortic prostheses in dogs. I. Biochemical and functional alterations," J. Lab. Clin. Med., 97, 345-359 (1981). Received April 6, 1990 Accepted March 11, 1991