Active tissue factor shed from human arterial smooth muscle cells adheres to artificial surfaces.

Through a series of in vitro assays, this study outlines a flow-mediated process by which active tissue factor (TF), the prime initiator of coagulation, may be transferred from the plasma membrane of vascular smooth muscle cells (VSMCs) to that of artificial surfaces such as those typically associated with intravascular implants. Studies with quiescent and activated rat VSMCs demonstrated that pathologically high shear stresses (tau(w) = 250 dyn cm(-2)) resulted in the loss of TF activity from the cell surface. Subsequent experiments with human VSMCs showed that VSMCs continuously release active TF into their extracellular medium, presumably in the form of lipid vesicles or microparticles, and that fluid shear stress (tauw = 50 dyncm(-2)) or chemical agonists (A23187) can significantly accelerate this release. Experiments with a wide array of polymeric and metallic materials showed that the TF shed from VSMCs was able to adhere to these surfaces and promote the activation of coagulation factor X (FX) at the material surface. Extracellular TF bound strongly to both uncoated and human plasma coated surfaces under a wide range of hemodynamic shear stresses (0-20 dyncm(-2)). When an extracellular, VSMC-derived TF mixture was perfused over Ti 6-4 surfaces, the adhesion of TF was found to be time-dependent, gradually accumulating on the material surface over time. Thus an important criterion in the design or success of intravascular devices may be related to their ability to interact with TF, shed from cell surfaces. This is especially important as TF may lead to thrombotic complications, the products of which may also increase cellular proliferation.

Residence-time dependent changes in fibrinogen adsorbed to polymeric biomaterials.

It has generally been accepted that biomaterials adsorbing the least amount of the plasma protein fibrinogen following exposure to blood will support less platelet adhesion and therefore exhibit less thrombogenicity. Several studies suggest, however, that the conformation or orientation of immobilized fibrinogen rather than the total amount adsorbed plays an important role in determining the blood compatibility of biomaterials. The purpose of this study was to investigate time-dependent functional changes in fibrinogen adsorbed to polytetrafluoroethylene (PTFE), polyethylene (PE), and silicone rubber (SR). Fibrinogen was adsorbed to these materials for 1 min and then allowed to 'reside" on the surfaces for up to 2 h prior to assessing its biological activity. Changes in fibrinogen reactivity were determined by measuring the adhesion of 51Cr-labeled platelets, the binding of a monoclonal antibody (mAb) directed against an important functional region of the fibrinogen molecule (the gamma-chain dodecapeptide sequence 400-411), and the ability of blood plasma to displace previously adsorbed fibrinogen. Platelet adhesion differed among the polymeric materials studied, and PTFE and PE samples exhibited a small decrease in adhesion with increasing fibrinogen residence time. Platelet adhesion to SR was the least among all materials studied and showed no variation with residence time. When using PTFE and SR as substrates, mAb recognition of adsorbed fibrinogen did not change with residence time whereas that on PE decreased slightly. The mAb binding was least to fibrinogen adsorbed to SR, which is in agreement with the platelet adhesion results. Finally, the ability of plasma to displace previously adsorbed fibrinogen decreased dramatically with increasing residence time on all materials. These in vitro studies support the hypothesis that fibrinogen undergoes biologically significant conformational changes upon adsorption to polymeric biomaterials, a phenomenon that may contribute to the hemocompatibility of the materials following implantation in the body.

Vascular cell attachment and procoagulant activity on metal alloys.

The attachment and growth of vascular smooth muscle cells on biomaterials used as components of devices implanted in the vascular space may influence the biocompatibility of such materials. The nature of the materials may affect the attachment and/or the activation of these cells' procoagulant responses. Therefore, the main objective of this study was to measure the strength of adhesion of these vascular cells to potential biomaterials (titanium, zirconium alloys, and stainless steel) by exposing them to a range of shear stresses (50-300 dyn cm(-2)) in a parallel plate flow chamber. The procoagulant responses of the cells were evaluated by measuring the tissue factor (TF) activity promoted by the different materials under flow conditions. The materials supported distinctly different levels of initial cell adhesion in static culture. However, the fraction of adherent cells did not decline significantly with incrementally increasing shear stress within the range tested. TF expression, as measured by factor Xa (FXa) production. was material-dependent. For example, cells cultured on Ti1313 exhibited more FXa production (13.2 nM 10(-5) cells) than Ti1313(DH) (8.5 nM 10(-5) cells) or stainless steel (2 nM 10(-5) cells). Thus, our studies indicate that the level of adhesion, strength of attachment and the expression of procoagulant activity of adherent vascular cells depend strongly on the nature of the underlying biomaterial.

Mechanical factors affecting hemostasis and thrombosis.

Both physical and chemical factors can influence the activity of platelets and coagulation factors responsible for the formation of thrombotic and hemostatic masses in the vicinity of an injured vessel wall. Studies performed in controlled shear devices (viscometers) have indicated that physical factors alone can induce platelet aggregation, even in the absence of exogenous chemical factors. The physical considerations which appear to be important for the local activation of hemostatic/thrombotic mechanisms appear to be related to the magnitude of the shear rate/stress, the duration of the applied physical force and the local geometry. Blood flow alone has multiple influences on platelet and coagulative mechanisms. It has been well established that at physiologically encountered shear conditions, increases in the local shear rate enhance the attachment of platelets to the vessel wall and the growth of platelet aggregates on adherent platelets. In contrast, increases in local shear conditions inhibit the production of fibrin formation on surfaces where tissue factor (TF) is exposed. At levels of shear rate/stress high as compared to normal physiological conditions, but comparable to those observed at the apex of severely stenosed vessels, platelet aggregate formation is dependent on the duration of the exposure time. Considerable advances in our understanding of flow-related mechanisms have evolved from the use of well-defined perfusion chambers employing parallel flow streamlines. However, processes leading to hemostasis and thrombosis generally occur in more complicated flow situations where flow streamlines are not parallel and in which abnormally high, as well as abnormally low, shear rates and shear stress levels may be encountered in close proximity to each other.

A computational analysis of FXa generation by TF:FVIIa on the surface of rat vascular smooth muscle cells.

A computational model was developed to investigate the contribution of classical mass transport and flow parameters to factor X (FX) activation by the tissue factor-factor VIIa complex (TF:VIIa) on one wall of a parallel-plate flow chamber. The computational results were compared to previously obtained experimental data for the generation of factor Xa (FXa) by TF:VIIa on the surface of cultured rat vascular smooth muscle cells. In this study, the complete steady-state convection-diffusion equation was solved using the commercial software package, FLUENT (Fluent Inc., Lebanon, New Hampshire). A user-defined subroutine interfaced with FLUENT implemented the surface reaction which was modeled using classical Michaelis-Menten reaction kinetics. The numerical solutions were obtained for 12 cases which used combinations of three wall shear rates and four reaction rates. The numerically obtained fluxes for a given reaction rate displayed a wall shear rate dependence which ranged from classical kinetic reaction control (no dependence) to pure diffusional control (maximum dependence). The experimental data, however, were not represented by numerical data generated using a single reaction rate. The three numerically obtained fluxes which corresponded most closely to the experimental fluxes were determined using three different Vmax values. This finding supports the hypothesis that there may be a direct effect of flow on the TF:VIIa complex or the cell membrane.

Factor Xa generation at the surface of cultured rat vascular smooth muscle cells in an in vitro flow system.

The purpose of the present investigation was to explore the effects of well-defined flow conditions on the activity of tissue factor (TF) expressed on the surface of cultured rat vascular smooth muscle cells. Cells were cultured to confluence on Permanox brand slides and stimulated to express TF by a 90 min incubation with fresh growth medium containing 10 percent calf serum. The stimulated cells were then placed in a parallel plate flow chamber and perfused with Hank's Balanced Salt Solution containing factor VIIa, factor X (FX), and calcium. The chamber effluent was collected and assayed for factor Xa (FXa) and the steady-state flux of FXa was calculated. The flux values were 68.73, 94.81, 139.75, 138.19, 316.82, and 592.92 fmole/min/cm2 at wall shear rates of 10, 20, 40, 80, 320, and 1280 s-1, respectively. The FXa flux depended on the wall shear rate to a greater degree than predicted by classical mass transport theory. The flux at each shear rate was three to five times less than that calculated according to the Leveque solution. These features of the experimental data imply nonclassical behavior, which may partially result from a direct effect of flow on the cell layer.

Hemocompatibility studies of surface-treated polyurethane-based chronic indwelling catheters.

The objectives of this research were to evaluate and compare the interactions of several polyurethane-based central venous catheter materials with blood. Specifically, measurements of fibrinogen adsorption, platelet adhesion, kallikrein generation, and fibrinopeptide A (FPA) release were performed. The catheter materials examined in this study included: platinum-cured, 50 shore A durometer, barium sulfate-filled, silicone (SI); Tecoflex EG85A-B20 polyurethane (PU); PU catheters whose outer surface had been impregnated with ion beam-deposited silver atoms (AgI and AgII); PU catheters coated with a hydrophilic, polyacrylic acid polymer (UC); PU catheters coated with an air-cured PTFE emulsion (CS); and PU catheters coated with an aminofunctional dimethylsiloxane copolymer (JG). The time course of fibrinogen adsorption from plasma to the SI, JG, PU, and CS materials was similar, with CS exhibiting the least amount of adsorbed fibrinogen after 1 h (65 +/- 4.7 ng cm-2) and PU the greatest (144 +/- 16.5 ng cm-2). After 90 min of contact, AgI and AgII exhibited the greatest number of adherent platelets, levels that were approximately two to three times higher than those on the other catheter materials. With the exception of UC and PU, which caused kallikrein generation levels approximately half that of the positive (glass) control, little kallikrein formation was observed for any of the materials relative to the negative control. Finally, FPA generation was greatest using the SI, CS, and PU materials, with the latter causing the production of almost four times the amount of FPA as the negative control. This preliminary assessment of the hemocompatibility of the various catheters suggests that the surface treatments did not adversely affect their interactions with blood components; further investigations of these materials are therefore warranted in order to completely characterize their behavior prior to use in clinical situations.

Initial hemocompatibility studies of titanium and zirconium alloys: prekallikrein activation, fibrinogen adsorption, and their correlation with surface electrochemical properties.

Two novel metal alloys, Ti-13Nb-13Zr and Zr-2.5Nb, have been engineered for applications in orthopedic implants because of their favorable mechanical properties, corrosion resistance, and compatibility with bone and tissue. These alloys also have the ability to form a hard, abrasion-resistant, ceramic surface layer upon oxidative heat treatment (diffusion hardening, DH). Previous studies have indicated that these and other ceramics cause limited hemolysis and exhibit remarkable structural integrity after extended exposure to physiological environments. Such observations suggest that DH Ti-13Nb-13Zr and ZrO2/Zr-2.5Nb could be used successfully as components in blood-contacting devices. Materials intended for such applications must possess properties that do not elicit adverse physiological responses, such as the initiation of the coagulation cascade or thrombus formation. In the present study measurements of prekallikrein activation, fibrinogen adsorption from diluted human plasma, and the strength of fibrinogen attachment as judged by residence-time experiments were performed to evaluate the potential hemocompatibility of these materials. The results of the prekallikrein activation and fibrinogen-retention studies correlated well with two electrochemical properties of the alloys, the open circuit potential and reciprocal polarization resistance. The results indicate that both the original and treated Ti and Zr alloys activate prekallikrein and adsorb as well as retain fibrinogen in amounts similar to other materials used as components of blood-contacting devices. On the basis of these studies, these alloys appear to be promising candidates for cardiovascular applications and merit further investigation.

Identification of quantifiable hemodynamic factors in the assessment of cerebral aneurysm behavior. On behalf of the Subcommittee on Biorheology of the Scientific and Standardization Committee of the ISTH.

Previous experimental and theoretical studies on the hemodynamics of saccular intracranial aneurysms have provided evidence that aneurysms tend to grow, thrombose and rupture when (1) wall shear stress and mural tension are increased compared to normal values, and (2) flow deviates from a laminar unidirectional pattern (for example flow recirculation). Aneurysm wall shear stress, however, is the only hemodynamic factor which has received special attention in terms of estimation. Additional flow-related parameters exist which could potentially bring increased insight into mechanisms for cerebral aneurysm behavior; they could also help categorize the severity of such malformations and design effective intravascular treatment techniques. The purpose of this paper is thus to present an overview of such hemodynamic factors that could assist in determining the geometries which present the greatest risks to patients. These parameters include (1) hemodynamic shear stress, (2) pressure and related stresses, (3) impingement force on the aneurysm wall, (4) inflow rate into the aneurysm, and (5) residence time of blood within the aneurysmal sac. In addition, these factors can also be currently estimated in an in vitro setting.

Correction of the platelet adhesion defect in delta-storage pool deficiency at elevated hematocrit--possible role of adenosine diphosphate.

Previous studies on patients with storage pool deficiency (SPD) who are specifically deficient in platelet dense granules (delta-SPD) have suggested a role for dense granule substances, in all likelihood adenosine diphosphate (ADP), in mediating thrombus formation on subendothelium at high shear rates. The role of dense granule substances in mediating platelet adhesion appears to be more complicated Previous studies in delta-SPD suggested an adhesion defect that was strongly influenced by the patient's hematocrit (Hct) value. To explore further the possibility that red blood cells (RBCs) may influence the role that platelet storage granules play in mediating adhesion at high shear rates, we have measured adhesion (and thrombus formation) throughout a preselected range of Hct values (30% to 60%) in normal subjects and in patients with delta-SPD. The present studies confirm the defect in platelet adhesion in patients with delta-SPD, most significantly at Hct values of 30% to 40%. This defect (but not that of thrombus formation) can be completely corrected by the addition of RBCs. The correction of the platelet adhesion defect by RBCs was specific for delta-SPD; it was not observed in either von Willebrand's disease or thrombasthenia. Studies performed on normal blood under conditions that could be expected to block any effect of ADP on adhesion and an analysis of the type of adhesion defect in delta-SPD suggest that ADP may be involved in the process required for platelet spreading on the subendothelium. The corrective effect of RBCs on platelet adhesion in delta-SPD appears to be chemical rather than physical in nature, possibly due to shear-induced release of RBC ADP or to other recently described properties of RBCs that enhance collagen-induced platelet interactions.

A numerical analysis of factor X activation in the presence of tissue factor--factor VIIa complex in a flow reactor.

A mathematical model has been developed to investigate previously obtained experimental findings relating to the activation of factor X by surface-bound tissue factor--factor VIIa (TF:VIIa) in a tubular flow reactor. In those experiments, factor X was perfused through a microcapillary tube over a range of flow (shear) conditions and the activated product, factor Xa, was measured at the outlet of the tube using a chromogenic assay. In the present study, the steady-state convection-diffusion equation with Michaelis-Menten kinetics used to describe the reaction at the wall was numerically integrated using an implicit method based on linear systems of ordinary differential equations. The results from the numerical analysis indicated that shear rate directly affects both Km and Vmax. Values of Km decreased from 151 to 16 nM as the shear rate increased from 25 to 2400 sec-1. Additionally, there was a twofold increase in Vmax from 1.4 to 3.0 pmol/cm2/min as the shear rate increased from 25 to 300 sec-1. These findings are in contrast with classical enzyme behavior and imply a direct effect of fluid flow on the kinetics of factor X activation.

Antiplatelet agents affecting the interaction of Tissue Factor-Factor VIIa complex with Factor X in a continuous-flow reactor.

The purpose of the present study was to examine the role of antithrombotic agents in the activation of Factor X in the presence of the Tissue Factor-Factor VIIa (TF-VIIa) complex in a continuous-flow reactor. Tissue Factor immobilized in a phospholipid bilayer on the inner surface of a capillary tube (internal diameter = 0.27 mm) was exposed to a perfusate containing Factors VIIa and X flowing at a flow rate of 12.7 microliters/min, corresponding to a wall shear rate of 100 s-1. Factor Xa (the activated form of Factor X) in the effluent was determined by a chromogenic assay. The effectiveness of two platelet aggregation inhibitors, alpha,alpha'-bis-[3-(N,N-diethylcarbamoyl)piperidino-p-xylene dihydrobromide (A-1) and alpha,alpha'-bis-[3-N-benzyl-N-methylcarbamoyl)piperidino]-p-xylen e dihydrobromide (A-4) in inhibiting Factor X activation is reported here. The results suggest that the Tissue Factor pathway, mediated through TF-VIIa complex, produces significantly lower levels of Factor Xa in the presence of compounds A-1 and A-4. On the basis of these findings, it appears that the anticoagulation action of these compounds reinforces their platelet aggregation-inhibitory properties. These carbamoylpiperidines (nipecotamides) therefore appear to be useful antithrombotic agents.

Computer modeling of intracranial saccular and lateral aneurysms for the study of their hemodynamics.

There is strong evidence indicating hemodynamic stress as an underlying cause for saccular intracranial aneurysm growth, thrombosis, and/or rupture. We examined flow fields encountered in models of cerebral aneurysms having a lateral (originating from the side of an artery, not at a branch point) geometric configuration. Shear stress and pressure gradients acting on aneurysm walls under a variety of flow and geometric conditions were evaluated. For this purpose, a two-dimensional finite-element computer model of lateral aneurysms in a steady-flow state was developed. Three idealized aneurysm shapes were studied, half-spherical, spherical, and pear-shaped. The ostium width of the cerebral aneurysm, relative to the radius of the parent artery and the Reynolds number, were also varied. Maximal shear stresses and maximum pressures (for an ostium width of 2 times the radius of the parent artery) were typically found at the downstream site of the ostium, rather than at the dome of the aneurysm. In general, the highest shear stresses and the lowest pressures (at the distal portion of the ostium) were obtained in the spherical aneurysm, whereas the lowest shear stresses and the highest pressures were found in the half-spherical aneurysm. The location of maximal stresses (shear and pressure) at the distal region of the ostium suggests that growth and/or rupture may well proceed from this point. Such findings are in contrast to the commonly held opinion that aneurysm rupture occurs at the dome. Careful pathological investigation will need to be performed to clarify this finding. The results of this preliminary investigation also indicate that the flow field in lateral aneurysms is highly dependent on a number of factors related to flow and geometric parameters. Geometry seems to be a significant mediator of local magnitudes of stress. Thus, the tendency for growth or thrombosis may be influenced by variations in size or shape.

Mathematical Analysis of Bleeding Time Data in Patients with Platelet Disorders and von Willebrand's Disease.

In a previous study the volume of blood obtained from bleeding time incisions was measured every 30 s from normal subjects (n = 15), patients with thrombasthenia (TSA, n=4), idiopathic thrombocytopenic purpura (ITP, n=4), von Willebrand's disease ((v)WD, n=3), Bernard-Soulier syndrome (BSS, n = 2), and δ and αδ-storage pool deficiencies (SPD, n=4 and 5, respectively) and the experimental results analyzed by empirical curve-fitting of the data. In the present investigation, a mathematical model based on blood flow physiology was developed to describe the rate of blood loss over time from these same patients as a function of two parameters, α, which describes the magnitude of vessel contraction following transection, and ß, the rate of vessel dilation to its nominal diameter. For the normal controls a third parameter, δ, was used to describe the rate of vessel closure due to the formation of a hemostatic plug. Optimal values for these parameters for the normal subjects and each patient group were determined by least-squared fitting of the experimental bleeding time data. For all subjects, values for the magnitude of vessel contraction were similar (α=0.65±0.02). However, values for ß were reduced in both TSA (ß=0.22±0.04) and (v)WD (ß = 0.30±0.03) and were increased relative to normal controls (ß=0.39±0.03) in BSS (ß=0.50±0.01) and both δSPD (ß=0.50±0.07) and αδSPD (ß=0.50±0.05). The initial rate of blood loss was also significantly greater in patients with BSS, ITP, δ-SPD, and αδSPD than in the normal subjects, as determined by a one-way analysis of variance. These results suggest that: (1) the initial contraction of severed blood vessels does not appear to be mediated by any plasma or platelet compounds absent in the various bleeding disorders considered in this study; and (2) the increased initial bleeding observed in SPD may reflect the absence of vasoactive agents, such as ADP or serotonin, released from platelet dense granules following platelet activation. These conclusions are consistent with those reported previously on the same patients and indicate that mathematical modeling of bleeding time measurements, based on assumptions of vascular and platelet reactivity, can provide insights into the complex series of events occurring at sites of vessel injury.

A perfusion chamber developed to investigate thrombus formation and shear profiles in flowing native human blood at the apex of well-defined stenoses.

The precipitating event leading to stroke, myocardial infarction, and/or sudden death may be related to the formation of mural thrombus at the site of a ruptured or superficially damaged stenotic plaque. The fluid dynamic properties at atherosclerotic plaques that may be implicated in this thrombus formation have been described in a wide variety of model systems in both the process of plaque rupture and the growth of platelet thrombi. In general, the local fluid dynamic conditions are complex and show major variations from flow in well-defined laminar flow systems. However, no studies have attempted to quantify the effect of stenosis-related disturbances on thrombus formation in native human blood and to compare them with the local fluid dynamics. We developed a parallel-plate perfusion chamber device in which thrombus formation is measured at the "apex" of eccentric stenoses and have correlated such measurements with values of the local fluid dynamics obtained by computer simulation. The extent of stenoses (reduction in the cross-sectional area of the blood flow channel) was 60%, 80%, and 89%, corresponding to "apex" wall shear rates of 2600, 10,500, and 32,000 sec-1, respectively. The wall shear rate in the laminar flow region proximal and distal to the stenoses was 420 sec-1. The surface of the stenosis was purified collagen type III fibrils that were exposed to flowing nonanticoagulated human blood drawn directly from an antecubital vein by a pump placed distally to the perfusion chamber. The resulting blood-collagen interactions were quantified by light microscopy by using a morphometric image analysis technique. Under all conditions studied, platelet thrombus formation at the "apex" was extensive.(ABSTRACT TRUNCATED AT 250 WORDS)

Platelet size distribution measurements as indicators of shear stress-induced platelet aggregation.

The mechanisms underlying shear stress-induced platelet aggregation (SIPA) were investigated by measuring changes in the platelet size distributions resulting from the exposure of human platelet-rich plasma (PRP) to well-defined shear stresses in a modified viscometer. Exposure of PRP to a shear stress of 100 dyne/cm2 for 1 min at 37 degrees C resulted in the loss of single platelets, an overall shift in the distribution to larger particle sizes, and the generation of platelet fragments. Treatment of PRP prior to shearing with a monoclonal antibody directed against platelet glycoprotein (GP) IIb-IIIa (integrin alpha IIb beta 3) at a concentration that completely inhibited ADP-induced platelet aggregation also inhibited SIPA. Furthermore, incubation of PRP with a recombinant fragment of von Willebrand factor (vWF) that abolishes ristocetin-induced platelet agglutination significantly inhibited but did not eliminate SIPA. Pretreatment of PRP with the tetrapeptides RGDS or RGDV, which constitute the GP IIb-IIIa peptide recognition sequences on fibrinogen and vWF, almost completely blocked platelet aggregation at 100 dyne/cm2, whereas the negative control peptide RGES had no discernible effect. Finally, incubation of PRP with a monoclonal antibody directed against the platelet vitronectin receptor (integrin alpha v beta 3) did not affect SIPA. These results indicate that both GP IIb-IIIa and GP Ib, the latter through its interaction with vWF, are required for SIPA at 100 dyne/cm2; that the interaction of GP IIb-IIIa with its adhesive ligands under shear stress can be inhibited by RGD-containing peptides; and that the vitronectin receptor on platelets, which shares the same beta 3 subunit as GP IIb-IIIa, plays no role in SIPA.(ABSTRACT TRUNCATED AT 250 WORDS)

Further studies on the presence of functional tissue factor activity on the subendothelium of normal human and rabbit arteries.

Although tissue factor (TF) activity has been observed on the subendothelial surface of rabbit aorta and human umbilical cord, immunofluorescent and in situ hybridization methods have failed repeatedly to demonstrate TF in the intima of human blood vessels. In the present study, TF activity on everted, de-endothelialized arteries was studied by two methods. One utilized a flow system and measured fibrin deposition and fibrinopeptide A formation. The other utilized a newly developed rotating probe system and measured the conversion of factor X to factor Xa in the presence of factor VIIa and Ca+2. The study attempted to control, or assess, the possibility that functional TF could have been exposed on the vessel surface by the procedures used to prepare the arterial segments. By both methods, TF activity was detected on the subendothelium of rabbit aortae and human umbilical arteries, and was unaffected by the length of storage or by inclusion of actinomycin D in the storage buffer. TF activity was also observed in the subendothelium of adult human ileo-colic, internal mammary, and renal arteries, studied by the rotating probe method. The latter may underestimate TF activity, as some of the factor Xa formed appears to bind to the subendothelial surface. TF activity (Xa formation) was detected on the luminal surface (subendothelium) of non-everted arteries, but increased activity was observed after eversion of the vessel. The source of the subendothelial TF, and its presence in normal subendothelium in vivo, requires further study. In addition, if any of the TF activity observed in this study was derived from injured endothelial or myointimal cells during preparation of the everted vessel segments, the techniques described could serve as a useful model for studying TF-induced thrombosis and factor Xa formation on injured blood vessels, and for evaluating the anti-thrombotic properties of TF-inhibitors.