Heparin-platelet Factor 4 Antibodies Research Articles

Antibodies to the Platelet Factor 4-Heparin Complex and Mortality in Chronic Hemodialysis Patients.

Background: Heparin-induced thrombocytopenia is a serious complication of heparin therapy that can lead to thromboembolism, cardiovascular events or death. Patients with this disorder develop antibodies to the platelet factor 4-heparin (PF4-H) complex. Hemodialysis patients are repeatedly exposed to heparin and are at risk for developing PF4-H antibodies. The clinical impact of asymptomatic PF4-H antibodies in patients on chronic hemodialysis is not known.Objective: To determine the association between asymptomatic PF4-H antibodies and mortality in a cohort of chronic hemodialysis patients repeatedly exposed to heparin.Methods: Pre-dialysis blood samples were drawn from 419 asymptomatic patients. All patients received unfractionated heparin (Baxter) while on dialysis. All samples were screened for PF4-H antibodies using an ELISA assay (GTI PF4 Enhanced, GTI Diagnostics). All positive and indeterminate samples were then tested using an IgG-specific PF4-H ELISA assay and a platelet serotonin-release assay. Participants were then followed up prospectively for thromboembolic events, cardiovascular events, or death.Results: During a median follow-up of 2.5 years there were 129 deaths. After controlling for important potential confounding variables, the relative risk of death was 2.92 (95% CI: 1.18-7.25; P= 0.02) in patients with IgG-specific PF4-H antibodies and 4.08 (95% CI: 1.26–13.2; P= 0.02) in patients with IgG-specific antibodies and an indeterminate serotonin-release assay. Conclusions: PF4-H antibody formation is associated with increased all-cause mortality in patients on chronic hemodialysis. Further investigation is needed to determine if anticoagulation with alternative agents would improve survival in this population.

Prospective Evaluation of Laboratory Tests for the Diagnosis of Heparin-Induced Thrombocytopenia.

The laboratory diagnosis of heparin-induced thrombocytopenia (HIT) relies on the demonstration of antibodies to the heparin-platelet factor 4 (H-PF4) complex. Assays are based on the functional ability of H-PF4 antibodies to activate platelets, or detect the antibody directly by immunological methods. Multiple assays in each category are currently in clinical use and newer, rapid immunological assays are becoming available. The aim of this study was to compare available methods for detecting H-PF4 antibodies in a prospective study of patients with clinically suspected HIT. Functional assessment included serotonin release assay (SRA) and lumi-aggregometry (LA). Immunological assessment included ELISA (GTI), and particle gel immunoassay (PGIA; Diamed and Akers). Circulating platelet microparticles (PMP) were assessed by flow cytometry. Patients were also assessed for the pre-test probability of HIT using the Warkentin 4-T scoring system. 151 patients were enrolled. 54/151 patients (35.8%) had a positive GTI ELISA, while 53/151 (35.1%) and 39/151 (25.8%), respectively, had positive Akers and Diamed PGAI tests. Only 15/149 (10.1%) patients had a positive SRA, while only 5/150 (3.3%) gave a positive result by lumi-aggregometry. There was a strong correlation between the ELISA OD values obtained in serum and plasma using both fresh (r=0.98) and frozen (r=0.99) samples, although slightly more positive results were obtained using serum. Differences were only seen with OD values around the cut-off of 0.4. The majority (77.8%) of H-PF4 antibodies detected by ELISA were neutralized by heparin in the ‘confirmatory' procedure. Weak antibodies (OD 0.4–0.5) were more likely to be non-neutralizable (5/12; 42%) than strong antibodies (OD>1.0; 4/23; 17%). 47 patients positive by ELISA were retested to determine the predominant immunoglobulin subclass. 15/47 (32%) were positive (OD>0.4) for IgG; 27/47 (57%) for IgM, and 12/47 (25%) for IgA. The Diamed assay more closely correlated with the GTI ELISA than the Akers test (82.1% vs. 56.7%, respectively). The PGIAs were only moderately correlated with each other (64%) with the Akers assay giving more “false positive” results relative to the ELISA. PMP were higher in patients with a positive ELISA (6.2 vs 4.7 × 106/ml) or positive SRA (5.5 vs. 5.1 ×106/ml) but this was not statistically significant due to the wide range of results. Of 119 patients assessed, 87 had a low pre-test probability of HIT (4-T score 0–3), 27 had an intermediate probability (4–5), and 5 had a high probability (6–8). The GTI ELISA was positive in 24, 56 and 80% of low, intermediate and high probability cases. The Akers PGIA was positive in 39, 41 and 40% respectively; the Diamed assay in 21, 33 and 40%, and the SRA in 7, 11 and 40%, respectively. This study was conducted in a patient population biased towards cardiovascular surgery, and confirms previously reported observations that immunoassays are more frequently positive than functional assays. The ELISA correlated better than the PGIA tests with the pre-test probability of HIT, although the Diamed test showed acceptable correlation with the ELISA. In contrast, the Akers assay correlated poorly with the ELISA, often producing positive results when the latter test was negative. We conclude that while the PGIA tests are rapid and convenient, further studies are needed to determine the basis for disparate results relative to the widely used ELISA.

The prevalence of antibodies to the platelet factor 4 -heparin complex and association with access thrombosis in patients on chronic hemodialysis

IntroductionHeparin-induced thrombocytopenia is a serious complication that can lead to thrombocytopenia, venous and arterial thrombosis. Patients with this disorder develop antibodies to the platelet factor 4-heparin (PF4-H) complex. Hemodialysis patients are repeatedly exposed to heparin and are at risk for developing PF4-H antibodies.We sought to determine the prevalence of PF4-H antibodies in a large cohort of patients on chronic hemodialysis and to evaluate the relationship between PF4-H antibodies and hemodialysis vascular access thrombosis in a case-control study. Material and methodsPre-dialysis blood samples were drawn on 419 patients; 107 cases with access thrombosis and 312 controls that never had access thrombosis. All samples were screened for PF4-H antibodies using an ELISA assay (GTI PF4 Enhanced, GTI Diagnostics). All positive and indeterminate samples were then tested using an IgG-specific PF4-H ELISA assay and a platelet serotonin-release assay. ResultsAntibodies to PF4-H were positive in 54 (12.9%) patients using the screening ELISA assay. Nine (2.1%) patients had IgG-specific PF4-H antibodies. None of the patient's had a positive platelet serotonin-release assay. No relationship between hemodialysis access thrombosis and PF4-H antibodies was noted using the screening ELISA assay (unadjusted odds ratio 0.63; 95% CI 0.30–1.30; P = 0.21), the IgG-specific ELISA assay (unadjusted odds ratio 0.83; 95% CI 0.17–4.06; P = 0.82) or indeterminate platelet serotonin-release assay results (unadjusted odds ratio 0.97;95% CI 0.10–9.44;P = 0.98). ConclusionsHemodialysis with repeated exposure to unfractionated heparin was associated with a moderately elevated prevalence of PF4-H antibodies. However, our results do not support a relationship between PF4-H antibodies and hemodialysis vascular access thrombosis.

High Prevalence of Antibodies to the Heparin-Platelet Factor 4 Complex in Hemodialysis Patients.

Heparin-induced thrombocytopenia is a serious complication of systemic heparin therapy. Patients with this disorder develop antibodies to the heparin-platelet factor 4 (H-PF4) complex. Patients with H-PF4 antibodies are at higher risk of venous and arterial thrombosis independently of the platelet count. Hemodialysis (HD) patients are repeatedly exposed to heparin to prevent thrombosis of the extra-corporeal circuit. However, thrombosis remains a frequent complication of vascular access, which accounts for up to 20% of all hospitalization in HD patients. Studies to date on H-PF4 antibody and vascular access thrombosis in HD patients have produced conflicting data but have been limited by small sample sizes. We sough to determine the prevalence of H-PF4 antibodies in a large cohort of HD patients and the correlation with vascular access thrombosis.Pre-dialysis blood samples were drawn on 419 HD patients; 107 cases with access thrombosis and 312 controls that never had access thrombosis. H-PF4 antibodies were measured twice using an ELISA (GTI PF4 Enhanced, GTI Diagnostics). The average of the two measurements was used in this analysis. All patients received unfractionated heparin while on dialysis. The mean age was 64 16 yrs with 62% male patients. Antibodies to H-PF4 were positive in 54 (12.9%) patients. H-PF4 antibodies were present in 10 (10.5%) of patients with confirmed access thrombosis and in 44 (15.1%) of controls. Hemodialysis with repeated exposure to systemic unfractionated heparin is associated with a high prevalence of H-PF4 antibodies. Although H-PF4 antibodies contribute to hypercoagulability leading to thrombotic complications, our results are not supporting a correlation between H-PF4 antibodies and vascular access thrombosis. Given the potential for venous and arterial thrombosis and the implications of our findings further investigations are needed in this population.

Detection and Significance of Heparin-Platelet Factor 4 Antibodies

Heparin-induced thrombocytopenia (HIT) represents an immune reaction to heparin, which if unrecognized and untreated, has the potential for catastrophic thrombotic events. The demonstration of antibodies directed against the heparin-platelet factor 4 (H-PF4) complex is an important component of the diagnosis of HIT. Typically, test results are not rapidly available, and the diagnosis therefore rests principally on clinical grounds. Nevertheless, the presence of H-PF4 antibodies provides important confirmation of the diagnosis, and may predict for adverse cardiac events even in the absence of HIT. Although more research is certainly required, the use of alternative (non-heparin) anticoagulants for cardiac surgery and other cardiovascular procedures would substantially reduce the incidence of HIT and could possibly improve patient outcomes.

Functional Consequences of Cellular Interactions with Heparin-PF4 Antibodies in Heparin-Induced Thrombocytopenia.

Heparin-induced thrombocytopenia (HIT) is a potentially devastating complication of heparin therapy. The condition is due to formation of antibodies to the heparin-platelet factor 4 (HPF4) complex that activate platelets and/or endothelial cells by binding to FcγIIa (CD32) receptors. It is not clear why only a sub-population of patients that form HPF4 antibodies also develops clinical HIT, and why only a fraction of those develop thrombosis. We hypothesized that some functionally active HPF4 antibodies (i.e. that activate platelets) may also activate leukocytes and endothelial cells by upregulating procoagulant activity; a property not shared by non-functional HPF4 antibodies. Citrated plasmas were obtained from patients clinically suspected of HIT. A commercially available ELISA (GTI, Brookfield, WI) combined with a 14C-serotonin release assay (SRA) was used to stratify patient plasmas into functional (F) and non-functional (NF) HPF4 antibody groups. Pooled normal plasma (NP) served as negative control. Some (n=20) GTI-ELISA positive samples (both F and NF) were also tested by flow cytometry using HPF4 coated microspheres (DiaMed-Caribbean, Inc., Miami, FL) for immunological antibody detection, and fluorescent annexin V binding to detect functional activity. The ability of HPF4 antibodies to cause platelet-monocyte aggregate (conjugate) formation was evaluated by two-color flow cytometry. The functional consequences of HPF4 antibody interaction with EaHy926 hybridoma endothelial cells were studied in the presence/absence of washed platelets and therapeutic heparin levels. Cell surface procoagulant activity (PCA) was measured by a single-stage clotting assay. Approximately 25% of ELISA positive plasmas were also positive by SRA. Although most ELISA positive samples were strongly positive in the microsphere flow cytometry assay (5–20 fold increase in mean fluorescence intensity [MFI] compared to NP) there was no direct correlation between the ELISA OD values and MFI. In the functional flow cytometric assay, not all the SRA positive samples were able to cause platelet activation.Platelet-monocyte aggregate formation in the presence of functional HPF4 antibodies was increased by 2–6 fold (MFI increase 50–300%). This was significantly higher than seen with the NF (p<0.05, n=6) or NP (p<0.01, n=6) plasma pools. Exposure of confluent EaHy926 cells to functional HPF4 antibodies and washed platelets caused a significant and time-dependent increase (20–35%) in cellular PCA as evidenced by the shortening of plasma clot time (235±15 vs. 330±12 sec, p<0.05). The clot time was not significantly different in samples incubated with NP or with HPF4 positive samples incubated without platelets. These observations suggest that in addition to platelet activation, a subset of functional HPF4 antibodies can cause platelet-monocyte aggregation and platelet-dependent endothelial cell interaction. This may lead to upregulation of cellular PCA (tissue factor) and thereby contribute to the pathogenesis of HIT. Non-functional HPF4 antibodies, even at comparable titer, appear to be devoid of such cellular effects. A prospective study to relate these observations to clinical findings in HIT patients is currently underway.

The Relationship between Systemic Anticoagulation during Cardiac Surgery and Heparin Platelet Factor 4 Antibody Formation.

Heparin-induced thrombocytopenia occurs in approximately 1–3% of patients receiving unfractionated heparin. However, up to 50% of patients receiving heparin for systemic anticoagulation during cardiac surgery with cardiopulmonary bypass develop antibodies to the heparin-platelet factor 4 (HPF4) complex detectable by enzyme-linked immunoassays (ELISA). This high incidence of seroconversion is not diminished by off-pump cardiac surgery (OPCAB), despite significantly lower doses of heparin used for surgical anticoagulation. We tested the hypothesis that HPF4 antibodies develop following cardiac surgery even when heparin is not used for surgical anticoagulation. Patients (n=133) were randomized (2:1) to receive bivalirudin (The Medicines Company, Parsippany, NJ) or heparin for surgical anticoagulation during OPCAB. There were no restrictions on the pre-operative or post operative use of heparin. Blood samples were obtained pre-operatively (baseline), and 7 and 30 days after surgery. Serum samples were stored frozen until tested with a commercially available ELISA for HPF4 antibodies (GTI Inc., WI). An optical density (OD) ≥ 0.4 was considered positive, while an OD >1.0 was deemed strongly positive. The percentage of patients having a positive HPF4 antibody test (OD ≥ 0.4), was similar at baseline (6.7 vs. 6.8%) and at 7 days (27.1 vs. 29.3%) in bivalirudin and heparin-treated patients, respectively. At 30 days, the frequency of HPF4 antibodies was lower in the patients who received bivalirudin for surgical anticoagulation (32.5% vs. 40.9%). The percentage of patients having a strongly positive ELISA (OD > 1.0) was 0% at baseline in both groups, 4.3 vs. 7.3% at 7 days and 4.8 vs. 9.1% at 30 days for bivalirudin vs heparin, respectively. These data suggest that HPF4 antibody formation during hospitalization for cardiac surgery is not dependent only on the heparin used for anticoagulation during surgery. Antibody positivity in baseline samples presumably results from pre-operative heparin use and is consistent with previous reports. Since bivalirudin is not capable of eliciting an HPF4 antibody response, post-operative seroconversion presumably reflects a combination of both pre- and post-operative heparin use. However, when heparin was used for surgical anticoagulation, the number of patients subsequently developing strongly positive HPF4 antibodies was almost twice that seen in the bivalirudin group. This suggests that the use of bivalirudin as a surgical anticoagulant may reduce the incidence and risk of HIT following cardiac surgery.

Frequency of Heparin-Platelet Factor 4 Antibodies among Patients with Acute Coronary Syndromes Presenting to the Emergency Department.

Heparin is widely used to treat patients that present with acute coronary syndromes (ACS). A significant, albeit under-recognized complication of heparin therapy is heparin-induced thrombocytopenia (HIT), which is estimated to occur in 1–3% of patients. A significantly higher proportion (20–50%) of patients undergoing cardiac surgery and other invasive cardiology procedures develop an antibody against the heparin-platelet factor 4 (HPF4) complex. Most patients remain asymptomatic, but are nevertheless at risk for HIT with subsequent heparin exposure. Patients with severe coronary artery disease frequently have multiple hospital admissions, and heparin exposures. HPF4 antibodies are transient, and most will become undetectable three months after their initial formation. Therefore, patients with a recent history of hospitalization would be at greater risk of HIT on heparin re-exposure. We tested this hypothesis in a prospective study of patients with ACS presenting to the Emergency Department. We studied 230 patients presenting to the ED with ACS. Patients were stratified into Group 1 (having a recent [<6 months] history of hospital admission), or Group 2 (having a more remote [or no] history). Plasma samples were tested for HPF4 antibodies with a commercially available ELISA and positive samples were re-tested with a serotonin release assay (SRA). Sixteen patients (7.0%) tested positive for HPF4 antibodies by ELISA. Group 1 subjects had a higher incidence of HPF4 antibodies (10.7%) than Group 2 patients (3.2%). Of the 16 patients positive by ELISA, 14 were also tested by SRA. Of these, 7 were positive, 6 were negative and 2 gave borderline results. We conclude that approximately 7% of all patients presenting with acute coronary syndromes have a pre-existing HPF4 antibody. The incidence of a positive test is tripled in patients that have been recently hospitalized with high probability of recent heparin exposure. A significant proportion of these patients has a functionally active HPF4 antibody, which would presumably carry a higher risk of clinical HIT if re-exposed to heparin therapy for ACS and subsequent cardiovascular procedures. Our data suggest that high-risk individuals can be identified on a history of recent hospital admissions, and should receive a non-heparin anticoagulant, e.g. a direct thrombin inhibitor, until their HPF4 antibody status can be established.

Successful Use of Bivalirudin in the Treatment of Patients Suspected, or at Risk of, Heparin-Induced Thrombocytopenia.

Heparin-induced thrombocytopenia (HIT) is a serious complication of heparin therapy. As thrombosis occurs in 50% of untreated patients, prompt treatment with a direct thrombin inhibitor (DTI) is recommended. Lepirudin and argatroban are currently approved for the treatment of HIT. However, their use is complicated by antibody formation with potential for anaphylactic shock (lepirudin), effects on the PT/INR that complicate transition to coumadin (argatroban), and significant dose adjustments in patients with renal (lepirudin) or liver (argatroban) impairment. The effectiveness of bivalirudin, another DTI, as a replacement for heparin has been well documented in percutaneous coronary intervention, but there are little published data on its use in treating HIT. We now report our experience with bivalirudin in 52 patients with clinical suspicion of HIT, or at increased risk of this complication. HIT was suspected on the basis of a falling platelet count and/or thrombosis in the setting of current or recent heparin therapy (n=49). Patients were considered at increased risk of HIT if they required ongoing intravenous anticoagulation in a setting associated with a high incidence of heparin-platelet factor 4 (HPF4) antibodies (n=3). Data were collected by retrospective chart review, and patients classified according to whether HIT was very likely (n=13), likely (n=17), unlikely (n=11) or very unlikely (n=11). Bivalirudin was given by intravenous infusion, typically at an initial dose of 0.15 - 0.20 mg/kg/h and adjusted to achieve an APTT of approximately 1.5–2.5 x baseline. Twenty-one patients had moderate or severe renal impairment. The infusion rate was significantly lower for patients with severe renal insufficiency, but was not different in those with mild renal dysfunction. ELISA-detectable HPF4 antibodies were present in 43/52 cases. Twenty-seven patients were significantly thrombocytopenic (<100,000 x 109/L) and 22 had thrombosis before therapy. Transition to warfarin was achieved in 44/52 patients with a median overlap of therapy of 4 (0.5–14) days. Bivalirudin therapy was continued for an average of 8.0 (3–47) days, and had a relatively minor impact on the PT/INR. The mean INR on monotherapy was 1.50 (1.23–2.18) with a mean change in INR due to bivalirudin therapy of 0.33 ± 0.22. Therapeutic APTTs were achieved in all patients, with approximately 92.5% of tests in the desired range. The average time to platelet recovery was 3.0 (1–10) days. There were no cases of major bleeding, no deaths attributable to HIT, and no patient required amputation. We conclude that bivalirudin provides safe and effective anticoagulation for patients with suspected HIT, as well as for those with an increased risk of HIT that require intravenous anticoagulation. Potential advantages of bivalirudin include the relatively minor effect on the PT, which facilitates transition to warfarin therapy, and its short half-life in patients at high risk of bleeding or who require invasive procedures at short notice.

Heparin antibodies and thromboembolism in heparin-coated and noncoated ventricular assist devices

Objective: Coating of ventricular assist devices (VADs) with heparin improves the biocompatibility and may reduce the need for systemic anticoagulation. However, heparins are associated with the risk of formation of heparin/platelet factor 4 antibodies (HPF4/A) and the development of heparin-associated thromboemboli. We analyzed the occurrence of HPF4/A and thromboembolism in patients with heparin-coated and noncoated VADs. Methods: One hundred patients were enrolled in the investigation. Fifty-seven received a heparin-coated (group 1) and 43 a noncoated (group 2) VAD. HPF4/A testing was performed before and 2 and 12 weeks after implantation by the heparin platelet factor 4 enzyme-linked immunosorbent assay. Results: There was no significant difference in the occurrence of HPF4/A in the 2 groups (P =.102). Before the operation, 21 of the patients in group 1 had positive test responses and 25 in group 2. No patient had HPF4/A after termination of systemic heparinization. In group 1 there was no significant difference in the incidence of recurrent pump thromboses in patients who had positive test responses for HPF4/A (n = 11) when compared with those who had negative test responses (n = 9, P =.89). Twenty-one patients had HPF/A but no thromboembolism. However, all 22 patients who had thromboembolism had HPF4/A. Conclusions: Heparin coating of the VAD surface does not enhance the occurrence of HPF4/A-associated immunologic or thrombogenic reactions. However, the presence of these antibodies is strongly associated with an increased risk of thromboembolism in patients with a VAD. (J Thorac Cardiovasc Surg 2001;121:331-5)

Prevalence and Persistence of Heparin/Platelet Factor 4 Antibodies in Patients with Heparin Coated and Noncoated Ventricular Assist Devices

Thromboembolism is a major complication in patients with ventricular assist devices (VAD). Anticoagulation with heparin, coumarin, and anti-platelet agents, particularly the development of biocompatible surfaces such as inner pseudo-endothelial layers or a coating with heparin, are intended to reduce these complications. However, the administration of heparin can lead to heparin induced thrombocytopenia type II (HIT II). Predominantly heparin/platelet factor 4 (HPF4) antibodies are responsible for the development of HIT II. The goal of the present investigation was to assess the prevalence of these antibodies in patients with heparin coated and noncoated VADs. Fifty-five patients were enrolled in the investigation. A heparin coated system was implanted in 30 patients, and a noncoated system was implanted in 25 patients. Antibodies were evaluated before, on days 7 and 14, and 3 months after implantation. Testing was performed with the Heparin/Platelet factor 4 enzyme-linked immunosorbent assay (ELISA) (Stago, France). In 40 of the 55 patients, the formation of HPF4 antibodies was observed (73%). In 35 of these patients (88%), HPF4 antibodies were present before surgery. There were no differences between the groups. In 11 patients (equal from both groups), the antibodies disappeared after termination of systemic heparinization. We conclude that in a rather high percentage of patients with VADs HPF4 antibodies are found. This finding may be explained by the repetitive and prolonged exposure of these patients to heparin. Immobilized heparin, as presently used in the carmeda coating, seems not to influence the formation and persistence of HPF4 antibodies. Further studies will have to prove whether HPF4 antibodies contribute to thromboembolic complications in these patients.

Presence of Autoantibodies to Interleukin-8 or Neutrophil-Activating Peptide-2 in Patients With Heparin-Associated Thrombocytopenia

Eighty-seven patients with heparin-associated thrombocytopenia (HAT) showed either a positive heparin platelet aggregometry test result and/or the presence of antibodies to heparin-platelet factor 4 (H-PF4) complexes by enzyme-linked immunosorbent assay (ELISA). Fifteen of these patients lacked antibodies to H-PF4, and plasma from these patients was analyzed for the presence of antibodies to PF4-related chemokines, Neutrophil-activating peptide-2 (NAP-2) and interleukin-8 (IL-8). Of these 15 patients, 6 showed antibodies to IL-8 and 3 to the platelet basic protein (PBP)-derived protein, NAP-2. Antibodies to IL-8 and NAP-2 were not observed in control patients (n = 38), patients with HAT and H-PF4 autoantibodies (n = 72), patients with autoimmune diseases (n = 21), or patients with non-HAT thrombocytopenia (n = 30). Five of these nine patients with anti-IL-8 or anti-NAP-2 developed thrombosis during heparin treatment, which is not statistically different from the patients with H-PF4 antibodies. The existence of autoantibodies to IL-8 and NAP-2 in HAT patients highlights the significance of chemokines in the pathogenesis of HAT. The contribution of heparin in vitro was minimal in patients with anti-IL-8 and anti-NAP-2 antibodies, suggesting a biologic difference from the majority of patients with HAT and anti-PF4 antibodies. It may be that antibodies to IL-8 and NAP-2 have weaker affinity for heparin and that the ELISA system may not reflect in vivo heparin-chemokine complex formation. Alternatively, antichemokine autoantibodies may predate heparin exposure, and the role of heparin in initiating HAT may be to mobilize the chemokines and to target them to platelets, neutrophils, or endothelial cells. Subsequent chemokine-binding autoantibodies then lead to cell activation resulting in thrombocytopenia and thrombosis.