Heparin Induced Thrombocytopenia: A Focus on Thrombosis

Abstract

Heparin induced thrombocytopenia (HIT) is an immune mediated disorder caused by antibodies that recognize complexes of platelet factor 4 and heparin. Thrombosis is a central and unpredictable feature of this syndrome. Despite optimal management, disease morbidity and mortality from thrombosis remain high. The hypercoagulable state in HIT is biologically distinct from other thrombophilic disorders in that clinical complications are directly attributable to circulating ultra-large immune complexes (ULICs). In some individuals, ULICs elicit unchecked cellular procoagulant responses that culminate in thrombosis. To date, the clinical and biologic risk factors associated with thrombotic risk in HIT remain elusive. This review will summarize our current understanding of thrombosis in HIT with attention to its clinical features, cellular mechanisms, and its management.

Graphical Abstract

INTRODUCTION

Unfractionated heparin (UFH) remains the mainstay of anticoagulation for indications for which there are no effective substitutes, such as anticoagulation for cardiopulmonary bypass (CPB) surgery, extracorporeal membrane oxygenator (ECMO) circuits, dialysis and mechanical prosthetic valves. The continued use of UFH in clinical practice places patients at risk for developing heparin induced thrombocytopenia (HIT), a potentially devastating immune disorder caused by antibodies to complexes of platelet factor 4 (PF4) and heparin (anti-PF4/H Abs).

Thrombosis is the most serious and life-threatening complication of HIT. Thrombosis occurs at presentation or complicates the course of illness in 20–64%.^{1, 2} Disease morbidity is further compounded by high rates of bleeding (~40%) associated with use of potent non-heparin anticoagulants for the prevention or treatment of HIT thrombosis. ² Even with use of alternative anticoagulants, thrombosis contributes to fatal outcomes in ~6–26% of HIT patients.^2–5

As thrombosis is central to disease morbidity and mortality in HIT, this review will summarize our current understanding of its risk factors, mechanisms, and treatment. For topics related to other aspects of the disease, such as the epidemiology^{3, 6}, diagnosis⁷or laboratory testing^{8, 9} in HIT the reader is referred to corresponding reviews on these topics.

Clinical Features & Risk Factors for Thrombosis

HIT is primarily a disease of hospitalized or recently discharged patients. HIT occurs in 0.5–1% of patients exposed to UFH for medical and/or surgical indications with a markedly lower incidence (0.1–0.5%) in patients receiving LMWH.^{3, 6, 10, 11} While the impact of direct oral anticoagulants (DOACs) has yet to be directly investigated, observational data suggests that the incidence of HIT and its disease burden in recent years has not been significantly affected. Data from a publicly available US database indicate that the incidence of HIT has remained fairly stable between years 2009 and 2013, with a diagnostic prevalence between the initial and last year of the study period (2013) showing only a modest decrease.³ Of various populations exposed to heparin, those undergoing cardiac surgery appear to particularly high risk for HIT, despite the fact that only a subset of HIT antibodies generated are platelet-activating.^{12, 13} While isolated cases of spontaneous HIT have been reported in recent years,^14–18 its incidence is likely rare and significantly less than that provoked by heparin exposure.

Clinical Features of Thrombosis

For the majority of patients, HIT presents in the context of recent exposure as a fall in platelet count, either as an absolute (<150,000/μL) or relative thrombocytopenia (>30% fall from baseline platelet count). Unlike other thrombocytopenic disorders, the low platelet counts in HIT do not increase bleeding risk, but rather serve as a marker of thrombotic risk. ¹⁹

The true incidence of thrombosis in HIT is uncertain due to few prospective studies of this disease and the challenges ascertaining HIT-related complications in critically ill patient populations with confounding risk factors. However, recent retrospective studies document the occurrence of thrombosis in ~20–64% of patients with HIT ^{1, 2} (See Table 1). New or progressive thromboses develop in ~19–40% of patients treated with alternative anticoagulants, while mortality varies depending on the study. Patients who develop isolated thrombocytopenia as initial disease presentation remain at significant risk for subsequent thrombosis occurring in 18–50%.^{5, 20}

TABLE 1:

Clinical Features of Thrombosis in HIT

Study	Type of Study (Total # of patients/# HIT patients)	Prevalence of thrombosis at time of diagnosis	Arterial v venous thrombosis events at presentation	Development of new/progressive thrombosis on treatment	Amputation	Death
Warkentin et al., 1996 20	Single Center, Retrospective (127/127)	51%	1:4	76%	NR	20%
Elalamy, et al., 20095	Muticenter Retrospective (114/49)	59%	1:1	43%	8%	53%
Kuter et al., 20171	Muticenter Retrospective (442/355)	20%	NR*	19%	3%	22%
Pishko et al., 2019 2]	Single Center Retrospective (310/42)	64%	1:4	36%	NR	26%
Gruel et al., 20204	Muticenter Retrospective (144/144)	40%	1:3	NR*	4%	6%

^*Not recorded

Timing of Thrombosis

Whereas thrombocytopenia follows a predictable course after seroconversion, with platelet counts usually declining 2–4 days after seroconversion, thrombosis is less predictable and can occur at any time after detection of circulating anti-PF4/H Abs.²¹ In a large retrospective review of 408 patients with HIT, approximately two-thirds of patients developed thrombosis either concurrently (26%) or several days after thrombocytopenia (40%), while one-third experienced thrombosis prior to a fall in platelet counts. ¹⁹ Thrombotic risk in HIT extends for several weeks beyond duration of heparin exposure due to the presence of circulating anti-PF4/H Abs that recognize PF4 bound to endogenous glycosamginoglycans (GAGs) or other platelet components.^22–24 Cross-reactivity of HIT antibodies with PF4 bound to cell-surface GAGs also explains why timing of heparin discontinuation has no measurable impact on disease progression in some patients.²⁵ Indeed, PF4-dependent, but heparin-independent platelet activation^26–29 may identify a subset of patients with “autoimmune” HIT whose disease manifestations are often delayed or severe.¹⁸

Types of Thrombosis

Thrombosis in HIT affects both arterial and venous beds, with venous thrombosis occurring three to four times more commonly than arterial thrombosis (Table 1).¹⁹ Deep venous thrombosis (DVT) and pulmonary embolism (PE) are the most common sites of venous thrombosis, while arterial thrombosis most often manifests as peripheral arterial embolism, followed by stroke and myocardial infarction. Arterial thrombosis may be more common after cardiac surgery,⁴ and spontaneous HIT may be more likely to be precipitated by recent orthopedic surgery.³⁰ Distinctive, but uncommon, presentations of thrombosis include skin necrosis, venous limb gangrene, bilateral adrenal hemorrhage and cerebral vein thrombosis (CVT).^31–35 Skin necrosis in HIT occurs at heparin injection sites and is one of the few clinical settings where HIT may present without accompanying thrombocytopenia.³¹ Venous limb gangrene is often precipitated by concurrent warfarin use and is caused by severe venous occlusion extending to venular beds.^{33, 36, 37} Adrenal hemorrhage is a thrombotic manifestation of HIT caused by occlusion of the sole adrenal central vein supplying the adrenal gland. This complication is most often seen in the post-operative setting and when bilateral, the patient presents with symptoms and signs of adrenal insufficiency (hypotension, nausea, vomiting and hyponatremia), if unrecognized, can progress to death from adrenal failure.^{32, 38}Rare presentations of CVT have been reported in spontaneous ^{34, 35}as well as drug-induced HIT. ^39–42

Risk factors for Thrombosis

Numerous studies have examined biologic and clinical risk factors predisposing individuals to thrombotic complications in HIT. To date, there is no evidence to support a role for conventional thrombophilic abnormalities, such as Factor V Leiden or deficiencies of antithrombin, protein C or protein S.^{43, 44} As discussed below, polymorphic variation in the extracellular domain of the FcγIIA receptor (H131R; histidine to arginine at the 131 amino acid position) and its functional regulation by tyrosine phosphatases, CD148 and the T-Cell Ubiquitin Ligand-2 (TULA-2), have been identified as potential genetic causes influencing predisposition to thrombosis in HIT.^{45, 46} While genome wide association (GWAS) studies have identified several single nucleotide polymorphisms (SNPs) and HLA associations in small cohorts of HIT as compared to control patients, ^47–50 these findings may be more germane to development of the immune response given the study design of comparing patients with HIT to those without. Other shortcomings of the GWAS studies, including differences in case ascertainment, lack of validation and absence of functional studies, presently limit the generalizability of these studies to understanding HIT pathogenesis.

Other clinical risk factors appear related to disease presentation. Several groups have identified cardiovascular disease and/or surgery as a major risk factors for arterial thrombosis.^{3, 4, 43, 51} Endothelial injury associated with central venous catheters places patients at increased risk for venous thrombosis.^{51, 52} Other clinical predictors of thrombotic risk include severe thrombocytopenia, with a >70% drop in platelet count increasing thrombotic risk by 8-fold, high antibody levels or titers in immunoassays or platelet activation assays and heparin-independent platelet activation seen in a subset of patients with autoimmune HIT which account for disease manifestations occurring after heparin discontinuation.^53–56

Pathogenesis of Thrombosis in HIT

The intense hypercoagulable state in HIT is triggered by cellular activating anti-PF4/H Abs that principally engage cellular Fcγ receptor, FcγRIIA (on platelets, monocytes and neutrophils) ^{57, 58} or indirectly activate endothelial cells through non-FcγR mechanisms.⁵⁹ The human FcγRIIA (CD32A) is a 40kDa, Type 1 transmembrane glycoprotein low affinity receptor for IgG that preferentially binds to IgG in immune complexes over monomeric IgG.^60–62Clustering of the FcγRIIA receptors by immune complexes initiates phosphorylation of the immunoreceptor tyrosine-based activation motif (ITAM) contained in the cytoplasmic tail of the receptor and activates downstream signaling via the spleen tyrosine kinase (Syk) resulting in release of intracellular Ca²⁺ stores, degranulation, cytokine production and cellular activation.⁶⁰

Although FcγRIIA was implicated in cellular activation by HIT antibodies in the 1980’s,⁵⁷ its essential role in disease pathogenesis was not proven until the development of the murine HIT model in 2001. As mice lack the human equivalent of FcγRIIA ⁶³ and murine PF4 does not cross-react with HIT antibodies, ⁶⁴ double transgenic mice, expressing hPF4 and hFcγRIIA, were generated to characterize the in vivo requirements for thrombosis. Single transgenic mice, expressing either the hPF4 or hFcγRIIA transgenes, did not develop thrombocytopenia or thrombosis when injected with a HIT like monoclonal antibody (KKO),⁶⁴ but double transgenic mice developed severe thrombocytopenia (>80% drop in platelet counts) and thrombosis in multiple vascular beds (heart, liver, kidneys and lungs) in response to the monoclonal antibody.⁶⁵ In addition to establishing the requirements for cellular FcγRIIA in HIT thrombosis, these murine studies were also first to demonstrate the vivo pathogenicity of circulating anti-PF4/H Abs.⁶⁵

Cellular contributions to HIT Thrombosis

The following sections will describe the individual cellular contributions as well as the cellular interactions that promote and propagate thrombosis in HIT.

Platelets

Platelets play a prominent role in virtually all aspects of HIT, from their conspicuous involvement in disease expression (thrombocytopenia occurs in >94% of patients with HIT) ¹⁹ to their central role in functional assays often used for diagnostic confirmation.⁵⁶ Early studies of HIT focused on effects of complement activation on platelet responses.^{66, 67} These studies demonstrated increased complement deposition on circulating platelets from HIT patients, ⁶⁶ complement fixation by HIT IgG, ^{66, 67} and requirements for classical pathway components in platelet activation.⁶⁶ Once platelet FcγRs were identified in 1987, ⁶⁸ the field shifted its focus to investigations of this cellular receptor as a target for HIT immune complexes. Kelton et al.⁵⁷ and Chong et al.⁵⁸ demonstrated that platelet activation by HIT antibodies, as measured by the release reaction, required both F(ab’)2 and Fc portions of HIT IgG, was blocked by antibodies to FcγRII and did not involve the glycoprotein (GP) receptors, Ib/V/ and IIb/IIIa IX.^{57, 58} Other studies showed that while GP IIb/IIIa inhibitors could prevent platelet aggregation responses by HIT antibodies ^{69, 70}, they did not prevent release of radiolabeled ¹⁴C-serotonin, ^{67, 71} which required ADP and P2Y12 or Gi-dependent signaling pathways.^{71, 72} In vivo studies using the murine HIT model have validated the importance of FcγRIIA signaling pathways by demonstrating the efficacy of the Syk kinase inhibitor, PRT-060318 (PRT318), in preventing HIT-induced thrombocytopenia and thrombosis.⁷³

Platelets from healthy donors show considerable interindividual variation in responses to HIT ULICs; platelets from some healthy donors consistently activate in response to HIT ULICs while platelets from other donors are poorly responsive.⁷⁴ This donor variability in platelet activation response is presumed to contribute to the heterogeneity seen in clinical disease. The mechanisms underlying this variability in platelet responses is not fully understood, but several recent studies have shown effects of genetic variants on FcγRIIA functional responses to HIT immune complexes. Of these genetic determinants, the FcγRIIA H131R polymorphism has garnered the most attention in the field. The H131R polymorphism shows differential affinity for binding the IgG₂ subclass; the 131R allotype binds minimally to human IgG₂ whereas the 131H binds IgG₂ as well as other subclasses. Early clinical studies were largely inconclusive as to the thrombotic risk conferred by the H131R polymorphism.^75–77 However, a recent study has revisited the functional consequences of the 131RR -allelic expression in the context of other endogenous IgG.^{4, 78} In these studies, Rollin and colleagues demonstrated that platelets from 131RR individuals have increased reactivity to HIT antibodies (generally, of the IgG₁ subclass) due to inability of the RR isoform to bind endogenous IgG₂. The 131HH allotype, on the other hand, binds native IgG₂ and therefore offers fewer binding sites for platelet activating HIT IgG₁.⁷⁸ Consequently, platelets from individuals expressing the 131RR allele have greater reactivity to HIT IgG than platelets expressing other isoforms. In a prospective study of 144 HIT patients whose genotype was characterized for several genetic variants, including FcγRIIA H131R, HPA-1a/b dimorphism affecting glycoprotein IIIa, and platelet endothelial cell adhesion molecule-1 (L125V) thrombotic risk was increased only in individuals expressing the 131RR polymorphism, which was present in 38% of patients with thrombosis as compared to 18 % without thrombosis (OR 2.9).⁴ Other studies have also examined genetic variants regulating FcγRIIA signaling. Two polymorphisms on CD148 (276QQ and 326RR), a tyrosine phosphatase that can modulate FcγRIIA signaling were shown to be protective of thrombosis.⁷⁹ Similar findings were noted in studies investigating TULA-2, a tyrosine phosphatase that dephosphorylates Syk.^{46, 80} In humans, TULA-2 expression was inversely correlated with platelet responsiveness to HIT antibodies, ⁸⁰ while murine TULA-2 deficiency was associated with heightened platelet reactivity and severe thrombocytopenia in response to immune complexes.⁴⁶

Once activated by HIT ULICs, platelets propagate thrombosis through generation of procoagulant microparticles, ⁸¹ upregulation of P-selectin that mediates formation of platelet leukocyte aggregates ^{82, 83} and release of intracellular PF4 and polyphosphates.⁸⁴ Released PF4 binds to cellular GAGs on platelets and/or extruded von Willebrand factor on the endothelium to form antigenic sites for circulating HIT antibodies.⁸⁵ See Figure 1.

Figure 1. Platelet contributions to thrombosis:

Binding of HIT ULICs to platelet FcγRIIA (1) initiates phosphorylation of ITAM motifs and downstream signaling via Syk kinase(2) leading to degranulation of alpha and dense granules leading to release of additional PF4, polyphosphate (PO4⁻_n), ADP, and release of soluble P-selectin and microparticles (3). Released ADP binds to G-coupled receptors, P2Y₁₂ initiating Gi-dependent intracellular signaling (4) resulting in further activation signals to generate “coated” platelets. Negative regulators include increased expression of TULA-2 and polymorphisms of the tyrosine phosphatase CD148 (5).

Monocytes

Findings of monocytes and neutrophil involvement in HIT pathogenesis could be traced to early studies by Kelton and colleagues ⁵⁷ who showed that addition of monocytes and neutrophils to platelets, at physiologic ratios, inhibited platelet activation by HIT antibodies. These observations suggested competitive binding of HIT antibodies to cellular GAG or FcγR. Indeed, subsequent studies confirmed differential binding of HIT ULICs to monocytes as compared to platelets. This increased binding serves as an important physiologic reservoir for HIT antibodies and has a moderating effect on thrombocytopenia in vivo. In the murine HIT model, depletion of monocytes using clodronate liposomes or gadolinium chloride markedly exacerbates thrombocytopenia, as monocytes are not available to bind HIT antibodies.⁸⁶ Increased ULIC binding to monocytes results in heightened procoagulant activity, as indicated by upregulation of TF mRNA, ^{87, 88} release of TF-bearing microparticles ⁸⁹ and increased cell surface TF expression in response to KKO or HIT IgG.^87–90 Monocyte involvement in thrombosis was also evident in the size of thrombi formed after cellular depletion. In the murine HIT model, monocyte depletion markedly inhibited thrombus formation after infusion of KKO,⁸⁶ while ex vivo depletion of monocytes from human blood attenuated platelet deposition and fibrin generation.⁹¹ ULIC-mediated effects on monocyte activation were dependent on FcγRIIA, and downstream signaling via Syk kinases, as inhibition of this pathway markedly reduced thrombin and fibrin formation.^{88, 91} In turn, thrombin generated from monocyte TF activates platelet protease-activated receptor 1 (PAR-1).⁹¹ Dual engagement of platelet PAR-1 and FcγRIIA leads to generation of “coated” platelets, ⁹¹ a subpopulation of highly activated platelets formed by dual activation with thrombin and collagen or thrombin and FcγRIIA.⁹² Monocytes activated by HIT ULICs additionally upregulate CD11b (Mac-1) expression which promotes formation of platelet-monocyte aggregates.⁸³ See Figure 2.

Figure 2. Monocyte contributions to thrombosis:

PF4 bound to cell-surface GAGs serves as an important physiologic reservoir for binding HIT antibodies (1). Circulating ULICs also engage cellular FcγRIIA (2) initiating downstream signaling via Syk kinase(3) leading to upregulation of cell surface TF and release of TF-bearing microparticles that promote thrombin (IIa) generation (4) which then activates platelet protease activated receptor-1 to generate coated-platelets. Activated monocytes upregulate CD11b/Mac-1 which binds to circulating platelets to form leukocyte-platelet aggregates (5).

Neutrophils

Similar to other FcγR expressing cells, neutrophils are readily activated by HIT ULICs. Cellular activation is accompanied by increased CD11b expression, ^{83, 93} degranulation ^{83, 93–95} and increased cell adhesion via L-selectin.⁹³ These activation events, however, do not appear as essential for thrombus formation as the ability of activated cells to release neutrophil extracellular traps (NETs). NETs are extruded strands of deoxyribonucleic acid (DNA) and granular material that serve important functions in innate immunity, by ensnaring pathogens.⁹⁶ NETs are also critical for thrombus formation,⁹⁷ as they directly engage platelets, red blood cells, and circulating hemostatic proteins such as PF4, ⁹⁸ fibronectin and von Willebrand Factor.⁹⁹ Two recent studies by Gollomp et al.⁹⁸ and Perdomo et al.⁹⁵ showed that neutrophils promote thrombosis in HIT through formation of NETS. These investigators showed that patients with HIT, as compared to control subjects, have evidence of NETosis in plasma, as indicated by increased levels of cell free DNA, citrullinated histones and myeloperoxidase.^{95, 98} Further, in microfluidic studies using human blood, HIT plasma or IgG binds NETs,⁹⁸ protects DNA from degradation by deoxyribonucleases^{95, 98} and induces neutrophil-dependent thrombus formation.⁹⁵ Using a cremaster injury model, Gollomp and colleagues showed that there was significant neutrophil and platelet accumulation in venous, but not arterial thrombi, findings which were reinforced in HIT mice lacking peptidyl arginine deiminase 4, an enzyme essential for NETosis.⁹⁸ Using a differing approach for assessing thrombosis, by evaluating spontaneous pulmonary embolism rather than laser-induced injury occlusion, Perdomo and colleagues found that inhibition of NETosis through chemical or genetic approaches abrogated thrombosis in HIT mice, a process dependent on neutrophil FcγRIIA, but appeared relatively independent of platelets.⁹⁵ It should be noted that while platelets were not considered essential for NETosis in both murine studies, their presence was a critical determinant of thrombus size.^{95, 98} These murine studies not only demonstrating a critical role for neutrophil involvement in HIT, but also importantly furnish a mechanism (NETs) by which these cells promote thrombosis. See Figure 3.

Figure 3. Neutrophil contributions to thrombosis:

Circulating ULICs engage cellular FcγRIIA (1) initiating neutrophil activation, degranulation (2) upregulation of L-selectin and CD11b/Mac-1 to promote adhesion and formation of leukocyte platelet aggregates (3). Activated neutrophils extrude NETs which bind PF4 and HIT ULICs (4).

Endothelial Cells

While there is persuasive clinical and laboratory data indicating endothelial involvement in HIT, the mechanisms by which HIT ULICs activate these cells is less clear. Of endothelial cells, only dermal microvascular cells express the activating FcγRIIA,¹⁰⁰ while a limited number of other endothelial beds (liver sinusoidal, placental and aortic endothelial cells) express the inhibitory receptor FcγRIIB.¹⁰¹ While there is no evidence for direct endothelial activation, there is ample support for ULIC-mediated bystander injury.

Histologic studies of the microvasculature from HIT patients undergoing limb amputations indicate multiple platelet thrombi, infiltrating endothelial cells and intraluminal hyperplasia of endothelial cells (ECs).¹⁰² Patient antibodies bind directly to ECs via cellular GAGs,^{59, 64, 103} promote complement deposition on cell surfaces,⁵⁹ and induce procoagulant activity.⁵⁹ EC activation by HIT serum or IgG is markedly enhanced by platelets, leading to increased expression of cellular adhesion proteins (E-selectin, ICAM-1, VCAM), cytokine release, (IL-1 β, IL6, PAI-1, and TNFα)and platelet deposition¹⁰⁴. These cellular effects of platelets are attenuated by selective inhibition of platelet GPIIb/IIIa receptors or ADP.¹⁰⁴ In microfluidic studies, KKO and/or HIT patient antibodies readily bind PF4-VWF complexes, promoting platelet adhesion and enlargement of thrombi within the microfluidic channels.⁸⁵ In microfluidic chambers coated with human umbilical vein endothelial cells (HUVECs) or endothelial cells from human aortas, chemical EC injury is accompanied by release of vWF from the EC glycocalyx, ¹⁰⁵ and increased PF4 binding to vWF strings leading to expression of antigenic sites for HIT antibodies.^{85, 105} In HIT transgenic mice, HIT antibodies exacerbate thrombus formation at peri-injury sites on the endothelium where released PF4 binds to the endothelium and thrombus occlusion can be modulated through use of ADAMTS13 or N-acetylcysteine.⁸⁵ These latter studies are in keeping with clinical findings showing that sites of endothelial injury, such as vessels containing central venous catheters are particularly prone to thrombosis.⁵¹

Taken together, these studies suggest that the endothelium promotes thrombosis in HIT through responses to direct or indirect cellular injury. Direct physical injury, perhaps through catheterization, surgery or underlying atherosclerotic disease, releases hemostatic vWF multimers that support thrombosis through binding platelets and providing additional binding sites for HIT antibodies. Intact endothelium could be activated indirectly by complement activation and/or secreted products of FcγRIIA bearing cells, leading to expression of adhesion markers and/or TF expression. See Figure 4.

Figure 4. Endothelial cell contributions to thrombosis:

Circulating ULICs or HIT antibodies bind to endothelial cell GAGs (1) triggering complement activation, deposition and EC activation(2). ECs express TF (3), and in the presence of activated platelets, upregulate cell-surface adhesive markers, E-selectin, ICAM-1 and VCAM (4). Activated ECs release vWF strings that bind PF4 to serve as new antigenic targets for HIT antibodies(5).

Treatment

Treatment in HIT is directed at lowering the intense thrombin generation that accompanies disease.¹⁰⁶ A number of on-label and off-label therapies have been used in the management of HIT and are briefly discussed below. For detailed information related to pharmacology, dosing and/or monitoring of therapies, the reader is referred to recent guidelines ¹⁰⁷ and reviews.^{108, 109}

Direct Thrombin Inhibitors (DTIs)

Argatroban and bivalirudin are FDA-approved medications for HIT management, the latter in the setting of percutaneous coronary interventions in HIT patients. In clinical trials leading to drug approval, argatroban reduced the risk of composite outcome (new thrombosis, death due to thrombosis, amputation due to thrombosis) compared to historical controls.¹¹⁰ Due to hepatic clearance, argatroban has a limited role in patients with severe liver disease. Bivalirudin, has a shorter half-life of approximately 25 min in patients with normal renal function. This agent has been studied in detail as a non-heparin alternative for cardiopulmonary bypass for patients with acute/subacute HIT requiring emergent cardiac surgery^{111, 112} but is not commonly used for this indication at most medical centers due to high rates of bleeding.

Important considerations when using DTIs is the concept of anticoagulation confounding due to baseline elevations of the activated partial thromboplastin time (aPTT) or international normalized ratio (INR). Baseline elevation of aPTT or INR can be caused by DTIs or HIT-associated complications of disseminated intravascular coagulation and contribute to under-dosing of DTI therapy leading to subsequent treatment failure from subtherapeutic anticoagulation.¹¹³

There is increasing recognition that bleeding complications from use of alternative anticoagulants also contributes to disease morbidity in HIT. Indeed, recent studies indicate bleeding complications occur in 38–44% of patients treated with non-heparin anticoagulants. ^{1, 2} The prothrombotic environment of HIT does not mitigate bleeding risk, as several studies show that HIT patients appear as susceptible to bleeding complications as those without disease. In a large retrospective study of ~300 patients treated with alternative anticoagulation for suspected or confirmed HIT, bleeding rates were similar in patients with or without disease, while critical illness, renal dysfunction and platelet counts <25 × 10⁹/L thrombocytopenia were more predictive of bleeding risk. ²

Heparin-derivatives

LMWH is contraindicated in HIT due to high-rates of cross-reactivity with HIT antibodies.¹¹⁴ Fondaparinux, a synthetic pentasaccharide containing the heparin binding sequence of antithrombin, shows minimal cross-reactivity with HIT antibodies in vivo ¹¹⁵ and by itself, is a rare cause of HIT.¹¹⁶ Fondaparinux is often used off-label in patients with HIT, ¹¹⁷ but is of limited utility in patients with renal disease due to renal clearance and a long half-life (15–20 hours).¹¹⁸ Danaparoid is a heparinoid that comprises primarily of heparan sulfate, but is not available in the United States.¹¹⁹ In addition to its anticoagulant effect, the drug inhibits formation of PF4-heparin complexes which may additionally contribute to its efficacy.¹²⁰

Direct oral anticoagulants (DOACs)

DOACs are increasingly being used off-label for the treatment of HIT. Of these drugs, rivaroxaban has the most published experience. A prospective observational study by Linkins and colleagues on the use of rivaroxaban for treatment of serologically confirmed HIT, while encouraging, had to close prematurely due to low patient accrual.¹²¹ A recent literature review of published and post-trial experience of DOACs by the same Hamilton group indicate that these agents are safe and effective for use in acute HIT, with findings of no complications of bleeding and thrombosis occurring in only 1/46 (2.2%) patients treated with rivaroxaban.¹²² Comparable findings were noted in patients treated with apixaban and dabigatran. Based on this and similar reports in the literature, ¹²³ the American Society of Hematology guideline panel on HIT provided conditional recommendations for use of DOACs in acute HIT in clinically stable patients who are considered average risk for bleeding.¹⁰⁷

Non-anticoagulant therapies

Despite maximal anticoagulation, some HIT patients develop refractory disease, characterized by severe and persistent thrombocytopenia, new and/or progressive thrombosis. Some of these patients respond to additional therapies directed at the immune response through treatment with intravenous immunoglobulin (IVIG) ¹²⁴ or therapeutic plasma exchange (TPE). IVIG was first reported as an adjunctive therapy in 1989 for thrombocytopenia.¹²⁵ Subsequent reports have shown that IVIG disrupts platelet activation by HIT antibodies by interfering with FcγRIIA- dependent platelet activation ^{126, 127} and is effective for treatment of thrombosis ^{124, 127, 128} and/or refractory disease.^{127, 129, 130} While there are concerns about thrombotic risk with IVIG ^{131, 132}, a recent report utilizing a large inpatient health care database suggests that treatment may be safe.¹³³ TPE is another modality that is often employed as an adjunctive therapy for management of acute or subacute HIT, particularly for management of emergent cardiac surgery ¹³⁴ or as salvage therapy for refractory disease.¹³⁵ The efficacy of TPE in HIT is presumed to be secondary to removal of anti-PF4/H Abs, ¹³⁶ but drawbacks include only transient effects on antibody removal and limited availability in the community setting. Emerging therapies at the pre-clinical stage include bacterial proteases to cleave IgG (IdeS or IgG‐degrading enzyme of Streptococcus pyogenes ¹³⁷ Syk ⁷³ and tyrosine kinase inhibitors ⁶² as well as PF4 tetramerization inhibitors.^{138, 139}

Conclusion

Thrombosis in HIT represents an orchestrated multicellular response to HIT ULICs. HIT ULICs trigger cell-specific procoagulant responses that are mutually reinforcing to generate a profoundly hypercoagulable state. At present, the reasons why some seropositive patients develop this potent cellular response, while most patients do not, remain elusive. While some of these differences can be explained by genetic variation in various effector mechanisms, ^{45, 46, 78, 80} they do not fully account for heterogeneity of disease expression. Additional studies focused on less characterized aspects of the disease, such as complement activation, structural studies of antigenic complexes and characterization of pathogenic and non-pathogenic antibodies are needed. New insights into mechanisms of thrombosis in HIT should translate to an improved understanding of other immune-mediate thrombotic disorders, such as the anti-phospholipid antibody syndrome and/or systemic lupus erythematosus.

Highlights:

• Heparin induced thrombocytopenia (HIT) remains a clinically relevant medical problem due to continued use of unfractionated heparin in hospitalized patients.

• The last two decades have witnessed significant progress in our understanding of the clinical manifestations and biologic mechanisms underlying thrombotic complications in HIT.

• The profound hypercoagulable state in HIT is generated by cellular activating antibodies that initiate cell-specific prothrombotic responses.

• Treatment outcomes in HIT remain suboptimal as current therapies do not interfere with cellular activating effects of HIT antibodies.

ABBREVIATIONS

anti-PF4/H

Abs anti-platelet factor 4 antibodies

DOAC

direct oral anticoagulant

EC

endothelial cell

GAG

glycosaminoglycan

HIT

heparin induced thrombocytopenia

HUVEC

human umbilical vein cells

IL

interleukin

IVIG

intravenous immunoglobulin

NET

neutrophil extracellular traps

PF4

platelet factor 4

Syk

spleen tyrosine kinase

TF

tissue factor

ULIC

ultra-large immune complexes

vWFvon

willebrand factor