Chapter 121: Acquired Qualitative Platelet Disorders

Platelet function may be adversely affected by drugs and by hematologic and nonhematologic diseases. Because the use of aspirin and other nonsteroidal antiinflammatory agents is pervasive, acquired platelet dysfunction is much more frequent than inherited platelet dysfunction. Acquired disorders of platelet function can be classified according to the underlying clinical conditions with which they are associated (Table 121–1).

It is important to have a balanced view of the clinical significance of acquired disorders of platelet function. On the one hand, their severity is usually mild. On the other hand, there are important exceptions to this rule, particularly when platelet dysfunction is associated with other hemostatic defects. If a patient does not present with a history of bleeding, it may be difficult to predict the risk of future bleeding. This is not surprising since even patients with thrombocytopenia may experience little or no spontaneous bleeding until their platelet count is less than 10 × 10⁹/L. Furthermore, clinical assessment of these disorders is made problematic by difficulties in standardization and interpretation of laboratory tests of platelet function, including platelet aggregometry. These tests are more useful in diagnosing platelet dysfunction than in predicting the risk of bleeding.^1,2

Drugs are the most common cause of platelet dysfunction (Table 121–2). For example, in an analysis of 72 hospitalized patients with a prolonged bleeding time (a test no longer considered reliable), 54 percent were receiving large doses of antibiotics known to prolong the bleeding time and 10 percent were taking aspirin or other nonsteroidal antiinflammatory drugs.³ Some drugs can prolong the bleeding time and either cause or exacerbate a bleeding diathesis. Other drugs may prolong the bleeding time but not cause bleeding, while many only affect platelet function ex vivo or when added to platelets in vitro. It is important for the hematologist to understand the clinical significance of these distinctions.

ASPIRIN AND OTHER NONSTEROIDAL ANTIINFLAMMATORY DRUGS

Aspirin

Aspirin irreversibly inactivates the enzyme cyclooxygenase (COX), also known as prostaglandin endoperoxide H synthase, by acetylating a serine residue at position 529.⁴ Two isoforms of COX have been identified (COX-1 and COX-2),⁵ as well as a splice variant of COX-1, COX-1b (COX-3), whose functional significance is uncertain.⁶ COX-1 is constitutively expressed by many tissues, including platelets, the gastric mucosa, and endothelial cells (Chap. 134 discusses the use of aspirin as an antithrombotic agent).⁵ COX-2 is undetectable in most tissues, but its synthesis is rapidly induced in cells such as endothelial cells, fibroblasts, and monocytes by growth factors, cytokines, endotoxin, and hormones.⁵ Platelets express only COX-1, whereas endothelial cells can express both COX-1 and COX-2.^7,8 In the cardiovascular system, COX products regulate complex interactions between platelets and the vessel wall. The platelet product of COX-1 mediated prostaglandin synthesis, thromboxane A₂ (TXA₂), produces vasoconstriction and is a receptor-mediated agonist for platelet aggregation and secretion.⁴ Thus, inactivation of COX-1 by aspirin prevents platelet synthesis of TXA₂, thereby inhibiting platelet responses that depend on this substance. Accordingly, platelet responses to adenosine diphosphate (ADP), epinephrine, low doses of collagen and thrombin, and arachidonic acid are affected (arachidonic acid completely), but there is almost no effect on the responses to higher doses of collagen or thrombin.^9,10 On the other hand, the endothelial cell prostaglandin (PG) product, prostacyclin (PGI₂), produces smooth muscle cell relaxation and vasodilation and increases the platelet content of cyclic adenosine monophosphate (AMP), thereby decreasing overall platelet reactivity.¹¹

Platelet PG synthesis in an adult is nearly completely inhibited by a single 100-mg dose of aspirin or by 30 mg taken daily for 7 to 10 days.⁴ Although single doses of aspirin irreversibly inhibit platelet and endothelial cell COX,¹² they have no lasting effect on PG synthesis by endothelial cells because of the ability of these cells to synthesize additional COX unaffected by aspirin.¹³ In vitro studies also suggest that the presence of erythrocytes contributes to agonist-stimulated platelet reactivity,¹⁴ an effect that can be inhibited by aspirin at doses greater than those required to inhibit platelet COX-1.¹⁵ A meta-analysis of clinical trials indicates that aspirin doses varying from 50 to 1500 mg daily are equally efficacious in preventing adverse cardiovascular and cerebrovascular events.¹⁶ This has led many to suggest that the lowest effective doses should be prescribed to minimize gastrointestinal toxicity. Nonetheless, even low doses of aspirin can be associated with significant gastrointestinal hemorrhage.^17,18,19

Aspirin is one of the relatively few drugs that prolongs the bleeding time in humans and appears to do so by blocking aggregation rather than adhesion. In normal individuals, the effect on the bleeding time is slight (generally no more than 1.2 to 2.0 times the preaspirin bleeding time),^20,21 observed in both males and females, and requires that almost all the COX in the circulating platelets be inhibited.¹¹ The sensitivity of the bleeding time to aspirin is dependent on such technical variables as the direction of the incision on the forearm and the degree of hydrostatic pressure applied to the arm,²² and hence the current view that the test is unreliable. The bleeding time may remain prolonged for 1 to 4 days after aspirin has been discontinued, and platelet aggregation tests may remain abnormal for up to a week until platelets affected by aspirin are replaced as the result of thrombopoiesis.²³

The significance of aspirin ingestion on the hemostatic competency of normal individuals appears to be minimal. Nevertheless, patients taking aspirin chronically report significant increases in bruising, epistaxis, and gastrointestinal blood loss.^17,18,19 Gastrointestinal blood loss appears to be the result of a direct effect of aspirin on the gastric mucosa.^24,25 Furthermore, there is an increase in the incidence of hemorrhagic stroke when aspirin is used in the primary and secondary prevention of vascular disease, as well as an increase in major gastrointestinal and other extracranial bleeding.²⁶ Aspirin may also increase bleeding in the mother and the neonate during parturition.²⁷ In addition, some studies show that aspirin taken preoperatively increases the amount of blood loss following cardiothoracic surgery.^28,29 In contrast, a retrospective analysis has documented the safety of performing epidural and spinal anesthesia in patients who had ingested aspirin.³⁰ Aspirin may increase the amount of blood loss following general surgery.³¹ The significance of aspirin ingestion in this setting was tested in the POISE-2 study³² in which patients at risk for vascular complications were randomized to aspirin or placebo prior to their noncardiac surgery. Although taking aspirin did not reduce the incidence of cardiovascular events, there was a small increase in hemorrhagic complications. This suggests that discontinuing aspirin prior to surgery is a useful practice, particularly prior to plastic or neurosurgical procedures in which the limits of tolerable bleeding are narrow.³³ On the other hand, patients taking aspirin and other antiplatelet agents for severe cardiovascular disease may be at risk for thrombosis if these medications are discontinued. Thus, the clinician must thoroughly weigh the potential risks and benefits of discontinuing aspirin prior to noncardiac surgery. This is especially true in patients with other hemostatic disorders; for example, aspirin precipitates hemorrhage in individuals with von Willebrand disease, hemophilia A, warfarin ingestion, uremia, and disorders of platelet function.^34,35,36 Infusion of desmopressin (DDAVP) has been effective in correcting a prolonged bleeding time caused by aspirin.^37,38

Resistance to the antiplatelet effects of aspirin (“aspirin-resistance”) is a controversial topic and whether it exists depends to large extent on whether resistance is considered from a biochemical or clinical perspective.³⁹ Biochemical resistance to the platelet inhibitory effects of aspirin, that is, the failure to achieve pharmacologic inhibition of TXA₂ production, is uncommon.³⁹ For example, when healthy subjects were given either standard or enteric-coated aspirin, 49 percent given a single dose of enteric-coated aspirin failed to inhibit TXA₂ synthesis, whereas the failure to inhibit TXA₂ synthesis was never seen in subjects given standard aspirin.⁴⁰ Nevertheless, subjects given enteric-coated aspirin eventually responded when taking it daily, implying that although some patients absorb enteric-coated aspirin preparations poorly, they will ultimately absorb sufficient amounts of aspirin to prevent platelet TXA₂ synthesis. Most commonly, aspirin-resistance occurs because patients are non-adherent with aspirin therapy, often because of gastrointestinal toxicity.⁴¹ Clinically, the term aspirin-resistance has been applied to patients who develop cardiovascular events despite taking aspirin. Given that aspirin treatment selectively inhibits platelet synthesis of only one endogenous platelet agonist, TXA₂, it is not surprising that aspirin does not completely abolish platelet-mediated vascular events.

Traditional Nonsteroidal Antiinflammatory Drugs

Unlike aspirin, nonsteroidal antiinflammatory drugs (NSAIDs), such as ibuprofen, naproxen, diclofenac, sulindac, piroxicam, indomethacin, and sulfinpyrazone, reversibly inhibit COX enzymes.⁴² Although these drugs can cause a transient prolongation of the bleeding time when given in therapeutic doses, this is usually not clinically significant.⁴³ Population studies have suggested that concurrent treatment with NSAIDs and anticoagulants increases the risk of bleeding complication, but many bleeding events were limited to the gastrointestinal tract where NSAIDs are known to induce gastritis and peptic ulcerations.⁴⁴ As evidence of the modest effect of NSAIDs on platelet function, ibuprofen has been given safely to patients with hemophilia A.^45,46 Nonetheless, care must be taken when ibuprofen is given to patients with hemophilia and HIV infection receiving zidovudine because increased bleeding has been reported in this circumstance.⁴⁷ Because ibuprofen, and probably other NSAIDs, binds to COX-1, blocking its acetylation by aspirin,⁴² coadministration of NSAIDs and aspirin may impair the irreversible, antithrombotic effects of aspirin on platelets.⁴⁸ For this reason, patients who require both medications should ingest aspirin at least 2 hours prior to the ingestion of traditional NSAIDs.

Coxibs (COX-2 Inhibitors)

COX-1 is present in the gastric mucosa where its products protect the integrity of the gastric lining cells. In inflammatory cells, COX-2 products such as PGE₂ and PGI₂ elicit an increased sense of pain and perpetuate the inflammatory process.⁴⁰ Thus, the coxibs (COX inhibitors), designed to be relatively more specific for COX-2 versus COX-1, were intended to reduce pain and inflammation with fewer gastric side effects than traditional NSAIDs.^40,42 However, clinical trials revealed that coxib administration was associated with cardiovascular toxicity (myocardial infarction, stroke, edema, exacerbation of hypertension), partly because of inhibiting PGI₂ synthesis.^{11,49,50,51,52} On the basis of these results, rofecoxib and valdecoxib were withdrawn from the market (valdecoxib was also associated with cases of Stevens-Johnson syndrome) and a black box warning regarding serious cardiovascular events was added to prescribing information for celecoxib, the only coxib now available in the United States.⁵⁰ Nonetheless, clinical evidence suggests there is no excess cardiovascular risk from daily doses of celecoxib of 200 mg or less.⁵¹ Traditional NSAIDs also inhibit COX-2 to a variable extent and several observational trials have revealed excess cardiovascular events associated with use of these drugs.^50,53,54,55 Thus, a warning has also been added to their prescribing information. If indicated, analgesics such as acetaminophen, sodium or choline salicylate and narcotics may be substituted for aspirin and NSAIDs for treating musculoskeletal pain.⁵⁰ One report suggests that acetaminophen can selectively inhibit COX-2,⁵⁶ but the clinical significance of this observation is not clear.

THIENOPYRIDINES

Ticlopidine, clopidogrel, and prasugrel are thienopyridines that are used as antiplatelet agents in arterial diseases (Chap. 134) with results at least comparable to aspirin in the secondary prevention of cerebrovascular and cardiovascular events.^16,57

Thienopyridines differ from aspirin in their mechanism of antiplatelet activity and their toxicity profile. All three thienopyridines are prodrugs that depend on oxidation by cytochrome P450 (CYP) enzymes in the liver (ticlopidine and clopidogrel) or in liver and intestine (prasugrel) to form the active metabolites that irreversibly inhibit the platelet P2Y₁₂ ADP receptor.^58,59,60,61 Ticlopidine at 250 mg twice a day, clopidogrel at 75 mg once per day, and prasugrel at 10 mg once a day inhibit platelet aggregation ex vivo in humans. The extent of this effect is equivalent to or greater than that of aspirin and the effect of thienopyridines and aspirin appears additive.^62,63 When given at their usual oral doses, the effect of thienopyridines on platelet aggregation and the bleeding time can be seen within hours of the first dose, but are not maximal for 4 to 6 days. A 300-mg loading dose of clopidogrel or 60 mg of prasugrel, followed by their usual daily doses, shortens the time required for their maximal antiplatelet effect to a few hours.^64,65 The common CYP polymorphism CYP2C19 results in lower levels of active clopidogrel and ticlopidine metabolites and has been reported to be associated with decreased platelet inhibition and an elevated risk for major adverse cardiovascular events.^55,66,67 Because the enzyme CYP3A is present in the intestine and can oxidize prasugrel to its pharmacologically active metabolite, intestinal metabolism may account for the rapid appearance and higher levels of the active metabolite in plasma after an oral dose.^61,68,69,70 Furthermore, prasugrel metabolism and inhibition of platelet function are not affected by CYP2C19 polymorphisms.^61,68,69,70

The clinical efficacy of prasugrel has been compared to clopidogrel in patients with acute coronary syndrome scheduled for percutaneous coronary intervention in the Triton-TIMI 38 trial. Patients who received prasugrel had a significantly decreased incidence of ischemic events compared to patients who received clopidogrel (9.9 percent vs. 12.1 percent, p <0.001).⁶⁹ However, major bleeding was also significantly increased in patients receiving prasugrel compared to clopidogrel (2.4 percent vs. 1.8 percent, p <0.03). Thus, although prasugrel appeared to be more efficacious than clopidogrel, this benefit was partially offset by a higher rate of hemorrhage.⁶⁹

The platelet inhibitory effects of thienopyridines persist for 4 to 10 days after the drugs have been discontinued, either because of their extended half-life after multiple dosing or their irreversible effect on platelets.⁵⁸ Ticlopidine administration is associated with potentially serious hematologic complications, including neutropenia (neutrophils <1200 × 10⁹/L in 2.4 percent of individuals)^58,71,72 and, less commonly, aplastic anemia, and thrombocytopenia.^73,74 In addition, at least one in 5000 patients develop a thrombotic thrombocytopenic purpura (TTP)-like syndrome.^75,76,77 Results from a large clinical trial suggest that hematologic complications may be less common with clopidogrel or prasugrel.⁵⁷ Clopidogrel may also be rarely associated with a TTP-like syndrome (one in 270,000),⁷⁸ although this rate is close to the TTP incidence in the general population. Because of its toxicity profile, ticlopidine has been replaced by the other thienopyridines in the United States.

Because aspirin and the thienopyridines inhibit platelet function by different mechanisms, their antithrombotic effects may be additive. In theory, this would be beneficial in the treatment of diseases associated with platelet activation such as ischemic heart disease, peripheral vascular disease, and ischemic strokes.^62,79,80 This hypothesis was tested in the CURE trial of patients with acute coronary syndromes.⁶² Although clopidogrel plus aspirin decreased the combined incidence of cardiovascular deaths, myocardial infarctions, and strokes from 11.4 percent to 9.3 percent, the benefit was partially offset by an increase in severe bleeding from 2.7 to 3.7 percent. Similarly, in the CHARISMA trial of a broad population of patients at risk for cardiovascular events, there were 94 fewer ischemic events in patients treated with both clopidogrel and aspirin, but this occurred at the expense of 93 more moderate or severe bleeding events.⁸¹ Furthermore, a meta-analysis of seven randomized controlled trials involving more than 39,000 patients confirmed that intracranial hemorrhage was more frequent in patients who received both clopidogrel plus aspirin compared to clopidogrel alone.⁸² Thus, except for special circumstances such as coronary artery stenting, it appears that the added benefit of dual antiplatelet therapy is small and has the added risk of increased bleeding.⁸³

OTHER ADENOSINE DIPHOSPHATE RECEPTOR ANTAGONISTS

Ticagrelor, cangrelor, and elinogrel are oral, reversible, nonthienopyridine P2Y₁₂ receptor antagonists. Because they are not prodrugs and do not require metabolic activation, the onset of their inhibitory activity is more rapid than that of the thienopyridines. A novel, and as yet unexplained, side effect of treatment with this class of the P2Y₁₂ antagonists is the occurrence of dyspnea which can complicate the management of patients with coronary artery disease.⁸⁴

Ticagrelor, the first drug of the class, has been approved for use in acute coronary syndromes. Its efficacy versus clopidogrel was tested in the PLATO trial in which patients with an acute coronary syndrome were randomized to treatment with either ticagrelor or clopidogrel.^85,86,87 At 1 year, the combined end point of death, myocardial infarction, and stroke was 9.8 percent in patients treated with ticagrelor compared to 11.7 percent in patients treated with clopidogrel.⁸⁸ Although stent thrombosis was also decreased in the ticagrelor-treated group, major bleeding not associated with coronary artery bypass surgery was increased in this group. The incidence of fatal intracranial hemorrhage was also greater in the ticagrelor-treated patients, but it was a rare event (0.1 percent of treated patients). In the ATLANTIC trial, patients suffering from an ST-segment elevation myocardial infarction were randomized to receive ticagrelor in the ambulance or in the catheterization laboratory.⁸⁹ Although initiating therapy before hospitalization was safe and lowered the incidence of stent thrombosis, there was no overall improvement in preventing major cardiovascular adverse events. Thus, ticagrelor, like prasugrel appears to be more efficacious than clopidogrel at preventing adverse cardiovascular events, but with more hemorrhagic complications.

THROMBIN RECEPTOR ANTAGONISTS

Thrombin is the most potent physiologic platelet agonist. Three G-protein–coupled thrombin receptors have been identified in humans (protease-activated receptors [PARs] 1, 3, and 4).⁹⁰ Although human platelets express both PAR-1 and PAR-4, the major platelet thrombin receptor is PAR-1 and can be activated by nanomolar concentrations of thrombin. PAR-4 signaling appears to be unnecessary for platelet activation if PAR-1 signaling is intact.⁹⁰ Vorapaxar is a potent, selective, long-acting, oral PAR-1 inhibitor generated from the naturally occurring muscarinic receptor antagonist himbacine.⁹¹ A high-resolution crystal structure of vorapaxar bound to PAR-1 revealed that the binding pocket for the drug is unusual for a peptide-activated G-protein–coupled receptor in that it consists of a superficial tunnel with little of the bound drug surface exposed to aqueous solvent, perhaps accounting for the very slow dissociation rate of vorapaxar from PAR-1.⁹²

The efficacy of vorapaxar for the secondary prevention of arterial thrombosis was examined in the phase III TRA 2P–TIMI 50 trial in which patients with a history of myocardial infarction, stroke, or peripheral arterial disease were randomized between vorapaxar and placebo.⁹³ Most patients were also taking either aspirin or a thienopyridine. Because of a high incidence of intracranial bleeding in the first years of the study, entry criteria were modified to eliminate patients with a history of a stroke. At 3 years, the incidence of the primary end point (cardiovascular death, myocardial infarction, and stroke) was significantly reduced in vorapaxar-treated patients (9.3 percent vs. 11.2 percent, p <0.001). However, moderate to severe bleeding, including intracranial bleeding, was significantly increased in the vorapaxar-treated patients (4.2 percent vs. 2.5 percent, p <0.001). Nonetheless, based on efficacy, vorapaxar received FDA approval in 2014. Atopaxar, a second PAR-1 antagonist, is currently being evaluated in clinical trials.⁹⁴ Atopaxar has a shorter half-life than vorapaxar, suggesting that potential bleeding complications might be easier to manage.

INTEGRIN α_IIbβ₃ RECEPTOR ANTAGONISTS

Drugs that specifically impair the function of the major platelet integrin α_IIbβ₃ (GPIIb/IIIa) have been developed for short-term use as antithrombotic agents in the setting of ischemic coronary artery disease.^95,96 Integrin α_IIbβ₃ mediates platelet–platelet cohesion by binding the divalent ligand fibrinogen, thereby crosslinking the integrin on adjacent platelets, causing the formation of platelet aggregates.⁹⁷ Thus, integrin α_IIbβ₃ is a viable therapeutic target to prevent arterial thrombosis. Abciximab, eptifibatide, and tirofiban are three FDA-approved structurally dissimilar integrin α_IIbβ₃ inhibitors that rapidly impair platelet aggregation. Abciximab is a human-murine chimeric Fab fragment, eptifibatide is a cyclic heptapeptide based on the sequence Lys-Gly-Asp (KGD), and tirofiban is an Arg-Gly-Asp (RGD)-based peptidomimetic. All three drugs have demonstrated efficacy in the management of patients with acute coronary syndromes, particularly in the setting of percutaneous coronary interventions (PCI) where iatrogenic artery wall injury occurs.⁹⁷

Inherited integrin α_IIbβ₃ abnormalities cause the bleeding disorder Glanzmann thrombasthenia (Chap. 120).^98,99 Thus, it is not surprising that integrin α_IIbβ₃ antagonists can predispose to bleeding. In EPIC, a clinical trial of abciximab in patients undergoing PCI, 14 percent of patients given abciximab experienced major bleeding compared to 7 percent of patients given placebo.¹⁰⁰ However, patients were also given aspirin and heparin. When the heparin dose was decreased in the subsequent EPILOG trial, the incidence of major bleeding in patients receiving abciximab decreased to 2.0 percent compared to 3.1 percent in the control group receiving heparin and aspirin alone.¹⁰¹ Nonetheless, in both EPIC and EPILOG, minor bleeding was significantly more frequent in patients given abciximab and standard-dose heparin compared to patients given standard-dose heparin alone, attesting to the ability of an integrin α_IIbβ₃ antagonist to impair normal hemostasis. In the PRISM-PLUS trial of tirofiban and the PURSUIT trial of eptifibatide, major and minor bleeding were slightly more frequent in patients receiving the study drug compared to controls.^102,103 Similarly, patients receiving the oral integrin α_IIbβ₃ inhibitors xemilofiban and sibrafiban for 30 and 28 days, respectively, frequently experienced mucocutaneous bleeding similar to that experienced by patients with congenital thrombasthenia.^104,105 Although short-term use of the parenteral integrin α_IIbβ₃ antagonists is often beneficial in patients with acute coronary syndrome of following PCI, paradoxically the long-term use of oral integrin α_IIbβ₃ inhibitors was associated with an increase in mortality.¹⁰⁶ The cause of this paradoxical effect is not clear, but has been attributed by some to an antagonist-induced conformational change in integrin α_IIbβ₃ simulating the effect of physiologic platelet agonists.¹⁰⁷

The risk of bleeding in patients undergoing PCI in the presence of integrin α_IIbβ₃ antagonists can be minimized by using heparin on a weight basis,¹⁰¹ by avoiding treatment of patients who are receiving warfarin at therapeutic doses, by early vascular sheath removal, and by meticulous care of vascular puncture sites.¹⁰⁸ Platelet transfusions can rapidly reverse the platelet function defect in patients receiving abciximab, presumably by decreasing the overall extent of integrin blockade. The ability of platelet transfusion to reverse the effects of the other integrin α_IIbβ₃ antagonists is less clear, but these drugs have very short half-lives if renal and hepatic function are normal.

Thrombocytopenia occurring within 24 hours of initiating therapy has been observed in small numbers of patients following the administration of all integrin α_IIbβ₃ antagonists.^{102,105,108,109} In EPIC, the incidence of platelet counts of less than 100 × 10⁹/L and of less than 50 × 10⁹/L in patients receiving abciximab for the first time was 3.9 percent and 0.9 percent, respectively.¹⁰⁹ Thrombocytopenia has also been reported in patients receiving eptifibatide, tirofiban, and a variety of small molecule RGD- and non–RGD-based integrin α_IIbβ₃ inhibitors with an incidence of up to 13 percent.^{102,105,109,110,111,112,113}

The mechanism responsible for thrombocytopenia following the administration of these drugs is uncertain, but may be related to the presence of preexisting antiintegrin α_IIbβ₃ antibodies that recognize epitopes exposed by the antagonist, or, in the case of abciximab, to murine sequences incorporated into the abciximab Fab fragment.¹¹⁴ The thrombocytopenia usually reverses readily when the drug is stopped, but it may also be reversed by platelet transfusion if clinically indicated.¹⁰⁸ Thrombocytopenia in patients receiving integrin α_IIbβ₃ antagonists must be differentiated from pseudothrombocytopenia as a result of drug-induced platelet clumping, from heparin-induced thrombocytopenia in patients receiving heparin concurrently, and from other causes of thrombocytopenia, depending on the clinical circumstances.^115,116 It is important to identify thrombocytopenia early because integrin α_IIbβ₃ antagonists are administered as long infusions and the drug should be stopped as soon as true thrombocytopenia has been confirmed. In most cases of profound thrombocytopenia, a platelet count obtained 2 to 4 hours after initiating therapy will provide evidence of a significant decrease in platelet count, although cases of delayed thrombocytopenia have been observed after treatment with abciximab.¹¹⁴

DRUGS THAT AFFECT PLATELET CYCLIC NUCLEOTIDE LEVELS OR FUNCTION

The pyrimidopyrimidine derivative, dipyridamole, inhibits platelet cyclic nucleotide phosphodiesterase, resulting in the intraplatelet accumulation of the inhibitory cyclic nucleotide cyclic AMP (cAMP). Dipyridamole may also inhibit the breakdown of cyclic guanosine monophosphate (cGMP), resulting in potentiation of the platelet inhibitory effect of nitric oxide.¹¹⁷ Although the platelet inhibitory effects of dipyridamole are seen in vitro, the clinical utility of dipyridamole has been controversial.^118,119 A meta-analysis failed to demonstrate the clinical benefit of adding dipyridamole to aspirin.¹⁶ However, many older dipyridamole trials used formulations with limited dipyridamole bioavailability.¹²⁰ In the European Stroke Prevention Study 2 (ESPS 2), dipyridamole was beneficial in preventing stroke and transient ischemic attack, but there was no difference in mortality between patients taking dipyridamole and placebo or among patients taking dipyridamole plus aspirin compared to either dipyridamole or aspirin alone.¹²¹ The basis for the benefit of dipyridamole in the ESPS 2 trial is unclear, but could be from a higher dipyridamole dosage or to the sustained-release dipyridamole preparation used in the trial.

Intravenous infusions of PGE₁, PGI₂, or stable PGI₂ analogues stimulate platelet adenylyl cyclase, causing an increase in platelet cAMP and a decrease in platelet responsiveness.¹²² These agents cause a transient inhibition of platelet shape change, aggregation, and secretion. However, their clinical utility is limited by their short half-life and side effects that include peripheral vasodilation.¹²³ Cilostazol, a phosphodiesterase III inhibitor has been approved in the United States for the treatment of peripheral vascular disease¹²⁴ and may have utility in the prevention of cardiac stent occlusion.¹²⁵ Nitric oxide (NO) and organic nitrates such as nitroglycerin inhibit platelet function in vitro, probably by activating guanylyl cyclase, thereby increasing cGMP.¹²⁶ Their effect on in vivo platelet function is uncertain. High concentrations of caffeine and theophylline also inhibit platelet phosphodiesterases in vitro.

ANTIBIOTICS

Penicillins contain a β-lactam ring and a unique side chain. Most cause a dose-dependent prolongation of the bleeding time in normal volunteers.¹²⁷ Because they reduce platelet aggregation and secretion, as well as ristocetin-induced platelet agglutination, they may affect both platelet adhesion and platelet activation. Tests of platelet aggregation are abnormal in 50 to 75 percent of individuals receiving large doses (at least several grams per day) of carbenicillin, penicillin G, ticarcillin, ampicillin, nafcillin, and azlocillin and in 25 to 50 percent of patients taking piperacillin, azlocillin, or mezlocillin.^127,128,129 Differences in the antiplatelet effects of these antibiotics probably relate to differences in blood levels and drug potency. Their effect on platelets is maximal after 1 to 3 days of administration and may remain for several days after the antibiotic has been stopped, suggesting that the effect of these antibiotics on platelets in vivo is irreversible.

Penicillins can impair the interaction of agonists and von Willebrand factor (VWF) with the platelet membrane.¹³⁰ Indeed, when many penicillins are incubated with washed platelets, albeit at concentrations higher than those attained in vivo, they inhibit the interaction of VWF and agonists, such as ADP and epinephrine, with their platelet receptors.¹³¹ The relative in vitro antiplatelet potency of the penicillins correlates well with their lipid solubility and with the inhibitory potency of the isolated side chains.¹³² Moreover, the inhibitory effect of penicillin G on platelet function in vitro is potentiated by the presence of probenecid.¹³³ When platelet function was tested after intravenous administration of penicillin, oxacillin or mezlocillin for 3 to 17 days to patients or normal volunteers, irreversible inhibition of agonist-induced aggregation was noted, along with a 40 percent reduction in low-affinity TXA₂ receptors.¹³⁴ Thus, penicillins probably inhibit platelet function by binding to one or more membrane components necessary for adhesive interactions with the vessel wall or for stimulus-response coupling.

Although clinically significant bleeding is associated with the use of carbenicillin, penicillin G, ticarcillin, and nafcillin, it is far less common than prolongation of the bleeding time.^127,135 Patients with coexisting hemostatic defects (e.g., thrombocytopenia, vitamin K deficiency, uremia) may be particularly prone to this complication. On the other hand, high doses of penicillin G did not increase gastrointestinal blood loss in a thrombocytopenic rabbit model.¹³⁶ In our experience, bleeding attributable to antibiotic-induced platelet dysfunction is uncommon and unpredictable. Because β-lactam–induced platelet dysfunction resolves with time following cessation of the drug, this class of drugs should only be considered as a cause of bleeding in the appropriate clinical setting. A similar pattern of platelet dysfunction has been reported with some cephalosporins or related antibiotics, but not with others.^127,137,138 Broad-spectrum antibiotics can also cause a bleeding diathesis attributable to killing of gut flora, resulting in vitamin K deficiency. Nitrofurantoin, a structurally unrelated antibiotic, may cause a mild prolongation of the bleeding time and impair platelet aggregation when blood levels of the drug are higher than 20 μM, as may occur in patients with renal insufficiency.¹³⁹ Miconazole, an antifungal agent, inhibits human and rabbit platelet COX in vitro and rabbit platelet COX after intravenous infusion.¹⁴⁰

ANTICOAGULANTS, FIBRINOLYTIC AGENTS, ANTIFIBRINOLYTIC AGENTS

Heparin predisposes to bleeding primarily through its anticoagulant effect, but it may also impair platelet function. For example, a bolus injection of heparin (100 U/kg) can cause a significant prolongation of the bleeding time in normal subjects and in patients prior to cardiopulmonary bypass, suggesting that therapeutic doses of heparin may impair platelet function.¹²⁶ Heparin likely impairs platelet function by inhibiting the generation and action of the potent platelet agonist thrombin. On the other hand, in vitro studies suggest that heparin can enhance platelet aggregation induced by other platelet agonists.¹⁴¹ Heparin binds to a single class of high-affinity binding sites on resting platelets and to an additional class of lower-affinity binding sites on fully activated platelets.¹⁴² High heparin doses also impair VWF-dependent platelet function, possibly by binding to the heparin-binding domain of VWF.¹⁴³ The contributions of these effects on platelet function to the bleeding complications of heparin therapy are uncertain.

Bleeding during fibrinolytic therapy is predominantly a result of the combined effects of structural lesions in blood vessels and the fibrin(ogen)olytic activity of the agent used. However, pharmacologic doses of streptokinase, urokinase, and tissue plasminogen activator (t-PA) can affect platelet function.¹⁴⁴ High concentrations of plasmin ex vivo cause platelet aggregation.¹⁴⁵ Moreover, marked increases in the urinary excretion of the TXA₂ metabolite 2,3-dinor-TXB₂ have been detected in patients receiving streptokinase or t-PA for coronary thrombolysis, suggesting that in vivo platelet activation had occurred during infusion of the drug.^146,147 Nevertheless, several in vitro studies indicate that plasmin generation has an inhibitory effect on platelet function. First, very high levels of fibrin(ogen) degradation products, coupled with very low levels of fibrinogen, may impair platelet aggregation.¹⁴⁸ Second, plasminogen can bind to platelets¹⁴⁹ and after its conversion to plasmin, enzymatically degrade platelet glycoprotein (GP) Ib, impairing the interaction of platelets with VWF.^150,151 Third, plasmin can inhibit platelet arachidonic acid metabolism.¹⁵² Fourth, t-PA promotes the disaggregation of platelet aggregates, presumably by inducing lysis of the fibrinogen that mediates aggregate formation.¹⁵³ Finally, after initial activation, platelets incubated with plasmin and recombinant t-PA in vitro become refractory to activation by other agonists.¹⁵⁴ Whether any of these in vitro and ex vivo observations apply to the in vivo situation and are clinically significant remains uncertain.¹⁵⁵ The antifibrinolytic drug, ε-aminocaproic acid, can increase the bleeding time when administered for several days at doses 24 g/day or greater.¹⁵⁰

CARDIOVASCULAR DRUGS

Administration of nitroprusside, which increases platelet cGMP,^{156,157,158,159,160} nitroglycerine,¹⁶¹ and propranolol,^162,163 can decrease platelet aggregation and secretion ex vivo. Nitroprusside can increase the bleeding time twofold when administered at infusion rates of 6 to 8 mcg/kg/min.^156,164 Inhalation of NO advocated for the treatment of pulmonary hypertension and the adult respiratory distress syndrome, can impair agonist-induced platelet aggregation ex vivo, although the clinical significance of these observations is unclear.^165,166,167 Calcium channel blockers such as verapamil, nifedipine, and diltiazem inhibit platelet aggregation when added at very high concentrations to washed platelets.¹²³ This effect is seen primarily with epinephrine-induced aggregation and does not appear to be related to calcium channel blockade.¹⁶⁸ At therapeutic doses, calcium channel blockers do not prolong the bleeding time, although one agent, nisoldipine, has been reported to inhibit agonist-induced calcium transients and platelet aggregation after ten days of oral administration.¹⁶⁹ At high concentrations, the antiarrhythmic drug quinidine has been reported to cause a mild prolongation of the bleeding time and to potentiate the effect of aspirin.¹⁷⁰

VOLUME EXPANDERS

Dextran is a neutral polysaccharide that is heterogeneous in molecular size. Two preparations with average molecular weights of 40,000 and 70,000 are in clinical use. Although dextran infusions may prolong the bleeding time of normal subjects and patients with von Willebrand disease, this phenomenon has not been observed in most normal subjects.^9,171,172 Infused dextran adsorbs to the platelet surface and can impair platelet aggregation, secretion, and procoagulant activity. The maximal effect of dextran may require several hours, suggesting that larger molecules with a slower rate of clearance are responsible.⁹ Curiously, the drug has no effect when added to platelet-rich plasma.⁹ Dextran infusion produces a modest reduction in plasma VWF antigen levels and ristocetin cofactor activity.¹⁷¹ Despite these effects on primary hemostasis, prospective studies indicate that dextran is not associated with significant postoperative bleeding, unless it is administered together with low-dose heparin.^173,174 Hydroxyethyl starch, another volume expander, while generally safe, may prolong the bleeding time and predispose to hemorrhage, particularly if it is administered in doses exceeding 20 mL/kg of a 6-percent solution. Lower doses of hydroxyethyl starch may contribute to bleeding if administered simultaneously with low-dose heparin or if given to patients with preexistent hemostatic defects or after major cardiothoracic surgery.^{175,176,177,178} Different hydroxyethyl starch preparations vary in the average number of hydroxymethyl groups per glucose unit, and this may affect both intravascular survival and effects on hemostasis.^179,180

PSYCHOTROPIC DRUGS, ANESTHETICS, AND COCAINE

Platelets from patients taking antidepressants or phenothiazines may exhibit impaired aggregation, but this is not associated with bleeding.^181,182 The effect on aggregation has been attributed to inhibition of intracellular signaling molecules such as protein kinase C (PKC).¹⁸³ Selective serotonin reuptake inhibitors such as paroxetine, have been shown to decrease platelet serotonin storage.¹⁸⁴ Fluoxetine does not appear to impair platelet aggregation in vitro and has only rarely been associated with clinical bleeding.^185,186 General anesthesia with halothane or propofol may cause a slight prolongation of the bleeding time, most likely the result of an effect on calcium signaling, but this has no adverse effect on surgical hemostasis.^187,188 In addition to an association with thrombocytopenia, cocaine has been reported to either inhibit^189,190 or stimulate platelet activation.¹⁹¹ It has been suggested that heroin decreases platelet NO production.¹⁹² The clinical relevance of these observations is unknown.

ONCOLOGIC DRUGS

Administering mithramycin to a total dose of 6 to 21 mg decreases platelet aggregation and is associated with mucocutaneous bleeding.¹⁹³ An ex vivo defect in platelet secretion and secondary aggregation has been reported in patients with solid tumors within 48 hours of receiving infusions of autologous marrow and high-dose chemotherapy consisting of cisplatin, cyclophosphamide, and either bis-chloroethylnitrosourea (BCNU) or melphelan.¹⁹⁴ Both daunorubicin and BCNU can inhibit platelet aggregation and secretion when added to platelet-rich plasma, but they have not been shown to cause clinically significant platelet dysfunction.^195,196,197 Administration of recombinant forms of thrombopoietin to thrombocytopenic patients with cancer results in the production of normally functioning platelets.^198,199 Dasatinib, the broad-spectrum protein tyrosine kinase inhibitor, impairs collagen-induced platelet activation in vitro and increases tail bleeding times in mice, perhaps explaining some bleeding episodes in patients with chronic myelogenous leukemia who have been treated with the drug.²⁰⁰ Ibrutinib, a Bruton tyrosine kinase (BTK) inhibitor efficacious in a wide variety of lymphoid malignanies,^201,202 is associated with hemorrhagic complications in up to half of patients, with significant hemorrhagic toxicity in 5 percent.^201,202,203 Exposing platelets to ibrutinib ex vivo can produce defective platelet adhesion.²⁰⁴ Furthermore, humans or mice lacking BTK have impaired ex vivo platelet function, although the impairment is quite mild.^205,206 Whether the hemorrhagic toxicity of ibrutinib is caused by platelet BTK inhibition or by an off-target effect remains to be determined.

MISCELLANEOUS AGENTS

The immunosuppressive drug cyclosporine has been reported to enhance ADP-stimulated platelet aggregation in vitro.²⁰⁷ It is unclear whether this contributes to the TTP-like syndrome associated with this drug. Antihistamines,²⁰⁸ the serotonin antagonist ketanserin,²⁰⁹ and certain radiographic contrast agents^210,211 can impair platelet aggregation responses ex vivo by unknown mechanisms.

FOODS AND FOOD ADDITIVES

Certain foods and food additives affect platelet function in vitro, and it is conceivable that some may affect hemostasis, particularly in association with other hemostatic defects. For example, diets rich in fish oils containing ω-3 fatty acids (eicosapentaenoic acid; docosahexaenoic acid) cause a slight prolongation of the bleeding time.²¹² These fatty acids act by reducing the platelet content of arachidonic acid and by competing with arachidonic acid for COX.^213,214 Easy bruising noted after eating Chinese food has been attributed to an antiplatelet effect of the black tree fungus.²¹⁵ A component of extract of onion can inhibit platelet arachidonic acid metabolism.²¹⁶ Ajoene, a component of garlic, is an inhibitor of fibrinogen binding and platelet aggregation.²¹⁷ Extracts of two commonly used spices, cumin and turmeric, also inhibit platelet aggregation and eicosanoid biosynthesis.²¹⁸

CHRONIC MYELOPROLIFERATIVE NEOPLASMS

Definition and History

Bleeding and thrombosis are significant causes of morbidity and mortality in the chronic myeloproliferative neoplasms, particularly in essential thrombocythemia, polycythemia vera, and primary myelofibrosis.^219,220,221 Thrombocytosis is a constant finding in essential thrombocythemia, but the differential diagnosis includes these other myeloproliferative neoplasms, including chronic myelogenous leukemia, as well as other diseases associated with reactive thrombosis (Chap. 119).^222,223 Most of the information about platelets, bleeding and thrombosis in the myeloproliferative neoplasms comes from studies of essential thrombocythemia and polycythemia vera.

Etiology and Pathogenesis

Several factors contribute to the hemostatic abnormalities in the myeloproliferative neoplasms: (1) Increased whole-blood viscosity in polycythemia vera: The engorgement of blood vessels associated with polycythemia is a risk factor for thrombosis and bleeding, particularly in postoperative situations.^224,225,226 (2) Intrinsic defects in platelet function: Many intrinsic platelet function defects have been reported in the myeloproliferative neoplasms, although their precise relationships to clinical bleeding are generally unclear.^227,228 (3) Elevated platelet counts: The contribution of an elevated platelet count, per se, to the risk of hemorrhage and thrombosis in myeloproliferative neoplasms is controversial, as the risk does not extend to patients with reactive thrombocytosis.^229,230 A number of retrospective studies indicate that the risk of abnormal hemostasis cannot be confidently predicted from the degree of thrombocytosis.²²⁷ On the other hand, acquired von Willebrand syndrome, which represents a potential major cause of bleeding in the chronic myeloproliferative neoplasms, is most frequently associated with extreme elevations of the platelet count (e.g., ≥1000 to 1500 × 10⁹/L)^231,232,233; in some, the VWF abnormality can be corrected, albeit transiently, by infusion of DDAVP or factor VIII/VWF concentrates, while in others it can be partially or completely corrected by cytoreductive therapy.²³⁴ (4) Leukocytosis may represent a risk factor for thrombosis in the myeloproliferative neoplasms.^221,235 In this context, leukocyte and/or endothelial dysfunction may contribute to the thrombotic phenotype in some individuals with polycythemia vera^236,237 or essential thrombocythemia²³² through leukocyte–platelet and leukocyte–endothelial cell interactions.^232,238,239

Under the light or electron microscope, platelets in these disorders may be larger or smaller than normal, may be abnormally shaped, and may exhibit a reduction in the number of storage granules.²⁴⁰ In essential thrombocythemia, platelet survival may be modestly reduced.²⁴¹ A number of functional and biochemical abnormalities have been described in platelets from patients with myeloproliferative neoplasms. The most frequently encountered functional abnormality is a decrease in platelet aggregation and granule secretion in response to epinephrine, ADP, or collagen.²²⁷ The defect in epinephrine-induced aggregation often includes absence of the primary wave of aggregation, which is unusual in other conditions. This is not simply the result of an elevated platelet count, because it is not encountered in reactive thrombocytosis.^222,242 Thus, loss of platelet responsiveness to epinephrine may help to support the presence of a myeloproliferative neoplasm in otherwise ambiguous cases, although the discovery of genetic abnormalities (e.g., JAK2, thrombopoietin receptor [MPL], calreticulin) is beginning to eliminate all ambiguity in the diagnosis of a myeloproliferative neoplasm (Chaps. 84 to 86).

Reduced platelet aggregation and secretion in the myeloproliferative neoplasms is associated with one or more of the following: decreased agonist-induced release of arachidonic acid from membrane phospholipids^243,244; reduced conversion of arachidonic acid to PG endoperoxides or lipoxygenase products²⁴⁵; reduced platelet responsiveness to TXA₂²⁴⁶; decreased numbers of α₂-adrenergic receptors associated with reduced or absent platelet responses to epinephrine^247,248; deficiency of integrin α₂β₁, resulting in variable changes in platelet responsiveness to collagen²⁴⁹; diminished stimulus–response coupling downstream of several agonists associated with reduced activation of phosphatidylinositide 3′-kinase, Rap1 and integrin α_IIbβ₃²⁵⁰; and deficiency of dense or α granules.^251,252 Reduction in platelet procoagulant activity has been reported in some patients with myeloproliferative neoplasms and thrombocytosis,²⁵³ as have specific platelet membrane abnormalities, including decreased expression and activation of integrin α_IIbβ₃,²⁵⁴ decreased amounts of the GPIb–V–IX complex, resulting in an acquired form of Bernard-Soulier syndrome²⁵⁵; decreased numbers of receptors for PGD₂²⁵⁶; increased numbers of FcγRIIa receptors²⁵⁷; an increase in GPIV (CD36) with^258,259 or without²⁶⁰ a corresponding decrease in GPIb; and impaired expression of MPL in polycythemia vera²⁶¹ and essential thrombocythemia.¹¹¹

On the other hand, evidence for in vitro platelet or coagulation hyperactivity has been reported in the myeloproliferative neoplasms. This includes spontaneous platelet aggregation in a patient with essential thrombocythemia and thrombosis,²⁶² increased thromboxane biosynthesis by platelets from patients with essential thrombocythemia²⁶³ or polycythemia vera,²⁶⁴ and increased “procoagulant imbalance” in patients manifested by increased endogenous thrombin potential²⁶⁵ and increased procoagulant activity in circulating microparticles.²⁶⁶

Several features of these protean in vitro platelet functional defects require emphasis relative to the clinical setting. First, none are unique to a particular myeloproliferative neoplasm. Second, their relative frequencies have varied widely in reported series. Third, none has been prospectively shown to be predictive of bleeding or thrombosis. Fourth, although the chronic myeloproliferative neoplasms comprise several distinct clinicopathologic entities, they represent clonal abnormalities of hematopoiesis.²⁶⁷ Consequently, megakaryocytes and their platelet progeny may acquire genetic, biochemical, and structural abnormalities as they develop from clones of abnormal progenitors. Examples of clonal defects in the chronic myeloproliferative neoplasms are acquisition of activating mutations in JAK2 (e.g., V617F in polycythemia vera, essential thrombocythemia, and myelofibrosis; or in exon 12 in polycythemia vera)^{268,269,270,272} or MPL (W515L/K in essential thrombocythemia and myelofibrosis).^273,274 Mutations in the calreticulin gene have been found in most of the essential thrombocythemia and myelofibrosis patients who lack activating mutations in JAK2 or MPL.^275,276 It is biologically plausible that mutations in these or other leukocyte and platelet proteins might influence hemostatic mechanisms, including the activation state of platelets.^277,278,279 However, the precise impact of their presence or allele burden on human platelet function and on thrombotic risk is only now beginning to be understood.^232,280 For example, most,^232,268 but not all,^237,281 studies have concluded that the presence of the JAK2 (V617F) mutation or a high JAK2 (V617F) allele burden confers increased thrombotic risk in essential thrombocythemia, the latter in part a result of higher hemoglobin values. On the other hand, essential thrombocythemia or myelofibrosis patients with calreticulin mutations tend to have higher platelet counts, lower hemoglobin and leukocyte values, and fewer thromboses compared to patients with JAK2 mutations.^{275,282,283,284,285} The same may hold true for rare patients with familial essential thrombocythemia or myelofibrosis and somatically-acquired calreticulin mutations.²⁸²

Clinical and Laboratory Features

Pathologic bleeding occurs in approximately one-third of patients with myeloproliferative neoplasms and contributes to mortality in 10 percent of those affected patients. Thrombosis also occurs in one-third of patients with myeloproliferative disorders, contributing to mortality in 15 to 40 percent of affected patients.^228,232 Most symptomatic patients experience either bleeding or thrombosis; however, some develop both complications during the course of their disease. Bleeding usually involves the skin or mucous membranes, but may also occur after surgery or trauma. Thrombosis can involve arteries or veins and may occur in unusual locations such as abdominal wall vessels or the hepatic, portal, and mesenteric circulations.^{286,287,288,289,290,291} Indeed, full-blown or latent chronic myeloproliferative neoplasms account for a substantial proportion of patients with splanchnic vein thrombosis.^{286,291,292,293,294} Individuals with essential thrombocythemia may experience ischemia and necrosis of the fingers and toes from digital artery thrombosis, microvascular occlusion in the coronary circulation, or transient neurologic symptoms, including headaches,²⁹⁵ because of cerebrovascular occlusion.²⁹⁶ A syndrome of redness and burning pain in the extremities, termed erythromelalgia, is strongly associated with essential thrombocythemia and polycythemia vera and is thought to be partly caused by arteriolar platelet thrombi, although it may also have vasculopathic and neuropathic components.^297,298 It has been difficult to predict the risk of bleeding or thrombosis in an asymptomatic patient,²²⁹ but an increase in leukocyte count^221,232,235 or the number of reticulated platelets in patients with thrombocytosis, thought to reflect an increase in platelet turnover, has been associated with an increased risk for thrombosis.²⁹⁹ Vascular complications are also more likely to occur in patients older than 60 years of age and, most importantly, in patients with other cardiovascular risk factors, such as diabetes, hypertension, hyperlipidemia and obesity.^{221,300,301,302,303}

Therapy

Therapy should be risk-adapted and considered for symptomatic patients, for patients with a history of thrombosis or bleeding, for those with standard cardiovascular risk factors, for patients older than 60 years of age, and for individuals about to undergo surgery. Readers are referred to expert recommendations for a summary of the treatment of essential thrombocythemia and polycythemia vera, with particular relevance to risk factors for hemostasis and thrombosis (Chaps. 84 and 85).^{221,232,301,303,304,305,306} Treatment includes phlebotomy to correct the polycythemia and maintenance of a normal red cell mass, with the goal to achieve a hematocrit of less than 45 percent,^235,307,308 as well as therapy of the underlying disorder.^{228,232,309,310} Platelet count reduction to less than 400 × 10⁹/L in patients with thrombocytosis, either by plateletpheresis or cytoreductive agents, has been considered to be a target value associated with clinical improvement in patients with essential thrombocythemia.^228,302,311

Effective cytoreductive agents include the ribonuclease reductase inhibitor hydroxyurea,³¹² interferon-α (most recently the pegylated form of interferon alfa-2a), and anagrelide.^{301,311,313,314} In a prospective, randomized trial of 114 “high-risk” individuals with essential thrombocythemia who were either older than 60 years of age or had a previous history of thrombosis, hydroxyurea significantly reduced the incidence of new thrombosis from 24.0 to 3.6 percent.³¹² Anagrelide, an imidazoquinazoline derivative, is thought to decrease platelet counts by impairing megakaryocyte maturation.³¹⁵ Anagrelide has essentially no effect on red and white cell counts and is not known to be leukemogenic. Nevertheless, 10 to 20 percent of patients experience neurologic, gastrointestinal, and cardiac side effects, in particular fluid retention, often necessitating discontinuation of the drug.^314,316,317 When hydroxyurea and anagrelide were compared head-to-head in a randomized trial of 809 patients with essential thrombocythemia (all of whom were taking aspirin), subjects in the anagrelide group showed an increased rate of arterial thrombosis, major bleeding, and transformation to myelofibrosis relative to the group treated with hydroxyurea; however, the anagrelide group showed a relative decreased rate of venous thrombosis.³¹⁸ Progression to myelofibrosis despite treatment with anagrelide has also been observed in a phase II study.³¹⁹ However, in a newer, although relatively small, randomized, phase III study of 259 previously untreated high-risk patients with essential thrombocythemia, anagrelide was found to be noninferior to hydroxyurea in the prevention of arterial or venous thrombotic complications.³²⁰ It should be noted that this study used a long-lasting anagrelide drug that is not currently available in the United States. During an episode of acute bleeding in the chronic myeloproliferative neoplasms, DDAVP infusion may temporarily improve hemostasis if the patient has an acquired storage pool defect or acquired von Willebrand syndrome.^252,321 In the case of acquired von Willebrand syndrome, cytoreduction to reduce the platelet count may also ameliorate the process, although this may take time and require more temporizing interventions including DDAVP or factor VIII/VWF concentrates.²³⁴

Low-dose aspirin (~80 to 100 mg/day) may be useful in patients with essential thrombocythemia and thrombosis, particularly those with erythromelalgia or with digital or cerebrovascular ischemia.^{231,232,298,322} However, the evidence to date remains largely anecdotal, and aspirin can exacerbate a bleeding tendency in patients with myeloproliferative neoplasms, particularly in those individuals with acquired von Willebrand syndrome or with World Health Organization (WHO)-defined prefibrotic myelofibrosis masquerading as essential thrombocythemia.^{221,301,303,323} Consequently, even though a single, daily, low-dose aspirin is recommended for thromboprophylaxis in essential thrombocythemia, a risk-adapted approach is advised.^221,305 In addition, because platelet volume and turnover may be enhanced in essential thrombocythemia and polycythemia vera, the platelets of some individuals may not achieve total COX-1 inhibition with a single daily dose of aspirin. In such circumstances, 12-hour dosing may be considered, although this protocol has not been formally evaluated in a prospective clinical trial.^324,325

In a double-blind, placebo-controlled study of 518 patients with polycythemia vera who were judged to have no contraindications to daily low-dose (100 mg) aspirin, subjects in the aspirin arm exhibited a reduced risk of nonfatal arterial and venous cardiovascular end points. Although aspirin was well-tolerated, there was no effect of aspirin on overall and cardiovascular mortality.³²⁶ As has been noted,³⁰⁷ this study population was heavily pretreated to normalize the platelet count, although some individuals may have had residual elevations in red cell mass. Consequently, the safety and efficacy of aspirin as observed in this study may not be relevant to all patients with polycythemia vera.

Pregnant women with essential thrombocythemia or polycythemia vera pose special challenges because of an apparent increased risk of unsuccessful pregnancy, thrombotic or bleeding complications, and potential teratogenicity of hydroxyurea.^305,327 In essential thrombocythemia, the risk of first trimester miscarriages may be higher among women with the JAK2 (V617F) mutation.³²⁸ Although evidence-based recommendations are not available, Barbui and Finazzi recommend a risk-adapted approach to management in pregnancy. High-risk women are defined as those with previous major bleeding or thrombotic episodes, previous pregnancy complications, or a platelet count greater than 1500 × 10⁹/L.³²⁹ Low-risk individuals are recommended to be maintained at a hematocrit of less than 45 percent and to receive aspirin, 100 mg/day during pregnancy and low-molecular-weight heparin, 4000 U/day for 6 weeks after delivery. Interferon-α, rather than aspirin, is considered if there has been previous major bleeding or if platelets are greater than 1500 × 10⁹/L. High-risk patients are recommended to receive low-molecular-weight heparin throughout pregnancy.

LEUKEMIAS AND MYELODYSPLASTIC SYNDROMES

Clinical and Laboratory Features

The most frequent cause of bleeding in patients with leukemia or a myelodysplastic syndrome is thrombocytopenia. However, abnormal platelet function in vitro has been described in acute myelogenous leukemia, and in some patients this may be clinically significant. In acute myelogenous leukemia and its variants, platelets may be larger than normal, abnormally shaped, and exhibit a marked variation in the number of granules. There may be decreased aggregation and serotonin release in response to ADP, epinephrine, or collagen, decreased surface P-selectin expression in response to platelet activation via the PAR-1 thrombin receptor, and decreased platelet procoagulant activity. These functional abnormalities may be caused by either acquired storage pool deficiency or a defect in the process of platelet activation through one or more signaling pathways.^{330,331,332,333,334} These defects are intrinsic to the platelet and probably relate to the fact that the megakaryocytes from which platelets were derived originated from a leukemic stem cell. Indeed, in a familial platelet disorder with a predisposition to acute leukemia, platelet dysfunction prior to the development of leukemia occurs, at least in part, because of downregulation of genes such as NF-E2 or ALOX12, themselves target genes of RUNX1, a transcription factor that is germline-mutated in these individuals.^335,336

As discussed in the section on oncology drugs, drugs used to treat acute leukemias may affect platelet function, at least in vitro.^200,337,338 Bleeding in the acute leukemias usually responds to platelet transfusions and to treatment of the underlying disease. Similar in vitro platelet abnormalities may be seen in the myelodysplastic syndromes, sometimes accompanied by clinical bleeding disproportionate to that expected for the degree of thrombocytopenia.^{330,339,340,341,342,343,344} In these syndromes, platelets may be less uniformly affected; perhaps because there is a residual population of normal platelets admixed with those from the malignant clone.

Reduced platelet aggregation has been reported in children with acute lymphocytic leukemia.³³¹ Unless the leukemia is biphenotypic, it is difficult to ascribe the platelet defect to the leukemic process itself. Platelets are normal in children with acute lymphoblastic leukemia in complete remission.³⁴⁵ Single cases have been reported of patients with acute B-lymphoblastic leukemia³⁴⁶ or Hodgkin lymphoma³⁴⁷ whose severe bleeding was attributed, in part, to acquired Glanzmann thrombasthenia associated with antiintegrin α_IIbβ₃ antibodies. Hairy cell leukemia is a lymphoproliferative disease in which platelet dysfunction may rarely complicate the clinical picture; bleeding is usually due to thrombocytopenia rather than platelet dysfunction.³⁴⁸ Some patients may exhibit storage pool deficiency or a defect in the process of platelet activation. These abnormalities have been reported to disappear following splenectomy,³⁴⁹ which usually corrects the thrombocytopenia as well. Acquired von Willebrand syndrome has been reported in association with hairy cell leukemia.³⁵⁰

DYSPROTEINEMIAS

Definition and History

Platelet dysfunction is observed in approximately one-third of patients with Immunoglobulin (Ig) A multiple myeloma or Waldenström macroglobulinemia, 15 percent of patients with IgG myeloma, and in occasional patients with monoclonal gammopathy of undetermined significance (Chap. 106).^351,352 In addition to platelet dysfunction, other causes of bleeding should be considered in these patients, including the hyperviscosity syndrome,³⁵³ thrombocytopenia, complications of amyloidosis such as amyloid angiopathy³⁵⁴ or acquired factor X deficiency^355,356), and rarely, a circulating heparin-like anticoagulant^357,358,359 or systemic fibrino(gen)lysis.^360,361 The monoclonal immunoglobulin may also affect in vitro coagulation tests by interfering with fibrin polymerization and with the function of other coagulation proteins. On occasion, paraproteins can impair in vivo hemostasis as well.

Etiology and Pathogenesis

The bleeding time may be prolonged in patients with dysproteinemias, even in the absence of clinical bleeding. The platelet defect is caused by the monoclonal protein. It has been suggested that some monoclonal immunoglobulins interact with the platelet surface to interfere nonspecifically with platelet adhesion or stimulus–response coupling. This concept is supported by the observations that platelet dysfunction is more common when the concentration of the paraprotein in plasma or on the platelet membrane is very high³⁶²; that platelet aggregation, secretion, clot retraction, and platelet procoagulant activity may all be affected; and that normal platelets can acquire these defects when incubated with the purified monoclonal immunoglobulin.³⁶³

In some cases, specific interactions of the monoclonal protein with platelets or with components of the extracellular matrix have been described. One reported IgA myeloma protein inhibited the ability of a suspension of aortic connective tissue to aggregate normal platelets.³⁶⁴ The bleeding time and bleeding diathesis of the patient from whom this myeloma protein was obtained were corrected by removal of the protein by plasmapheresis. In another patient with IgDλ myeloma, λ dimers were found to bind to the A1 domain of VWF, inhibiting shear-induced platelet aggregation.³⁶⁵ In still another patient, an IgG myeloma protein bound specifically to the platelet integrin β₃ subunit. Both the intact immunoglobulin and its F(ab’)2 fragment inhibited the binding of fibrinogen to activated integrin α_IIbβ₃, thus inducing a thrombasthenic-like state.³⁶⁶ A number of patients with myeloma, monoclonal gammopathy of undetermined significance, lymphoma, or chronic lymphocytic leukemia have been reported to have an acquired form of von Willebrand disease in which the level of plasma VWF is reduced or the high-molecular-weight multimers of VWF are selectively reduced.^{321,352,367,368,369,370,371,372,373,374}

Therapy

When clinically significant platelet dysfunction occurs in a patient with a dysproteinemia, cytoreductive therapy should be considered as a means to reduce the production and plasma level of the monoclonal immunoglobulin.^351,352 Plasmapheresis can also control bleeding by reducing the level of the abnormal protein and can be lifesaving during acute bleeds.^352,375,376 Cryoprecipitate, DDAVP and/or plasmapheresis may be transiently effective in patients with acquired von Willebrand syndrome.^{321,368,377,378} However, high-dose intravenous gamma globulin (IVIG) appears to be particularly effective in individuals with acquired von Willebrand syndrome associated with an IgG monoclonal gammopathy of undetermined significance, although intermittent infusions may be required at approximately 3-week intervals (Chap. 126).^{352,369,370,371,379,380,381} The reported experience with rituximab for the latter condition is extremely limited, but so far disappointing.³⁸²

ACQUIRED VON WILLEBRAND SYNDROME

Acquired von Willebrand syndrome is a relatively rare disorder that typically occurs in the setting of an autoimmune or clonal hematologic disease.^{231,381,383,384,385} It is being increasingly recognized in conditions associated with high shear and turbulence in the circulation, such as severe aortic stenosis, hypertrophic obstructive cardiomyopathy and circulatory assist devices.^{381,386,387,388,389,390} It also can occur in association with a number of other unrelated medical conditions,³⁸¹ including Gaucher disease,³⁹¹ hypothyroidism^392,393 and Noonan syndrome.³⁹⁴ As discussed above, it can represent one cause of bleeding in multiple myeloma,^321,377 Waldenström macroglobulinemia,³⁹⁵ monoclonal gammopathy of undetermined significance,³⁷⁰ low grade non-Hodgkin lymphoma,^396,397 chronic lymphocytic leukemia,³⁹⁸ and chronic myeloproliferative neoplasms, the latter particularly in association with very high platelet counts.²³³

The pathophysiology of acquired von Willebrand syndrome involves a reduction in circulating VWF (and its associated factor VIII molecule), generally because of rapid VWF turnover in the circulation.³⁸¹ VWF levels and multimer patterns may simulate type I, II or III von Willebrand disease. In lymphoproliferative disorders, a specific, often nonneutralizing anti-VWF antibody is present,^{321,352,377,399} whereas in autoimmune disorders, anti-VWF antibodies are part of a generalized autoimmune response.⁴⁰⁰ In other situations, the syndrome may result from increased adsorption of VWF by tumor cells (e.g., Wilms tumor, osteosarcoma⁴⁰¹) or platelets (myeloproliferative neoplasms),^{315,350,402,403,404} increased VWF proteolysis (e.g., aortic stenosis, ventricular assist devices), or decreased VWF production (hypothyroidism).^381,405,406

Mucocutaneous bleeding should raise the suspicion of acquired von Willebrand syndrome in patients without a prior personal or family history of bleeding. This is especially important in patients with a known autoimmune, lymphoproliferative, or myeloproliferative disorder.^315,381 Diagnostic evaluation includes measurements of factor VIII activity, VWF antigen, ristocetin cofactor activity, and VWF multimer analysis.⁴⁰⁷ The presence of an in vitro inhibitor may, or may not, be detected depending on whether the antibody binds to VWF and neutralizes its function or merely leads to accelerated VWF clearance by the reticuloendothelial system.³¹⁵ An abnormally high ratio of VWF propeptide to von Willebrand antigen may be present as a result of the rapid clearance of von Willebrand antigen but not VWF.⁴⁴

Given the uncommon prevalence of this syndrome, reports of patient management have been retrospective and largely anecdotal. Treatment should be reserved for patients with active bleeding or those who are likely to bleed if left untreated.^352,381 Infusions of DDAVP^321,398,400 or factor VIII/VWF concentrates^408,409 may be useful, although the rapid clearance of VWF may limit efficacy. Treatment has included glucocorticoids or rituximab in patients with lupus,^231,385,410 and recombinant factor VIIa,⁴¹¹ or high-dose IVIG.^190,384,412 High-dose IVIG is particularly effective when acquired von Willebrand syndrome is associated with a lymphoproliferative disorder or, as discussed in the section “Dysproteinemias.” with an IgG monoclonal gammopathy of undetermined significance. IVIG likely acts by delaying VWF clearance via reticuloendothelial cell blockade, although other mechanisms have been postulated.^{297,369,370,379,380,397,413} Treatment of the underlying disease can be effective in some situations^381,414,415 (e.g., hypothyroidism with thyroid replacement,^416,417 Gaucher disease with enzyme replacement therapy,³⁹¹ and extreme thrombocytosis with cytoreduction^{233,315,323,418}). As with inherited von Willebrand disease, longstanding acquired von Willebrand syndrome can be associated with and complicated by gastrointestinal tract arteriovenous malformations, resulting in severe bleeding.⁴¹⁹ An example of such gastrointestinal bleeding is found in patients with severe aortic stenosis and referred to as Heyde syndrome. In this situation, valve replacement can correct the hemostatic defect.^387,420

UREMIA

Definition and History

In the predialysis era, hemorrhage occurred in approximately 50 percent of uremic patients and was a cause of death in approximately 30 perccent.^421,422 With the advent of dialysis, the frequency of spontaneous hemorrhage in patients with renal failure has decreased.⁴²² Experience with percutaneous renal biopsy in several thousand patients with renal disease supports the notion that the hemostatic defect in patients with renal disease is usually mild. Although the incidence of small perirenal hematomas following biopsy may be as high as 85 percent when patients are examined by computerized tomography, gross hematuria is observed in only 5 to 10 percent of cases and is usually transient.^423,424 Severe bleeding following biopsy requiring surgical intervention is even less common and usually can be attributed to factors other than a uremic hemostatic defect, such as needle lacerations of the kidney or spleen, anomalous vessels, heparin anticoagulation, or the presence of amyloid in the kidney.

Etiology and Pathogenesis

The hemostatic defect in uremia has been attributed to defects in platelet function and appears to be multifactorial.⁴²⁵ One prominent factor is renal failure-associated anemia.⁴²⁶ A lowered hematocrit ex vivo induces a defect in platelet adhesion that can be corrected by increasing the hematocrit to 30 percent or more.⁴²⁷ In uremic patients, successful treatment of anemia with red blood cell transfusion or recombinant human erythropoietin (EPO) results in partial or complete correction of prolonged bleeding times when the hematocrit is increased to 27 to 32 percent.^{428,429,430,431} The effect of anemia on primary hemostasis is not unique to uremia. In normal individuals, the bleeding time correlates with the hematocrit and bleeding times can be prolonged in patients with severe anemia of any etiology.⁴²⁶ Red cells may have a beneficial effect on hemostasis both because they displace platelets toward the periphery of the column of circulating blood⁴³² and they may enhance platelet reactivity.¹⁴

Because correction of anemia does not always return the bleeding time to normal, other factors present in renal failure may perturb platelet function.⁴²⁷ Ristocetin-induced platelet aggregation, a surrogate for VWF binding to the platelet GPIb–IX–V complex, may be decreased in uremia. However, plasma VWF concentrations are normal or elevated in renal failure⁴³³ and qualitative VWF abnormalities have not been uniformly observed.^434,435 Mixing studies using uremic platelets and normal plasma, and vice versa, do not demonstrate consistent quantitative or qualitative abnormalities in GPIb–IX–V.^434,435,436 Nonetheless, uremic plasma can inhibit the adhesion of normal platelets to deendothelialized human umbilical artery segments, whereas uremic platelets adhere normally in the presence of normal plasma.⁴³⁴ Because the defective adhesion appears independent of VWF, an unidentified component of uremic plasma may be responsible for the adhesion defect.⁴³⁴ Uremic platelets also exhibit markedly reduced spreading on the subendothelium of rabbit vessels, a defect attributed to impaired VWF binding to platelet integrin α_IIbβ₃.⁴³⁷ Because VWF binding to integrin α_IIbβ₃ requires platelet stimulation, this observation suggests a uremia-induced defect in platelet signal transduction.

There are a number of reports describing defective agonist-induced platelet activation in uremic patients, including reduced fibrinogen binding, aggregation, and secretion. These abnormalities may be retained after platelets are separated from uremic plasma and in some cases, uremic plasma imparts the defect to normal platelets.⁴³⁸ Furthermore, the ability of activated platelets to express procoagulant activity is reduced in uremia.⁴³⁹ These functional defects likely result from uremia-induced abnormalities in platelet biochemistry, including reduced agonist-induced increases in cytoplasmic free calcium,⁴⁴⁰ reduced release of arachidonic acid from platelet phospholipids,⁴²¹ and reduced conversion of released arachidonic acid to PG endoperoxides and TXA₂.^138,441,442

A number of dialyzable and nondialyzable substances have been reported to be responsible for the platelet function defects in uremia,⁴⁴³ but urea itself is not responsible. Ex vivo platelet aggregation can be inhibited by small dialyzable substances, such as guanidinosuccinic acid and phenolic acids, as well as by poorly characterized “middle molecules” at concentrations found in uremic plasma.^444,445 Venous and arterial segments from uremic patients have been reported to produce more PGI₂ than segments from normal individuals, an abnormality not corrected by dialysis.⁴⁴⁶ Altered NO metabolism has been observed in uremia.^447,448 In a uremic rat model, defective platelet adhesion was normalized by an inhibitor of NO formation,⁴⁴⁹ suggesting that increased NO synthesis by endothelial cells or platelets is at least partially responsible for the defective platelet function.⁴⁵⁰ Why renal failure increases NO synthesis is not entirely clear, although exposing endothelial cells to guanidinosuccinic acid can mimic the effects of NO, suggesting that retained guanidinosuccinic acid may be the relevant substrate.⁴⁵¹ Uremia has been reported to upregulate the y⁺L system for L-arginine transport into platelets, enabling platelets to maintain or enhance NO synthesis, even in the face of low circulating L-arginine concentrations.^452,453 By contrast, some substances found in high concentrations in uremic plasma, such as urea and parathyroid hormone, appear to play no role in platelet dysfunction.⁴⁵⁴

Concurrent medications and thrombocytopenia must always be considered when a patient with renal failure exhibits a bleeding tendency. Aspirin can prolong the bleeding time inordinately in uremia. Unlike aspirin’s effect on COX, this effect is transient and correlates with blood levels of aspirin.^34,35 Bleeding may be potentiated by the administration of heparin during hemodialysis; in this situation, the use of an ethylene-vinyl alcohol copolymer hollow fiber dialyzer or intermittent saline infusion and high blood flow rates may eliminate the need for heparin.⁴⁵⁵ β-Lactam antibiotics that prolong the bleeding time may have a greater effect in uremic patients and increase the occurrence of bleeding.⁴⁵⁶

Mild thrombocytopenia has been reported in chronic renal failure, particularly in patients on dialysis,⁴⁵⁷ as a result of diminished marrow production and decreased platelet survival.⁴⁵⁸ Serum thrombopoietin levels in hemodialysis patients are increased,^457,459 perhaps reflecting increased platelet turnover or a decrease in megakaryocyte mass. But when platelet counts are greater than 100 × 109/L, it is necessary to consider whether a systemic disease or medication, such as multiple myeloma, systemic vasculitis, hemolytic uremic syndrome, eclampsia, renal allograft rejection, or heparin, could be responsible for bleeding in a uremic patient.

Clinical and Laboratory Features

Despite dialysis, abnormal platelet function in uremia remains a clinical issue because it may contribute to bleeding following surgery or trauma or in conjunction with anatomic lesions of the gastrointestinal tract.^441,455 The bleeding time has often been used as an indication of hemorrhagic risk in uremia, but critical reviews of the literature indicate that it is not appropriate to use for this purpose.^460,461

Therapy

Abnormal platelet aggregation is common in uremic patients, but by itself is not an indication for therapeutic intervention.⁴²⁵ The frequency of excessive bleeding after biopsies or other surgical procedures in uremic patients who have not received specific treatment is not known, but may be uncommon. Thus, if bleeding does complicate a procedure, a thorough search for causes of bleeding other than uremia should be initiated without assuming that uremia is the etiology. However, when therapy for a uremic bleeding diathesis is necessary, the uremic platelet defect can usually be successfully treated.

There are several therapeutic maneuvers that can either partially or completely correct an abnormal bleeding time in uremic patients and anecdotal observations indicate that they may also improve hemostasis. Because prospective studies comparing various treatment regimens have not been performed, the choice of therapy should be based on the severity of the bleeding, the anticipated severity of the hemostatic stress imposed by surgery or trauma, the predicted duration of the therapeutic effect, and the risks of therapy.

The mainstay of therapy is dialysis. Intensive dialysis can correct the bleeding diathesis in many patients, but is only partially effective in others.⁴⁶² Peritoneal dialysis and hemodialysis are equally effective.^462,463 If a patient undergoing dialysis bleeds, it may be worthwhile to increase the intensity of the dialysis.

In uremic individuals, increasing the hematocrit by transfusion or treatment with recombinant human EPO to 27 to 32 percent is often associated with diminished clinical bleeding.^{428,429,430,464,465} A number of reports suggest that EPO has an effect on platelets independent of an increase in hematocrit,⁴³¹ perhaps the result of an increase in the number of young platelets in the circulation.⁴⁶⁶

DDAVP, a vasopressin analogue whose pressor effects are substantially less than its antidiuretic effects, causes the release of VWF from tissue stores, has been reported to shorten the bleeding time in 50 to 75 percent of patients with uremia. In many cases, surgery has been carried out safely after administration of this drug, although no controlled trials have been performed.⁴⁶⁷ DDAVP is usually administered intravenously in a dose of 0.3 mcg/kg over 15 to 30 minutes (maximum dose: 20 mcg) but it is also effective at this dose when given subcutaneously.⁴⁶⁷ Alternatively, the drug can be given intranasally.⁴⁶⁸ Improvement in the bleeding time is seen within 30 to 60 minutes of administration, lasts for approximately 4 hours, and roughly correlates with the rise in the plasma levels of VWF and the appearance in the circulation of high-molecular-weight VWF multimers.⁴⁶⁷ In some patients, the drug has been given repeatedly at 12- to 24-hour intervals, although tachyphylaxis can occur.⁴⁶⁹

Side effects of DDAVP have been mild and uncommon and have included a 10 to 15 percent decrease in mean arterial pressure, a 20 to 30 percent increase in pulse rate, facial flushing, water retention, and hyponatremia leading to seizures; the latter is more common after repeated administration and when fluids are given freely.⁴⁶⁷ Water retention and hyponatremia have not been observed in patients whose kidneys cannot respond to the hormone. Several uremic and nonuremic individuals with atherosclerosis have been reported to develop stroke or myocardial infarction after DDAVP administration, although such complications appear to be rare.^470,471 If dialysis is not effective, DDAVP is the treatment of choice for uremic bleeding, particularly if only a short-term effect is required.⁴⁶⁷

Conjugated estrogens at a dose of 0.6 mg/kg intravenously for 5 days have also been reported to shorten the bleeding time in most, but not all, uremic individuals, both in uncontrolled studies and in randomized, double-blind studies.^{34,472,473,474} They may also be useful in some patients with uremia who bleed from gastrointestinal telangiectasia.⁴⁷⁵ No changes in the plasma levels or multimer distribution of VWF have been noted with this treatment and it has been postulated that the active component in conjugated estrogens, 17β-estradiol, acts through an estrogen receptor mechanism.⁴⁷⁶

Lastly, uncontrolled studies suggest that infusions of cryoprecipitate can shorten the bleeding time in uremic patients and ameliorate bleeding.²⁴³ However, others have reported inconsistent results,⁴⁷⁷ and because of concerns of viral contamination, cryoprecipitate is very rarely used for this indication.

ANTIPLATELET ANTIBODIES

Definition and History

Antibody binding to platelets in several pathologic conditions, including immune thrombocytopenia (ITP), systemic lupus erythematosus (SLE), and platelet alloimmunization can cause thrombocytopenia as a result of decreased platelet survival. Less commonly, bleeding times may be shorter than expected for the degree of thrombocytopenia, suggesting enhanced platelet function.⁴⁷⁸ On occasion, platelet function is impaired in ITP.^{479,480,481,482,483}

Etiology and Pathogenesis

The mechanism by which autoantibodies or alloantibodies impair platelet function is likely antibody binding to specific platelet GPs. Most antiplatelet antibodies are directed against integrin α_IIbβ₃,^{479,480,481,482} but antibodies directed against GPIb–IX–V, integrin α₂β₁, and GPIV have been detected as well.^484,485 In most instances, the functional consequences of antibody binding are obscured by the presence of thrombocytopenia. However, patients have been reported with normal platelet counts, absent platelet aggregation, autoantibodies against integrin α_IIbβ₃, and a bleeding diathesis reminiscent of Glanzmann thrombasthenia.^{479,480,481,482,486,487,488,489} Similarly, autoantibodies against GPIb and integrin α₂β₁ have been detected that selectively inhibit ristocetin-induced platelet aggregation^490,491 and collagen-induced platelet aggregation,^492,493 respectively. A patient with ITP has also been identified whose anti-GPVI autoantibody produced GPVI shedding from the platelet surface and platelets unresponsive to stimulation by collagen.⁴⁹⁴

Besides interfering with platelet function, some autoantibodies can activate platelets and induce aggregation and secretion in vitro. Such antibodies can activate platelets through immune complex binding to platelet Fc receptors, by depositing sublytic quantities of the membrane attack complex of complement (C5b-9) on the cell surface,⁴⁹⁵ or by binding to a specific membrane antigen.²⁴⁶ The prototypic example of this phenomenon is heparin-induced thrombocytopenia in which antibodies bound to neoepitopes exposed on the platelet factor 4 molecule by heparin activate platelets after binding to platelet Fc receptors (Chap. 118).⁴⁹⁶

Clinical Laboratory Features and Therapy

Platelet dysfunction should be suspected in any patient with ITP or SLE who has mucocutaneous bleeding with a platelet count that is not ordinarily associated with bleeding (e.g., equal to or greater than approximately 30 × 10⁹/L). Likewise, this scenario has been described occasionally in patients with Hodgkin disease,^347,480 non-Hodgkin lymphoma and myeloma,^497,498 and hairy cell leukemia.⁴⁹⁹ The clinical spectrum of autoimmune platelet dysfunction may also include some individuals with “easy bruising” and a normal platelet count. These patients may have ITP with “compensated thrombocytolysis,” as a substantial proportion have circulating antiplatelet antibodies and large platelets.⁵⁰⁰

Patients with antiplatelet antibodies may exhibit defective platelet function in vitro, even if they do not manifest a prolonged bleeding time or excessive bleeding. These deficits include impaired platelet aggregation to ADP, epinephrine, and collagen,^{501,502,503,504} as well as impaired adhesion to the subendothelial matrix.²⁰ The most frequently reported abnormalities are absence of platelet aggregation in response to low concentrations of collagen and absence of the second wave of aggregation in response to ADP or epinephrine. This pattern is identical to that seen in individuals with inherited storage pool disease. In fact, both ITP and SLE may be associated with an acquired form of storage pool disease manifested by a reduced platelet content of dense- and α-granule components.^432,505 In one report, platelets in ITP also exhibited an activation defect manifested by impaired conversion of arachidonic acid to TXA₂.⁵⁰⁶

Because antibody-mediated platelet dysfunction and bleeding almost always occur in the setting of ITP, therapeutic efforts should be directed to the treatment of these disorders.

CARDIOPULMONARY BYPASS

Definition and History

Circulating blood through an extracorporeal bypass circuit during cardiac surgery induces a variety of hemostatic defects. The most significant of these are thrombocytopenia, platelet dysfunction, and hyperfibrinolysis.^507,508,509 At their extreme, these defects can result in substantial postoperative bleeding that may last hours to days after bypass. Approximately 5 percent of patients experience excessive postoperative bleeding after extracorporeal bypass; roughly half of the bleeding is from surgical causes; much of the remainder is caused by qualitative platelet defects and hyperfibrinolysis.

Etiology and Pathogenesis

Thrombocytopenia is a consistent feature of bypass surgery.^126,508 Typically, platelet counts begin to decrease to approximately 50 percent of presurgical levels within the first half hour after the initiation of bypass, but thrombocytopenia can occur within 5 minutes and often does not nadir for the first few days.^507,509,510 The major factor responsible for thrombocytopenia is hemodilution from priming the pump with colloid or crystalloid solutions, but it is often more profound than can be accounted for by hemodilution alone.^509,510,511 Platelet adhesion to artificial surfaces in the circuit has been demonstrated by scanning electron microscopy.⁵¹² The mechanism of this interaction is uncertain, but it may be a result of the deposition of fibrinogen on the bypass circuit and platelet adhesion mediated by integrin α_IIbβ₃.⁵¹³ Less-common causes of thrombocytopenia during bypass are disseminated intravascular coagulation, sequestration of damaged platelets in the liver, and heparin-induced thrombocytopenia.⁵¹⁴ Like antibodies against the complex of platelet factor 4 and heparin that are commonly detected after bypass surgery and can be responsible for heparin-induced thrombocytopenia,⁴⁹⁶ antibodies against protamine and protamine/heparin complexes are commonly detected as well.^82,515,516 Such antibodies may contribute to the thrombocytopenia and possibly to thromboembolic events following cardiopulmonary bypass.⁵¹⁷

Qualitative platelet defects are the primary nonstructural hemostatic defects induced by the bypass circuit^508,518 and are manifest as abnormal ex vivo platelet aggregation, decreased ristocetin-induced platelet agglutination, deficiency of platelet α and δ granules, release of soluble CD40 ligand, and the generation of platelet microparticles.^{507,509,510,519,520,521,522} The severity of these abnormalities correlates with the duration of extracorporeal bypass⁵²³ and they generally resolve within 2 to 24 hours.⁵⁰⁸

Bypass-induced defects in platelet function are likely caused by platelet activation and fragmentation,^521,524 hypothermia, contact with fibrinogen-coated synthetic surfaces, contact with the blood–air interface, cardiotomy suction and retransfusion of cardiotomy suction blood, and platelet exposure thrombin, plasmin, ADP, or complement.^{513,519,525,526,527,528} Drugs such as heparin, protamine, integrin α_IIbβ₃ antagonists, and aspirin, as well as the production of fibrin degradation products, can also impair platelet function.^{126,529,530,531} Controversy exists about the significance of these defects in vivo. Some investigators suggest that the entire qualitative platelet defect is a result of the use of heparin during bypass and its inhibitory effect on thrombin activity⁵²⁹; however, this would not account for the bleeding diathesis that can exist hours after heparin reversal.

Hyperfibrinolysis may also contribute to the bleeding diathesis associated with cardiopulmonary bypass.^532,533 This is likely from thrombus formation in the pericardial cavity followed by local, and subsequently systemic, fibrinolysis.⁵³² The relevance of hyperfibrinolysis to postbypass bleeding is bolstered by the efficacy of antifibrinolytic therapy in minimizing cardiopulmonary bypass surgery blood loss.

Therapy

A preoperative evaluation of cardiac surgical candidates should include a history of bleeding in either the patient or family member. Some authors recommend a screening prothrombin time, partial prothrombin time, and bleeding time even in individuals with no history of bleeding.⁵³⁴ However, the validity of this approach is controversial.⁵³⁵ Regardless, prophylactic transfusion of allogeneic blood components is not indicated.^508,536,537 Preoperative administration of recombinant human EPO has been reported to reduce the need for allogeneic blood transfusion in undergoing elective open-heart surgery.^{538,539,540,541} Cell savers are now often used during bypass surgery and the collected washed autologous red blood cells are reinfused after completion of cardiopulmonary bypass. In addition, blood collected from chest tube drainage has been reinfused to minimize allogeneic transfusions.⁵⁴¹ The safety of transfusing large quantities of blood by this technique has not fully been established.^528,542

A number of maneuvers have been taken to reduce the hemostatic abnormalities associated with cardiac surgery. These include coating the artificial surfaces of cardiopulmonary bypass devices with heparin,^{543,544,545,546,547} using centrifugal rather than roller pumps,⁵⁴⁸ use of a number of pharmacologic agents,⁵⁴⁹ and performing coronary artery surgery without bypass.^301,550 Off-pump coronary artery bypass surgery appears to preserve platelet function, but concerns have been raised about adverse thromboembolic events postsurgery because of the concurrence of normal platelet function, late thrombin generation, and reduced fibrinolysis.^551,552,553 Several pharmacologic maneuvers have been tried to assist in the management of postoperative bleeding. Postoperative patients with a prolonged bleeding time and excessive blood loss may respond to DDAVP, as evidenced by a shortening of the bleeding time. However, results of trials using this agent have been contradictory, some studies showing a reduced blood loss and most showing no benefit.^554,555,556 Based on the assumption that platelet activation during bypass could be a major cause of postoperative platelet dysfunction, infusion of platelet activation inhibitors such as PGE₁, PGI₂, or stable PGI₂ analogues have been carried out in animal models and in humans. By increasing platelet cAMP and reducing platelet responsiveness, these agents prevent bypass-induced thrombocytopenia and platelet dysfunction. However, randomized trials using PGI₂ and its analogue, iloprost, have not shown a clear overall benefit, in part because of significant toxicity, including hypotension.^123,557 Recombinant factor VIIa has been recommended to treat uncontrolled postoperative bleeding that has not responded to routine hemostatic therapy.⁵⁵⁸ However, the off-label use of recombinant factor VIIa is associated with an increased risk for arterial and venous thromboembolism⁵⁵⁹ and a retrospective case-matched review of patients who had received recombinant factor VIIa perioperatively during major cardiac surgery indicated that it was associated with worse survival.²²³ Nonetheless, it remains a potentially useful therapeutic consideration in view of the prognosis of uncontrolled postoperative hemorrhage.⁵⁶⁰ Based primarily on the cardiovascular complications encountered by patients in two randomized studies of the use of parecoxib/valdecoxib to treat pain after cardiac surgery, the use of coxibs as well as traditional NSAIDs appears contraindicated in this setting.^561,562

Inhibiting fibrinolysis using ε-aminocaproic acid or tranexamic acid during cardiopulmonary bypass can reduce mediastinal blood loss and transfusion requirements.⁵⁴⁹ Aprotinin (Trasylol), a broad-spectrum protease inhibitor, was also used for this purpose, but observational studies,^563,564,565 as well as a blinded clinical trial,⁵⁶⁶ revealed that its use is associated with serious end-organ damage and a higher mortality than the use of ε-aminocaproic acid or no antifibrinolytic agent.

The most important determinant of blood loss following cardiopulmonary surgery is the surgical procedure itself. If excessive nonsurgical postoperative bleeding occurs, one should verify that the patient is no longer hypothermic and that heparin has been fully reversed. At this point, the administration of pharmacologic agents, along with judicious transfusions of platelets, cryoprecipitate, fresh frozen plasma and red blood cells is appropriate.

MISCELLANEOUS DISORDERS

Measurements of hemostatic function are frequently abnormal in patients with end-stage and fulminant liver disease and result from decreased coagulation factor production, fibrinolysis, dysfibrinogenemia, thrombocytopenia as a result of hypersplenism and thrombopoietin deficiency, as well as disseminated intravascular coagulation (DIC).^425,567 However, the clinical consequences of these laboratory abnormalities have been reassessed because they do not take into account that both anti- and prohemostatic pathways are perturbed in liver disease.^568,569 Thus, hemostasis in liver disease is considered to be “rebalanced,” although it remains unstable, with patients prone to both bleeding and thrombosis. Chronic liver disease can be associated with a prolonged bleeding time and reduced platelet aggregation and procoagulant activity,^567,570,571 but there is no evidence for a platelet function defect specific to liver disease.⁵⁷² Rather, they are the result of multiple factors, including thrombocytopenia, hypofibrinogenemia, and anemia, none of which imply an intrinsic defect in platelet function.⁵⁷³ Regardless, the prolonged bleeding in these patients may respond to infusion of DDAVP,⁵⁷⁴ but clinical relevance of this observation is uncertain.⁵⁷⁵

Patients with DIC may exhibit reduced platelet aggregation and acquired storage pool deficiency.^576,577 These result from platelet activation in vivo by thrombin or other agonists. Alternatively, elevated levels of fibrin(ogen) degradation products and the low fibrinogen levels that accompany DIC may contribute to the platelet defect. Although purified low-molecular-weight fibrinogen-degradation products can impair platelet aggregation, this effect requires concentrations of degradation products unlikely to occur in vivo.⁵⁷⁸ Moreover, it is difficult to assess the significance of platelet dysfunction in most patients with DIC because of the simultaneous presence of thrombocytopenia and other hemostatic defects.

Decreased platelet aggregation and secretion in response to ADP and epinephrine has been reported in Bartter syndrome, a group of rare inherited disorders characterized by severe restrictions of salt reabsorption by the thick ascending limb of Henle, perhaps caused by excessive PGE₂ synthesis.^{579,580,581,582} However, reviews of series of patients with Bartter syndrome make no mention of hemostatic problems⁵⁸⁰ so that the clinical significance of the platelet aggregation abnormalities is doubtful.

In addition to thrombocytopenia, platelet dysfunction has been observed in some patients with hemorrhagic fevers caused by Dengue, Hanta, Lassa, Junín, and Ebola viruses.⁵⁸³ There are also isolated reports of a slight prolongation of the bleeding time and/or ex vivo platelet function defects in a number of other clinical conditions. These include nonthrombocytopenic purpura with eosinophilia,^584,585,586 atopic asthma and hay fever,⁵⁸⁷ acute respiratory failure,⁵⁸⁸ and Wilms tumor elaborating hyaluronic acid.⁵⁸⁹ The clinical significance of these associations is not clear.