Andreas Greinacher, Institut für Immunologie und Transfusionsmedizin, Ernst-Moritz-Arndt-Universität Greifswald, Sauerbruchstr, 17487 Greifswald, Germany.
Tel.: +49 3834 865 482; fax: +49 3834 865 489.
Summary. Heparin-induced thrombocytopenia (HIT), typically occurring in the second week of heparin therapy, is an antibody-mediated adverse drug reaction associated with increased thrombotic risk. The most important antigens are located on platelet factor 4 (PF4)/heparin complexes. All HIT is caused by platelet-activating antibodies, but not all PF4/heparin-reactive antibodies cause HIT. Thus, tests have high negative, but only moderate, positive predictive value. Cessation of heparin requires substitution with an alternative anticoagulant, but as these drugs have increased bleeding risk, they should be used in therapeutic doses only if HIT is considered very likely. Avoiding/postponing coumarin is crucial in minimizing microthrombotic complications. Recent studies of HIT immunobiology suggest that HIT mimics immunity against repetitive antigens, as they are relevant in microbial defense. Thus, understanding HIT may help unravel why host defenses can trigger autoimmunity.
Heparin-induced thrombocytopenia (HIT) is an immunologic adverse effect of heparin therapy. Antibody-mediated platelet activation and consequent thrombin generation result in a fundamental paradox: despite thrombocytopenia induced by an anticoagulant, the major clinical effect in HIT is enhanced risk for venous and/or arterial thrombosis. Prompt diagnosis and introduction of alternative non-heparin anticoagulants are important for preventing further complications. Diagnosis of HIT can be problematic, particularly in patient populations with a high incidence of thrombocytopenia. So-called ‘HIT tests’ have a high negative predictive value but a rather low positive predictive value, and must be interpreted in the clinical context. Diagnosis should be based on both the clinical picture and detection of pathogenic HIT antibodies, that is, HIT is a clinicopathologic syndrome .
HIT usually features a platelet count fall >50% (from the highest value after day 4 of heparin treatment) and often new thrombosis, typically occurring 5–14 days after start of prophylactic or therapeutic dose heparin . Thromboembolic complications predominantly affect the venous system. Other rare complications include skin necrosis, adrenal hemorrhagic necrosis, or post-intravenous heparin bolus anaphylactoid reactions. The more unusual a new thrombosis during heparin treatment seems, the more HIT should be considered. HIT-associated thrombosis often occurs close to the 50% platelet count decrease and can even precede it by 1–2 days (∼30% of thrombi). In case of re-exposure to heparin within 1 (rarely up to 3) months, HIT can occur by day 1 of heparin treatment (rapid-onset HIT) when heparin-dependent antibodies are still present in the patient’s plasma. Delayed-onset HIT, which begins several days to a few weeks after heparin exposure, is caused by antibodies that activate platelets independently of heparin, mimicking an autoimmune disease. Rarely, HIT occurs in the absence of preceding heparin (‘spontaneous’ HIT) .
Most frequently, HIT antigens are found on PF4 bound to heparin or other sulfated polysaccharides , forming linear multimolecular clusters . Relative size, amount, and stability of the PF4/heparin complexes influence immunogenicity (UFH>LMWH∼fondaparinux). Anti-PF4/heparin antibodies bind to PF4 via their F(ab) domains. The pathogenic immunoglobulin (Ig) isotype in clinical HIT is IgG. As serial PF4 molecules become aligned, several IgG molecules bind, leading to formation of large immune complexes that cross-link the platelet FcγIIa receptors, triggering platelet aggregation  and an associated platelet procoagulant response. Thus, thrombocytopenia does not result from reticuloendothelial system phagocytosis, but rather from intravascular platelet activation , with release of procoagulant platelet-derived microparticles (catalytic surface for enhanced thrombin generation). The magnitude of thrombocytopenia correlates with thrombotic risk. Antibodies to other heparin-binding proteins (interleukin-8; neutrophil activating protein-2) have a minor role in HIT. HIT antibodies also bind to endothelial cells, leading to procoagulant changes on microvascular endothelial cells . Further, HIT antibodies can activate monocytes  and neutrophils, forming platelet—monocyte—neutrophil aggregates. These pancellular activating effects could contribute to the prothrombotic nature of HIT.
Diagnosis of HIT
Assays either measure antibody binding to PF4/heparin complexes or detect heparin-dependent platelet activation by patient serum. The classic activation test is the serotonin-release assay (SRA) , which–like the heparin-induced platelet activation (HIPA) test —uses washed platelets. Unlike standard platelet aggregometry, sensitivity of washed platelet activation assays for detecting clinically-relevant antibodies is as high (≈99%) as the polyspecific antigen assays (IgG/A/M) and the IgG-specific enzyme-immunoassay (EIA) . Therefore, laboratory testing is excellent for ruling out HIT (high negative predictive value). However, the diagnostic specificity of the assays ranges from ≈50–75% (polyspecific IgG/A/M EIA) to ≈55–90% (IgG-EIA) to 95–99% (washed platelet activation assays) . Washed platelet activation assays have the most favorable sensitivity/specificity tradeoff.
In all assays, the strength of the reaction correlates directly with the probability of clinical HIT [14–16]. Thus, HIT is more probable if the patient presents with a strong positive EIA (>1.0 optical density (OD)), but unlikely in patients with a weak positive result (0.4–1.0 OD). The particle gel immunoassay [15,17] reveals sensitivity and specificity intermediate between the washed platelet activation assays and the EIA.
Higher test specificity of the activation assays results in part from the inhibitory step (high heparin concentrations), which disrupts PF4/heparin complexes. This maneuver also enhances specificity of the EIAs . This is especially useful in patients with lupus or antiphospholipid syndrome in whom anti-PF4 (but not anti-PF4/heparin) antibodies of unclear biological relevance are often found .
The diagnosis of HIT requires the combination of pretest probability (clinical assessment) and a positive test for platelet-activating antibodies, or inferred from a strong-positive antigen test result. One tool providing standardized assessment of the clinical context is the 4Ts score. A low score (≤3 points) predicts a negative laboratory assay . Combining the clinical score with a laboratory test for HIT provides the highest predictivity for HIT .
A diagnostic algorithm for HIT
- 1Assess pretest probability for HIT (e.g., using the 4Ts). Patients with a low pretest probability (score ≤3) need no further testing and heparin can be maintained.
- 2If the EIA is negative, HIT is very unlikely and heparin can be maintained. A positive result in a screening EIA indicates the presence of anti-PF4/heparin antibodies. If an IgG EIA is weakly positive (OD <1.0), the antibodies are most likely non-platelet-activating. A confirmatory step using high heparin should be performed; if reactivity is not inhibited, HIT is very unlikely and heparin can be maintained.
- 3An IgG EIA (OD >1.0) indicates an increased risk for platelet-activating antibodies. These sera should ideally be assessed by a washed platelet activation assay. Demonstration of platelet-activating antibodies makes HIT very likely. A negative functional assay makes HIT unlikely and heparin can be maintained/restarted.
- 4Clinical reassessment should support a final confirmation or exclusion of the diagnosis. However, it is a misconception to automatically retest patients whose EIA is negative (as the EIA is positive even during the earliest phase of HIT) .
If there is high clinical suspicion for HIT, stopping heparin alone is insufficient. To prevent new thrombosis, non-heparin anticoagulant therapy is required. Vitamin K antagonists (VKA) must not be given in acute HIT (they can induce venous limb gangrene), and vitamin K should be given if HIT is recognized only after VKA treatment has been started .
Three drugs are approved for anticoagulation in HIT: the two direct thrombin inhibitors (DTIs), lepirudin and argatroban, and the heparinoid, danaparoid. Also, the DTI, bivalirudin, and the anti-factor Xa inhibitor, fondaparinux, are rational therapies for HIT .
All alternative anticoagulants confer significant risk for major bleeding (0.8–1.25% per treatment), and no antidote is available. In only 1/10 of the patients, clinically suspected to have HIT, is the diagnosis finally confirmed by demonstrating heparin-dependent, platelet-activating antibodies. (even fewer in ICU patients). Therefore, in patients with low/moderate clinical probability for HIT, our practise to reduce the risk of bleeding is to use prophylactic dose alternative anticoagulation, pending the results of laboratory testing.
Direct thrombin inhibitors
DTIs are usually monitored by APTT. However, a major problem is inappropriate dose reduction, when low prothrombin levels (e.g., because of VKA therapy) result in falsely elevated APTTs during DTI treatment . This often results in catastrophic outcomes. The ecarin chromogenic assay overcomes this issue by providing a linear dose-response curve for DTIs independent of prothrombin levels.
Lepirudin (recombinant, bivalent, irreversible DTI; t1/2∼90 min) approved dosing is too high; recommended dosing must be adjusted downwards even further in patients with renal dysfunction [22,24]. Lepirudin induces antibodies in about 40–70% (re-exposure) of patients, which may reduce elimination, thereby prolonging lepirudin half-life. IgG-mediated anaphylactic reactions are rare and avoidable by omitting the bolus.
Argatroban (synthetic, monovalent, reversible DTI, t1/2∼45 min) substantially increases the INR, complicating argatroban-VKA overlap. Functional clotting assays cannot be interpreted readily during argatroban treatment . In patients with reduced liver perfusion, 75–90% dose reduction to 0.2–0.5 μg kg−1 min−1 is important .
Bivalirudin (synthetic, bivalent, reversible DTI, t1/2∼25 min)  is cleaved by thrombin, which avoids major drug accumulation; the dose still needs ∼50% reduction in patients with renal and/or hepatic dysfunction. Experience with bivalirudin in HIT is anecdotal.
Indirect FXa inhibitors
Danaparoid is an AT-dependent heparinoid  with predominant anti-factor Xa activity (t1/2∼24 h). The dose–response relationship is predictable, and thus monitoring of therapeutic-dose treatment (anti-FXa activity) is required in patients with severely impaired renal function. In vitro cross-reactivity with HIT antibodies is of minor clinical relevance. Therapeutic dosing is indicated when HIT is strongly suspected or confirmed, with prophylactic dosing appropriate when suspicion for HIT is moderate.
Fondaparinux (AT-dependent FXa inhibitor; t1/2∼17 h, considerably prolonged in patients with renal impairment) may be used in patients with low or moderate suspicion of HIT. Experience in treatment of acute HIT with thrombosis with therapeutic-dose fondaparinux is anecdotal.
Most limb amputations in HIT result from microthrombosis due to severe DIC or (more often) VKA therapy. Thus, avoiding/postponing VKA use until platelet count recovery is a fundamental tenet of HIT management. Inappropriate DTI dose reductions/cessation due to VKA-related APTT prolongation is the rationale for giving vitamin K when HIT is recognized after VKA use .
Immunology of HIT
Currently, the most interesting aspect of HIT is its immunobiology. HIT neither exhibits features typical of a primary immune response (initial formation of IgM antibodies followed by a more delayed IgG response) nor the serological features of a secondary immune response (stronger and more persistent formation of IgG antibodies) . Most HIT patients form IgG antibodies between days 4 and 10, even with first heparin use. HIT antibodies do not persist, however . This profile seems more compatible with a non-T-cell dependent B-cell response, as described for immune reactions against viral antigens with repetitive epitopes . Indeed, repetitive epitopes in HIT are expressed as structures with a distance of 4–6 nm contained within 100–150 nm size, linear, ridge-like clusters of PF4/heparin . This is within the range of repetitive viral epitopes found to cause T-independent B-cell activation. However, other arguments favor a T-cell-dependent immune reaction in HIT. T-cell independent B-cell responses should be primarily IgM whereas in HIT IgG predominates, and in a mouse model, the immune response against PF4/heparin was T-cell dependent . This suggests that there has been previous contact(s) between the immune system and the ‘HIT antigens’.
Our current working hypothesis is that early exposure to PF4 complexes, perhaps induced by endogenous non-heparin factors, leads to PF4 clustering and a T-cell-dependent antibody class switch of B-cells. Later in life, these B-cells (possibly, marginal zone B-cells ) become again transiently activated when PF4 clusters are induced by heparin treatment, together with a proinflammatory milieu. This would explain why major surgery is a risk factor for HIT (through PF4 release and inflammation). Understanding the immunobiology of HIT may help explain why host defenses can result in autoimmunity.
Our current knowledge on HIT is based on important contributions of many researchers and clinicians. For those who do not find their work cited here, please accept my apologies for word count limitations permitting only a very limited number of references.
Disclosure of Conflict of Interests
A. Greinacher has given lectures sponsored by: Organon Int, Schering-Plough, Essex Pharma (drug: danaparoid), Mitsubishi Pharma (drug: argatroban), Glaxo-Smith-Kline (drug: argatroban, fondaparinux), Pharmion (drug: lepirudin). He has served as a consultant for: Organon Int, Schering-Plough, Essex Pharma, Mitsubishi Pharma, Pharmion, Bayer, Novartis.
He has received funding for research projects by Organon Int, Schering-Plough, Mitsubishi Pharma, Pharmion.