Challenges in Predicting Thromboembolic Risk in Acute Leukemia: Clinical Profiling and Limitations of Current Scoring Systems

Introduction

Patients with newly diagnosed acute leukemia (AL) are prone to an increased risk of thrombotic events, although the mechanisms are still not completely understood. The reported incidence of leukemia-associated thrombosis ranges from 2% to 25%, depending on the phenotype and characteristics of the cohort. Prior studies addressing traditional risk factors for this complication are scarce and heterogeneous. In other cancers, the Khorana score and the ISTH disseminated intravascular coagulation (DIC) score have been used, with limited applicability in AL. In this study, we aim to identify clinical and laboratory factors that can predict thrombosis in AL patients, as well as validate prior scores and report their clinical profiles and associated findings related to these events.

Methods

This is a retrospective cohort study conducted at a reference cancer center in Brazil with newly diagnosed patients with acute leukemia between 2009 and 2022. Two patients were excluded from the analysis due to prior warfarin use. All AL subtypes were included. Data on epidemiological and laboratory parameters at diagnosis and at the moment of each thrombotic event were collected. The primary endpoint was the cumulative incidence of symptomatic venous thromboembolic events (VTE) or arterial thrombosis within the first 60 days from diagnosis, with death as a competing event. Correlation between baseline parameters and VTE/thrombotic events was assessed through logistic regression. Local ethics committee approved this study.

Results: Overall, 421 patients were included, 49% diagnosed with AML, 23% with APL, and 28% with ALL or MPAL. The 60-day cumulative incidence of thrombosis was 15% (95% confidence interval [CI] 12-18%), with incidence rates of 21%, 11%, and 20% in ALL, AML, and APL, respectively. The original phenotype was associated with VTE incidence (p=0.031). Most events occurred within the first 30 days (72%), while the remaining cases were detected at disease presentation (24%) or between 30-60 days from diagnosis (4%). Arterial events were rare, comprising 8 out of 69 events, all of which were cerebrovascular, and more frequent in APL (32%). One patient presented with VTE and an arterial event simultaneously. Regarding venous thrombosis, the upper limbs were the most common site (47%), followed by catheter-related events (20%) and lower limbs (13%). Prophylactic anticoagulation was administered in 15% of patients with thrombosis. Two patients with AL and thrombosis were receiving combined estrogen-progesterone therapy at presentation. In the ALL subset, 44% of thrombotic events occurred after the use of asparaginase. Cerebral venous thrombosis (CVT) represented 25% of the events in this subgroup. Overall, thrombosis recurred in 5 patients. At the date of thrombosis, 33% of subjects still displayed blasts in the peripheral blood, and 25% presented less than 30x10⁹/L platelets. In univariate analysis, there was no association between baseline clinical (age, sex, body mass index) and laboratory (traditional biochemical, hematologic, or coagulation parameters) and the risk of thrombosis. Patients presented low, intermediate, and high Khorana risk scores in 3%, 95%, and 2% of cases, respectively, with no statistical association with thrombosis (p=0.19). The ISTH DIC score was “positive” (≥5) in 86%, and it was also not associated with thrombotic risk in our cohort (p=0.806). Another recently described tool, the SiAML-thrombosis score (Owattanapanich et al. Thrombosis Journal, 2023), which includes different cut-offs for WBC, platelets, and D-dimer, was also not predictive of thrombosis (p=0.926).

Conclusion

In this real-life clinical cohort, we could not identify laboratory parameters associated with an increased risk of thrombosis, despite its high overall incidence. Existing scores showed limited utility. While these seem not to predict thrombosis, these events may occur even in the presence of low platelet counts, prolonged prothrombin time, or hypofibrinogenemia. This suggests a complex coagulopathy leading to hypercoagulability in AL. Therefore, identifying novel biomarkers of coagulation, particularly those related to global coagulation or thromboinflammation, may contribute to a better characterization of thrombotic and bleeding risks in these patients. New prospective studies aimed at exploring different coagulation parameters are warranted in this setting.