Mechanisms

The inter-relationships of the coagulant and anticoagulant proteins are shown in Figure 1. The coagulation system is a highly regulated cascade of surface-associated interacting enzymes and cofactors that generate the remarkably potent enzyme, thrombin, at sites of vascular injury [1]. Thrombin proteolytically converts soluble fibrinogen to insoluble fibrin, activates factor XIII (the transglutaminase enzyme) that causes the formation of a crosslinked insoluble clot, and activates both platelets and endothelial cells by proteolytic digestion of its unique cellular receptor on these cells. Thrombin promotes its own formation, through activation of factors VIII and V, and inhibits its own formation through activation of the protein C system. The major deterrents to pathologic thrombin generation are a group of natural anticoagulant systems, including antithrombin III (AT III, a serine protease inhibitor), the vitamin K-dependent protein C system, and the newly described tissue factor pathway inhibitor [2]. The remarkable completeness of inhibitory control is indicated by the fact that each step in the enzymatic cascade can be blocked by an inhibitor pathway.

View larger version (15K): [in this window] [in a new window]   Figure 1. Anticoagulant inhibitory systems and the control of blood coagulation. The orchestrated interactions of the four separate systems are shown: antithrombin III (AT III), activated protein C (PCa), tissue factor pathway inhibitor (TFPI), and plasmin. PL = the cellular phospholipid surface on which various macromolecular complexes assemble; II = prothrombin; and Fg =fibrinogen.

In addition to these anticoagulant systems, the fibrinolytic system (a second, highly regulated enzymatic cascade) generates the broadly active serine protease plasmin by the action of tissue plasminogen activator on plasminogen. Plasmin digests and dissolves the fibrin clot. This system plays an important role in preventing pathologic thrombus formation and is also closely regulated by a number of protease-inhibitor systems including several plasminogen activator inhibitors and plasmin inhibitors (for example, ₂-plasmin inhibitor, ₂-macroglobulin, and C1-inactivator).

The major site of control of these coagulant and anticoagulant interactions is probably at the vascular endothelial cell surface. As shown in Figure 2, the normal function [constant and steady state] of the endothelium in preventing thrombus formation is controlled by a number of membrane-related activities including 1) expression of thrombomodulin—a protein "receptor" for thrombin that converts thrombin into an activator of protein C; 2) proteoglycans containing heparan sulfate—that bind and activate antithrombin III; and 3) assembly of a plasmin-generating system. We suggest that impaired regulation of these endothelial cell functions contributes substantially to the development of primary hypercoagulable states.

View larger version (26K): [in this window] [in a new window]   Figure 2. Endothelial cell thromboregulation. Constitutive nonthrombogenic functions of the endothelium include thrombomodulin-bound thrombin generation of activated protein C, heparin antithrombin III, and plasmin generation. After an inflammatory stimulus, "activated" endothelium expresses tissue factor, cell adhesion molecules (CAM), plasminogen activator inhibitor (PAI-1), and platelet-activating factor (PAF). Plg = plasminogen.

Inherited Disorders

Abnormalities of the antithrombin III-heparan-sulfate proteoglycan system, the protein C-protein S system, or the surface fibrinolytic system are all inherited disorders (Table 1). Abnormalities of the tissue-factor pathway inhibitor system have not yet been reported as a cause of an inherited hypercoagulable state; however, they probably will be in the future.

Antithrombin III Deficiency

Antithrombin III is a vitamin K-independent hepatocyte-synthesized protease inhibitor that irreversibly neutralizes factors XIIa, XIa, IXa, Xa, and thrombin by forming a complex with the serine protease (see Figure 1) [3]. This process is dramatically increased in the presence of heparin. After complex formation, heparin is released and binds to another antithrombin III molecule. In vivo, heparan sulfate on the endothelial cell markedly enhances antithrombin III-inhibitory function at the vascular surface [4].

Two major types of inherited antithrombin III deficiency have been described. The most common abnormality results from the decreased synthesis of a biologically normal molecule [5]. Functional and antigenic levels of circulating antithrombin III are the same in these heterozygous patients and are approximately 50% of normal values. The molecular abnormality is usually due to a nonsense mutation, splice-site substitution, or rarely, gene deletion [6]. Less frequently, functional deficiencies of antithrombin III associated with specific molecular abnormalities have been described [7]. In these patients, the biologic activity of the protease inhibitor is decreased, although the antigen level is normal; the molecular abnormality may involve a defect in heparin binding [8] or in the thrombin-binding domain [9].

Antithrombin III deficiency has classically been suspected when a patient has recurrent, familial, and juvenile deep-vein thrombosis with or without pulmonary embolism [10]. In many patients, the initial thromboembolic event occurs after exposure to identifiable thrombosis risk factors (for example, pregnancy, postpartum complications, trauma, surgery, immobilization, or oral contraceptives). Thrombosis may occur at an unusual site such as in the brachial or mesenteric veins or in the cavernous sinus. Arterial thrombosis can occur. The severity of the disease may vary substantially within and among families in a manner not correlated with the level of deficient protease inhibitor. A heterozygous mutant may be clinically silent in one person and may be devastating, leading to a pathologic event, in another. This may reflect differences in the genetic background in the two persons. Interaction of several genes may be necessary for overt clinical thrombosis [11].

A deficiency in antithrombin III levels is inherited as an autosomal dominant trait, and both sexes are affected equally. The prevalence in the general population is approximately 1 in 2000 to 1 in 5000. The prevalence in a large group of patients presenting with a history of venous thrombosis was 2.8% [10]. A crucial question is the prevalence of thrombotic disease in patients with antithrombin III deficiency. A meta-analysis [12] of 62 families showed a prevalence of venous thrombosis of 51% in persons with antithrombin III deficiency (with a range of 15% to 100%). However, objective testing for the diagnosis of venous thrombosis was done only in 17% of the patients [12]. The prevalence of arterial thrombotic disease in these patients was 2%. In contrast, the prevalence of objectively proven venous thrombosis in a well-studied cohort of 67 patients with a specific antithrombin deficiency (antithrombin III Hamilton) was only 19% [12]. Eighty percent of the initial thromboembolic events in the affected family members occurred in relation to a documented thrombosis risk factor.

Acquired antithrombin III deficiency may be seen in various clinical settings including disseminated intravascular coagulation, liver disease, the nephrotic syndrome, L-asparaginase chemotherapy, and oral contraceptive use. These are usually easily recognizable and should not be confused as the initial presenting manifestation of the hereditary deficiency state. The short-term use of antithrombin III concentrate as a therapeutic replacement in these conditions is expensive but may be clinically valuable in situations such as fulminant hepatic failure [13], fatty liver of pregnancy complicated by disseminated intravascular dissemination, or after L-asparaginase therapy [14].

Deficiencies in the Protein C and Protein S System

Vascular endothelium, except in the brain [15], contains thrombomodulin, a receptor that binds thrombin and alters its substrate specificity. Thrombomodulin-bound thrombin is a potent activator of protein C. Protein Ca (an active vitamin K-dependent serine protease), in association with protein S (a membrane-bound vitamin K-dependent cofactor), is a physiologic anticoagulant because it inactivates factor V_a and factor VIII_a [16]. This system is a major regulator of blood fluidity and prevents thrombus formation, particularly at the capillary level where there is a relatively high density of thrombomodulin receptors [17]. The clinical manifestations and the inheritance patterns of these deficiency states resemble those seen in antithrombin III deficiency. Homozygous or doubly heterozygous protein C deficiency has been reported [18] in newborns with purpura fulminans. Coumadin-induced skin necrosis has been associated in some patients with heterozygous protein C or protein S deficiency [19, 20]. In this disorder, necrosis and skin infarction appear on the extremities, trunk, or breast within a few days of initiating warfarin therapy. Treatment includes heparin, vitamin K, plasma, or protein C concentrates, although it is not clear that these agents will reverse acute lesions. The pathogenesis is related to a transient hypercoagulable period associated with exaggerated protein C deficiency occurring at a time when the remaining vitamin K-dependent clotting factors are still at relatively normal levels because of their longer plasma half-lives.

Acquired deficiency of protein C or protein S or both may be seen in severe liver disease, disseminated intravascular coagulation, the nephrotic syndrome, the acute respiratory distress syndrome, pregnancy, postoperative states, or after L-asparaginase therapy. C4b binding protein (an acute-phase reactant) binds protein S and may, thus, substantially deplete the functional, free circulating protein S level [21]. This may partly explain hypercoagulable states in association with acute inflammatory processes. Decreases in protein S levels have been reported in some patients infected with human immunodeficiency virus. These patients may be at risk for thrombotic disease [22].

Disorders of Plasmin Generation

Dysplasminogenemia [23], hypoplasminogenemia [24], decreased synthesis or release of tissue plasminogen activator [25], and increased concentrations of plasminogen activator inhibitor [26] have all been reported as rare causes of recurrent familial thromboembolic disease associated with impaired fibrinolysis. Some of the conditions may be acquired, as probably occurs in a group of young survivors of myocardial infarction with concomitant hypertriglyceridemia [26]. Assay kits, both immunologic and functional, are available for these proteins; however, routine screening is not cost-effective and is not indicated.

Dysfibrinogenemias

A number of abnormal fibrinogens have been associated with thromboembolic complications. The abnormal fibrinogen molecule in several of these patients is resistant to lysis by plasmin [27]; however, most patients with dysfibrinogenemia have a bleeding disorder because of defective fibrin formation. Thrombin and reptilase clotting times are abnormal. In general, functional assays for the ability of fibrinogen to clot indicate a much lower level than the plasma antigenic level.

Homocystinuria (Cystathionine Synthase Deficiency)

The development of premature arteriosclerosis with peripheral vascular, cerebral vascular, and coronary artery disease, as well as venous thromboembolism have all been associated with homocystinuria [28, 29]. In patients with homozygous cystathionine synthase deficiency, severe vascular disease may appear in childhood. Sixty percent of these patients have thromboembolic events before the age of 40 years [28]. Persons who have heterozygous homocystinuria (1 in 70 of the normal population) may develop premature occlusive arterial disease [28]. Homocysteine abnormalities have been found in 20% to 40% of persons presenting with premature peripheral vascular disease or stroke. Recent data [30] suggest that increased plasma homocysteine levels are an independent risk factor for the development of coronary artery disease [30].

Homocysteine infusions in animals have been associated with endothelial damage [31]. At the cellular level, homocysteine down-regulates endothelial thrombomodulin function [32] and may also impair vascular surface plasmin generation by inhibiting binding of tissue plasminogen activator to the endothelial cell receptor [33]. Homocysteine also alters lipoprotein(a) (a lipoprotein associated with accelerated atherogenesis) and augments the deposition of the lipoprotein on fibrin surfaces [34]. These multiple effects may explain the thrombogenic and atherogenic potential of increased plasma homocysteine levels. It seems reasonable to screen patients who present with unexplained premature vascular disease, particularly of the cerebral or the peripheral circulation, because high doses of vitamin B₆ decrease the levels of plasma homocysteine in heterozygous persons. In homozygous patients, thromboembolic events decreased after pyridoxine treatment [29]. Therapeutic efficacy, in heterozygous persons with milder forms of the disease, remains to be proven.

Other Possible Inherited Hypercoagulable States

Heparin cofactor II is a potent inhibitor of thrombin in the presence of heparin [35]. The inhibitor is also activated by dermatan sulfate, which is present in skin and in various connective tissues. Heparin cofactor II forms a stable enzyme-inhibitor complex and functions independently of antithrombin III. Venous thrombotic disease has been described in a family with 50% of the normal levels of heparin cofactor II [36]. Another heparin-binding plasma protein, histidine-rich glycoprotein, also binds plasminogen and may modulate plasmin generation at vascular and cell surfaces. An inherited increase in levels of histidine-rich glycoprotein has been reported in a family with thromboembolic disease [37]. It remains to be proven whether a causal relation exists between heparin cofactor II deficiency or increased levels of histidine-rich glycoprotein and clinical thrombosis.

Secondary Hypercoagulable States

Hypercoagulable states can be secondary to a large number of heterogeneous disorders (Table 2). In some of these, platelet abnormalities or endothelial dysfunction or both, mediated perhaps by cytokine activation, may be important. Based on extensive in-vitro and in-vivo experimental observations in the literature, we hypothesize that endothelial activation leads to loss of the normal anticoagulant surface functions with conversion to a proinflammatory thrombogenic phenotype. This is mediated in part (see Figure 2) by the surface expression of tissue factor, adhesion receptors for leukocytes and platelets, platelet-activating factor, and by secretion of plasminogen activator inhibitor 1.

Although failure of normal endothelial steady-state functions underlie most primary (inherited) hypercoagulable states, endothelial activation and acquisition of a vascular thrombogenic phenotype account for most secondary (acquired) hypercoagulable states. In recent years, substantial advances have occurred for a number of these troubling clinical disorders and are presented briefly.

The Antiphospholipid Syndrome

The antiphospholipid syndrome occurs because of the appearance of circulating autoantibodies to negatively charged phospholipids, the best characterized of which is cardiolipin [38]. Clinical features include venous and arterial thrombosis, fetal wastage, thrombocytopenia and, occasionally, livedo reticularis, pulmonary hypertension, valvular heart disease, various neurologic disorders, and retinal artery or retinal vein thrombosis [39]. The antiphospholipid antibodies are usually detected by positive tests for syphilis (a false-positive test; that is, the patient does not have syphilis), circulating lupus anticoagulant, and IgG anticardiolipin. Different antibodies exist and are not present in all patients. The enzyme-linked immunosorbent assay for anticardiolipin antibody is the most sensitive and specific diagnostic (>80%) test [38]. The lupus anticoagulant causes prolongation of the activated partial thromboplastin time and the Russell viper venom time. The antiphospholipid syndrome may be independent of an underlying collagen vascular disorder or part of systemic lupus erythematosus, drug-induced syndromes, and the acquired immunodeficiency syndrome.

A recent literature review and meta-analysis [40] has shown that as many as one half of patients with lupus or lupus-like disorders have either lupus anticoagulant or antiphospholipid antibodies. An association between antiphospholipid antibodies and a history of thrombosis or ischemic neurologic disease, as well as fetal loss, has been noted. In patients without lupus, however, no strong association with thromboembolic disease has been detected. However, well-documented cases of thromboembolic disease have been described in patients who have presented with clear-cut hypercoagulability and antiphospholipid antibodies; these patients have no other risk factors. The pathophysiologic mechanisms leading to thromboembolic disease in these patients are not fully clarified. Some reports [41] suggest that these autoantibodies may inhibit endothelial-cell prostacyclin production or block endothelial-cell thrombomodulin-mediated protein C activation.

Thrombotic disease in these patients should be managed with anticoagulant agents. Therapy is not indicated in asymptomatic patients who present with laboratory abnormalities; however, antithrombotic prophylaxis is indicated for major surgical procedures. Low-dose subcutaneous heparin and aspirin (80 mg) have been used successfully in pregnant women with previous recurrent abortions and are recommended instead of prednisone-containing regimens [42]. Rarely, patients with the antiphospholipid syndrome and a high titer of cardiolipin antibodies have presented with diffuse acute noninflammatory visceral and peripheral vascular occlusions [43]. Heparin anticoagulation, plasmapheresis, and intravenous globulin should be used in these life-threatening episodes.

Increased Levels of Plasma Factor VII and Fibrinogen

Several epidemiologic studies [44] have shown an association between increased or "high normal" levels of factor VII coagulant activity and plasma fibrinogen with the risk for ischemic heart disease. This supports the well-recognized role of thrombosis in the pathogenesis of coronary vascular occlusion. The concept that increased factor VII coagulant activity predisposes persons to a prethrombotic state has been supported by data [45] showing a positive correlation between levels of plasma VII coagulant activity and the concentration of activation peptide (F₁ +2), a fragment generated from prothrombin after the conversion to thrombin. This correlation supports data [46] suggesting that the factor VII-tissue factor pathway is the crucial physiologic variable controlling the basal activation state of coagulation.

A major determinant of factor VII activity is dietary fat intake [47]. Epidemiologic studies suggest that smoking is a major determinant of the fibrinogen level [48]. Thus, two cardiovascular risk factors appear to directly influence coagulation activity. A recent meta-analysis [49] suggests that fibrinogen may be an independent cardiovascular risk factor. Fibrinogen is an important determinant of blood viscosity [50], influences platelet aggregability [51], and may interact with endothelial cells after an inflammatory stimulus [52]. Thus, low-dose anticoagulation for the primary prevention of ischemic heart disease may be appropriate for certain high-risk patients. Current ongoing trials may answer this question in the next few years.

Anti-cancer Drugs

Antineoplastic agents may be associated with a clinically heterogeneous group of vascular abnormalities ranging from fatal thrombotic thrombocytopenia purpura to recurrent venous thrombosis. Many of these disorders reflect direct effects of drugs or drug metabolites on the endothelium. This clinical spectrum of vascular toxicity includes pulmonary veno-occlusive disease (bleomycin), hepatic veno-occlusive disease (conditioning regimens for bone marrow transplantation and cyclophosphamide), the Budd-Chiari syndrome (methotrexate), myocardial infarction (vinca alkaloids), thrombotic thrombocytopenic purpura (mitomycin), the Raynaud phenomenon (vinblastine and bleomycin), and venous thrombosis (combination adjuvant chemotherapy for breast cancer) [53]. Among patients older than 50 years, the incidence of venous thrombosis during chemotherapy for stage II breast cancer was 10% [54].

Heparin-induced Thrombopathy

Thrombocytopenia develops in approximately 1% to 5% of patients receiving heparin [55]. In most of these patients, the pathogenic mechanism is unknown and the clinical consequences are minimal. In some situations, however, the heparin appears to act as a hapten and initiates an immune response against a platelet-heparin complex. In some patients, paradoxical life-threatening arterial or venous thrombosis at multiple sites may develop and may necessitate immediate cessation of heparin. The pathophysiologic mechanism of this acquired hypercoagulable state is not fully understood. However, studies [56] suggest that immunologic endothelial cell injury and activation initiate the thrombosis. In addition, intravascular platelet aggregates may form, contributing to thrombosis and tissue injury. For this reason, platelet transfusions may worsen the problem and should be avoided. Management is difficult because of the need to stop administration of the offending drug, heparin. If adequate anticoagulation with warfarin has not been achieved, ancrod (derived from the Malayan pit viper; Venacil, Abbott Laboratories, Chicago, Illinois), if available, should be given, leading to causes defibrination and anticoagulation [57]. In the near future, other rapid-acting, intravenous nonheparin anticoagulants, such as recombinant hirudin or hirudin analogs, will be useful in these clinical situations.

The Myeloproliferative Syndromes

Polycythemia vera, essential thrombocythemia, and agnogenic myeloid metaplasia are all myeloproliferative syndromes that have been associated with hypercoagulability, presumably because of increased whole blood viscosity or thrombocytosis or both. Abnormal platelet function in these disorders probably contributes to the thrombotic process. In addition, hepatic vein or mesenteric venous thrombosis may be associated with primary myeloproliferative disorders, even in the absence of abnormal erythrocyte or platelet counts. In one prospective study [58] of 20 patients with the Budd-Chiari syndrome, myeloproliferative disorders were detected in 16 patients. Conventional criteria for primary myeloproliferative disorders were met in only 2 of the 16 patients. Similar data have been obtained by at least three other groups studying a total of 99 patients with portal or mesenteric thrombosis. Therefore, an occult myeloproliferative disorder without peripheral blood changes may be a major cause of mesenteric or portal system venous thrombosis, particularly in young women.

Cancer

The association of cancer and hypercoagulability has been recognized for more than a century. Patients who develop deep venous thrombosis with no identifiable risk factor (such as surgery, immobilization, trauma, or pregnancy) have a substantial likelihood of having or developing clinically overt cancer. A recent prospective study [59] indicated that the incidence of cancer was as high as 10% in these patients.

The pathophysiologic processes underlying this clinical association are not fully understood. However, it has been shown that tumor cells interact with thrombin- and plasmin-generating systems and can directly influence thrombus formation. Careful diagnostic evaluation is clearly indicated in patients who present with deep-vein thrombosis and no identifiable risk factor. A work-up should include serum carcinoembryonic antigen and prostate-specific antigen in men, mammography in women, and repeated tests for fecal occult blood. The diagnostic value of computed tomography scans and gastrointestinal endoscopy, in otherwise asymptomatic patients, has not been proven.

Diagnosis

When should the patient who presents with an isolated acute venous thrombosis be considered as a potential carrier of a primary hypercoagulable state? More than 90% of the time, the pathologic event will not be explained by an abnormality of the proteolytic inhibitor system or defects in the fibrinolytic pathway. Screening all patients with deep-vein thrombosis is not cost effective [60]. The prevalence of inhibitor deficiency states in 680 consecutive patients with a history of venous thrombosis was 7.1% (antithrombin III deficiency, 2.8%; protein C deficiency, 2.5%; protein S deficiency, 1.3%; and combined protein C and protein S deficiency, 0.4%) [61]. In this study, it was assumed that the defect was hereditary when the same defect was detected in at least one additional family member.

Screening

In view of the low prevalence of inherited abnormalities in the general population and the low mortality rate of untreated undiagnosed patients, we recommend that only those patients who have recurrent, familial, or juvenile deep-vein thrombosis or thrombosis in an unusual location (for example, mesenteric or cerebral) be screened routinely for inherited abnormalities. If a patient meets these criteria, diagnostic evaluation should include tests of antithrombin III, protein C, protein S, and fibrinogen levels. The best single screening test for antithrombin III deficiency is a functional antithrombin-heparin cofactor assay. Protein C levels should be measured both antigenically and functionally using a chromogenic substrate or a factor Xa clotting assay [62]. Warfarin therapy decreases circulating protein C levels; thus, to properly diagnose a deficiency state in an anticoagulated patient, it is necessary to measure the ratio of protein C antigen to prothrombin antigen. Alternatively, family studies can be useful in this situation. In general, it is best to wait 1 to 2 weeks after warfarin has been discontinued and to then measure the protein C level in a patient suspected of having the deficiency state. Total protein S antigen, bound and free, should be measured to evaluate a deficiency state. Functional assays are not readily available. Familial thromboembolic disease with initial episodes appearing in the juvenile period—in the absence of protein C, protein S, or antithrombin III deficiency—indicates the possibility of a hereditary dysfibrinogenemia. Screening of patients and family members should include thrombin and reptilase times, as well as functional and antigenic fibrinogen levels. In general, functional assays for fibrinogen will indicate a much lower level than will plasma antigenic levels.

Premature arterial thrombosis (for example, a cerebrovascular accident or myocardial infarction before the age of 40 years) indicates the possibility of heterozygous homocystinuria, which has a prevalence of 1 in 70 in the general population. In many of these patients, abnormal homocystine metabolism is shown only after a methionine-loading test [63]. If a positive test is found, family members should be screened as well because the sequelae of this disease can be clinically devastating.

Thrombosis in the mesenteric or portal circulation in the absence of underlying anatomic abnormalities indicates the possibility of an occult myeloproliferative disease. Making the diagnosis is difficult, but it can be made by bone marrow biopsy, cytogenics, and bone marrow culture showing growth of erythroid precursors in the absence of exogenous erythropoietin [64]. Rarely, paroxysmal nocturnal hemoglobinuria occurs with mesenteric thrombosis.

Bauer and colleagues [65] have proposed a "biochemical" definition of the prethrombotic state based on recently developed assays for circulating byproducts of the activated coagulation cascade. They have suggested that showing circulating evidence of factor Xa activity (quantified by measuring the levels of the prothrombin activation peptide, fragment₁ +2) with normal or slightly increased evidence of thrombin activity (quantified by measuring levels of the fibrino-peptide A) may define persons at risk for thrombosis. Limited clinical studies^[66] to date, however, have not supported the use of currently available assays for this purpose.

Family Testing

If evaluation of a patient with thromboembolic disease shows a primary hypercoagulable state (for example, deficiency of antithrombin III, protein C, protein S, or dysfibrinogenemia), family testing is warranted, because at least 50% of first-degree relatives will also have the abnormality. Although routine prophylaxis of asymptomatic persons is not indicated, they should be counseled about the need for prophylaxis in high-risk situations, such as surgery, prolonged immobilization, and pregnancy. They should also receive counseling about avoiding other risk factors, such as cigarette smoking and use of oral contraceptives.

Management

Treatment decisions about patients with primary hypercoagulable states revolve around three questions: when to start, which agents to use, and what to do about pregnancy? The management of asymptomatic patients with deficiencies of antithrombin III, protein C, or protein S remains to be defined. Because thrombotic risks and thrombotic mortality in untreated asymptomatic patients are low, and risks for bleeding using life-long anticoagulation are high, we recommend that prophylaxis for venous thrombosis be restricted to recognizable high-risk situations. Long-term prophylaxis with oral anticoagulants is recommended only for patients who have had two or more documented, spontaneous thrombotic events or a single life-threatening event [67]. In general, as with any patient with thromboembolic disease, heparin is used for acute events, and warfarin is used for chronic prophylaxis. Because of the potential risk for warfarin-induced necrosis in patients with protein C or protein S deficiency, it is best to begin oral therapy in these patients when they are on full doses of heparin.

Purified, human antithrombin III (Thrombate III; Miles Biological, West Haven, Connecticut) is now available for replacement therapy. Because of its cost, potential for infectious complications, and need for frequent intravenous administration, it is not routinely recommended for antithrombin III-deficient patients. In some patients with an acute thrombotic event, it may be a useful adjunct to heparin, especially if adequate anticoagulation is difficult to achieve with heparin alone. Antithrombin III replacement is also indicated for short-term therapy in some patients with acquired antithrombin III-deficiency states, such as fulminant hepatic failure, fatty liver of pregnancy with disseminated intravascular coagulation, and L-asparaginase chemotherapy.

Management of pregnancy in patients with primary hypercoagulable states remains an important clinical problem. Prophylactic anticoagulation is indicated for women who have a history of thrombosis [68]; for women with no history, anticoagulation is probably also indicated, although clinical data are lacking. The risk for thrombosis during pregnancy is greater for women with antithrombin III deficiency and less for women with protein S deficiency [69]. Coumadin is no longer used during pregnancy because of potential teratogenicity [70]; therefore, heparin should be used throughout pregnancy. Recent studies [71] have shown the efficacy and safety of low-molecular-weight heparin for thrombosis prophylaxis in pregnancy. The drug does not cross the placenta, may be administered as a single daily injection, and appears to be associated with less thrombocytopenia. For asymptomatic women with no previous thrombotic events, heparin can be started at the time the pregnancy is diagnosed [72]. For planned pregnancies in women receiving long-term prophylaxis, warfarin should be discontinued and heparin begun before pregnancy. Alternatively, for women who are antithrombin III deficient, replacement therapy can be used before pregnancy and heparin can be used after the onset of pregnancy.