The need for anticoagulation in patients with thrombocytopenia is a common, vexing problem and a scenario in which clinicians have to commit to therapy without the benefit of solid trial data. Instead, the decision to administer or withhold anticoagulation depends on the assessed risks of thrombosis and bleeding. To make an informed decision, patients and clinicians must consider the clinical context, the presence of additional risk factors, and the potential consequences of thrombosis and bleeding. Management decisions result in a trade-off since administration of anticoagulation will increase the risk of bleeding, and omission of anticoagulation will increase the risk of thrombosis. In addition to balancing the absolute risks for individual patients, several principles should be considered when deciding on anticoagulation in the setting of thrombocytopenia: 1) a low platelet count does not protect from thrombosis; and 2) in general, thrombotic complications are more dangerous than bleeding complications.
A 75-year-old woman with immune thrombocytopenia (ITP) develops atrial fibrillation with a CHADS2 score of 3. Her platelet count is 19 × 109 /L. You contemplate administering anticoagulation.
As the population ages, the incidence of atrial fibrillation in patients with ITP will continue to increase.1 Additionally, as the management of stroke prevention becomes more aggressive,2 the need for anticoagulation in patients with ITP will become a more common occurrence. In general, older ITP patients have an increased annual risk of bleeding compared with younger patients: 10.4 percent in those older than 60 years compared with 0.4 percent in those younger than 40 years.3 In a systematic review of adults with ITP, the overall risk of severe bleeding was 9.6 percent (95% CI, 4.1-17.1%)4 and the risk of intracranial hemorrhage was approximately 1 to 2 percent.4,5 Conversely, ITP patients may have an increased risk of thrombosis compared with age-matched controls,6,7 and some ITP treatments, including intravenous immune globulin8 and thrombopoietin receptor agonists,9,10 have been associated with an increased thrombotic risk.
Data on the use of anticoagulation in ITP are lacking; thus, treatment decisions should be based on the best available evidence, typically from studies in nonthrombocytopenic patients and patient values. For this patient, stroke prevention would likely take priority at the expense of bleeding,11 especially since the risk of thrombosis (including stroke) may be particularly high in patients with ITP,12 and the bleeding risk can be mitigated by improving the thrombocytopenia. For patients with normal platelet counts and a CHADS2 score of 3, the annual risk of stroke is 5.9 percent (95% CI, 4.6-7.3).13 In the context of ITP and ITP treatments, that annual risk might be approximately twofold higher. With full-dose anticoagulation (e.g., apixaban) in non-ITP patients, the annual risk of stroke is reduced to 1.3 percent14 ; the risk reduction is likely to be similar or greater in patients with ITP. The risk of major bleeding with apixaban is approximately 2.1 percent for nonthrombocytopenic patients. Add to that the risk of bleeding attributable to thrombocytopenia (approximately 10%), and the risk of major bleeding for this patient might be closer to 15 percent4 ; however, raising the platelet count above 30 to 50 × 109 /L would partially mitigate her risk of bleeding.
Given the risks of bleeding and stroke in the patient, the treatment approach was to increase the platelet count using ITP therapies (starting with low-dose corticosteroids) until it is above 50 × 109 /L, and then to administer anticoagulation.
A 34-year-old man who recently underwent autologous stem cell transplantation for relapsed diffuse large B-cell lymphoma develops deep vein thrombosis in his right leg. His platelet count is 15 × 109 /L. You contemplate administering anticoagulation.
Cancer-associated thrombosis is common, occurring in up to 19 percent of patients depending on the tumor type and cancer treatments.15,16 Anticoagulation is required to improve immediate and long-term symptoms and to reduce the risk of recurrence. In patients without cancer, the risk of recurrence after unprovoked venous thromboembolism (VTE) is 5 percent at one year.17 The risk of recurrence is higher in patients with cancer — 20.7 percent (95% CI, 15-25.8%) at one year as reported in one study.18 However, risk estimates are limited because anticoagulation is rarely stopped, and life expectancy is reduced in this patient population. On the other hand, the risk of bleeding on anticoagulants in patients with cancer is also increased — up to 12.4 percent (95% CI, 6.5%-18.2%) at one year18 and even higher due to the severe thrombocytopenia.
In terms of management approach, in this patient, his one-year risk of VTE recurrence is likely higher than his risk of bleeding, the thrombocytopenia is anticipated to be short-lived, the consequences of thrombosis are more dangerous than bleeding in general, and his platelet counts and bleeding symptoms can be closely monitored, the decision was made to administer a platelet transfusion and start anticoagulation once the platelet count is greater than 50 × 109 /L. Low-molecular weight heparin (LMWH) is recommended for patients with cancer-associated thrombosis19 and the dose can be adjusted for severe thrombocytopenia.20 Data from a recent randomized trial supported the use of the direct oral anticoagulant edoxaban for patients with cancer-associated VTE21 ; however, patients with thrombocytopenia were excluded from that trial and the generalizability of those data is limited.
Future Research: Clinical trials are needed to address pressing clinical questions related to the use of anticoagulation in patients with thrombocytopenia. For example, in patients with ITP (platelets <50 × 109 /L) and atrial fibrillation, what is the risk of severe bleeding with full-dose anticoagulation? In designing a trial to address this question, patients could be stratified by the severity of the thrombocytopenia (bleeding risk) and CHADS2 score (thrombotic risk); the intervention could be to administer or withhold anticoagulation; and the primary outcome could be bleeding (as assessed by an ITP-specific bleeding tool22 ). While a randomized trial design could be used, the feasibility and buy-in from clinicians and patients would be important limitations. An alternate approach could be a prospective longitudinal registry, where the decision to administer or withhold anticoagulation is made by the clinician and patient as per clinical practice, and bleeding and thrombosis outcomes can be captured over time. Registries and other well-designed observational studies are needed therefore to inform these difficult management scenarios.
Dr. Balitsky and Dr. Arnold indicated no relevant conflicts of interest.