Chapter 118: Heparin-Induced Thrombocytopenia

Heparin-induced thrombocytopenia (HIT) is a complication of heparin therapy in which there is a fall in platelet count and an unusually high incidence of arterial and/or venous thromboembolic complications in association with heparin therapy.

Although clinical usage of heparin as an anticoagulant began in the late 1950s, it was not until the early 1970s that a small percentage of treated patients were noted to develop a complication consisting of thrombocytopenia with paradoxical, life-threatening thromboemboli (for a historical review see Ref. 1). In the 1980s, it became clear that HIT was caused by immunoglobulin (Ig) G antibodies that activate platelets. It was also recognized that HIT could be divided into two types, the classic immune-mediated prothrombotic disease that is the focus of this chapter (formerly called HIT type II), and a benign nonimmune condition associated with a mild, immediate, and transient drop in platelet count and no increased risk of thrombosis (formerly called HIT type I).² In this chapter, “HIT” means the immune-mediated form of the disease.

In the 1970s and 1980s, it became clear that HIT antibodies activated both platelets and endothelial cells.^3,4 Further analysis showed that blocking platelet FcγRIIA inhibited platelet activation by HIT sera in vitro,⁵ suggesting that platelet activation involved an immune complex. In the early 1990s, this complex was identified as heparin bound to the platelet-specific chemokine, platelet factor 4 (PF4).⁶ Over the past 20 years, additional insights into the mechanism(s) underlying this immune complex disorder have emerged that have advanced our understanding of why this disorder is particularly prothrombotic and occurs in a only a small subset of patients. Additionally, advances have been made on the clinical side with respect to prevention, diagnosis, and treatment.

The frequency of HIT in heparin-treated patients ranges from less than 0.1 percent to 5.0 percent, depending on patient- and heparin-specific risk factors. These include the patient population, gender, nature of the heparin used, and duration of heparin exposure (Table 118–1).

The most important determinant of risk is the patient population. In a meta-analysis of seven prospective studies, the incidence of HIT was greater among surgical than medical patients (odds ratio [OR]: 3.25; 95 percent confidence interval [CI]: 1.98 to 5.35).⁷ The incidence of HIT approaches 5 percent in patients who receive unfractionated heparin (UFH) after major orthopedic surgery.⁸ Patients undergoing surgery with cardiopulmonary bypass have a very high frequency of anti-PF4/heparin antibody seroconversion (50 to 75 percent by postoperative day 10), but a lesser incidence of HIT (0.5 to 1.0 percent).^8,9,10 HIT occurs in 0.5 to 1.0 percent of medical patients⁷ and in less than 0.1 percent of pregnant women^11,12 and children.¹³ In a randomized trial of trauma patients, major trauma was associated with a significantly greater incidence of HIT than minor trauma (2.2 percent vs. 0.0 percent, p = 0.01) despite identical heparin exposure.¹⁴

Female sex is also a risk factor for HIT. A meta-analysis found an approximately twofold greater risk of HIT in women than men (OR: 2.37; 95 percent CI: 1.37 to 4.09). Analyses of a German database and a randomized trial of UFH versus low-molecular-weight heparin (LMWH) after orthopedic surgery yielded similar findings.⁷

HIT is more common with UFH than LMWH in surgical patients. In a meta-analysis of 15 studies, primarily involving orthopedic surgery patients, the incidence of HIT with UFH and LMWH was 2.6 percent and 0.2 percent, respectively.¹⁵ Data are conflicting on whether the risk of HIT is reduced with LMWH in medical patients.^7,16,17 Fondaparinux, a synthetic pentasaccharide anticoagulant, is associated with a nearly negligible risk of HIT, although several cases of fondaparinux-associated HIT have been reported.¹⁸

Duration of heparin exposure also influences the risk of HIT. In a meta-analysis of 3529 patients receiving UFH thromboprophylaxis for 6 or more days, the incidence of HIT was 2.6 percent.¹⁵ Review of a hospital database indicated that briefer courses induce a substantially lower incidence of HIT (0.2 percent).¹⁹

High-quality data on the impact of dose and route of administration of heparin on the risk of HIT are lacking. Some studies suggest a lower rate of HIT with prophylactic dose subcutaneous UFH than therapeutic dose intravenous UFH,¹⁹ but these analyses are confounded by differences in the patients that receive these treatments including the clinical indication for heparin. Rarely, HIT has been reported with very low doses of heparin such as with use of heparin flushes or heparin-bonded catheters.^20,21

The development of HIT antibodies is nonclassical in that these antibodies typically begin as IgG and not IgM,²² may disappear after a few months, and may not reappear with heparin reexposure.²³ It has been proposed that the initial antigen exposure involves PF4 complexed with bacterial wall (Fig. 118–1) and that antibodies against this complex may be an important antimicrobial defensive mechanism.²⁴ These antigenic complexes, as well as circulating PF4–heparin complexes, are detected by splenic marginal B cells that subsequently produce pathogenic HIT antibodies.²⁵

The nature of the antigenic heparin–PF4 complex has been partially defined. At the concentrations reached at sites of injury, PF4 exists as a tetramer. Crystal structure analysis shows that this tetramer is encircled by a ring of positive charge (Fig. 118–2),²⁶ and heparin is thought to bind to this region.²⁷ There are two closely spaced HIT antibody recognition domains (Fig. 118–2).²⁸ These domains are distinct from the heparin-binding domain. About half of patients have antibodies that react with one or the other HIT antigenic domain and one-third of patients do not have antibodies that react to either domain, suggesting that there are other HIT antigenic sites on PF4. Studies of antihuman PF4 monoclonal antibodies suggest that unlike nonpathogenic anti-PF4 antibodies, HIT antibodies markedly increase their binding affinity for PF4 maintained in a tetrameric as opposed to dimeric state.²⁹ Moreover, tetrameric PF4 and UFH need to be at approximately 1:1 molar ratio for optimal HIT antigenicity.^30,31 At this ratio, ultralarge complexes (>670 kDa) of PF4 and heparin form that appear as visible colloidal complexes, and these are likely to be the antigenic source in HIT.^32,33 At higher or lower ratios, PF4 predominantly forms smaller and less antigenic PF4–heparin complexes. Ultralarge complexes are inefficiently formed with LMWH, offering a potential explanation as to why this class of agents is associated with a lower incidence of HIT compared with UFH. Fondaparinux does not form large complexes with PF4, explaining the negligible incidence of HIT with this agent and supporting its potential utility in the prevention or treatment of HIT.

A passive immunization murine model of HIT has been used to demonstrate the following components are necessary to induce both thrombocytopenia and a prothrombotic state in HIT: the presence of human PF4 in platelets, FcγRIIA on the surface of platelets and possibly other vascular cells, and the presence of a pathogenic HIT-like antibody.³⁴ Heparin has a more complex relationship to the pathogenesis of HIT (Fig. 118–3). In the passive immunization model of HIT, wherein a pathogenic HIT-like antibody is infused into mice expressing both human PF4 and FcγRIIA, heparin infusion is not needed to cause thrombocytopenia and a prothrombotic state.³⁵ The explanation for a lack of need to infuse heparin in this model is as follows: One important potential role of heparin in the pathogenesis of HIT is to form soluble, circulating PF4–heparin complexes that induce anti-PF4–heparin antibody production by presentation of the complex to splenic marginal B cells (see Figs. 118–1 and 118–3).²⁵ Passive immunization with HIT antibodies in these mice obviates the need for delivery of soluble antigenic complexes to the spleen. Another role for infused heparin is counterintuitive in that it may prevent HIT by partially or completely removing surface-bound PF4 (Fig. 118–3).³⁵ If the level of circulating, free human PF4 in the mice was initially low relative to the level of infused heparin, all surface-bound PF4 and detectable surface antigenicity would be removed and the circulating HIT antibodies would have no targets on platelets and other vascular cells (Fig. 118–3). HIT therefore would not develop in spite of the presence of heparin and circulating HIT antibody. On the other hand, if the level of circulating free human PF4 in the mice was initially high relative to the infused heparin, not all surface-bound PF4 would be removed. Circulating HIT antibodies could then target and activate platelets and other vascular cells, leading to thrombocytopenia and thrombosis (Fig. 118–3).

Most patients likely begin with little PF4 bound to surface glycosaminoglycans (GAGs). After therapeutic heparinization, the level of surface PF4 goes down markedly so that the platelets cannot be targeted by anti-PF4–heparin HIT antibodies. However, patients with high levels of surface PF4 and significant surface PF4 antigenic complexes remaining after heparinization may be at risk for binding of pathogenic anti-PF4–heparin HIT antibodies to platelets and other vascular cells. Bound pathogenic antibodies lead to thrombocytopenia by clearance of antibody-coated platelets by the reticuloendothelial system. Bound antibodies also lead to platelet activation through FcγRIIA and the formation of procoagulant platelet microparticles that contribute to thrombosis.³⁶

As part of this activation, HIT antibodies also bind to endothelial cells likely via PF4–surface GAG complexes,^4,37 leading to local vascular activation and contributing to further local thrombosis. Additionally, HIT antibodies activate PF4-targeted monocytes^38,39 and neutrophils.⁴⁰ Monocyte activation may involve its surface Fcγ receptors⁴¹ with subsequent increased tissue factor expression and other changes consistent with a prothrombotic and inflammatory state. Monocyte and neutrophil GAGs are more complex than GAGs on the surface of platelets, which are mostly chondroitin sulfate⁴² and have relatively low affinity for PF4.⁴³ Monocytes and neutrophils bind PF4 with greater avidity and are more resistant to removal of bound PF4 by circulating heparin than platelets. In HIT, they may be preferentially targeted and activated relative to the platelets, contributing to the prothrombotic state of this immune thrombocytopenia.⁴⁴

Are there any genetic polymorphisms that are associated with an increased risk of developing HIT or developing thrombosis after HIT begins? No clear linkage has been shown with known thrombophilic polymorphisms including Factor V^Leiden, Prothrombin^G20210A, MTHFR^C677T, α_IIbβ₃ and α₁β₂.⁴⁵ Studies addressing a functional FcγRIIA^R/H131 polymorphism had varied outcome.^46,47 It is unclear whether patients with HIT have a higher density of FcγRIIA on their platelets.⁴⁸ High IgG affinity for the heparin–PF4 complex appears to affect the risk of developing HIT.

The model shown in Fig. 118–3 suggests that individuals with high PF4 content in their platelets and/or sustained platelet activation as might be seen in patients with significant atherosclerosis, a postsurgery state, or trauma would be most likely to develop HIT after heparinization and HIT antibody development. However, a relationship between formation of HIT antibodies and the degree of atherosclerosis in patients undergoing cardiopulmonary bypass surgery was not noted.⁴⁹ If levels of PF4 on the surface of circulating cells determine risk of developing HIT, this would offer a potential method for prescreening patients prior to heparinization and eliminate those at increased risk of HIT or as a useful tool in heparinized patients who develop thrombocytopenia to see whether they potentially can develop HIT. Theoretically, only those heparinized individuals with detectable surface PF4 would be potential candidates for developing HIT if they concurrently develop pathogenic antibodies.

The clinical hallmark of HIT is development of thrombocytopenia in the setting of a proximate heparin exposure. The combination of thrombocytopenia and heparin exposure in hospitalized patients is common and has poor specificity for HIT.⁵⁰ Therefore, other clinical clues must be sought in estimating the clinical likelihood of HIT. These include timing, degree of platelet count fall, nadir platelet count, presence of thromboembolism or hemorrhage, and the likelihood of other causes of thrombocytopenia.

TIMING

The platelet count in HIT characteristically begins to fall 5 to 10 days after initial heparin exposure.²³ There are three exceptions to this rule: (1) in rapid-onset HIT, patients with recent heparin exposure (within the previous 90 days) and preformed anti-PF4–heparin IgG experience a fall in platelet count immediately upon reexposure; (2) in delayed-onset HIT, clinical manifestations develop a median of 10 to 14 days after heparin is discontinued^51,52; and (3) a small number of patients with spontaneous HIT have been reported. These patients present with a thrombotic thrombocytopenic disorder reminiscent of HIT in the absence of recognized heparin exposure.⁵³ Both delayed-onset HIT and spontaneous HIT occur in the absence of circulating heparin and may involve pathogenic HIT antibodies that recognize complexes of PF4 and endogenous GAGs on blood and vascular cells.

DEGREE OF FALL IN PLATELET COUNT

The percentage fall in platelet count is measured from the peak platelet count after initiation of heparin to the nadir platelet count. Most patients with HIT experience a 50 percent or greater fall in platelet count; a more modest decline (30 to 50 percent) occurs in approximately 10 percent of patients.⁵⁴

NADIR PLATELET COUNT

As opposed to most other forms of drug-induced immune thrombocytopenia, thrombocytopenia associated with HIT is characteristically mild or moderate. The median nadir platelet count is approximately 60 × 10⁹/L and rarely falls below 20 × 10⁹/L in the absence of concomitant disseminated intravascular coagulation (DIC).⁵⁴ The nadir platelet count in HIT need not meet the traditional definition of thrombocytopenia (<150 × 10⁹/L). For example, patients with postoperative thrombocytosis may experience a subsequent greater than 50 percent decline in platelet count attributable to HIT that does not fall below this threshold.⁵⁵

THROMBOSIS

Thromboembolism is the presenting feature in up to 25 percent of patients with HIT and complicates approximately half of all cases.^56,57 Lower-extremity deep vein thrombosis and pulmonary embolism are the most common thrombotic manifestations, outnumbering arterial events by approximately 2:1.⁵⁶ Major venous obstruction can lead to limb gangrene. Catheter-associated upper extremity deep venous thrombosis is common.⁵⁸ Arterial thromboembolism most frequently involves the extremities, but may also manifest as stroke or myocardial infarction.⁵⁶ Thrombosis of other vascular beds including cerebral sinuses, mesenteric vessels, and adrenal veins is well-documented as is thrombotic occlusion of vascular grafts, fistulas, and extracorporeal circuitry.

HEMORRHAGE

In contrast to most other forms of drug-induced immune thrombocytopenia, spontaneous hemorrhage is rare in HIT, even when thrombocytopenia is severe. In a prospective study, bleeding complications were not increased in HIT patients compared with nonthrombocytopenic controls.⁵⁹

UNUSUAL CLINICAL MANIFESTATIONS

Rare sequelae of HIT include anaphylactoid reactions after intravenous heparin bolus, transient global amnesia, and skin necrosis at subcutaneous heparin injection sites.^60,61 Curiously, these phenomena may occur in the absence of thrombocytopenia. Nonnecrotizing erythematous injection site lesions are generally caused by delayed type IV hypersensitivity rather than HIT.⁶²

OTHER CAUSES

The likelihood of other etiologies of thrombocytopenia must be carefully considered in patients with suspected HIT. Common causes of hospital-acquired thrombocytopenia include infection; drugs other than heparin; DIC; dilution; and intravascular devices and extracorporeal circuits such as intraaortic balloon pumps, cardiopulmonary bypass, and extracorporeal membrane oxygenation.⁶³

Clinical scoring systems have been developed to permit estimation of the probability of HIT based on the aforementioned features. The most extensively studied of these systems, the 4T score,⁶⁴ classifies the probability of HIT as low, intermediate, or high on the basis of four criteria: Thrombocytopenia, Timing, Thrombosis or other sequelae, and the likelihood of other causes of thrombocytopenia (Table 118–2). In a meta-analysis of 13 studies, the negative predictive value of a low probability 4T score was 99.8 percent (95 percent CI: 97.0 to 100.0). The positive predictive value of an intermediate and high probability 4T score was 14 percent (95 percent CI: 9 to 22) and 64 percent (95 percent CI: 40 to 82), respectively.⁶⁵ The 4T score is limited by moderate interobserver agreement.⁶⁶ An alternative scoring system, the HIT Expert Probability (HEP) Score, exhibited improved reliability and favorable operating characteristics in a retrospective study, but remains to be prospectively validated.⁶⁷

In light of the complexity and limited positive predictive value of clinical diagnosis,⁶⁵ clinicians rely heavily on laboratory testing to aid in diagnosis. Laboratory assays for HIT fall into two categories: immunoassays and functional assays.

IMMUNOASSAYS

These assays detect the presence of circulating anti-PF4–heparin antibodies, irrespective of whether they are able to activate platelets and cause disease. The prototypical immunoassay is the solid-phase enzyme-linked immunosorbent assay (ELISA), in which dilute patient serum is added to microtiter wells coated with complexes of PF4–heparin (or PF4–polyvinylsulfonate).⁶ The polyspecific ELISA detects circulating anti-PF4–heparin IgG, IgM, and IgA. At the manufacturer-recommended cutoff, the sensitivity and specificity of this assay for HIT are 94 to 100 percent and 81 to 93 percent, respectively.^22,68,69,70

A key limitation of the polyspecific ELISA is its specificity. False-positive results are common and may result from detection of nonpathogenic anti-PF4–heparin antibodies⁶⁹ or antiphospholipid antibodies against either PF4⁷¹ or PF4-bound β₂-glycoprotein I.⁷² Specificity may be improved by raising the optical density (OD) cutoff. OD is directly associated with the 4T and HEP scores,⁶⁷ the risk of thrombosis,⁷³ and the likelihood of a positive functional assay.⁷⁴ In a Canadian study, only one of 37 patient samples exhibiting a weakly positive OD (0.40 to 0.99) demonstrated heparin-dependent platelet activation compared with 33 out of 37 samples with a strongly positive OD (>2.0).⁷⁴ In a recent analysis of 1958 patients, increasing the cutoff from a manufacturer-recommended threshold of 0.4 to 0.8 OD units increased specificity from 85 percent to 93 percent with a slight reduction in sensitivity from 100 percent to 98 percent.⁷⁵

Several modifications have been made to the PF4–heparin ELISA with the goal of improving specificity. Because pathogenic antibodies are primarily of the IgG class, detection systems specific for IgG have been developed. In a pooled analysis of studies comparing the IgG-specific and polyspecific ELISA, the former showed greater specificity (93.5 percent vs. 89.4 percent) at the cost of reduced sensitivity (95.8 percent vs. 98.1 percent).⁷⁶ Another modification involves the addition of a high heparin confirmatory step, in which reduction of the OD by 50 percent or more with the addition of excess heparin (100 U/mL) is considered to affirm the presence of heparin-dependent antibodies.⁷⁷ This method improves specificity, but false-positives remain common and false-negatives may also occur, particularly at high OD values.^78,79

Another limitation of the PF4–heparin ELISA is turnaround time. Although the analytical turnaround time of the ELISA is only approximately 2 hours, the assay is most cost-effective when multiple samples are run in batch. Consequently, many laboratories perform the ELISA only once or twice a week, leaving clinicians to make critical initial management decisions without the benefit of laboratory results. This drawback of the ELISA has spawned the development of several rapid immunoassays, which are designed to accommodate single samples and yield results in minutes. Table 118–3 summarizes the properties of these rapid assays.^{68,80,81,82,83,84,85,86} The latex particle-enhanced immunoturbidimetric assay and chemiluminescence assays are instrument-based and must be performed on proprietary analyzers. A rapid particle immunofiltration assay is approved in the Unites States, but published data suggest that it has unacceptable diagnostic accuracy.⁸⁷ A modified version of the assay⁸⁸ requires further study.

FUNCTIONAL ASSAYS

Functional assays are more specific than commercial immunoassays because they detect only the subset of antibodies capable of inducing platelet activation in a heparin-dependent manner. The prototypical functional assays are the ¹⁴C-serotonin release assay (SRA) and the heparin-induced platelet-activation assay (HIPA). In the SRA, various concentrations of heparin and heat-inactivated patient serum are added to washed donor platelets radiolabeled with ¹⁴C. A positive test is signified by heparin-dependent release of ¹⁴C-serotonin.⁸⁹ The HIPA is based on a similar principle, but uses visual assessment of platelet aggregation as an end point.⁹⁰ The sensitivity and specificity of the SRA and HIPA are said to exceed 95 percent, but universally accepted reference standards against which to measure their performance do not exist.⁶³

Washed platelet functional assays are technically demanding. Both the SRA and HIPA require reactive donor platelets and the SRA requires radioisotope. Because these reagents are impracticable for most clinical laboratories, functional assays are performed at only a small number of reference laboratories around the world. Even among such laboratories, test methodology, result interpretation, and reporting are not well-standardized.⁹¹

Novel immunoassays and functional assays for HIT designed to overcome the limitations of assays currently in use are in development.^92,93

NONHEPARIN ANTICOAGULANTS

Management of HIT requires immediate withdrawal of heparin, including cessation of heparin flushes and removal of heparin-coated catheters. However, discontinuation of heparin alone is insufficient to prevent thromboembolism. Historical studies of untreated patients document a 5 to 10 percent daily risk of thrombosis in the first 48 hours after heparin is stopped and a 30-day cumulative incidence of thrombosis of approximately 50 percent.^57,94 Discontinuation of heparin must therefore be accompanied by initiation of a rapid-acting, parenteral, nonheparin anticoagulant.⁹⁵ Table 118–4 summarizes the properties of nonheparin anticoagulants used to treat HIT.

Argatroban and bivalirudin are direct thrombin inhibitors. Argatroban is the only FDA-approved drug for treatment of HIT available in the United States. Its approval was based on two open-label single-arm studies in which argatroban-treated subjects were compared with untreated historical controls.^96,97 In a pooled analysis of these two studies, argatroban reduced the relative risk of new thrombosis by two-thirds. The incidence of major bleeding was approximately 1 percent per day.⁹⁸ An important limitation of these studies was that serologic confirmation of HIT was not required for enrollment. Indeed, 36.4 percent of subjects were found to be anti-PF4–heparin antibody-negative on post hoc testing,⁹⁹ suggesting that a sizable proportion of the study population did not have HIT.

Bivalirudin is a hirudin analogue. It is approved for patients with and without HIT undergoing percutaneous vascular procedures. It is not approved for treatment of HIT, although it has been used off-label for this indication, particularly in patients with critical illness and multiorgan failure¹⁰⁰ and those undergoing cardiac surgery.¹⁰¹ Published evidence supporting its use is limited to retrospective single-center cohort studies.^100,102,103

Two other direct thrombin inhibitors have been studied as treatments for HIT. Lepirudin, a recombinant hirudin, was shown to reduce the risk of thromboembolism compared with untreated historical controls, but is no longer available.⁹⁴ A randomized clinical trial of desirudin closed because of poor accrual after only 16 subjects had been randomized.¹⁰⁴

Danaparoid and fondaparinux are indirect factor Xa inhibitors. Danaparoid is approved for treatment of HIT in multiple jurisdictions, but is no longer marketed in the United States and drug shortages have limited its availability elsewhere. In an open-label randomized trial, 42 patients with HIT complicated by thrombosis were allocated to receive either danaparoid or dextran 70. Significantly more subjects in the danaparoid arm were judged to have complete recovery from thrombosis at hospital discharge (56 percent vs. 14 percent; p = 0.02).¹⁰⁵ In vitro crossreactivity of HIT antibodies with danaparoid occurs in some patients, although the clinical relevance of this phenomenon has not been established.¹⁰⁶

Fondaparinux in not approved for treatment of HIT, nor is it recommended in the 2012 American College of Chest Physicians Guidelines.⁹⁵ Its use is supported by several small case series and retrospective cohort studies.^{107,108,109,110} In a pooled analysis involving 71 patients, no new thrombotic events were reported. Four patients suffered major hemorrhage.⁶³ A small number of cases of HIT induced or exacerbated by fondaparinux have been reported, although the attribution to fondaparinux in at least some of these cases remains uncertain.¹¹¹ Fondaparinux is more convenient to use than other agents given the ease of once-daily administration and a potential lack of need for monitoring (see Table 118–4). This may, in part, account for its increasing use. In a recent multicenter German registry of patients with suspected HIT, a greater proportion of patients were treated with fondaparinux (40 percent) than with danaparoid (23.6 percent) or argatroban (16.4 percent).¹¹²

Oral direct inhibitors of thrombin (e.g., dabigatran) and factor Xa (e.g., rivaroxaban, apixaban, edoxaban) do not induce platelet aggregation or PF4 release in the presence of HIT-positive sera in vitro¹¹³ and constitute biologically rational approaches to the treatment of HIT. However, clinical evidence is limited to a small number of case reports^114,115 and use of these agents for HIT cannot be recommended outside of a clinical trial. One such trial of rivaroxaban in patients with suspected HIT began enrollment in 2013 and is ongoing.¹¹⁶

An important toxicity of anticoagulants used for treatment of HIT is major bleeding, a risk compounded by the absence of effective reversal agents. Novel therapeutic approaches that target pathways proximal to activation of coagulation may provide effective antithrombotic therapy without the degree of bleeding risk associated with anticoagulants. Candidate strategies include a desulfated form of heparin with minimal anticoagulant activity¹¹³ and small molecule PF4 antagonists,¹¹⁷ which interfere with formation of PF4–heparin complexes; inhibitors of FcγRIIA-mediated platelet activation by HIT immune complexes; and inhibitors of splenic tyrosine kinase and Ca²⁺¹¹⁸ and diacylglycerol-regulated guanine nucleotide exchange factor I,¹¹⁹ which disrupt intracellular transduction triggered by immune complex binding.

WHO TO TREAT

Because of the frequency of heparin use and thrombocytopenia among hospitalized patients, the modest specificity of immunologic assays, and clinicians’ fears of missing a case of HIT, overdiagnosis and unnecessary treatment with nonheparin anticoagulants of patients without HIT is common.¹²⁰ Inappropriate use of these agents is associated with increased costs and bleeding risk.¹²¹ In light of the very-high negative predictive value (99.8 percent) of a low-probability 4T score,⁶⁵ a reasonable first step toward reducing unnecessary treatment is to avoid use of nonheparin anticoagulants in patients with a low-probability 4T score. Heparin should be discontinued and a nonheparin anticoagulant initiated in patients with an intermediate- or high-probability 4T score until the results of HIT laboratory testing become available.^65,95,122

TRANSITIONING TO A VITAMIN K ANTAGONIST

Warfarin and other vitamin K antagonists should not be prescribed as the initial anticoagulant in patients with acute HIT because their use increases the risk of venous limb gangrene as a result of rapid lowering of protein C activity.¹²³ For patients receiving a vitamin K antagonist at the time HIT is diagnosed, the vitamin K antagonist should be discontinued and its effects reversed with vitamin K. A vitamin K antagonist may be initiated once the platelet count has recovered to a stable plateau. Large loading doses (e.g., warfarin >5 mg/day) should be avoided. The vitamin K antagonist should be overlapped with a parenteral nonheparin anticoagulant for at least 5 days and until the international normalized ratio (INR) has reached its intended target.^63,95 If the patient is being transitioned from argatroban to warfarin, guidelines regarding the appropriate INR target should be followed,¹²⁴ because both argatroban and warfarin increase the INR. This target will vary according to the sensitivity of the prothrombin time reagent to argatroban used in each institution.

DURATION OF ANTICOAGULATION

Patients with HIT-associated thromboembolism are typically treated with therapeutic anticoagulation for 3 to 6 months. The optimal duration of anticoagulation in patients with HIT without thrombosis (i.e., isolated HIT) is unknown. In a historical series of untreated patients with isolated HIT, the cumulative incidence of thromboembolism at 30 days was 53 percent.⁵⁷ Most events occurred within 10 days of heparin cessation, corresponding to the platelet recovery phase. It is therefore generally accepted that anticoagulation be continued in patients with isolated HIT until platelet count recovery. Some authorities recommend longer courses (e.g., 4 weeks).¹²²

PLATELET TRANSFUSION

There is a long-held concern that platelet transfusion may precipitate thrombosis in HIT by “adding fuel to the fire.” Two case series challenge this dogma. Collectively, these series included 41 patients with suspected HIT who underwent platelet transfusion. None developed thrombosis during extended followup.^125,126 Nevertheless, because HIT is characteristically prothrombotic rather than prohemorrhagic, prophylactic platelet transfusion is rarely indicated. Transfusion may be considered in the setting of clinically significant bleeding, high bleeding risk, or diagnostic uncertainty.

HEPARIN REEXPOSURE IN PATIENTS WITH A HISTORY OF HEPARIN-INDUCED THROMBOCYTOPENIA

In general, heparin reexposure should be avoided in patients with a history of HIT because of the risk of reoccurrence.¹²⁷ An exception to this rule is the use of intraoperative heparin in patients with a history of HIT who are undergoing cardiovascular surgery. The HIT immune response wanes over time. Functional assays become negative at a median of 50 days after heparin cessation, whereas anti-PF4–heparin antibody titers decline more slowly and are no longer detectable in 60 percent of patients by day 100.²³ HIT laboratory testing can be used to determine the safety of heparin reexposure during cardiovascular surgery. Patients with a negative immunologic and functional assay may safely receive UFH during surgery. This was first demonstrated in 10 patients with a history of HIT undergoing cardiac surgery, none of whom developed clinical reoccurrence.¹²⁸ In a newer report, 11 of 17 such patients developed recrudescence of anti-PF4–heparin antibodies, but only one developed HIT.¹²⁹ Heparin should be strictly avoided in patients with a positive functional assay. If possible, surgery should be delayed in these individuals until functional and immunologic assays become negative. If surgery cannot be delayed, a nonheparin anticoagulant (e.g., bivalirudin) should be used.¹⁰¹ Appropriate intraoperative anticoagulation of patients with a functional assay that has become negative, but an immunologic assay that remains positive is uncertain. The 2012 American College of Chest Physicians Guidelines recommend a nonheparin anticoagulant in this setting.⁹⁵ However, intraoperative heparin was used uneventfully in three such patients undergoing urgent heart transplantation.¹³⁰ When heparin is administered to patients with HIT, it should be limited to the intraoperative setting. Pre- and postoperative exposure should be scrupulously avoided, though patients with a history of HIT who are (inadvertently) reexposed to longer courses of heparin do not always develop recurrent HIT.¹²⁹

In light of its documented efficacy and safety in large coronary angiography trials, bivalirudin is recommended over heparin in patients with a history of HIT who require percutaneous vascular procedures, irrespective of the results of HIT laboratory testing.⁹⁵

HEMODIALYSIS

Although approximately 10 percent of patients on chronic hemodialysis develop circulating anti-PF4–heparin antibodies,¹³¹ the incidence of HIT in this population is less than 1 percent.¹³² Ongoing heparin exposure during dialysis in patients with a history of HIT is contraindicated. Alternative strategies including regional citrate, saline flushing, danaparoid, argatroban, and vitamin K antagonists use have been reported.⁹⁵

PREGNANCY

HIT is rare (<0.1 percent) in pregnant women exposed to heparin.^11,12 When it does occur, initiation of a nonheparin anticoagulant is warranted. The largest published experience is with danaparoid. A retrospective cohort of 30 women with acute HIT received danaparoid during pregnancy.¹³³ Five patients developed thrombosis and three developed major bleeding. Danaparoid does not cross the placenta and there was no measurable anti-Xa activity in the cord blood of six neonates who were tested after delivery. If danaparoid is unavailable, fondaparinux may be considered though evidence supporting its use in pregnant women with HIT is limited to case reports^134,135 and partial transplacental passage has been demonstrated.¹³⁶