Bivalirudin for the treatment of patients with confirmed or suspected heparin-induced thrombocytopenia

Authors


  • Manuscript handled by: I. Pabinger
  • Final decision: F. R. Rosendaal, 14 April 2014

Summary

Background

Heparin-induced thrombocytopenia (HIT) is an adverse immune-mediated response to unfractionated heparin and, less commonly, low molecular weight heparin. It is associated with a high thrombotic risk and the potential for limb and life-threatening complications. Argatroban is the only approved and currently available anticoagulant for HIT treatment in the USA.

Objectives

To report safety and efficacy outcomes with bivalirudin for HIT treatment.

Methods

We retrospectively examined records from our registry of patients with a suspected, confirmed or previous history of HIT and who had received bivalirudin for anticoagulation in a single tertiary-care center over a 9-year period.

Results

We identified 461 patients who received bivalirudin: 220 (47.7%) were surgical patients, and 241 (52.3%) were medical patients. Of this population, 107 (23.2%) were critically ill, and 109 (23.6%) were dialysis-dependent. Suspected, confirmed and previous history of HIT were reported in 262, 124 and 75 patients, respectively. Of 386 patients with suspected or confirmed HIT, 223 patients (57.8%) had thrombosis at HIT diagnosis. New thrombosis was identified in 21 patients (4.6%) while they were on treatment with therapeutic doses of bivalirudin. No patient required HIT-related amputation. Major bleeding occurred in 35 patients (7.6%). We found a significant increase in major bleeding risk in the critically ill population (13.1%; odds ratio 2.4, 95% confidence interval 1.2–4.9, P = 0.014). The 30-day all-cause mortality rate was 14.5% (67 patients), and eight of 67 (1.7%) deaths were HIT-related.

Conclusion

Bivalirudin may be an effective and safe alternative option for the treatment of both suspected and confirmed HIT, and appears to reduce the rate of HIT-related amputation.

Introduction

Heparin-induced thrombocytopenia (HIT) is a highly prothrombotic, potentially lethal adverse immunogenic reaction to unfractionated heparin (UFH) or low molecular weight heparin (LMWH) [1]. It is triggered by the release of procoagulant microparticles from platelet α-granules resulting from activation of platelets by means of an interaction between their FcγIIa receptors and immune complexes consisting of heparin-dependent IgG antibodies [2]. These antibodies are bound to multimolecular heparin–platelet factor 4 (PF4) complexes. Rapidly acting non-heparin anticoagulants that inhibit thrombin or its generation constitute the mainstay for treatment of HIT [2].

Because the production of the direct thrombin inhibitor (DTI) lepirudin was discontinued in 2012, and the heparinoid danaparoid is not available in the USA, argatroban is the only available Food and Drug Administration (FDA)-approved anticoagulant for the prevention and treatment of HIT. Argatroban, a hepatically cleared DTI, has a half-life of 40–50 min, and requires dose reduction both in patients with hepatic impairment and in the critically ill [3]. It also has a significant effect on the prothrombin time (PT)/International Normalized Ratio (INR), which makes bridging to warfarin more complex [4]. Two drugs used off-label for the treatment or prevention of HIT in the USA include the DTI bivalirudin and the indirect factor Xa inhibitor fondaparinux.

Bivalirudin has US FDA approval as an alternative anticoagulant for patients undergoing percutaneous coronary intervention (PCI), and is recommended over argatroban in patients with HIT by the American College of Chest Physicians (ACCP) for this indication [5]. It is theoretically an attractive anticoagulant for HIT therapy, given its favorable pharmacologic profile, which includes: predominantly proteolytic elimination (only 20% renal clearance), immediate action, lack of the need for an initial bolus, a low risk for immunogenicity, a short half-life of ~ 25 min, the potential for rapid dose titration, and minimal interference with the PT/INR [5]. We have used bivalirudin almost exclusively at our tertiary-care institution as an alternative to argatroban since 2002, and present our 9-year experience demonstrating it to be a safe and effective alternative treatment for HIT.

Materials and methods

Study design and patients

We examined 461 consecutive patient records identified from the Department of Pharmacy Registry of all hospitalized patients who received bivalirudin for a suspected, confirmed or previous history of HIT at the Cleveland Clinic from 1 January 2002 to 31 December 2010. Exclusion criteria included bivalirudin for indications other than HIT. We performed a retrospective review of the patients' demographic, clinical, laboratory and radiologic characteristics and outcomes for up to 30 days following bivalirudin treatment, utilizing the method of Feinstein et al., and recorded the data on a standardized form with REDCap (Research Electronic Data Capture) electronic data capture tools, a secure, web-based application, hosted at our institution [6, 7]. The study was approved by our institutional review board.

Management of HIT

At our institution, acute HIT was suspected on the basis of the presence of well-defined clinical features as described by King, Kelton, and Warkentin, and confirmed by vascular medicine consultative services [8-10]. According to our protocol, venous duplex ultrasound scans of both upper and lower limbs were obtained for most patients with suspected or confirmed HIT. Pretest probability scores were calculated by use of one of the following clinical scoring systems: Lillo-Le Louet et al., Lo et al., or Cuker et al. [11-13]. Serologic testing for PF4–heparin antibodies was obtained, UFH (or LMWH) was discontinued, and an alternative anticoagulant was started in patients with an intermediate to high clinical suspicion of HIT. Functional assays (serotonin release assay [SRA] and/or heparin-induced platelet aggregation [HIPA] tests) were performed in most patients with a high clinical suspicion and/or positive immunogenic assay result. Bivalirudin was initiated at doses of 0.05–0.1 mg kg−1 h−1 and titrated upwards or downwards by 10–20% increments to obtain an activated partial thromboplastin time (APTT) target of 1.5–2.5 times the patient's baseline. The starting dose was further reduced in the setting of renal insufficiency. If HIT was excluded, we stopped bivalirudin and reintroduced UFH or LMWH. If HIT was confirmed or considered to be indeterminate, on the basis of laboratory testing or clinical suspicion, we transitioned patients to warfarin after platelet recovery by using bivalirudin bridging. In addition, bivalirudin was administered to hospitalized patients with a previous history of HIT who required parenteral anticoagulation.

Laboratory testing for HIT

We used two types of assay to aid in the diagnosis of HIT: immunogenic and functional.

Immunogenic assays utilized anti-PF4–heparin ELISA, and were performed in all patients with suspected HIT. From 2002 until 2005, we used the GTI PF4 polyspecific kit (currently Lifecodes PF4 enhanced ELISA; Lifecodes, Waukesha, WI, USA), and from 2005 on, the GenProbe Lifecodes PF4 IgG ELISA kit. Tests showing an OD value of ≥ 0.400 and inhibition of a positive reaction by ≥ 50% in the presence of excess heparin were interpreted as positive in our laboratory.

Functional assays measure platelet activation in the patient's serum in the presence of heparin. Most patients with a high clinical suspicion for HIT or a positive immunogenic assay result underwent either HIPA or SRA, or both. HIPA is an in-house test that uses donor platelets. Aggregation of normal platelets in the patient's plasma assessed by turbidimetry at 0.3 U mL−1 and 0.5 U mL−1 heparin is interpreted as positive. The SRA was sent to a centralized outside referral laboratory (Blood Center of Wisconsin, Milwaukee, WI, USA), and ≥ 20% release of serotonin with low-dose heparin plus < 20% release in the presence of a high concentration of heparin was considered to be positive.

Data collection

We abstracted age, gender, body mass index (BMI), length of hospitalization, clinical setting, concomitant illnesses, dialysis dependence, concomitant/recent use of other antithrombotic agents, UFH or LMWH exposure details (type, dosage, duration, and indication) and laboratory values, including renal and liver function, estimated creatinine clearance (Cockcroft–Gault formula), platelet counts (baseline, nadir, and recovery time) and immunogenic and functional assay results for HIT, from the medical records. We recorded patients' outcomes, including thrombosis, bleeding, amputation, and death. For patients with suspected HIT, we retrospectively calculated the 4Ts score (thrombocytopenia, timing, new thrombosis or alternative diagnosis) [12]. We obtained available data for follow-up from the index hospitalization up to 30 days following the diagnosis and treatment of HIT.

Definitions

We defined confirmed HIT according to the ISTH Platelet Immunology SSC Working Group recommendation as the presence of both a positive immunogenic assay result and functional assay result (either SRA or HIPA) in a patient with suspected acute HIT [14]. A patient with acute HIT suspected but not confirmed according to the above definition was classified into the suspected HIT group. A thrombotic event was defined as any new venous or arterial thrombus that developed during or after UFH or LMWH exposure, in a patient with clinical suspicion of HIT. New thrombosis (while the patient was on bivalirudin) was defined as any different venous or arterial thrombosis or progression of a pre-existing thrombosis that occurred after the patient had received 48 h of therapeutic doses of bivalirudin. Objective confirmation by means of the available corresponding imaging study was required in either case.

We classified bleeding according to the modified ISTH criteria [15-17]. We defined major bleeding as overt bleeding associated with a fall in the hemoglobin level of ≥ 2 g dL−1, or leading to a transfusion of two or more units of packed red blood cells or whole blood, or bleeding that occurred in a critical site – intracranial, intraspinal, intraocular, pericardial, intra-articular, intramuscular with compartment syndrome, or retroperitoneal – or surgical site bleeding that required a second intervention, or contributed to death. Clinically relevant non-major bleeding was defined as overt bleeding not meeting the criteria for major bleeding, but associated with a medical intervention, an unscheduled contact with a physician, (temporary) cessation of anticoagulation, or discomfort for the patient, such as pain or impairment of activities of their daily life [15, 16].

We defined as critically ill any patient who required intensive care treatment for ≥ 24 h during the index hospital stay. Chronic renal failure was defined as a glomerular filtration rate of < 60 mL min−1 1.73 m–2 for > 3 months. Chronic liver disease was defined as the presence of biochemical, radiologic and/or histologic features of cirrhosis.

Statistical analysis

We examined each variable for normal distribution, presented continuous variables as mean ± standard deviation or as the median (25th and 75th percentiles) when appropriate, and categorical variables as proportions. For continuous variables, we used Student's t-test in the case of bi-level categorical independent variables, and one-way anova in the the case of multi-level categorical independent variables. Data were analyzed with the chi-square test for categorical variables. Variables with a skewed distribution were logarithm-transformed before the analysis, and then retransformed to their natural units to represent them, and were compared by use of the Mann–Whitney U-test and the Kruskal–Wallis H-test. spss for Windows version 20.0 (SPSS, Chicago, IL, USA) was used for the analysis. All reported P-values were two-sided, and a P-value of < 0.05 was considered to be statistically significant.

Results

Among 461 patients included in our study, 241 (52.3%) were medical and 220 (47.7%) were surgical. Among the patients, 107 (23.2%) were critically ill, and 109 (23.6%) were dialysis-dependent. Of these, 75 (16.3%) had a previous history of HIT, 262 (56.8%) had suspected HIT, and 124 (26.9%) had confirmed HIT (Table 1; see Table S1 for additional details).

Table 1. Demographic and clinical characteristics of patients
CharacteristicsAll patients (n = 461)Previous history of HIT (n = 75)Suspected HIT (n = 262)Confirmed HIT (n = 124)P-value for comparison between HIT groups
  1. CG GFR, Cockcroft–Gault formula-based glomerular filtration rate; HIT, heparin-induced thrombocytopenia. Continuous variables are presented as mean ± standard deviation or as the median (25th and 75th percentiles).

Age (years)62.5 ± 14.561.3 ± 14.561.9 ± 14.864.4 ± 13.60.226
Males, n (%)262 (56.8)43 (57.3)147 (56.1)72 (58.1)0.932
Body mass index (kg m−2)30.0 ± 8.030.6 ± 8.229.7 ± 8.330.1 ± 7.10.767
Clinical setting, n (%)
Intensive care107 (23.2)14 (18.7)67 (25.6)26 (21.0)0.361
Medical241 (52.3)40 (53.3)140 (53.4)61 (49.2)0.724
Surgical220 (47.7)35 (46.7)122 (46.6)63 (50.8)0.724
Comorbidities, n (%)
Active cancer53 (11.5)5 (6.7)35 (13.3)13 (10.5)0.254
Chronic liver disease35 (7.6)6 (8.0)20 (7.6)9 (7.3)0.981
Chronic renal failure124 (26.9)18 (24.0)74 (28.2)32 (25.8)0.727
Known thrombophilic condition27 (5.9)11 (14.7)12 (4.6)4 (3.2)0.002
Dialysis dependence109 (23.6)11 (14.7)69 (26.3)29 (23.4)0.230
Patients treated with an alternative anticoagulant before bivalirudin initiation (fondaparinux, lepirudin, or argatroban)31 (6.7)5 (6.7)17 (6.5)9 (7.3)
Baseline laboratory parameters
Platelet count on admission (1000 μL−1)186.0 (135.0, 243.0)197.5 (137.5, 276.0)185.5 (132.8, 248.0)179.0 (138.0, 231.0)0.287
Serum creatinine (mg dL−1)1.1 (0.8, 1.9)1.0 (0.8, 2.0)1.2 (0.8, 1.8)1.2 (0.9, 2.0)0.125
CG GFR (mL min−1 1.73 m–2)55.2 (31.2, 86.5)55.7 (32.0, 96.6)57.1 (31.4, 90.0)52.0 (29.2, 73.3)0.383

The clinical and serologic profiles of patients with suspected or confirmed HIT (n = 386) are shown in Tables 2 and 3, and Fig. 1 (see Table S2 for additional details). The most common heparin exposure was intravenous UFH (272 patients, 70.5%). The most common indication for UFH or LMWH was antithrombotic prophylaxis (153 patients, 39.6%).

Table 2. Clinical profile of patients with patients with suspected or confirmed heparin-induced thrombocytopenia (HIT)
CharacteristicsSuspected or confirmed HIT (n = 386)Confirmed HIT (n = 124)Suspected HIT (n = 262)P-value for comparison between suspected and confirmed HIT groups
  1. ACS, acute coronary syndrome; DVT, deep venous thrombosis; i.v., intravenous; LMWH, low molecular weight heparin; PE, pulmonary embolism; s.c., subcutaneous; SVT, superficial vein thrombosis; UFH, unfractionated heparin. Continuous variables are presented as mean ± standard deviation or as the median (25th and 75th percentiles).

Duration of heparin exposure (days)7.0 (4.0, 12.0)7.0 (3.0, 10.0)7.0 (4.0, 13.0)0.345
Type of heparin use and route of administration, n (%)
UFH i.v.272 (70.5)74 (59.7)198 (75.6)0.001
UFH s.c.138 (35.8)52 (41.9)86 (32.8)0.081
LMWH s.c.50 (13.0)24 (19.4)26 (9.9)0.010
UFH for cardiopulmonary bypass or during surgery123 (31.9)44 (35.5)79 (30.2)0.294
Indication for heparin exposure, n (%)
DVT74 (19.2)22 (17.7)52 (19.8)0.624
PE30 (7.8)9 (7.3)21 (8.0)0.795
ACS46 (11.9)15 (12.1)31 (11.8)0.940
Atrial fibrillation/flutter67 (17.4)22 (17.7)45 (17.2)0.891
Antithrombotic prophylaxis153 (39.6)54 (43.5)99 (37.8)0.280
Thromboembolic complications at the time of diagnosis of HIT, n (%)223 (57.8)87 (70.2)136 (51.9)0.001
Venous thromboembolic events, n (%)213 (55.2)85 (68.5)128 (48.9)0.207
DVT182 (47.2)73 (58.9)109 (41.6)0.002
Proximal lower extremity DVT69 (17.9)27 (21.8)42 (16.0)0.169
Distal lower extremity DVT73 (18.9)33 (26.6)40 (15.3)0.008
Upper extremity DVT106 (27.5)41 (33.1)65 (24.8)0.090
PE30 (7.8)13 (10.5)17 (6.5)0.171
SVT76 (19.7)25 (20.2)51 (19.5)0.872
Visceral7 (1.8)3 (2.4)4 (1.5)0.539
Arterial thrombotic events, n (%)38 (9.8)15 (12.1)23 (8.8)0.878
Table 3. Heparin-induced thrombocytopenia (HIT) laboratory and pretest probability scoring details
CharacteristicsSuspected or confirmed HIT (n = 386)Confirmed HIT (n = 124)Suspected HIT (n = 262)P-value for comparison between suspected and confirmed HIT groups
  1. HIPA, heparin-induced platelet aggregation; PF4, platelet factor 4. Continuous variables are presented as mean ± standard deviation or as the median (25th and 75th percentiles).

Platelet count nadir (1000 μL−1)60.0 (36.0, 88.0)51.0 (31.0, 79.0)65.0 (41.0, 92.0)0.003
Anti-PF4 ELISA (OD)1.3 ± 1.12.2 ± 1.10.8 ± 0.80.000
≥ 0.4, n (%)293 (75.9)123 (99.2)170 (64.9)0.000
0.4–1.0, n (%)105 (27.2)21 (16.9)84 (32.1)
≥ 1.0, n (%)188 (48.7)102 (82.2)86 (32.8)
Serotonin release assay, n (%)
Positive74 (19.2)74 (59.7)
Indeterminate9 (2.3)9 (3.4)
HIPA, n (%)
Positive71 (18.4)71 (57.3)
Indeterminate36 (9.3)15 (12.1)21 (8.0)
4Ts score5.0 (4.0, 6.0)
Low probability (0–3), n (%)27 (10.3)
Intermediate probability (4–5), n (%)145(55.3)
High probability (6–8), n (%)90 (34.4)
Figure 1.

Flow chart depicting laboratory testing and the 4Ts score details of the different heparin-induced thrombocytopenia (HIT) groups. *One patient with HIT-related thrombosis was added to the confirmed HIT group as an exception, because the serotonin release assay (SRA) was positive three times (94–95% and 0–1% release with low-dose and high-dose heparin, respectively, and 99% and 0% release with low-dose and high-dose low molecular weight heparin, respectively), though anti-platelet factor 4 levels were negative three times (maximum 0.383 OD). †SRA negative and no heparin-induced platelet aggregation (HIPA) (n = 21), HIPA negative and no SRA (n = 120), HIPA indeterminate and no SRA (n = 10), both SRA and HIPA negative (n = 72), SRA indeterminate and HIPA negative (n = 6), SRA negative and HIPA indeterminate (n = 8), SRA and HIPA indeterminate (n = 3), and no functional assay (n = 22). ‡HIPA negative (n = 35), HIPA intermediate (n = 15), and no HIPA (n = 3). §SRA negative (n = 2), and no SRA (n = 48).

Of the 223 patients (57.8%) who developed thrombotic events, 213 (55.2%) had venous thromboembolism (VTE), 38 (9.8%) arterial thrombosis, and four (1.0%) skin necrosis. The upper extremity was the most common location for deep vein thrombosis (DVT) (106 patients, 27.5%). The frequency of thrombotic events was significantly higher in patients with confirmed HIT than in those with suspected HIT (70.2% vs. 51.9%; P = 0.001).

Patients were treated with bivalirudin for a median of 9 days (range, 5–15 days), and achieved a therapeutic APTT within a median period of 12 h (range, 5–23.5 h). The platelet count recovered to > 150 000 μL−1 after a median of 4 days of therapy (range, 2–7 days) in 225 (58.3%) suspected or confirmed HIT patients. The platelet count did not recover to > 150 000 μL−1 during bivalirudin therapy in 153 patients (39.6%; confirmed HIT [40 patients, 10.4%]), and the remaining eight patients did not have platelet count nadirs below 150 000 μL−1. The outcomes of bivalirudin treatment are summarized in Table 4 (see Table S3 for additional details on bleeding events).

Table 4. Treatment results and outcomes
CharacteristicsAll patients (n = 461)Previous history of HIT (n = 75)Suspected HIT (n = 262)Confirmed HIT (n = 124)P-value for comparison between HIT groups
  1. HIT, heparin-induced thrombocytopenia.

Bleeding, n (%)46 (10.0)10 (13.3)29 (11.1)7 (5.6)0.134
Major bleeding35 (7.6)7 (9.3)22 (8.4)6 (4.8)0.386
Fatal bleeding7 (1.5)2 (2.7)5 (1.9)0 (0)0.242
Non-major clinically relevant bleeding11 (2.4)3 (4.0)7 (2.7)1 (0.8)0.323
New thrombosis, n (%)21 (4.6)1 (1.3)11 (4.2)9 (7.3)0.139
Amputation, n (%)0 (0)0 (0)0 (0)
Transition to warfarin, n (%)271 (58.8)38 (50.7)139 (53.1)94 (75.8)0.000
All-cause 30-day mortality, n (%)67 (14.5)10 (13.3)45 (17.2)12 (9.7)0.138
Death resulting from HIT-related thrombosis, n (%)1 (0.2)1 (0.8)

There were 46 (10%) clinically relevant bleeding events related to bivalirudin: 35 were major, including seven fatal, and 11 were non-major. New thrombosis was documented in 21 (4.6%) of the patients while they were receiving therapeutic doses of bivalirudin. There were no HIT-related amputations. During the 30-day follow-up, there were 67 all-cause deaths (14.5%), and eight (1.7%) were HIT-related. One death resulted from pulmonary embolism, and seven were resulted from bleeding related to bivalirudin therapy (two intracranial hemorrhages; two gastrointestinal bleeds; and three diffuse alveolar hemorrhages).

More than half of the patients (271 of 461, 59%) were changed to warfarin. The proportion of patients with a change to warfarin was significantly higher in the confirmed HIT group than among patients with suspected HIT or with a previous HIT history. The reasons for not changing the treatment of 190 patients were as follows: exclusion of HIT (68 patients), death (57 patients), move to a hospice (five patients), high bleeding risk (22 patients), change to fondaparinux (five patients), refusal (one patient), and unclear reasons (32 patients).

We examined our data to evaluate for predictors (chosen on the basis of on clinical significance) of treatment outcomes (bleeding events, new thrombosis/HIT-related amputation, and 30-day all-cause mortality) as summarized in Table 5. There were no significant differences in treatment outcomes in patients by age, BMI, gender, or the presence of renal insufficiency, or across various HIT groups (see Table S4 for additional details). Patients in the critical care setting had a significantly higher frequency of major bleeding events (odds ratio [OR] 2.4, 95% confidence interval [CI] 1.2–4.9, P = 0.014) and significantly higher 30-day mortality rates (OR 1.9, 95% CI 1.1–3.4, P = 0.02). A lower nadir of platelet counts was significantly associated with all-cause 30-day mortality and major bleeding events.

Table 5. Predictors of the outcomes in patients with confirmed or suspected heparin-induced thrombocytopenia treated with bivalirudin
  N Major bleeding eventNew thrombosis30-day mortality
Yes P Yes P Yes P
  1. Continuous variables are presented as mean ± standard deviation. Categorical variables are presented as n (%).

Critical care setting
Yes46114 (13.1)0.0144 (3.7)0.6423 (21.7)0.02
No21 (5.9)17 (4.8)44 (12.6)
Dialysis dependence
Yes46113 (11.9)0.067 (6.4)0.2132 (29.9)0.000
No19 (6.2)11 (3.6)33 (10.9)
Chronic liver dysfunction
Yes4613 (8.6)0.823 (8.6)0.248 (22.9)0.16
No32 (7.5)18 (4.1)59 (14.0)
Platelet count nadir38650.6 ± 24.6(vs. 67.6 ± 44.2 in the no-event group)0.04957.8 ± 33.4 (vs. 66.8 ± 43.7 in the no-event group)0.3655.7 ± 34.8 (vs. 68.4 ± 44.3 in the no-event group)0.047

Discussion

We have used bivalirudin almost exclusively to treat patients with a suspected, confirmed or previous history of HIT. Our study from a single center demonstrates that bivalirudin may be an effective and safe alternative anticoagulant for treating HIT (with and without associated thrombosis).

The 2012 ACCP guidelines recommend non-heparin anticoagulants (lepirudin, argatroban, or danaparoid) for the treatment and/or prevention of HIT [18]. That document reviewed the results of pooled analyses of prospective cohort studies comparing lepirudin and argatroban with historical controls, and found that both agents were more effective at preventing new thrombosis in patients with HIT with thrombosis (HITT) and those with isolated HIT [18-23]. Argatroban was also found to significantly reduce the frequency of death resulting from thrombosis. Neither agent reduced the risk of limb amputation, but both led to an increased risk of major bleeding in patients with HITT (lepirudin, 15.4%; argatroban, 8%) and in those with isolated HIT (lepirudin, 14%; argatroban, 4%), and fatal bleeding events occurred in 1.2% and 0.4% of patients who received lepirudin and argatroban, respectively [18]. The authors raised concerns that the argatroban trials might have overestimated its efficacy, as laboratory confirmation of HIT was not required. The ACCP guidelines also suggested that danaparoid is efficacious in the treatment of HIT; however, it is not available in the USA [24, 26]. The guidelines concluded that there are no high-quality prospective head-to-head comparative trials to establish the superiority of any of these agents, and that there is little evidence supporting the use of other agents, including bivalirudin, in HIT.

Despite the several attributes of bivalirudin that make it an ideal agent in the treatment of HIT, and its increased familiarity among cardiovascular physicians because of its widespread use in PCI, there are only a few retrospective studies that have evaluated the role of bivalirudin in the treatment of HIT. Kiser et al. [27] studied safety, efficacy and dosing requirements in a retrospective study of 37 patients with suspected or proven HIT. Their new thrombotic rate was 3% (1/37), and the clinically significant bleeding rate was 5% (2/37). In a single-center retrospective study of 138 patients with suspected or known HIT treated with bivalirudin (92 patients) or argatroban (46 patients), Skrupky et al. [28] found that both agents were similar in achieving therapeutic anticoagulation, and preventing new thromboembolic events (bivalirudin, 8%; argatroban, 4%) and bleeding events (bivalirudin, 9%; argatroban, 11%). Dang et al. [29] concluded, in their study of 42 patients with suspected or proven HIT treated with bivalirudin (24 patients), argatroban (13 patients), or lepirudin (five patients), that bivalirudin achieved faster therapeutic anticoagulation than either argatroban or lepirudin (8.5 h vs. 14 h and 24 h, respectively), although the differences were not statistically significant. They also found that argatroban had longer treatment durations (mean days: argatroban, 15.7; bivalirudin, 7.1; lepirudin, 5.2), and the lepirudin group had lower bleeding rates (lepirudin, 20%; argatroban, 62%; bivalirudin, 58%). Those studies support the concept that bivalirudin is a potential alternative anticoagulant for the treatment of HIT. However, the small sample size and lack of uniform diagnostic criteria for HIT in those studies are major limitations.

To our knowledge, our study is the largest series evaluating the safety and efficacy of bivalirudin for anticoagulation in HIT. Our new thrombosis rate was 4.6%, and the HIT-related death rate was 1.7%. None of our patients required HIT-related amputation after initiation of bivalirudin, in contrast to previous studies reporting amputation rates in patients receiving fondaparinux, lepirudin and argatroban of 4.8%, 6% and 14%, respectively [18, 30-34]. Our all-cause 30-day mortality rate and rate of death related to thrombosis were 14.5% and 0.2%, respectively, in patients treated with bivalirudin. These rates are similar to those found for argatroban (all-cause mortality rate, 17.6%; death related to thrombosis, 0.3%) in the ARG-911 trial, and higher than those for lepirudin (all-cause mortality rate, 7.3–9.8%) [19-21]. The major bleeding rate for bivalirudin was 7.6% (35/461) in our study, which is comparable to major bleeding rates for fondaparinux (6.4%), lepirudin (14.0–15.4%), argatroban (4.0–8.0%), and danaparoid (3.8%) [18, 30]. We found a fatal bleeding rate of 1.5% (7/461). Major bleeding risk was increased among patients in a critical care setting. In our series, bivalirudin was safely and effectively administered in patients with advanced age, obesity, recent surgery, and renal or liver dysfunction, without a significant increase in the major bleeding rate. The duration of bivalirudin treatment (median, 9.0 days) was intermediate as compared with previous studies of lepirudin (mean, 15.8 days) and argatroban (mean, 6.6 days) [18]. We had a lower proportion of patients with a treatment change to warfarin (58.8%) than in the argatroban trials (62%) and lepirudin trials (83%) [18]. Mortality, high bleeding risk and exclusion of HIT were the most common reasons for a lack of change to warfarin.

We found two major differences from previously reported epidemiologic features of HIT [35-37]. In our population, we found a male sex predominance, in contrast to previous data suggesting an increased risk in females [35]. We also found similar proportions of surgical and medical patients in our study, in contrast to the literature, which has reported a higher risk of HIT in cardiac and orthopedic surgical patients with exposure to UFH or LMWH than in medical or obstetric patients [37-40]. Our data suggest that acutely ill medical patients may also be at high risk for developing HIT. The factors contributing to this observed epidemiologic variation from previous studies include heightened awareness of HIT at our institution, the differences between the definition of HIT used in our study and those used in earlier studies, and advances in the laboratory diagnostic methods. The high incidence of thrombosis (57.8%) with a greater proportion of venous thromboembolic events in our registry population is supported by previous studies that have also reported thrombotic rates ranging from 17% to 55% in patients with HIT [41-43]. The high incidence of VTE also reflects an aggressive duplex ultrasound screening protocol for patients with suspected or confirmed HIT at our institution that routinely included the upper extremity, as the upper extremity was the most common DVT location site. Hong et al. [44] reported that the presence of a central venous catheter (CVC) was a crucial factor in increasing the risk of an upper limb DVT in patients with HIT as compared with HIT patients without a CVC (9.7% vs. 0%, P = 0.00035). On the basis of our data, we advocate that duplex ultrasound evaluation include both the upper and lower limbs in patients suspected of having HIT, especially in those individuals with a CVC.

Our study has the following limitations. First, this is a retrospective case series from one institution, and is uncontrolled. Second, our follow-up data are limited to 30 days. Third, a major limitation of our study is that a definitive diagnosis of active HIT could be confirmed in only 32.1% (124/386) of this population of patients with intermediate to high clinical suspicion for acute HIT, underscoring the challenges in diagnosing this disorder and the need to establish a ‘gold standard' for its diagnosis [45]. Some of the suspected HIT patients without serologic confirmation may not have had HIT. Patients without serologic confirmation, but whose clinical presentation was still consistent with HIT, were often managed as HIT patients by our clinicians. Nonetheless, as it is recommended that a non-heparin anticoagulant be administered to patients with an intermediate to high clinical suspicion of HIT while the results of confirmatory testing are awaited, this series provides important real-world data on the safety and efficacy of bivalirudin in both patients with suspected HIT and those with HIT. Thus, our data on the efficacy and safety of bivalirudin in both suspected and confirmed HIT patients are important for guiding strategies at a time when we face challenges in terms of both overdiagnosis and under-recognition of HIT [2].

In conclusion, we believe that our single-center retrospective study provides evidence that bivalirudin may be a safe and effective alternative anticoagulant for the treatment of both suspected and confirmed HIT.

Addendum

L. Joseph study conception, data collection, data analysis and interpretation, article drafting and revisions, and final approval. A. I. Casanegra study conception, data collection, data interpretation, article revisions, and final approval. M. Dhariwal data collection and final approval. M. A. Smith data collection and final approval. M. G. Raju data collection, article revisions, and final approval. M. A. Militello data acquirement, article revisions, and final approval. M. P. Gomes data interpretation, article revisions, and final approval. H. L. Gornik data interpretation, article revisions, and final approval. J. R. Bartholomew study conception, data interpretation, article drafting and revisions, and final approval.

Acknowledgements

We acknowledge all of the staff physicians and fellows of the Cleveland Clinic Vascular Medicine consultative service for their service and diligent patient care, which made this work possible.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest (J. R. Bartholomew was a consultant for the Medicines Company in the year 2011).

Ancillary