Portal de Periódicos da CAPES

Guidelines on the diagnosis and management of heparin‐induced thrombocytopenia: second edition

2012; Wiley; Volume: 159; Issue: 5 Linguagem: Inglês

10.1111/bjh.12059

1365-2141

Henry G. Watson, Simon Davidson, David Keeling,

Intramuscular injections and effects

The guideline was drafted by a writing group identified by the Haemostasis and Thrombosis Task Force of the British Committee for Standards in Haematology (BCSH). The 2006 guideline (Keeling et al, 2006) was reviewed along with additional information published since 2005. A search was performed of PubMed and Embase using the term 'heparin induced thrombocytopenia' combined with 'diagnosis', 'treatment' and 'clinical presentation'. The search covered articles published from January 2006 to April 2012. References in recent reviews were also examined. The writing group produced the draft guideline, which was subsequently revised by consensus by members of the Haemostasis and Thrombosis Task Force of the BCSH. The guideline was then reviewed by a sounding board of approximately 50 UK Haematologists, the BCSH, and the British Society for Haematology Committee and comments incorporated where appropriate. The 'GRADE' system was used to quote levels and grades of evidence, details of which can be found at: http://www.bcshguidelines.com/BCSH_PROCESS/EVIDENCE_LEVELS_AND_GRADES_OF_RECOMMENDATION/43_GRADE.html. The objective of this guideline is to provide healthcare professionals with clear guidance on the clinical features of heparin-induced thrombocytopenia (HIT), the indications for monitoring of patients on heparins for HIT, the investigation of suspected HIT and the treatment of HIT. The pathophysiology of HIT has been described in several reviews (Warkentin, 2003; Kelton, 2005; Greinacher et al, 2010). HIT is caused by the development of IgG antibodies directed against a complex of platelet factor 4 (PF4) and heparin. The antibodies primarily recognize a heparin-induced conformational change in the PF4 tetramers (Horsewood et al, 1996) which is affected by the chain length and degree of sulphation of the heparin. This partially explains the differences in incidence of HIT observed with different preparations. Theoretically, the optimal concentration of heparin to produce conditions that favour the development of HIT are thought to be associated with prophylactic rather than therapeutic doses of heparin. The IgG/PF4/heparin complexes bind to and activate platelets through their Fc receptors and may also generate thrombin by other actions (Qian et al, 2010) resulting in a prothrombotic condition that is associated with venous and arterial thrombosis. The frequency of HIT in different settings has been comprehensively reviewed (Lee & Warkentin, 2004; Linkins et al, 2012). It is important to distinguish between the frequency of antibody detection, antibody formation with thrombocytopenia (HIT), and HIT with thrombosis. The incidence of HIT is greater with bovine than with porcine heparin and for thromboprophylaxis is greater with unfractionated heparin (UFH) than with low molecular weight heparin (LMWH) (Martel et al, 2005). All heparins used in the United Kingdom are of porcine origin. The frequency of HIT is greater in surgical than medical patients. In trauma cases the severity of injury and the need for major surgery strongly affects the risk of developing HIT (Lubenow et al, 2010). In orthopaedic patients given subcutaneous prophylactic heparin, the incidence is approximately 5% with UFH and 0·5% with LMWH (Warkentin et al, 2000; Lee & Warkentin, 2004). We previously recommended platelet count monitoring in orthopaedic patients receiving LMWH thromboprophylaxis. This has become a significant issue with the move to extended thromboprophylaxis in hip and knee surgery. The American College of Chest Physicians (ACCP) recently made a 2C recommendation that platelet count monitoring be restricted to those where the risk is >1% (Linkins et al, 2012), whereas previously monitoring was recommended where the risk was >0·1% (Warkentin & Greinacher, 2004a; Keeling et al, 2006; Warkentin et al, 2008a); we agree with this approach. In medical patients given therapeutic porcine UFH the risk of HIT is approximately 0·7% (Lee & Warkentin, 2004) and in medical patients given subcutaneous UFH a rate of 0·8% was reported (Girolami et al, 2003). A study in medical patients given LMWH for prophylaxis or treatment reported an incidence of 0·8% (Prandoni et al, 2005). This was surprising given that, in a meta-analysis, LMWH had been found to carry a 10-fold lower risk than UFH (Martel et al, 2005), and while this analysis contained mostly orthopaedic studies, other studies in medical patients had shown a similar pattern (Lindhoff-Last et al, 2002; Pohl et al, 2005). The results reported by Prandoni et al (2005) was the principal reason our previous guideline (Keeling et al, 2006) recommended monitoring the platelet count in medical patients receiving LMWH in contrast to the 2004 ACCP guideline (Warkentin & Greinacher, 2004a). The 2008 ACCP guideline (Warkentin et al, 2008a) reconsidered this paper but at that time concluded that it overestimated the incidence of HIT and still did not recommend routine platelet count monitoring in medical patients receiving LMWH (Warkentin et al, 2008a). The 2012 ACCP guidelines (Linkins et al, 2012) do not recommend routine platelet count monitoring in medical patients receiving LMWH as the risk is under the new 1% threshold. This is a particularly important issue given the move towards higher rates of thromboprophylaxis for medical patients. Further, a recent study has suggested that higher rates of venous thromboembolism (VTE) prophylaxis do not increase the rate of HIT and that surveillance in patients on VTE prophylaxis may have a very low yield (Jenkins et al, 2011). A recent analysis of 25 653 medical in-patients found rates of ≤0·2% in patients on prophylactic LMWH, treatment dose LMWH, and prophylactic UFH, but 0·7% on treatment dose UFH (Kato et al, 2011). The risk of HIT is very low in obstetric patients given LMWH. A systematic review identified 2777 pregnancies in which LMWH was given (Greer & Nelson-Piercy, 2005). In the 2603 given LMWH as prophylaxis there were two cases of thrombocytopenia not thought to be related to heparin, and in the 174 given LMWH as treatment there was one case of thrombocytopenia also not thought to be related to heparin treatment. If HIT develops the platelet count typically begins to fall 5–10 d after starting heparin, although in patients who have received heparin in the previous 3 months it can have a rapid onset due to pre-existing antibodies. Occasionally, the onset can occur after more than 10 d of heparin exposure but it is rare after 15 d. In patients undergoing cardiopulmonary bypass a significant fall in platelet count is very common in the 72 h post-surgery (Nader et al, 1999). In these patients platelet recovery followed by a secondary fall in counts between post-operative days 5–14 is much more suspicious of HIT than a low count that persists beyond 4 d (Selleng et al, 2010). A very rare prothrombotic disorder characterized by thrombocytopenia that is similar to HIT, but occurs without heparin exposure has been described (Warkentin et al, 2008b). In HIT the platelet count normally falls by >50%; the median nadir is 55 × 109/l (Warkentin & Kelton, 2001; Warkentin, 2003). Severe thrombocytopenia (platelet count 1·5) have likelihood ratios of 20 and 10, respectively, for HIT post-cardiac surgery (Warkentin & Greinacher, 2004b). We recognize that, in routine clinical practice, most clinicians do not have access to platelet activation assays (HIPA and SRA). Ideally, the diagnosis of HIT should be confirmed by a washed-platelet assay but in reality, the vast majority of clinicians will manage the patient using pre-test probability assessment of HIT together with the results of an antigen assay (Nellen et al, 2012). In the United Kingdom, the alternative anticoagulants licensed for use in HIT are danaparoid and argatroban. Fondaparinux and bivalirudin have UK licences but not for this specific indication. Off-licence use of fondaparinux is becoming widespread and will be considered. The use of bivalirudin will be considered for the specific cases of percutaneous coronary intervention (PCI) and cardio-pulmonary bypass (CPB). The main principle of treatment is that patients with a high suspicion of, or proven, HIT discontinue UFH or LMWH and commence treatment with an alternative non-cross reacting anticoagulant. The initial anticoagulant treatment of HIT should be the same whether or not it is already complicated by thrombosis at the time of diagnosis. LMWH is not an appropriate alternative if HIT develops during treatment with UFH because there is cross-reactivity in vivo in approximately 50% of cases. Argatroban and bivalirudin are both non-cross reacting. Danaparoid demonstrates cross reactivity in vitro (Pouplard et al, 1997) which is only rarely evidenced in vivo (Keng & Chong, 2001) while fondaparinux is highly immunogenic but is not well recognized by anti-fondaparinux-PF4 antibodies generated during exposure, suggesting that it should be associated with a low risk of developing HIT (Warkentin et al, 2005b). Warfarin, especially when used in isolation, can increase the risk of microvascular thrombosis in HIT and its introduction should be delayed until there has been substantial resolution of the thrombocytopenia. It should then be introduced with overlap of the alternative anticoagulant (Warkentin et al, 1997; Smythe et al, 2002). Where argatroban is being used care is required in the interpretation of the International Normalized Ratio (INR). Bleeding is uncommon in HIT. Uneventful and efficacious platelet transfusion has been documented in a series of four patients with suspected HIT based on good clinical and laboratory evidence (Hopkins & Goldfinger, 2008). There is some residual concern that platelet transfusions could theoretically contribute to thrombotic risk (Greinacher & Warkentin, 2004). Based on this, it is reasonable to consider platelet transfusion for patients with HIT and bleeding but prophylactic platelet transfusion is generally not advised. Whichever alternative anticoagulant is used, it is important to administer it in appropriate therapeutic doses as discussed below, as there is evidence for treatment failure in cases where doses deemed appropriate for prophylaxis in other circumstances have been used in active HIT. This pertains to all cases whether or not they are complicated by thrombosis at the time of diagnosis. The evidence for this is the high failure rate of a prophylactic dose of danaparoid (750 u b.d. or t.i.d.) in comparison to dose adjusted lepirudin, or higher ('therapeutic') doses of danaparoid (2500 u bolus followed by continuous infusion) in the Heparin Associated Thrombocytopenia (HAT) studies (Farner et al, 2001). Major bleeding commonly complicates the treatment of HIT with an alternative anticoagulant (Greinacher et al, 2000). Clinical decision-making should address the likely risks and benefits of the available treatment strategies. Probably because of depletion of the natural anticoagulants proteins C and S, vitamin K antagonists (VKAs) can worsen the prothrombotic state in HIT. In view of this, it is suggested that VKAs should be discontinued and reversed at the diagnosis of HIT and that warfarin is only restarted after the platelet count has risen into the normal range and then using low dose rather than high dose initiation regimens (Linkins et al, 2012). Danaparoid is a heparinoid composed of heparan sulphate, dermatan sulphate and chondroitin sulphate. It indirectly inhibits Xa and, to a lesser degree, thrombin. It has a predictable dose response and a long half-life of approximately 24 h. Danaparoid does not prolong the prothrombin time (PT) and has a minimal effect on the activated partial thromboplastin time (APTT), which cannot be used to monitor it. If monitoring is required, a specific anti-Xa assay calibrated for danaparoid should be used. The chromogenic anti-Xa assay is not affected by factors that may affect the APTT, such as lupus anticoagulant or warfarin. Monitoring may be of value only in patients with severe renal impairment and body weight >90 kg (Farner et al, 2001). Danaparoid is approved in the European Union for use in two distinct dosing regimens. Published data report use of a low dose ('prophylactic') regimen of 750 anti-Xa units b.d. or t.i.d. subcutaneously and a higher dose ('therapeutic') regimen, which consists of a bolus injection followed by a reducing dose continuous infusion (bolus determined by weight 1250–3750 units i.v. followed by 400 u/h for 4 h then 300 u/h for 4 h then 200 or 150 u/h as a maintenance dose). Two small studies of 40 patients (Tardy-Poncet et al, 1999; Schenk et al, 2003) reported favourable outcomes using 600–800 anti-Xa units b.d. or 10 u/kg b.d. However, larger studies showed that low dose danaparoid regimens are associated with a higher rate of new thrombotic events than therapeutic doses of lepirudin or danaparoid (Farner et al, 2001). Patients with HIT complicated by thrombosis were given a full dose regimen while those with HIT without thrombosis were given lower doses. Efficacy data on 294 patients (danaparoid 126) showed that at 42 d there were no differences between treatments for the composite end point of death, amputation or new thrombosis. There was a non-significant increase in new thrombotic events in patients given danaparoid at low doses compared to full dose danaparoid or dose-adjusted lepirudin. Patients treated with low dose danaparoid were significantly more likely to reach the combined endpoint than those treated with lepirudin (P = 0·02). These data suggest that low dose danaparoid is insufficient treatment in patients with active HIT. On the other hand, full dose danaparoid appears equivalent to dose-adjusted lepirudin at preventing new thrombosis. In a randomized study of 42 patients, danaparoid was significantly superior to dextran (Chong et al, 2001). Argatroban is a direct thrombin inhibitor that is administered intravenously. The key feature that makes it attractive in the management of HIT is its hepatic metabolism in a condition that is often complicated by established or developing renal impairment. The data describing its role in HIT are two non-randomized open-label studies (Lewis et al, 2001, 2003) where it was compared with historical controls, most often treated by discontinuation of heparin along with oral anticoagulation using a coumarin. The quality of these studies was further compromised by the fact that around a third of the patients included in the analysis were found to be HIT antibody negative on retrospective testing (Walenga et al, 1999) and by the fact that some of the patients included had a remote rather than an immediate history of HIT. The combined data from these studies describe the outcomes for 882 patients with HIT, of whom 697 were treated with argatroban (1·7–2·0 μg/kg/min for 5–7 d) to achieve an APTT ratio of 1·5–3·0, compared with 185 historical controls (Lewis et al, 2006). Argatroban treatment resulted in a significant reduction in the primary end point of a composite of death due to thrombosis, amputation secondary to HIT-associated thrombosis, or new thrombosis within 37 d of baseline for both patients with HIT without thrombosis at diagnosis [Hazard Ratio (HR), 0·33; 95% CI 0·2–0·54, P < 0·001] and with thrombosis at diagnosis (HR, 0·39; 95% CI 0·25–0·62, P < 0·001). More argatroban-treated patients remained thrombosis-free during the 37-d follow-up, again for patients both with and without thrombosis at the time of diagnosis and fewer died from thrombosis (P ≤ 0·001). Major bleeding, defined by a fall in Hb of ≥20 g/l, or that led to transfusion of ≥2 units of blood or that was into the central nervous system, retroperitoneum or a prosthetic joint was similar in both groups with no significant excess in the argatroban recipients. There has been discussion about the efficacy of argatroban in preventing amputation in HIT patients. Comparing amputation rates in patients treated with argatroban and lepirudin with controls suggests that, while lepirudin reduces amputation rates, argatroban has little benefit compared with controls; relative risk (RR) 0·7 for lepirudin and 1·26 for argatroban (Warkentin et al, 2008a). It has been suggested that the effect of argatroban on the INR may result in premature discontinuation of argatroban (Bartholomew & Hursting, 2005). Alternatively, it has been suggested that, in the argatroban studies, severe ischaemic changes or gangrene were already established prior to the introduction of therapy so that these should not really be considered treatment failures (Lewis et al, 2006). Argatroban requires no dose adjustment in renal failure but it is contraindicated in severe hepatic failure and expert opinion suggests dose adjustment in critically ill patients in the intensive care setting (Alatri et al, 2012). Monitoring of argatroban therapy is most easily performed using an APTT test. The target range quoted in the summary of product characteristics (SmPC) is that used in the two multicentre studies (Lewis et al, 2001, 2003) an APTT ratio of 1·5–3·0 but not exceeding 100 s. A consensus meeting suggested each laboratory should generate its own dose response calibration curve though failed to recommend what argatroban concentration should be targeted (Alatri et al, 2012) but a Scientific Sub-Committee of the International Society for Thrombosis and Haemostasis communication found that the APTT ratios were similar when comparing seven different reagents (Gray & Harenberg, 2005). For otherwise uncomplicated patients, standard initial dosing is with 2 μg/kg per min as a continuous infusion with dose adjustment based on the APTT. Clinical experience has resulted in advice for dose reduction in critically ill patients and the SmPC suggests an initial dose of 0·5 μg/kg per min. Dosing schedules based on clinical scoring systems for evaluation of critically ill patients, such as Acute Physiology and Chronic Health Evaluation (APACHE) II, Sequential Organ Failure Assessment (SOFA) and Simplified Acute Physiologic Score (SAPS) II, have been proposed (Alatri et al, 2012) and a simplified dosing schedule for this group of patients is given in Table 3. 90 kg–3750 u Argatroban Standard dose in routine patients without liver failure APTT ratio 1·5–3·0 APTT repeated within 2 h of any dose adjustment and at least once daily APTT ratio 1·5–3·0 APTT repeated within 4 h of any dose adjustment and at least once daily Argatroban causes prolongation of the PT and this needs to be considered in the transition of patients to warfarin therapy. Warfarin and argatroban should be overlapped for at least 5 d and an INR of ≥4 should be observed for two consecutive days before argatroban is discontinued. An upper range target for the INR in this situation is not given but at very high INR levels the patient may be over-anticoagulated. We suggest that, at an INR > 5, the argatroban infusion should be discontinued for 4 h and the INR repeated. Including two recent reports (Goldfarb & Blostein, 2011; Warkentin et al, 2011) there are six case series totalling 71 patients demonstrating that fondaparinux is not only an effective anticoagulant in the setting of HIT, but it appears to have a low risk of overall complications (Greinacher, 2011), Combining all 71 patients reported in these cohorts, no new thrombotic events occurred after initiating treatment with fondaparinux (95% CI, 0–5·1%), which looks promising that fondaparinux can provide effective anticoagulation in patients with HIT (Greinacher, 2011). The dosing of these patients was variable – some patients were given prophylactic doses of fondaparinux (2·5 mg OD) whilst others were given daily therapeutic doses dependent on their weight ( 100 kg; 10 mg). Based on the inferior outcomes associated with the use of prophylactic doses of danaparoid, we would suggest that therapeutic doses should be given with consideration of age and renal function. HIT is uncommon in pregnancy and, in particular, the rates of HIT in patients receiving LWWH are so low that routine monitoring of this population is not indicated (Greer & Nelson-Piercy, 2005). There are few data on the diagnostic process in pregnancy and so a general clinical approach using a scoring system, such as 4Ts, combined with laboratory testing as indicated above seems reasonable. In strongly suspected HIT and in proven HIT, heparin exposure should be discontinued and an alternative anticoagulant started. There are data on the use of danaparoid, argatroban and fondaparinux in HIT in pregnancy. The largest number of reports is on the use of danaparoid (Lindhoff-Last et al, 2005). In 51 pregnancies given danaparoid for heparin intolerance or HIT (n = 32), 37 healthy infants were delivered in mothers given danaparoid up to term, and danaparoid was discontinued in a further 14 pregnancies prior to delivery for a variety of reasons not neccesarily related to danaparoid treatment. There were four maternal bleeding events during pregnancy; two of these, which were fatal, were due to documented placental problems. There were three fetal deaths which were not attributable to danaparoid. There are a small number of case reports documenting the use of argatroban in pregnacy (Young et al, 2008; Ekbatani et al, 2010; Tanimura et al, 2012). In two of these cases, argatroban was used in combination with fondaparinux with successful pregnacy outcomes (Ekbatani et al, 2010; Tanimura et al, 2012), and in another, argatroban was used continuously for 6 weeks, again with a good pregnancy outcome (Young et al, 2008). The option for subcutaneous injection favours the use of danaparoid and fondaparinux, especially in situations where prolonged anticoagulation is required; there are encouraging data using the latter in pregnancy (Knol et al, 2010). For patients with HIT and thrombosis we would regard HIT as a transient reversible risk factor and recommend anticoagulation with warfarin for 3 months (Keeling et al, 2011; Kearon et al, 2012). For isolated HIT not complicated by thrombosis we recommend therapeutic anticoagulation for 4 weeks to cover the main period of thrombosis risk, as suggested from observational studies (Warkentin & Kelton, 1996; Arepally & Ortel, 2006). Although recurrence is rare, where a patient with previous HIT requires a period of anticoagulation or anticoagulant prophylaxis an alternative to UFH or LMWH should be prescribed. Fondaparinux and danaparoid may be used, as may new anticoagulants such as dabigatran, rivaroxaban and apixaban, depending on the clinical circumstances, e.g., dabigatran, rivaroxaban and apixaban may be used as per licensed indications, such as orthopaedic surgery. Danaparoid and argatroban have both been used (Fischer, 2004). Suitable regimens for the use of both drugs in renal replacement therapy are given in Table 4. In cardiac surgery, the depth of experience with UFH, the established near-patient monitoring, and the rapid reversal indicate that its use should be considered. There is therefore a rationale and some data that support the safe use of UFH in patients with previous HIT. Firstly, in patients who develop typical HIT, there is no relationship between the day of onset and previous heparin exposure. Further, in patients with rapid onset HIT, there is an association with recent heparin exposure (previous 100 d) but not with more remote heparin exposure. Finally, HIT antibodies are transient with a median time to disappearance of 50–80 d. These data suggest that the antibodies that mediate HIT are transient, that there is no anamnestic immune response in HIT and that acute onset HIT represents recurrence due to renewed heparin exposure. There are reports of successful heparin re-exposure to permit cardiac and vascular surgery in patients with previous HIT (Potzsch et al, 2000; Warkentin & Kelton, 2001; Nuttall et al, 2003). In patients with recent or current HIT who require cardiac surgery the risk associated with further heparin exposure is probably much greater and therefore it should be avoided if possible. Several strategies, some including the use of UFH offset by the use of an anti-platelet agent, such as tirofiban or epoprostenol, have been reported (Koster et al, 2000a,b, 2001; Mertzlufft et al, 2000; Aouifi et al, 2001). The number of patients included in these reports is small and the experience confined to very few centres. The 2012 ACCP guideline (Linkins et al, 2012) favours the use of bivalirudin for cases of HIT where early cardiac surgery is required, primarily based on the results of two prospective cohort studies assessing bivalirudin in off-pump and on-pump cardiac surgery (Dyke et al, 2007; Koster et al, 2007). Amongst 100 (51 off-pump and 49 on-pump) patients successful clinical outcomes defined by absence of death, Q-wave myocardial infarction, repeat revascularization surgery, and stroke were observed in 88% and 86% respectively of patients at day 30. The largest series of lepirudin use in this context reported thrombosis-free survival in 54 of 57 (95%) patients (Koster et al, 2000b). Excessive blood loss and slow drug elimination was seen in the four patients with pre-existing renal failure but there were no haemorrhagic deaths. In 53 patients managed using a fixed dose danaparoid regimen severe post-operative bleeding occurred in 21% of patients. In addition clots were seen in the operative field in a third of patients (Magnani et al, 1997). There are published protocols for the use of lepirudin, bivalirudin and danaparoid in cardiac surgery (Warkentin & Greinacher, 2003; Poetzsch & Madlener, 2004; Warkentin & Koster, 2005). Appropriate bivalirudin concentrations for anticoagulation in this setting have been established and these can be monitored by a validated activated clotting time (ACT) measurement. If the postoperative period is complicated by renal failure, problems with the prolonged half-life of the drugs and the absence of an antidote may emerge. There is extensive experience of the use of bivalirudin for percutaneous coronary intervention (PCI) in the UK and it is licensed for use in patients requiring PCI who do not have HIT (Mahaffey et al, 2003). While the advice and information in these guidelines is believed to be true and accurate at the time of going to press, neither the authors, the British Society for Haematology nor the publishers accept any legal responsibility for the content of these guidelines.