Thrombosis in AML? Yes, but when to worry?

In this issue of Blood, Libourel et al report the incidence and risk factors for thrombosis in newly diagnosed adults with acute myeloid leukemia (AML). Their data suggested a thrombosis prevalence of 8.7% in younger and 10.4% in older adults with AML; patients at greater risk exhibit disseminated intravascular coagulation (DIC) at diagnosis and high D-dimer levels in particular.¹ 

AML is a heterogeneous disease based on host (age and comorbidities) and leukemia (cytogenetics and molecular genetics) characteristics.^2,3  Despite understanding more about pathophysiology, the care team remains faced with many uncertainties concerning choice of best antileukemia therapy in each clinical setting and determining optimal supportive care strategies. Although subtypes can vary widely with regard to risk of death during induction chemotherapy and relapse risk overall, AML in most patients is intrinsically resistant. The search for better therapies involving immunology, including stem cell transplant, pathophysiological understanding, and genomic data, is in progress. The study by Libourel et al provides useful insights into the issue of leukemic cells and their byproducts deranging normal physiology, specifically, the surprisingly common promotion of clotting.

Most of the focus in improving AML outcomes has been on antileukemic therapies. Less attention has been paid to the supportive care needed to manage patients receiving these therapies. One approach to decrease induction deaths is to substitute hypomethylating agents for standard “3+7” induction therapies in older adults, which could be acceptable in some older patients.⁴  The use of hematopoietic growth factors as part of the adjunctive therapy during AML induction, eschewed due to concerns of stimulating the AML clone, has not routinely been shown to reduce treatment-related mortality.⁵ 

Treatment-related morbidity and mortality in AML induction are presumed to be due almost exclusively to infection and bleeding. The advent of modern antibiotics and judicious use of platelet transfusions have led to a reduction in treatment-related mortality in select patients with multiple significant comorbidities and/or frailty from 20% in the 1980s to 5% to 10% currently.⁶  Thrombotic events have been considered rare and have not received a great deal of attention. However, by performing a prospective study including a validation cohort in adult patients with newly diagnosed AML, Libourel et al more accurately determined the risk of thrombosis in this population. This systematic approach to diagnosis, collection of data, and rigorous definition of thrombosis yielded a higher than expected value, even when asymptomatic catheter-associated thrombosis was discounted.

The study also provided proof of the authors’ hypothesis that DIC at diagnosis would be predictive for the onset of thrombosis. Each individual DIC parameter, including prothrombin time, D-dimer, antithrombin, and α-2-antiplasmin, was significantly associated with thrombosis risk in younger patients with AML, with the most striking association being that with D-dimer levels. The DIC score⁷  requires knowledge of the platelet count, the D-dimer level, the prothrombin time, and the fibrinogen level. The risk of thrombosis in those with a DIC score >5 is over 20% in younger adult patients with AML and up to 40% in older patients.

What is the basis for the association between DIC at diagnosis and thromboembolism during induction therapy in AML? The coagulopathy associated with acute promyelocytic leukemia (APL) is complex, involves DIC, fibrinolysis, and proteolysis, and is generally associated with bleeding, but there are patients with APL who present predominantly with clots at diagnosis.⁸  In the current study, high leukocyte count and monocytic lineage morphology were associated with DIC in both age groups of AML. Perhaps neither of these is surprising because ample tissue factor is present in such situations to promote clotting. The precise mechanism of clotting in AML needs to be elucidated, especially regarding the influence of potentially acquired or inherited thrombophilic factors such as a factor V Leiden mutation. Nonetheless, this paper opens eyes to the potential magnitude of the problem in AML and focuses on those patients where the probability might be increased, thereby generating a more rapid and thorough diagnostic evaluation.

A critical issue is whether anticoagulants should be used to prevent venous thromboembolism in newly diagnosed patients with AML defined as high risk based on a high D-dimer or a high DIC score at diagnosis. It would be premature to draw conclusions about prophylaxis or therapy based solely on this one study, nor should clinicians adopt routine use of anticoagulants in patients even in high-risk settings in AML due to the possible increased risk of bleeding in these thrombocytopenic patients who also have epithelial destruction secondary to chemotherapy. Nonetheless, the observations and correlations made in the Libourel et al paper should stimulate a prospective randomized clinical trial of anticoagulation in patients at high risk for clotting based on the presence of DIC at diagnosis. Thanks to this research, the prior probability of thrombosis in a patient with AML can be assessed using relatively simple blood tests at diagnosis and thus focus our “worry” about an important (and not as infrequent as previously thought) complication upon those at greatest risk.