Drug-Induced Thrombosis—Experimental,Clinical,and Mechanistic Considerations

Abstract

Awareness of the dangers of drug-induced thrombosis has recently been heightened and led to demand for improved testing methodology. For example, reports indicating that some selective inhibitors of cyclooxygenase-2 (COX-2) increase the risk of myocardial infarction and atherothrombotic events caused the withdrawal of rofecoxib from global markets and the issuance of warnings concerning the usage of other COX-2 inhibitors. Drugs may exert a prothrombotic state by a variety of mechanisms–those affecting the vessel wall, the blood flow, and/or different blood constituents. Our review serves as an update to that of Gerhard Zbinden published in 1976 by presenting recently acquired data that more fully elucidate the different mechanisms by which drugs are believed to induce thrombogenic effects and discussing new methods used to detect these without losing sight of the classical pathology of thrombosis. We offer correlations between experimental findings and clinical data and conclude that, because drugs may induce a prothrombotic state by a variety of mechanisms, they should be tested for these using appropriate experimental methods and animal models.

Introduction

Although pathologic hemostasis was described as early as 2650 BC by the Chinese physician Huang Ti (Anning, 1957), not until 1856 AD did the German pathologist Rudolph Virchow publish his triad of factors contributing to the development of thrombosis (Virchow, 1856). The three components included abnormalities of blood vessel wall, blood constituents, and blood flow, which have become known collectively as “Virchow’s triad” (Figure 1). Even though 150 years have passed since Virchow published his triad, it remains the best overall explanation of the participants in the pathogenesis of thrombosis. A current update of the three factors accounts for deviations in the endothelium and endocardium (“abnormalities of vessel wall”); in platelets and the coagulative and fibrinolytic pathways (“abnormalities in blood constituents”); and in hemorheology and turbulence at bifurcations, large vessels burdened by irregular atheroma, and stenotic regions (“abnormalities in blood flow”) (Chung and Lip, 2003).

The notion that drugs play an important role in the development of thrombosis has continually gained widespread attention. Thirty years ago, extensive investigations were undertaken to explain the apparent thrombogenic effect of steroidal oral contraceptives (Zbinden, 1976). During the last 2 years, researchers have again become attentive to drug-induced thrombosis following reports that some selective inhibitors of COX-2 increase the risk of myocardial infarction and atherothrombotic involvements led to the withdrawal of rofecoxib from global markets and the issuance of warnings about similar developments during usage of other COX-2 inhibitors.

Here we provide a detailed update of an earlier comprehensive review of different mechanisms by which drugs induce thrombogenic effects (Zbinden, 1976) by examining recent data acquired through mechanistic explorations and the new methods used to detect thrombogenic effects. We present the subject in a relevant, contemporary factual manner without losing sight of the classical pathology of thrombosis as we review the different mechanisms and factors involved in drug-induced thrombosis.

This review follows the basics for the induction of thrombosis as stated by Virchow’s triad, and starts with an overview of the physiological mechanisms of thrombosis. Afterwards we review the different pathways of drug-induced thrombosis. This review is accompanied by a series of figures, demonstrating the effect of drugs on the 3 components of Virchow’s triad, exemplifying the contribution of the endothelial vessel wall, the coagulation cascade and platelet aggregation to thrombosis induction.

Overview of Physiologic Mechanisms of Thrombosis

Vascular Mechanism of Thrombosis

Injury to the vessel wall in general and to the endothelial lining specifically may be the major event that precipitates the thrombotic process (Slauson and Cooper, 2002). Platelets, white blood cells and fibrin rarely adhere to normal intact endothelium. In contrast, the deeper parts of the vessel wall have considerable thrombogenic properties. If a vessel is injured, the injured vessel along with platelets and circulating coagulation factors combine in a series of interlacing reactions resulting in vasoconstriction and the formation of a coagulum consisting of platelets and fibrin.

The endothelial cells do not merely form a passive layer between the blood and the rest of the blood vessel, but perform vital functions in hemostasis, managing tissue fluid and leukocyte movement into the vessel wall, and in regulating vascular tone (Blann, 2003). Thrombosis is likely to be promoted by a damaged endothelium that releases high levels of prothrombotic von Willebrand factor and loss of anticoagulant membrane thrombomodulin. There are also numerous examples of molecules (e.g., thromboxane, angiotensin, and platelet activating factor) that may participate in both pathological processes.

Blood Flow

Altered rheological properties augment the thrombotic process in 2 ways: by creating altered turbulence to accelerate the cellular and enzymatic reactions important in thrombogenesis and by inducing further injury to the vessel wall (Slauson and Cooper, 2002). Normally, blood flow in arteries and veins is laminar. In laminar flow, shear stresses are maximal at the vessel wall, causing endothelial cells to elongate and align in the direction of flow. Flow regulates endothelial related vascular reactivity by inducing the release of endothelial defenses such as nitric oxide, prostacyclin and tissue plasminogen activator (Lowe, 2003). The synthesis and release of prothrombotic and proinflammatory endothelial mediators such as tissue factor, von Willebrand factor, endothelin, intercellular adhesion molecule-1 and vascular cell adhesion molecule-1 is also shear-dependent. Leukocytes and platelets may be displaced from their central flow streamlines by more numerous and more deformable red blood cells. This displacement concentrates them near the vessel wall, making them available for adherence and activation. In addition, high shear activates platelets and increases their adherence to exposed subendothelium.

In addition to turbulence, slowing of the blood flow and stasis increase the probability for thrombosis (Slauson and Cooper, 2002). The increased viscosity of the blood favors the interaction of coagulable proteins and platelets with the endothelium. In addition, stasis reduces the removal of activated clotting factors by delaying their clearance in the liver and delays the dilution of activated clotting factors. The stasis may also produce local hypoxemia, leading to initiation of extrinsic coagulation.

Blood Constituents

There are basically 3 functions of the blood that may become activated: the formation of fibrin coagulum by the coagulation cascade; platelet aggregation and adherence; and fibrinolysis (Slauson and Cooper, 2002).

Coagulation Cascade

Blood coagulation involves a series of proteolytic reactions resulting in the formation of a fibrin clot. Most factors are enzymes (proteins) or coenzymes that are synthesized independently. They are generally present in the plasma in an inactive form and must be activated to become biologically functional in the clotting mechanism. This activation is accomplished either by the activated form of the clotting factor’s predecessor or, more often, by a complex composed of activated and nonactivated components. Initiation of clotting occurs either by surface-mediated reactions, or through tissue factor expression by cells. The 2 systems converge into the common pathway resulting in the formation of fibrin clot upon action of thrombin on fibrinogen. Then, Factor XIII, activated by thrombin, crosslinks and stabilizes the fibrin clot into an insoluble fibrin gel (Slauson and Cooper, 2002; Gorbet and Sefton, 2004). A simplified scheme of the coagulation cascade is given in Figure 2.

The intrinsic pathway: after contact with a negatively charged surface, a complex consisting of 4 proteins (hemocoagulation factors XII and XI, high-molecular-weight kininogen and prekallikrein) results in contact activation of factor XII, thus initiating the coagulation cascade through the intrinsic pathway (Yarovaya et al., 2002). The importance of the intrinsic pathway to the normal blood coagulation is not clear. Patients deficient in the initial components of the intrinsic pathway—FXII, high-molecular weight kininogen, or prekallikrein—have no bleeding tendency (Monroe and Hoffman, 2006). This might be explained by the fact that the occurrence of negatively charged surfaces in vivo is limited. Contact activation is initiated after exposure of subendothelial surface following damage to the endothelial wall (Gorbet and Sefton, 2004).

The extrinsic pathway: tissue factor (TF) is the physiologic vascular trigger needed to initiate coagulation (Stassen et al., 2004). TF binds to factor VIIa and accelerates the activation of factor IX and factor X by factor VIIa. In healthy vessels, TF is mainly located in the extracellular matrix around endothelial cells, forming a protective lining around blood vessels, and activates blood coagulation after vascular injury. Endothelial cells can be induced to produce TF by endotoxin, thrombin, fibrin, cytokines, shear, and hypoxia. Monocytes and natural killer cells have also been found to upregulate TF expression on endothelial cells.

Inhibitors of coagulation: most activated coagulation factors are serine proteases. Plasma contains several protease inhibitors; the most important one is antithrombin III. It neutralizes its target enzymes, FXa and thrombin, by forming a complex with the enzyme, thereby blocking the enzyme’s active site (Gorbet and Sefton, 2004). TF-VIIa is not efficiently inhibited by antithrombin, and has its own inhibitor—tissue factor pathway inhibitor. A larger pool of tissue factor pathway inhibitor exists on the luminal surface of the vascular endothelium. The binding of thrombin to thrombomodulin on endothelial cells activates protein C, a vitamin K dependent protein, which in its active form inactivates FVa and FVIIIa. For protein C activation, protein S, another vitamin K dependent protein, is a necessary cofactor.

Platelets

The platelet’s main role in hemostasis is to preserve the integrity of the vascular wall through formation of a platelet plug (Gorbet and Sefton, 2004). Platelets become activated when they contact any thrombogenic surface such as injured endothelium, subendothelium, and artificial surfaces. Platelet activation is initiated by the interaction of an extracellular stimulus with the platelet surface. This interaction involves the binding of the agonist to specific receptors on the platelet plasma membrane.

Platelet activation results in many physiologic responses (Gorbet and Sefton, 2004):

• Platelet release of stored biologically active compounds.

• Expression of P-selectin on the platelet membrane. P-selectin plays an important role in mediating adhesion of activated platelets to neutrophils, monocytes and lymphocytes.

• Release of prostaglandins and thromboxane B₂.

• Drastic shape change, promoting platelet-platelet contact and adhesion.

• Formation of platelet microparticles with procoagulant activities.

The hallmark of platelet activation is the metamorphosis of glycoprotein IIb-IIIa (GPIIb-IIIa) from its resting to its active state, in which it serves as a receptor for the soluble ligands, fibrinogen and von Willebrand factor (Topol et al., 1999). Engagement of these ligands by GPIIb-IIIa mediates platelet cohesion, and serves as a central event in thrombus formation and thrombosis. GPIIb-IIIa also recognizes other ligands, including fibronectin, thrombospondin and vitronectin. Like fibrinogen and von Willebrand factor, GPIIb-IIIa activation is required for productive binding of these adhesive proteins as soluble ligands, and this activation modulates platelet aggregation and mediates platelet adhesion to the subendothelial matrix.

Fibrinolysis

Fibrinolysis is the biochemical process underlying thrombolysis since both processes depend on the degradation of the fibrin network that holds the blood clot (Longstaff and Thelwell, 2005). Fibrinolysis can be represented as a 2-step process: (I) the activation of plasminogen to plasmin by a plasminogen activator and (II) is the degradation of the structural protein, fibrin. The participants in stage (I) are assembled on the surface of fibrin fibers and potent plasma inhibitor system limits the propagation of plasmin action in the circulation.

Endothelial Damage

Primary Mechanical Endothelial Injury

As described previously, one of the fundamental properties of the normal, intact, nonactivated endothelial cell (EC) is its lack of promotion of activation of either the extrinsic or intrinsic coagulative pathways or the adherence of unstimulated platelets and leukocytes (Slauson and Cooper, 2002). These nonthrombogenic properties of the lining of the vessel walls have been known since the days of Virchow. Damage to the endothelial layer of vessels initiates the formation of thrombosis.

An example of this kind of damage occurs after intravenous infusion of contrast media, which limits the application of angiography for diagnostic purposes (Gospos et al., 1980, 1981; Gottlob, 1981; Raininko, 1981; Gospos et al., 1983a; Kropelin et al., 1983; Gospos et al.,1983b, 1983c; Palomaki et al., 2003). Ultrastructural techniques (scanning and transmission electron microscopy) used to analyze the effect of contrast media, such as Verographin, on the aortic endothelium of New Zealand white rabbits and Wistar rats have demonstrated that, immediately after the injection of X-ray contrast media, the luminal surface of the ECs exhibits typical patterns of damage, with increased numbers of microvilli and blebs and nuclear portions protruding sharply into the vessel lumen. The often-observed desquamation and denudation of the EC from the monolayer is accompanied by the presence of a microthrombus on the vessel surface (Aliev et al., 2003). These typical damage patterns, induced by Verographin injection, are demonstrated in Figure 3. These damage patterns appear to be a nonspecific reaction to injurious stimuli, and lead to adhesion of blood cells and fibrin films accumulation on the vessel lumen.

The degree of lesion formation induced by contrast media is dependent on several factors:

1. The specificity of contrast media: in a study comparing the effects of diatrizoate, ioxitalamate, metrizamide, ioxaglate, iopamidol, and a hyperosmotic sorbitol and sodium chloride solution on the endothelium of the vena cava of the rat, endothelial damage was demonstrated immediately after the injection of metrizamide, diatrizoate, and ioxitalamate. Ioxaglate induced only minor endothelial changes, whereas, after the administration of both hyperosmotic solutions, no endothelial damage was seen (Aliev et al., 2003).

2. The dose of contrast media and rate of injection: intravenous administration of high doses of contrast media induced pulmonary edema in nonanesthetized rats (Mare and Violante, 1983a, 1983b). The degree of edema formation was dependent on the rate of contrast media injection as well as dose.

3. Rate of proliferation of endothelial cells: by using autoradiography to measure the rate of DNA synthesis it was shown that the degree of endothelial lesion due to contrast medium is directly proportional to the rate of proliferation of ECs (Gospos et al., 1980, 1981, 1983a, 1983b, 1983c).

The negative side effects of the contrast medium Verographin are eliminated by a single injection of heparin prior to the diagnostic procedure (Figure 4). The accumulation of heparin on the endothelial surface increases the resistance of ECs to injury stimuli by normalizing EC receptor activity that was disrupted after the X-ray contrast media injection. The presence of heparin sulfate and other heparin binding proteoglycans on the cell surface is able to sequester this compound to the cell surface, resulting in increased membrane stabilization. Heparin, as a glycosaminoglycan, has features that promote the accumulation on the EC surface and block the direct relationships between injury stimuli and the EC (Aliev et al., 2003).

Other drugs that cause direct endothelial damage are of the antineoplastic group exemplified by 5-fluorouracil (5-FU), which causes severe changes in the arterial endothelium in rabbits (Kinhult et al., 2003). The thrombogenic effect of 5-FU is one of the pathophysiological mechanisms contributing to 5-FU-induced cardiotoxicity. 5-FU impairs the antioxidant defense capacity and also affects the production of nitric oxide, a short-lived free radical, which is involved in the endothelial cell function. The free radical generation induced by 5-FU treatment leads to lipid peroxidation and endothelial cell membrane damage. This damage is prevented by probucol, a lipid-regulating agent that has strong antioxidant properties by promoting endogenous antioxidants. Probucol inhibits the expression of vascular cell adhesion molecule-1 on the surface of the endothelial cell and protects the endothelium against the negative effects of oxidized low-density lipoprotein.

Primary lesion of the endothelial lining also contributes to hepatic venoocclusive disease, seen after treatment with the alkylating agent busulfan (Brisse et al., 2004). The pathophysiological hypothesis for hepatic venoocclusive disease involves a primary lesion of the endothelium lining the sinusoids and terminal hepatic venules. Venular microthromboses can be identified histologically, inducing alterations of the hepatic microcirculation. The thromboses observed after treatment with busulfan (Brisse et al., 2004) have been related to these abnormalities of the microcirculation, leading to a marked decrease in the intrahepatic portal flow. In animal studies bleomycin has been implicated in morphologic damage to vascular endothelium of the lung, which results in pulmonary thrombosis (Caine et al., 2003). In a preclinical experimental study, bleomycin was also shown to cause multifocal vasculitis with venous occlusion in rhesus monkeys (Burkhardt et al., 1976). In this study, light and electron microscopic studies demonstrated that bleomycin causes characteristic endothelial changes in capillaries and small arterioles with cumulative administration. Endothelial lesions ranged from vacuolization to detachment and necrosis (Gerl, 1994); this investigation, however, provided no data on potential long-term vascular alterations.

In recent years the use of drug-eluting stents (DESs) has been proposed as a potential intervention for reducing the frequency of instent restenosis. A DES releases single or multiple bioactive agents into the blood stream (Schwartz et al., 2002). Experience with stented porcine coronary arteries suggests that sudden death may be more common during usage of drug-eluting devices, principally due to platelet-rich coronary stent thrombosis. Such death typically occurs in the first 24 hours after implant but may occur later if healing is impaired. A case of late stent thrombosis, severe localized hypersensitivity reaction, and secondary development of acute myocardial infarction, occurring 18 months after insertion of 2 Cypher stents for unstable angina, was recently reported. The stent was coated with poly-n-butyl methacrylate and polyethylene-vinyl acetate copolymer containing 140 μg sirolimus. Analysis of the pathological specimens implicated a relationship between the stent malapposition and inflammation and thrombosis. Hypersensitivity to the polymer as the most likely mechanism for the thrombosis was hypothesized. The United States Food and Drug Administration (US FDA) recently issued a warning about subacute thrombosis and hypersensitivity reactions to these sirolimuseluting Cypher stents (Virmani et al., 2004a).

Preclinical animal studies of stent safety have demonstrated that, in contrast to the local reaction of the artery to bare-metal stents, where full healing of the neointima is seen by 28 days, evidenced by the presence of complete surface endothelialization, the absence of fibrin, and only a few chronic inflammatory cells around the stent struts, delayed healing occurs when a DES is used. The slowed healing is characterized by the presence of persistent fibrin, incomplete coverage of the stent struts by smooth muscle and the proteoglycan matrix, increased inflammation around some stent struts, and incomplete endothelialization. The inflammatory reaction continues to intensify between 28 and 90 days in DES with nonbioerodable polymers, in contrast to the reaction seen with bare-metal stents. The inflammation consists mostly of T-lymphocytes and eopsinophils, accompanied by excessive neointimal thickening; thrombosis is infrequent (Virmani et al., 2004b). Needs exist for enhanced surveillance of patients with DES to detect and avoid late complications and for development of tests to prescreen individuals with potential reactions to polymers (Virmani et al., 2004a).

Changes in the Expression of Procoagulant and Anticoagulant Properties of Endothelial Cells

Recently, abundant evidence has accumulated that the endothelial cell (EC) consists not merely of a passive lining for blood vessels, but metabolically active cells that possess both procoagulant and anticoagulant properties and wield considerable influence on the coagulative system (Slauson and Cooper, 2002). Inflammatory cytokines, such as tumor necrosis factor (TNF) and interleukin-1 (IL-1), cause alterations in coagulation that are mediated by their effect on ECs. TNF and IL-1 induce the formation of endothelial-cell procoagulants and downregulate the production of the anticoagulant thrombomodulin, both in vitro and in laboratory animals (Tolcher et al., 1995).

Cytotoxic chemotherapy for the treatment of breast cancer is associated with an increased incidence of venous thrombotic events (Tolcher et al., 1995). Granulocyte-macrophage colony-stimulating factor administration, which is used to ameliorate the myelosuppresive effects of dose-intensive chemotherapy regimens, increases the risk of arterial and venous thrombotic events in breast cancer women. Granulocyte-macrophage colony-stimulating factor does not directly modulate the endothelial cell functions related to procoagulant production and hemostasis-thrombosis in vitro, but it induces release of the secondary cytokines TNF and IL-1 (Sisson and Dinarello, 1988; Hazenberg et al., 1989; Stehle et al., 1990; de Vries et al., 1991), thus leading to increased incidence of thrombotic events.

By incubating murine macrophages with varying doses of cis-platinum, Pogrebniak et al. (1991) have shown increased production of TNF, which was preceded by transcription induction of the TNF gene. The mechanism for this selective chemotherapy-induced cytokine amplification remains unclear. TNF production was also increased after bleomycin treatment in tumor-bearing rat spleen cells in vitro (Micallef, 1993). This amplification of TNF production is one of the contributors to the increased thrombosis seen after treatment with these drugs (Gerl, 1994).

One of the important roles of ECs in the thrombotic cascade is the induction of tissue factor (TF) (Semeraro and Colucci, 1997), a transmembrane protein that serves as a high-affinity receptor for factor VIIa, which initiates the extrinsic blood-coagulation series of events (Nemerson, 1988). Drugs that affect TF induction will have a prominent effect on thrombosis. A drug that influences TF is rapamycin, which inhibits tumor growth by an antiangiogenic mechanism that involves blockage of vascular-endothelial-growth-factor signaling by ECs (Guba et al., 2002) and upregulates TF expression in ECs. Expression of TF is further up-regulated by the stimulation of TF cell signaling pathways by vascular-endothelial-growth-factor and rapamycin (Guba et al., 2005). The costimulatory effect of rapamycin and local vascular-endothelial-growth-factor secretion by tumors results in excessive endothelial TF expression leading to thrombosis in tumor-containing vessels. This unique finding takes advantage of what was thought to be a negative side effect of the drug (Barone et al., 2003; Robson et al., 2003; Trotter 2003; Fortin et al., 2004; Paramesh et al., 2004), which can be used to shrink existing tumors by clotting the existing blood supply.

Another drug that compromises the antithrombotic capacities of vascular ECs is recombinant tissue-type plasminogen activator (t-PA) (Jen et al., 1996). t-PA has been used clinically to treat patients with acute myocardial infarction and deep vein thrombosis (Shi et al., 1992). During this thrombolytic therapy, t-PA causes activation of a large amount of plasminogen and a transient increase of plasmin in the blood. This excess of plasmin combined with activation of plasminogen on the cell surface causes proteolysis at specific cell-cell contacts and cell matrix sites. This generates retraction of the EC monolayer and exposure of the subendothelial cell matrix (Conforti et al., 1994). Plasmin also acts as a feedback regulator of the fibrinolytic system. By binding to specific receptors on the ECs and inhibiting the protein kinase activity of these cells, plasmin down-regulates endogenous t-PA production (Shi et al., 1992). These 2 mechanisms lead to increased tendency towards thrombus formation.

Antibodies from patients with heparin-induced thrombocytopenia/thrombosis (HIT/T) have been shown to interact with and activate ECs (Reilly and McKenzie, 2002). A recent report indicated that HIT antibodies directly activate microvascular ECs (Blank et al., 2002). HIT antibodies bind to microvascular EC in vitro via the F(ab′)₂ end of the antibody. Activation of EC was assessed by release of IL-6, von Wille-brand factor, and thrombomodulin. This activation leads to increased elaboration of surface adhesion molecules, such as P-selectin, E-selectin, and vascular cell adhesion molecule-1. Increased adherence of monocytes to microvascular EC was observed in the presence of HIT antibodies and was abrogated with preincubation with anti-adhesion molecule antibodies. These observations suggest that HIT antibody-induced EC activation contributes to thrombosis seen in patients with HIT/T (Reilly and McKenzie, 2002).

Prostacyclin (PGI₂) is one of the products of the EC with enormous importance in vascular homeostasis. Released in low levels by resting ECs, PGI₂ can prevent or reverse aggregation of platelets and induce vasodilation. A cyclooxygenase-2 (COX-2)-produced derivative of arachidonic acid, PGI₂ is influenced by COX inhibitors. Great interest in PGI₂ arose especially after rofecoxib, a COX-2 inhibitor, was withdrawn by its manufacturer from markets worldwide, five years after its launch, after the APPROVe (Adenomatous Polyp Prevention on Vioxx) trial showed twice the risk of thrombotic events in patients administered 25 mg of rofecoxib compared with placebo after 18 months of treatment (Kasliwal et al., 2005; US FDA, 2004).

In platelets, which contain only the COX-1 isoform, the major isomerase coupled to COX-1 is thromboxane (TxA₂) synthase; thus, the major arachidonic-acid product of COX activity in platelets is TxA₂ (Krotz et al., 2005). Normally the vasculature maintains a healthy balance between COX-2-mediated PGI₂ and COX-1-dependent TxA₂.

COX-2 inhibitors exert 2 potentially negative effects on the vascular system. They do not inhibit the production of TxA₂ thereby increasing platelet aggregation, and they block PGI₂, tipping the delicate homeostatic balance towards aggregation of platelets and vasoconstriction (Meyer et al., 2005). These possible adverse effects were recently investigated following a localized catheter-induced intravasal injury in an animal model. While knockout mice lacking PGI₂ receptors developed significantly increased TxA₂ biosynthesis and platelet aggregation, knockout mice lacking both the TxA₂ and PGI₂ receptors did not show this exaggerated response. The authors concluded that endogenous PGI₂ modulates the vascular action of TxA₂ and platelet activation in vivo (Cheng et al., 2002).

Although theoretically the risk of atherothrombotic events should increase with the selectivity of the drug for COX-2, this property of a COX-2 inhibitor alone does not suffice for a thrombotic event to occur with statistical significance. In patients who are already at an increased risk of atherothrombotic events due to their underlying disease, however, the prothrombotic risk may become most pronounced. An example of this assumption is found in case reports (Meyer et al., 2005) describing retinal vein occlusion after administration of rofecoxib (Figure 5). In the retina predisposed locations exist for development of ocular thrombosis, such as the lamina cribrosa and arteriovenous crossings, where endothelial-cell edema and proliferation develop, similar to the intravasal injury induced by a catheter (Cheng et al., 2002). This damage to the ECs constitutes a predisposition to thrombosis.

Mitomycin is another drug that directly damages vascular ECs of the kidney and also inhibits production of PGI₂. The effect of mitomycin on the biosynthesis of PGI₂ was tested in culture of human umbilical cord vein endothelial cells (Duperray et al., 1988). A 30% inhibition of the thrombin-stimulated PGI₂ synthesis by mitomycin was observed. This inhibitory capacity of mitomycin on PGI₂ synthesis promotes aggregation of platelets and local intravascular coagulation. These properties led to the usage of mitomycin as an experimental model of thrombotic microangiopathy upon direct infusion into rat kidneys (Cattell, 1985).

Platelets

Increased Platelet Adhesiveness

Although the major factor responsible for adhesion of platelets to vessel walls is endothelial damage coupled with exposure of subendothelial structures, the platelets themselves contribute to the process through an adhesive ability of their own. Any increased tendency of platelets to adhere to the subendothelial surface may enhance the thrombotic process.

In order to test platelet adhesiveness ex vivo, discrete vascular lesions must be induced. Such induction has occurred by usage of a Laser Argon thrombosis model in mesenteric microvessels. The application of a laser beam on a vessel wall results in limited EC injury and the adhesion of platelets onto the exposed extracellular matrix, followed by the formation of a thrombus (Weichert et al., 1983), which, along with embolization, is assessed by direct observation. A drug that increases platelet adhesiveness would be expected to produce larger thrombi earlier and prolong microembolization. Using this method, investigators found that contrast media increase platelet adhesiveness as well as exert their direct deleterious effects on ECs (Aguejouf et al., 2000).

Another system to measure platelet adhesion is an ex vivo model using a parallel-plate flow chamber connected to the femoral artery of the rat. The assessment of adhesion occurs by direct visualization of the attachment of platelets to a testing surface (e.g., fibrinogen-coated). This method was used to discover that recombinant t-PA induces increased adhesion of platelets, which can be suppressed by coadministration of vitamin E, known to reduce adhesion (Jen et al., 1996).

There are several possibilities for the mechanism by which thrombolysis induces platelet activation (Jen et al., 1996):

1. The effect may be directly on platelets, as plasmin-induced platelet aggregation is accompanied by the cleavage of aggregin, a putative adenosine diphosphate receptor on the platelet membrane. This effect is also indirectly mediated by intracellulary activated calpain (Puri et al., 1990).

2. The effect may be directed by some plasma factors; for instance, fibrinolytic treatment can cause platelet-dependent thrombin generation.

3. The effect can be mediated through the compromise of antithrombotic capacity of ECs, as described in the previous section (“endothelial damage”).

The methodology of scanning electron microscopy revealed that t-PA causes dramatic changes in the morphology of platelets, causing pseudopod and cluster formation that contribute to the increased adhesion (Jen et al., 1996). Vitamin E administration suppressed these morphology changes. These morphological changes were correlated with the number density of adherent platelets. Adherent platelets were found to undergo postcontact morphological changes in a dynamic way, and cells of different morphology stand different levels of flow shear stress. These facts indicate that t-PA exerts its effects on platelets via its ability to enhance shape change by involving cytoskeletal rearrangements.

Platelet Aggregation

Aggregation of platelets is a key step in the process of hemostasis and formation of thrombus. Serotonin plays a crucial role in the latter, since it is able to induce and amplify the aggregation of platelets (Vanhoutte, 1991; Willerson, 1991; Cerrito et al., 1993). Platelets actively accumulate serotonin into dense granules and possess phospholipase Cβ-linked serotonin receptors, whose stimulation leads to activation of the phosphatidylinositide turnover, a rise in free cytoplasmic Ca²⁺, phosphorylation of proteins and a shape change reaction associated with platelet activation (Pletscher, 1987; McNicol and Israels, 2003); thus, drugs that influence levels of serotonin in the blood alter the tendency toward thrombosis. Antidepressants, particularly selective serotonin reuptake inhibitors (SSRIs), attenuate platelet activation by depleting serotonin storage and may decrease the risk of hypercoagulability (Sauer et al., 2001).

The immediate effect of SSRI treatment, however, is an increase in serotonin near specific serotonin-receptor subtypes in discrete regions of the body where relevant physiologic processes are regulated (Stahl, 1998). Due to this immediate effect, the risk of thrombosis may be increased in SSRI-treated patients (Kurne et al., 2004). Enhanced platelet aggregation induced by serotonin has also been demonstrated in patients with schizophrenia treated with chlorpromazine (Boullin et al., 1975a, 1975b; Orr and Boullin, 1976; Boullin et al., 1978; Orr et al., 1981), fluphenazine (Orr and Boullin, 1976; Orr et al., 1981), flupenthixole (Orr et al., 1981), trifluoperazine (Orr et al., 1981), and haloperidol (Orr et al., 1981). However, the suggested increase of platelet aggregation induced by antipsychotics has never been associated with clinical cases of venous thromboembolism (Hagg and Spigset, 2002).

Drugs may also induce sensitization of platelets against endogenous aggregators. For example, in one study (Aguejouf et al., 2000), platelets exhibited an enhanced response to adenosine diphosphate after injection of nonionic contrast media (iopamidol and iohexol) into Wistar rats (Aguejouf et al., 2000). In this study, aggregation was measured using a method that depends upon changes in electrical impedance caused by platelet accretion onto electrodes (Cardinal and Flower, 1980). Three parameters were studied: latency time; maximal amplitude of aggregation measured as an impedance and expressed in ohms; and speed of aggregation in ohms per minute.

Ionic contrast media (ioxaglate and diatrizoate), unlike nonionic contrast media, caused inhibition of platelet aggregation. Their mechanism of inhibition remains unknown, although it may be related to the minimal activation of the platelet GPIIb-IIIa receptor (Chronos et al., 1993). Platelets require fibrinogen binding to the activated GPIIb-IIIa receptor for the platelet aggregation to occur. Even at the highest concentrations of contrast media, only a small increase in the fibrinogen receptor activation was seen, and this was to a level below that required for platelet aggregation to occur. It might be also related to the contrast media viscosity, acting to slow down platelet-platelet contact and hence the rate of aggregation. Despite their anti-platelet properties, ionic contrast media exert a prothrombotic activity in vivo. It is likely that other mechanisms may occur in blood vessels: thrombin generation; leukocyte activation; and decreased red cell deformability with subsequent modifications of microvessel rheology (Aguejouf et al., 2000).

The increasing use of engineered carbon nanoparticles in nanopharmacology for selective imaging, sensor tools, or drug-delivery systems has increased the potential for blood platelet–nanoparticle interactions. A comprehensive effort was recently undertaken to determine the effects of nanoparticles on human platelet aggregation (Radomski et al., 2005). This assessment was performed both in vitro, by usage of lumi aggregometry, phase-contrast, immunofluorescence, transition electron microscopy, flow cytometry, zymography, and pharmacological inhibitors of platelet aggregation, and in vivo, by measurement of the rate of vascular thrombosis in the rat after induction by ferric chloride.

The nanoparticles stimulated platelet aggregation (Figure 6) and accelerated the rate of vascular thrombosis. Flow cytometry experiments and the fact that platelet aggregation is sensitive to inhibition with ethylenediaminetetraacetic acid have led to the assumption that nanoparticle-induced platelet aggregation likely results from activation of GPIIb/IIIa, the membrane glycoprotein complex that constitutes the fibrinogen receptor and thus mediates aggregation. The aggregation stimulated by nanoparticles appears to be protein kinase C independent. In addition, it is aspirin insensitive, indicating that thromboxane does not exert significant effects on this aggregation. The aggregation was inhibited by prostacyclin and S-nitroso-gluthatione, demonstrating that platelets exposed to particles retain the ability to be regulated by endogenous inhibitors of vascular hemostasis.

The nanoparicles’ shape is a major determinant in the ability of particles to cause platelet aggregation (Radomski et al., 2005). Preparations of nanotubes activated platelets, while nanospheres did not exert significant effects on platelet aggregation. It is possible that carbon nanotubes mimic molecular bridges involved in platelet-platelet interactions, thus stimulating aggregation, while nanospheres may not support cell-cell communication. Another determinant is the particle surface. Since the binding of GPIIb/IIIa integrin to its protein ligand depends on the multiple electrostatic interactions, the nanotube surface charge could play a role in direct interactions between these particles and GPIIb/IIIa. In addition, transition metals on the surface of particles can cause increased generation of reactive oxygen species, leading to platelet activation. Finally, the physicochemical nature of particles resulting from contamination with elements such as Ni and I and multiple transition metals, or the structure of the manufactured carbon nanomaterials, may underlie their ability to aggregate platelets.

Platelet Activation and Release of Platelet Content

Platelets are secretory cells; as a consequence of their aggregation and activation, their constituents are released into the local environment. Because the release reaction (Grette, 1962)—a liberation of adenosine diphosphate, serotonin, histamine, and enzymes—promotes further growth of platelet aggregates, it holds considerable toxicological importance.

One of the commonest causes of life-threatening drug-induced platelet activation and thrombosis is heparin-induced thrombocytopenia/thrombosis (HIT/T) (Reilly and McKenzie, 2002). Platelet factor 4 (PF4) and heparin form complexes on the surface of activated platelets, complexes that are recognized by anti-heparin–PF4 antibodies. These antibodies have been found to activate human platelets in vitro via Fc[gamma]RIIA. The binding of HIT antibodies to activated platelets likely promotes microparticle release and cell–cell interactions (platelet–platelet, platelet–leukocyte, and platelet–vessel wall), causing a predisposition to thrombosis.

Thrombogenic Substances Acting on the Coagulative System

Increased Concentration of Clotting Factors and Decreased Concentration of Inhibiting Factors

Perhaps the most controversial pharmaceuticals affecting thromboembolic risks are hormone-replacement-therapy (HRT) drugs and combined oral contraceptives (COC). Hormone replacement therapy was increasingly promoted for 40 years to improve the quality of life and reduce the risks of osteoporotic fractures and coronary heart disease (Lowe, 2004). In recent years, observational studies, randomized trials, and systematic reviews of such trials have shown that HRT does not reduce, but actually increases, cardiovascular risk. This therapy increases by twofold the relative risk of venous thromboembolism (VTE) and by 50% that of fatal or disabling stroke while also increasing the early risk of myocardial infarction and having no protective effect against coronary heart disease after longer-term use.

Increasing evidence exists that activation of blood coagulation is associated with risk of VTE and constitutes a plausible mechanism through which exogenous estrogens (COC, oral HRT) increase risks of VTE. There is evidence that oral estrogen results in increases in levels of prothrombin fragments 1+2, a marker of thrombin generation, and fibrinopeptide A, a marker of fibrin production. Increases in these markers of thrombin and fibrin production with conjugated equine estrogen are dose-dependent (Peverill, 2003). There is also evidence for an increase in thrombin and fibrin production with combined oral HRT in most, but not all, studies.

In oral contraceptive users, plasma levels of practically all coagulation factors increase. In particular, Factor VII and X may increase by up to 170% of their baseline values, whereas fibrinogen levels usually increase by 10–20% of baseline values (Martinelli et al., 2003). In oral HRT users, several epidemiological studies have reported higher factor VII levels, but, in contrast to COC, there are few reported studies of the effects of oral HRT on other coagulation factors (Lowe, 2004).

Plasma from oral contraceptive users is resistant to the anticoagulant action of activated protein C. Acquired activated protein C-resistance is an independent risk factor for venous thrombosis. The molecular basis of the observed acquired resistance to activated protein C during oral contraceptive intake is unknown and can only in part be explained by the decrease in protein S levels observed during oral contraceptive intake. In women using drugs with high estrogen content a decrease in protein S concentration of up to 50% of their baseline was observed (Martinelli et al., 2003). To a lesser extent, antithrombin levels may also be reduced. There is consistent data that oral HRT also significantly impairs all pathways of coagulation inhibition except protein C, about which discrepant results have been reported (Lowe, 2004).

In summary, in users of oral HRT and COC, activation likely results from an imbalance between increased levels of coagulation factors (Lowe et al., 1999; Sidelmann et al., 2003) and decreased inhibition of coagulation. A summary of associations of coagulative factors, inhibitors, and activation markers with risk of VTE and usages of COC and oral HRT is given in Table 1. It is currently unclear how these effects are brought about at the molecular level of the estrogen receptor. It is likely that these effects at the cellular level are under genetic control, because the hemostatic system of some women appears to be more sensitive to the effect of estrogens than that of other women (Rosendaal et al., 2002).

Another compound that induces a decrease in the levels of inhibitors of coagulation is L-asparaginase, a chemotherapeutic agent commonly used in the treatment of both adult and pediatric acute lymphoblastic leukemia. Its mechanism of action is the inhibition of protein synthesis by the hydrolysis of asparagine, which is necessary for lymphoid cells. Administration of L-asparaginase causes a reduced synthesis of proteins, among them antithrombin III. This reduction causes thrombosis, a major complication of L-asparaginase therapy (Beinart and Damon, 2004).

Cyclophosphamide, methotrexate, and 5-fluorouracil chemotherapy, a regimen used in the treatment of breast cancer, causes a statistically significant decrease in levels of protein C and protein S (Rogers et al., 1988). Possible explanations for these chemotherapy-induced abnormalities include impairment of vitamin K metabolism, inhibition of DNA/RNA synthesis leading to a decrease in protein synthesis by the liver, and initiation of intravascular coagulation.

Another drug used in the treatment of breast cancer and found to induce reduction in levels of antithrombin and protein C is tamoxifen (Mannucci et al., 1996; Pemberton et al., 1993). In clinical studies the frequency of tamoxifen-associated thromboembolic events has ranged from 1% to 3% (Reddy and Chow, 2000). Tamoxifen is an antiestrogen that has weak estrogenic effects, which may contribute to its prothrombotic activity (Caine et al., 2003). Existing data indicates that the risk of thrombosis during tamoxifen therapy is increased in patients with the factor V leiden mutation.

An increase in the levels of activated clotting factors is not necessarily induced by increased production. Corticosteroids, for example, cause an increase in these levels by decreasing their clearance rate via reticuloendothelial blockade (Kwaan and McFadzean, 1956).

Circulating antiphospholipid antibodies, including the immunoglobulin lupus anticoagulants and anticardiolipin antibodies, have been associated with an increased risk of thrombosis (Greaves, 1999). Patients with antiphospholipid syndrome have exhibited evidence of persistent activation of coagulation; an increased plasma concentration of markers of thrombin generation occurs, such as the prothrombin fragment f1 · 2 and fibrinopeptide A. This increase contrasts with the in vitro anticoagulant effects of these autoantibodies. Anticardiolipin antibody levels have frequently been reported to be elevated in patients taking conventional antipsychotics (Canoso and de Oliveira, 1988; el-Mallakh et al., 1988; Canoso et al., 1990; Metzer et al., 1994; Schwartz et al., 1998; Steen and Ramsey-Goldman, 1988) and clozapine (Davis et al., 1994). Venous thrombosis appears to be associated with the use of antipsychotic drugs in psychiatric patients (Thomassen et al., 2001).

Factor XIII (FXIII) is a thrombin-activated protransglutaminase responsible for fibrin-clot stabilization and longevity. Deficiency in FXIII has been associated with diffuse bleeding and wound-healing disorders in humans. Several studies were conducted to evaluate the safety and pharmacokinetics of recombinant human FXIII A2 dimer (rFXIII) (Ponce et al., 2005). As a study model cynomolgus monkeys were used because their coagulative parameters are similar to those of humans (Seaman and Malinow, 1968). In this model, rFXIII-mediated toxicity is expressed as an acute systemic occlusive coagulopathy. The proposed sequence of events involves an initial activation of rFXIII in plasma, leading to subsequent formation of acellular granular emboli possibly consisting of aggregated polymers of fibrin, fibrinogen, α2-plasmin inhibitor, α2-macroglobulin, and/or fibronectin. Accumulation of these complexes in small vessels might obstruct blood flow and result in local tissue ischemia and coagulative necrosis in the surrounding tissues (Figure 7).

The necrosis could subsequently cause the release of systemic inflammatory mediators including TF, thereby eliciting shock, inflammation, and activation of the clotting cascade. This observed occlusive coagulopathy is distinct from that of disseminated intravascular coagulation. Classical disseminated intravascular coagulation ultimately involves loss of control over the production of systemic thrombin and systemic plasmin via imbalances in the clotting and fibrinolytic systems (Bick et al., 1999). In contrast, rFXIII-mediated toxicity is attributable to the formation of cross-linked protein complexes (including fibrinogen polymers) in plasma by activated rFXIII that ultimately occlude small vessels and result in activation of a secondary systemic coagulative cascade.

Another factor that may enhance the blood-coagulation cascade is externalization of phosphatydilserine on the outer side of the erythrocytic membrane (Marcus, 1966; Zwaal et al., 1977; Zwaal, 1978; Mannhalter et al., 1984), which may act as a catalytic surface on which various coagulative factors can interact, resulting in a condition of hypercoagulability. Drugs that cause peroxidation of the erythrocytic membrane may cause this externalization. An example is phenylhydrazine, which was used in the past in the treatment of polycytemia vera (Owen, 1924, 1925). Vascular thrombosis occurs as a complication of phenylhydrazine treatment more frequently than with any other type of therapy. In 3 of 7 cases in one study (Brown and Giffin, 1926), peripheral thrombosis occurred during excessive destruction of blood. Erythrocytes with abnormal externalization of phosphatydilserine are also susceptible to adherence to endothelial cells, causing microvascular occlusions in addition to the hypercoagulable state.

Inhibition of Fibrinolysis

Any inhibition of the fibrinolytic system, which constitutes the ultimate weapon against the consequences of intravascular coagulation, could prove hazardous. Administration of corticosteroids exemplifies drug treatment that may inhibit the fibrinolytic activity of the blood. Dexamethasone administered to rats increased plasma plasminogen activator inhibitor-1 in a dose-dependent manner. In addition, treatment with dexamethasone decreased tissue plasminogen activator activity in tissue extracts. This decreased fibrinolytic activity presumably neutralizes the inhibition of arterial thrombosis, which is mediated through restraint of platelet aggregation (van Giezen et al., 1994).

Estrogens have significant effects on fibrinolysis. During oral contraceptive use, tissue plasminogen activator, plasminogen, plasminantiplasmin complexes and D-dimer were increased, and plasminogen activator inhibitor, plasminogen activator inhibitor-1 activity and tissue plasminogen activator antigen levels were decreased (Martinelli et al., 2003; Lowe, 2004). However, this increase in the fibrinolytic parameters was not accompanied by a change in clot lysis time, suggesting that oral contraceptive-induced increased endogenous fibrinolytic activity is counteracted by enhanced capacity of the coagulation system to down-regulate fibrinolysis via thrombin-activatable fibrinolysis inhibitor, which has been associated with an increased risk of thrombosis. Thrombin-activatable fibrinolysis inhibitor levels were found to be slightly higher in oral contraceptive users, but this finding was not statistically significant.

Release of Tissue Factor

Tissue factor is an important initiator of the coagulative pathway. While it is derived largely from ECs, it can also be produced in other types of cells like fibroblasts, smooth muscle cells, injured cells of various types, and stimulated monocytes. Drugs that stimulate monocytes can thus induce thrombosis through tissue factor (TF) expression. For example, two recent publications presented platelet factor-4 (PF4)-dependent tissue factor synthesis in monocytes in the presence of human HIT antibodies (Arepally and Mayer, 2001; Pouplard et al., 2001).

Tissue factor mRNA was detected after treatment of whole blood with HIT antibodies, PF4, and heparin (Reilly and McKenzie, 2002). Antibodies isolated from the sera of patients with HIT bound to monocytes in a PF4-dependent manner. This binding increases expression of IL-8, a proinflammatory cytokine, and expression and synthesis of cell surface tissue factor. The role of tissue factor is further shown by the abrogation of procoagulant activity in the presence of anti-tissue factor antibodies. The increase in tissue factor activity appears because of de novo synthesis, as the activity increases only after 6 to 12 hours’ incubation.

Thrombogenic Substances Inducing Changes in Blood Flow

Vasoconstriction

A deficiency in magnesium in tissue-culture media has been shown to potentiate the contractile responses of arteries to norepinephrine, acetylcholine, serotonin, angiotensin, and potassium (Turlapaty and Altura, 1980). Renal dysfunction can cause hypomagnesemia, and drugs such as cisplatin can induce arterial spasm by this mechanism (Vogelzang et al., 1985). Cisplatin induces tubular injury, which leads to hypomagnesemia by causing a decrease in the maximal rate of reabsorption of the divalent cation. In two series, 76 and 87% of patients treated with cisplatin developed hypomagnesemia, with median time to onset of 63 days (Gerl, 1994).

A more direct mechanism for vasoconstriction is exemplified by corticosteroids, which are known to decrease cerebral blood flow by a direct vasoconstrictive effect on cerebral blood vessels (Gerrits et al., 1974). Corticosteroid hormones play an important role in the control of vascular smooth muscle tone by their permissive effects in potentiating vasoactive responses to catecholamines through glucocorticoid receptors (Yang and Zhang, 2004). Increased cortisol response has been associated with an increase in arterial contractile sensitivity to norepinephrine and vascular resistance. Glucocorticoids regulate vascular reactivity by acting on both endothelial and vascular smooth muscle cells, and both glucocorticoid receptor protein and mRNA have been identified in endothelial and vascular smooth muscle cells.

1. Effect on endothelial cells: corticosteroids induce a reduction in the production of the endothelium-derived factor nitric oxide (NO), a powerful vasodilator. The mechanism of impaired endothelium-dependent dilatation in glucocorticoid-treated humans and rats involves abnormalities of the endothelial NO synthesis, but not the result of cellular events downstream of NO. The synthesis of NO from L-arginine in endothelial cells is catalyzed by the constitutively expressed endothelial isoform of NO synthase (eNOS). Glucocorticoids are thought to provoke genomic responses, mediated by the intracellular glucocorticoid receptors that are capable of interacting with glucocorticoid responsive elements. Studies of cultured endothelial cells suggest that glucocorticoids act at various sites in the NO synthesis pathway, affecting eNOS synthesis.

Glucocorticoids also affect prostacyclin synthesis. Prostacyclin is a powerful vasodilator, a potent inhibitor of platelet aggregation and an inflammatory mediator. The effects of glucocorticoids on endothelial prostacyclin synthesis may be developmentally regulated, with a down-regulation of cyclooxygenase-1 in fetal cells and an inhibition of phospholipase A2 and arachidonic acid release in adult cells.

1. Effect on vascular smooth muscle: glucocorticoids, acting through glucocorticoid receptors, potentiate vascular smooth muscle reactivity to vasoconstrictors by regulating Ca²⁺ homeostasis. They enhance agonist-mediated pharmacomechanical coupling by increasing Ca²⁺ mobilization and Ca² sensitivity of myofilaments.

Another compound that exerts a deleterious effect on blood vessels is not intrinsically a drug, but actually the herbal supplement, ephedra, which recently was removed from the U.S. marketplace due to a heightened concern that dietary supplements containing it may present “an unreasonable risk of illness or injury.” One of those injuries is myocardial infarction caused by in situ coronary thrombosis (Sachdeva et al., 2005).

Ma huang, an alkaloid of ephedra, increases the availability of catecholamines at adrenergic synapses in the brain and heart, directly stimulating α-and β-adrenergic receptors (Sapru and Theoharides, 1996). The underlying mechanism of myocardial infarction in patients taking ma huang is thought to be coronary vasospasm (Cockings and Brown, 1997; Zaacks et al., 1999). Enhanced sympathetic activity increases platelet reactivity, which likely contributes to a prothrombotic state in the presence of ephedra alkaloids.

Blood Stasis

Slowing of the blood flow and stasis increase the probability of thrombosis, though the latter alone probably does not cause thrombosis unless accompanied by endothelial damage or a hypercoagulable state of the plasma. If a state of hypercoagulability exists, however, thrombosis will develop preferentially in blood vessels with slow blood circulation (Wessler, 1962). Drugs that cause stasis must, therefore, be considered accessories to such an event. For example, drug-induced arterial hypotony and peripheral vasodilatation may lead to venous stasis in neuroleptica-treated patients and thus cause thrombogenesis (Pantel et al., 1997). Antipsychotics bind to α-adrenergic receptors, and exert hypotension by blocking these receptors.

Conclusions

Drugs can induce a pathological hypercoagulable state through the actions of a variety of mechanisms. These and examples of the various drugs that induce a thrombogenic state are given in Table 2. Thrombogenesis may result from alterations in the blood flow, the endothelium that covers the blood vessels, and the blood constituents themselves, such as coagulative factors and platelets. Of special interest is the fact that a drug that has an antithrombotic effect by one mechanism can paradoxically induce a prothrombotic effect by a different mechanism. Examples are t-PA and heparin; both are used to counteract the thrombotic state by affecting the coagulation system but as a side effect may activate the thrombotic process through platelet adhesiveness and activation.

In order to test drugs for possible negative thrombotic side effects, inclusion of all known thrombotic mechanisms in the study design is essential. This should be done using appropriate testing systems accompanied by challenge tests to reveal hidden prothrombotic properties of the drug by, for example, inducing vascular lesions or administering adenosine diphosphate to check increased sensitivity of platelets to adhesion and aggregation. Testing the drug both in vivo and in vitro and by using the appropriate animal model for the thrombotic mechanism examined must also be performed. These combined approaches would reduce the risks associated with the usage of pharmaceutical agents that might induce thrombosis.