Heparin-induced thrombocytopenia : a clinical and economic review

Introduction Heparin-induced thrombocytopenia is a serious immune-mediated, drug adverse effect. Research over the last 20 years has vastly expanded knowledge, awareness and management of this condition. This review includes findings mostly from primary source clinical and economic literature from this time period to synthesise a concise summary. We have highlighted key points regarding the pathophysiology, epidemiology and presentation, diagnosis and management, as well as the economic impact of heparin-induced thrombocytopenia. It is established that heparininduced thrombocytopenia is relatively uncommon overall, despite the widespread use of heparin. However, delayed recognition leads to significantly increased morbidity, mortality and disability. Treatment requires prolonged courses of expensive alternative anticoagulants. All of these impose substantial costs to patients, hospitals and society. The aim of this review was to assess the clinical and economic factors influencing heparin-induced thrombocytopenia. Conclusion Heparin-induced thrombocytopenia remains a significant clinical problem and is associated with morbidity, mortality and disability. Using heparin products can reduce (but not eliminate) the incidence of heparininduced thrombocytopenia. Properly administered treatment with factor Xa inhibitors or direct thrombin inhibitors can significantly decrease the direst complications of this condition and mitigate its economic impact. Introduction Heparin-induced thrombocytopenia (HIT) is a prothrombotic complication mediated by auto antibodies against heparin-platelet complexes. It causes prolonged morbidity, is fatal in 5%–10% of cases and causes lifelong disability in another 10% of cases1. Over the last two decades in particular, research has advanced our understanding, diagnosis and management of HIT. Nonetheless, the incidence of HIT is unlikely to decline because heparins remain a key therapy for thromboembolic and ischemic disorders. Besides, heparins are recommended and are widely used for prophylaxis of venous thromboembolism (VTE)2,3. Increasing adherence with this recommendation can be expected because it is tracked as a marker of quality improvement with potential reimbursement implications, given the resource constraints in healthcare. Not surprisingly, a growing body of literature, over the last decade, has begun to highlight the economic impact of HIT. This review draws largely on original primary source publications. We distilled this literature for the most salient findings in order to present a consolidated clinical and economic review of HIT. Discussion Epidemiology Up to 8% of patients receive heparins form HIT antibodies4. However, less than 10% of these patients develop HIT and an even smaller proportion experience symptomatic thrombosis5. Similarly, less than 1% of patients in the intensive care unit (ICU) develop HIT although multifactorial thrombocytopenia is common (30%–50%) in this setting6,7. These may be occurring due to platelet and antibody polymorphisms among other, yet undefined, patient factors8. Nonetheless, certain risk factors and specific patient populations at risk have been defined5,7: 1) HIT is far more likely with unfractionated heparins (UFH) than with low molecular weight heparins (LMWHs)9; 2) there is increased risk of HIT with bovine UFH (relative to porcine UFH), therapeutic versus prophylactic doses and prolonged duration of therapy of >5 days and 3) there is a 2:1 female to male gender predilection and an increased incidence in elderly and postsurgical (relative to medical) patients10,11. Cardiovascular or orthopaedic surgery patients are peculiar, in that, 30%–50% develop HIT antibodies although clinical HIT occurs in less than 3% and 5% patients, respectively12. HIT is uncommon in uremic, paediatric and obstetric patients. Interestingly, other risk factors for hypercoagulability such as cancer, diabetes, lupus, antiphospholipid syndrome, infection and trauma, are present in greater than 60% of patients with HIT13. M ed ic in e * Corresponding author E-mail: udehc@ccf.org 1 Center for Critical Care, Department of Cardiothoracic Anesthesia, Cleveland Clinic, Cleveland, Ohio 2 Department of Cardiothoracic Anesthesia, Anesthesiology Institute, Cleveland Clinic, Cleveland, Ohio 3 Department of Outcomes Research, Anesthesiology Institute, Cleveland Clinic, Cleveland, Ohio 4 Center for Critical Care, Department of Outcomes Research, Anesthesiology Institute, Cleveland Clinic, Cleveland, Ohio


Introduction
Heparin-induced thrombocytopenia (HIT) is a prothrombotic complication mediated by auto antibodies against heparin-platelet complexes.It causes prolonged morbidity, is fatal in 5%-10% of cases and causes lifelong disability in another 10% of cases 1 .Over the last two decades in particular, research has advanced our understanding, diagnosis and management of HIT.Nonetheless, the incidence of HIT is unlikely to decline because heparins remain a key therapy for thromboembolic and ischemic disorders.Besides, heparins are recommended and are widely used for prophylaxis of venous thromboembolism (VTE) 2,3 .Increasing adherence with this recommendation can be expected because it is tracked as a marker of quality improvement with potential reimbursement implications, given the resource constraints in healthcare.Not surprisingly, a growing body of literature, over the last decade, has begun to highlight the economic impact of HIT.This review draws largely on original primary source publications.We distilled this literature for the most salient findings in order to present a consolidated clinical and economic review of HIT.

Discussion
Epidemiology Up to 8% of patients receive heparins form HIT antibodies 4 .However, less than 10% of these patients develop HIT and an even smaller proportion experience symptomatic thrombosis 5 .Similarly, less than 1% of patients in the intensive care unit (ICU) develop HIT although multifactorial thrombocytopenia is common (30%-50%) in this setting 6,7 .These may be occurring due to platelet and antibody polymorphisms among other, yet undefined, patient factors 8 .Nonetheless, certain risk factors and specific patient populations at risk have been defined 5,7 : 1) HIT is far more likely with unfractionated heparins (UFH) than with low molecular weight heparins (LMWHs) 9 ; 2) there is increased risk of HIT with bovine UFH (relative to porcine UFH), therapeutic versus prophylactic doses and prolonged duration of therapy of >5 days and 3) there is a 2:1 female to male gender predilection and an increased incidence in elderly and postsurgical (relative to medical) patients 10,11 .Cardiovascular or orthopaedic surgery patients are peculiar, in that, 30%-50% develop HIT antibodies although clinical HIT occurs in less than 3% and 5% patients, respectively 12 .HIT is uncommon in uremic, paediatric and obstetric patients.Interestingly, other risk factors for hypercoagulability such as cancer, diabetes, lupus, antiphospholipid syndrome, infection and trauma, are present in greater than 60% of patients with HIT 13 .

Pathophysiology
Two types of HIT have been described: HIT type 1 and HIT type 2. HIT type 1 is now often called heparin-associated thrombocytopenia.It is non-immune mediated, has minimal risk of thrombosis and is seen in 10%-30% of patients receiving heparin.It is caused by a direct aggregating effect of heparin on platelets manifesting in mild thrombocytopenia within a few days of exposure.Platelet count rarely drops to <100 × 10 9 /L and rebounds within a few days after discontinuing all heparin 14 .HIT type 2 (the focus of the rest of this review) is seen in 1%-3% of exposed patients 15 .It is a true immunologic disorder associated with a risk of arterial and venous thromboses 20-40 times to that seen in control populations 16 .This high risk persists for weeks even after the platelet count normalises 1 .A complex immune response underlies HIT type 2. It begins with heparin's strong affinity for platelet factor 4 (PF4), a chemokine, stored in the α-granules of the platelets.The affinity of various heparin preparations to PF4 depends on the degree of sulphation, chain length and molecular weight of the heparin product.These molecular differences probably explain the increased incidence of HIT with UFH as compared with LMWHs 17 .
Ordinarily, heparin and PF4 are devoid of immunogenicity but when bound, the resultant macromolecule is antigenic and attaches to the surface of activated platelets.The mechanism for antigenicity appears to involve a conformational change in platelet surface-bound PF4 which then stimulates immunoglobulin G (IgG) production 18 .The avidity of IgG binding to the neoantigen is optimal when there is about a 1:1 ratio of PF4 to heparin molecules 19 .This may, in part, explain the relatively low (<3%) incidence of clinical HIT despite the presence of heparin-PF4 antibodies in 20%-50% of cardiac surgery patients 20 .It has been postulated that surface-bound PF4 levels are depleted by excess concentrations of UFH used during cardiopulmonary bypass, leading to suboptimal conditions for neoantigen formation and subsequent antibody binding 18 .
IgG binding to the platelet-surface bound heparin-PF4 neoantigen creates a self-amplifying process by inducing intense platelet activation and the release of procoagulant factors [21][22][23] .Further, the immune complexes interact with monocytes and endothelial cells causing endothelial damage and tissue factor expression.Platelet activation is a pivotal feature of clinical HIT and the resulting cascades lead to intravascular thrombogenesis (Figure 1).The immune complexes are cleared by the reticuloendothelial system causing moderate thrombocytopenia, but spontaneous bleeding is not a feature of HIT.As long as heparin is withdrawn, the antibody response is extinguished over time with antibodies undetectable after 3-4 months 24 .

Presentation
Three patterns have been described in the presentation of HIT-rapid, delayed or typical.In 25% of patients, a rapid onset characterised by an abrupt decrease in platelets within a few hours of heparin exposure, is seen 6 .This form is mostly attributable to a prior immunising heparin exposure in the preceding 100 days with residual circulating levels of HIT antibodies.Although uncommon, delayed-onset HIT can occur after 40 days but is not always associated with thrombocytopenia 25 .Thrombocytopenia occurring 5-10 days after exposure to heparin is the typical pattern for about 65% of patients without prior heparin exposure or whose last exposure was >100 days prior.After surgery with cardiopulmonary bypass, patients experience a transient consumptive and dilutional thrombocytopenia which typically starts to recover by postoperative day four.Therefore in these patients, thrombocytopenia that persists or a recurrent decline observed after four days should raise suspicion for HIT.
Approximately 50%-75% of patients will eventually develop a thrombosis if untreated, and indeed, a thrombotic event is recognized before thrombocytopenia in 20% of cases.Risk for thrombosis positively correlates with the severity of thrombocytopenia which in turn predicts a worse prognosis 11 .Development of thrombosis increases mortality to 20%-30% 26 .Thrombotic complications include venous (17%-55%), arterial (3%-10%) and even intracardiac thrombi, with a higher likelihood of occurrence in a localised vascular injury site 1,[27][28][29] .Interestingly, venous thromboses predominate in the medical and orthopaedic populations while arterial and venous thromboses probably occur in equal frequencies in cardiac and vascular surgery patients 11 .Thrombosis could manifest as critical limb and organ ischemia, strokes, dural sinus thrombosis, pulmonary embolism and myocardial infarction.Nonetheless, the commonest site is in the lower extremities with upper extremity thrombosis being commonly associated with central venous catheters 30,31 .Rarely, adrenal vein thrombosis occurs, presenting with fever, abdominal pain and hypotension.Such patients require urgent steroid supplementation.Necrotic skin lesions at heparin injection sites may occur, but this is uncommon.

Diagnosis
Prompt diagnosis of HIT is important in view of the risk of severe morbidity and mortality if untreated-there is a 5% daily risk of thrombosis if untreated 32 .Given this urgency, plus the slow turnaround of laboratory testing, the diagnosis of HIT is initially primarily clinical.HIT should be suspected whenever thrombocytopenia and/or thrombosis occur in the context of a temporal relationship to heparin exposure.Suggestive clinical criteria have been defined: unexplained thrombocytopenia up to <50% of the baseline that occurs 4-10 days after starting heparin, skin necrosis at heparin injection sites and anaphylactoid reactions following an intravenous heparin bolus 33,34 .However, the picture is often confounded by the concurrence of multiple other reasons for thrombocytopenia.Additionally, a positive diagnosis has major implications that include initiation of a prolonged expensive course of alternative anticoagulation, attendant risks of bleeding and treatment of any thrombotic complications.Therefore, confirmatory tests are necessary.
Available tests fall into two categories-immunological and functional assays 35 .The immunoassays detect the presence of the heparin-PF4 complex antibody 36,37 .Examples include the PF4-enzyme-linked immunosorbent assay (ELISA) and particle gel immunoassays 38 .The PF4ELISA assay utilises photometric analysis at 405 nm as a quantita-tive indicator of HIT antibodies.An optical density (OD) >0.4 is considered positive-the higher the OD, the more strongly positive the result.These immunoassays are widely available, relatively quick and highly sensitive (90%-97%) 39 .However, specificity (<80%) and positive predictive values (50%) are poor due to cross-reactivity with other antibodies 40 .Consequently, a strongly positive OD (>2) does not necessarily indicate the presence of plateletactiva ting IgG antibodies.This creates potential for inappropriate therapy, particularly if the clinical picture is uncertain and OD weakly positive (0.4-1.0).
Therefore, general and diseasespecific clinical prediction models have been developed to define the pretest probability of HIT and aid the decision to proceed with testing and interpreting of results [41][42][43] .The 4T test has been validated and is widely used; it assigns points for clinical characteristics that include thrombocytopenia, timing, thrombosis and other causes for thrombocytopenia (Table 1).A score of ≤3 indicates a low pretest probability of HIT [0%-3% chance of HIT] and lab testing is not indicated or recommended.A score of 4-5 indicates an intermediate risk, while a score ≥6 indicates a high risk of HIT.
Functional assays measure the ability of circulating heparin-PF4  All authors contributed to the concep on, design, and prepara on of the manuscript, as well as read and approved the fi nal manuscript.
All authors abide by the Associa on for Medical Ethics (AME) ethical rules of disclosure.
antibodies to aggregate or to activate normal platelets; they are highly specific, technically demanding, time-consuming and require selected donor platelets 38 .As such, they are mostly limited to reference laboratories for use in confirming positive immunoassays.Examples include heparin platelet aggregation test, flow cytometry, heparin-induced platelet activation and serotonin release assay (SRA) 44,45 .SRA, the gold standard test, is highly specific and sensitive for HIT 45 .It is complex and expensive because it utilises radioactive isotopes.First, platelets from donors, previously shown to be reactive to a panel of heparin-PF4 antibodies, are incubated for uptake of 14 C serotonin.Next, they are washed to remove free serotonin.Then, they are mixed with patient serum before being exposed to therapeutic and supra-therapeutic concentrations of heparin.Platelet activation is measured by the percentage of serotonin radio-isotope released.The test is positive if at least 20% is released in the presence of therapeutic heparin levels (0.1 U/ml) with substantially reduced release in the presence of supra-therapeutic (100 U/ml) heparin concentrations.The test is indeterminate if there is >20% release with high heparin concentrations.
The complexity involved in making a certain diagnosis of HIT, as discussed, underscores why HIT has been described as a clinicopathological syndrome.Clinical suspicion and symptoms must be interpreted in context of subsequent laboratory findings and vice versa.It is often helpful, if not imperative, to involve a haematologist in the management of these patients.

Prevention and management
Prevention of HIT is hampered by the lack of inexpensive and easily reversible alternatives to heparins.Consequently, if clinically appropriate for prophylaxis of VTE, limiting the duration of the use of UFH or substituting LMWHs for UFH, particularly in female postsurgical patients, are  1 Other clinical HIT prediction models may also be used. 2Any locally available immunoassay may be used for initial testing. 3Any locally available platelet activation/ aggregation test may be used for confirmation. 4If SRA is indeterminate or at any other juncture, consider consultation with a haematologist.the only two practical options available for reducing the incidence of HIT at present.Although there is such a paucity of data, the direct factor Xa inhibitor-fondaparinux may be considered as an option, since just like LMWHs, it is administered by a subcutaneous injection, once or twice daily 46,47 .Close monitoring of platelet count in the highest risk patients facilitates prompt diagnosis and early treatment which helps in limiting morbidity.Fortunately, more options are now available and a more widely accepted approach to management has now been established (Figure 2).Once HIT is suspected or diagnosed, all sources of heparin must be immediately discontinued and the patient must be evaluated for new thromboses or necrosis.Platelet transfusion for thrombocytopenia is not recommended due to concerns about precipitating a thrombotic complication.Platelets can be, and have been, safely used for active bleeding or invasive procedures.Merely discontinuing heparin is insufficient, given the ongoing risk of new thromboses 27 .Therefore, unless contraindicated (such as by active bleeding), systemic anticoagulation with an alternative agent is imperative.
Several direct factor Xa inhibitors and direct thrombin inhibitors (DTIs) have been introduced and approved for anticoagulation in HIT.Danaparoid is approved for HIT but is unavailable in the USA.Lepirudin is no longer being manufactured.Bivalirudin and argatroban are currently approved for use in HIT, and there is growing evidence supporting the use of Fondaparinux 46,47 .These are all parenteral medications and the role, if any, for the newer oral anticoagulants-rivaroxaban, apixaban and dabigatran, remains undefined.Bivalirudin (short halflife and predominant proteolytic metabolism) is US Food and Drug Administration-approved for use in cardiopulmonary bypass in patients with active or suspected HIT.After adequate systemic anticoagulation is achieved, transition to long-term anticoagulation with warfarin is started.Some crucial points need emphasis here.Warfarin is started in low doses and only after the platelet count has recovered to normal.This is to reduce the risk of early hypercoagulability caused by depletion of protein C and avoid precipitating venous limb gangrene or skin necrosis.Also, DTIs can prolong the International normalised ratio (INR), so there should be at least a five day overlap with warfarin to ensure that a therapeutic INR is achieved before the DTIs are discontinued.Duration of anticoagulation is at least three months (if no thrombotic complications occur) or six months (if thromboses do occur).

Economic considerations
The economic impact of HIT is ultimately borne by society and it encompasses the additional acute costs imposed on hospitals and third party payers for its diagnosis and management.Total hospital costs are increased when patients develop HIT and even more so if heparininduced thrombotic thrombocytopenia (HITT) ensues [48][49][50] .Patients, at the very least, require longer hospital stay for parenteral administration of alternative anticoagulation as well as management of any thrombotic complications.Among medical patients in the US who developed HIT, hospital costs were reported higher by an average of about US $41,000 compared to those without HIT 49 .Likewise, European researchers found a mean incremental cost of €3500-€9000 with higher costs in surgical patients than in medical patients, and the cost drivers were extended hospital stays and cost of alternate anticoagulants 50,51 .Similarly, a small US case-control study found that HIT patients had a two week increase in length of stay, incurring an average loss of $14,000-$20,000 per patient, depending on the payer 52 .These additional costs carry greater impact when hospital reimbursement is based on diagnosis related groups (DRG) as with US Medicare patients.In such systems, a global payment is made per DRG, and specific items such as laboratory, medications and diagnostic procedures are not separately reimbursable.Medicare does make additional payments for certain complications, but the sums are small and rarely cover their true cost.Leykum et al. estimated additional DRG reimbursements for HIT at $820.05 and for HITT at $1749.98 53 .
Given these extra costs, researchers have performed economic analyses of the prevention and management of VTE, considering the different costs and incidence of HIT with different drugs [54][55][56] .The most commonly compared have been UFH and LMWH because the unit cost of UFH is far cheaper, but it has a much higher incidence of HIT as compared with LMWH.Does the reduced incidence of HIT and differences in other complication rates justify using the more expensive LMWH?One study addressed this concern in a retrospective analysis of over 10,000 patients who received either UFH or LMWH 49 .The incidence of HIT was higher for UFH at 0.51% as compared with 0.084% for LMWH, and costs with LMWH were $13.88 less per patient as compared with UFH.In another clinical study, researchers analysed data from a prospectively collected registry of 349 patients with serologically confirmed HIT 48 .They found no difference in length of stay, but overall hospital costs for patients exposed to UFH were about double the costs for those exposed to LMWH ($113,100 versus $56,325).
However, these cost comparisons of actual clinical practice were limited by the variables that were analysed.For example, neither UFH nor LMWH could be compared to no prophylaxis, because use of prophylaxis is considered standard of care nor were the costs considered in a Licensee OA Publishing London 2013.Creative Commons Attribution Licence (CC-BY) mortality or quality of life context.These and other limitations can be overcome with the use of a decision analytical model.One such study analysed the cost effectiveness of enoxaparin versus UFH in a cost per quality adjusted life year context (QALY) 53 .QALY's are used to consider both the quantity and quality of the years of life gained by an intervention.A year in full health is given a value of 1, death a value of 0 and alternate health states are assigned values in between.Enoxaparin was found to be the cost-effective option with savings of $4550.17/QALY.
Another analysis was a three way comparison of enoxaparin, UFH and no prophylaxis with an averted death being the outcome measure 57 .Enoxaparin was also found to be the cost-effective option, completely dominating UFH which was both more costly and less effective.The incremental cost per death averted between enoxaparin and no prophylaxis was only $9100, a very compelling result supporting prophylaxis.These data support the use of enoxaparin for VTE prophylaxis.It is noteworthy that similar cohort simulation models of enoxaparin versus fondaparinux for VTE prophylaxis, in major orthopaedic surgery of the lower extremities, have found fondaparinux to be both more effective and cost saving [58][59][60] .
The discussion so far only covers cases of confirmed HIT.However, suspected HIT is far more common and involves real costs to an institution regardless of subsequent HIT confirmation.Whereas, thrombocytopenia can occur for many reasons, and clinicians now maintain such a high index of suspicion for HIT that testing with the widely available immunoassays is common practice.One study estimated that HIT is over-diagnosed by as much as 100% using such tests 61,62 .This risks overtreatment because a positive ELISA would prompt initiation of anticoagulation with a DTI or direct factor Xa inhibitors before confirmatory functional assays are completed.In 2006, the daily cost of therapy with different DTIs, based on average wholesale price, ranged from $533-$1749 63 .This daily cost would then be incurred until confirmatory test results are available which could take as long as one to two weeks if the tests are not performed locally.Thus, the cost of testing is compounded by the high acquisition costs of DTIs plus the risk of bleeding complications 64 .It is therefore crucial to ensure appropriate use of DTIslimiting them to those with an intermediate or high clinical probability of HIT.Protocols to support this have been tested and shown to be costeffective by avoiding unnecessary drug costs, averting the cost of DTIassociated bleeding and ensuring prompt cessation of treatment after negative confirmatory testing 65 .
There are also acute costs to the patients of mortality and prolonged morbidity, including costs of longterm anticoagulation as well as chronic costs of disability from strokes and limb amputation.These substantial costs also ultimately accrue to society.For instance, it has been estimated that for the first two years after trauma-related amputation, mean healthcare costs in 2002 were about $91,000 with an estimated lifetime cost of about $509,000 66 .There is about a 10% rate of amputation among HIT patients who develop thrombotic complications 64 .Even considering this in the context of the US alone gives an indication of the enormity of the costs involved.For instance, there are over 5 million ICU admissions annually in the US with an estimated 0.3%-0.5% incidence of HIT 6,7,67 .Roughly extrapolating the HIT-related amputation rate to just this group would result in well over 1000 attributable amputations with consequent costs of several tens of millions of dollars annually.

Conclusion
Given the rising healthcare costs, aging population and resource constraints, costly and potentially avertable conditions are a high priority target.This review provides evidence to support that HIT remains a significant clinical problem, and it is associated with prolonged and expensive morbidity, mortality and disability.Judicious use of heparin products can reduce (but not eliminate) the incidence of HIT.Vigilance and selective use of testing will ensure prompt diagnosis.Appropriately directed treatment with factor Xa inhibitors or direct thrombin inhibitors can significantly decrease the direst complications and mitigate the economic impact.Prophylaxis of venous thromboembolism using LMWHs or fondaparinux is costeffective; however, acquisition costs and the costs of treating any associated complications must be considered while choosing the prophylactic agent.

Review Page 4 of 9
Compe ng interests: none declared.Confl ict of interests: none declared.

Figure 2 :
Figure 2: Algorithm for diagnosis and management of HIT.1 Other clinical HIT prediction models may also be used.2Any locally available immunoassay may be used for initial testing.3Any locally available platelet activation/ aggregation test may be used for confirmation.4If SRA is indeterminate or at any other juncture, consider consultation with a haematologist.