Conference Coverage

Risk model could help predict VTE in acute leukemia



– A new clinical prediction model can determine the risk of venous thromboembolism in patients with leukemia, according to investigators.

Alejandro Lazo-Langner, MD, of Western University in London, Canada Will Pass/MDedge News

Dr. Alejandro Lazo-Langner

The scoring system, which incorporates historical, morphological, and cytologic factors, was internally validated at multiple time points over the course of a year, reported lead author, Alejandro Lazo-Langner, MD, of the University of Western Ontario, London.

“It is important that we can predict or anticipate which patients [with acute leukemia] will develop venous thrombosis so that we can develop preventions and aim for better surveillance strategies,” Dr. Lazo-Langner said at the annual congress of the European Hematology Association. Venous thromboembolism (VTE) risk modeling is available for patients with solid tumors, but a similar prognostic tool for leukemia patients has been missing.

To fill this practice gap, Dr. Lazo-Langner and colleagues conducted a retrospective cohort study involving 501 patients with acute leukemia who were diagnosed between 2006 and 2017. Of these patients, 427 (85.2%) had myeloid lineage and 74 (14.8%) had lymphoblastic disease. VTE outcomes of interest included proximal lower- and upper-extremity deep vein thrombosis; pulmonary embolism; and thrombosis of unusual sites, such as splanchnic and cerebral. Patients were followed until last follow-up, VTE, or death. Single variable and multiple variable logistic regression were used sequentially to evaluate and confirm potential predictive factors, with nonparametric bootstrapping for internal validation.

After last follow-up, 77 patients (15.3%) had developed VTE; specifically, 44 patients had upper-extremity deep vein thrombosis, 28 had lower-extremity deep vein thrombosis or pulmonary embolism, and 5 had cerebral vein thrombosis. The median time from leukemia diagnosis to VTE was approximately 2 months (64 days). Out of 20 possible predictive factors, 7 were included in the multivariable model, and 3 constitute the final model. These three factors are platelet count greater than 50 x 109/L at time of diagnosis (1 point), lymphoblastic leukemia (2 points), and previous history of venous thromboembolism (3 points).

Dr. Lazo-Langner explained that leukemia patients at high risk of VTE are those with a score of 3 or more points. Using this risk threshold, the investigators found that the overall cumulative incidence of VTE in the high-risk group was 44.0%, compared with 10.5% in the low-risk group. Temporal analysis showed a widening disparity between the two groups, from 3 months (28.8% vs. 6.3%), to 6 months (41.1% vs. 7.9%), and 12 months (42.5% vs. 9.3%).

When asked if treatment type was evaluated, Dr. Lazo-Langner said that treatment type was evaluated but proved unfruitful for the model, which is designed for universal use in leukemia.

“We did include a number of different chemotherapy regimens,” he said. “The problem is, because we included both AML [acute myeloid leukemia] and ALL [acute lymphoblastic leukemia] lineage, and the cornerstone of treatment is different for both lineages. It’s difficult to actually include what kind of chemotherapy [patients had]. For instance, it is known that anthracyclines increase risk of thrombosis, but in both lineages, you use anthracyclines, so you really cannot use that as a predictor.”

Looking to the future, the next step will be validation in other cohorts. If this is successful, then Dr. Lazo-Langner speculated that clinicians could use the scoring system to direct monitoring and treatment. For example, patients with high scores and low platelet counts could receive earlier transfusional support, while all high-risk patients could be placed under more intensive surveillance and given additional education about thrombosis.

“I think recognizing symptoms early is important,” Dr. Lazo-Langner said, “and that would be training not only clinicians, but also nursing personnel and the patients themselves to be aware of the symptoms, so they can actually recognize them sooner.”

The study was funded by the Canadian Institutes of Health Research. Dr. Lazo-Langner is an investigator with the Canadian Venous Thromboembolism Clinical Trials and Outcomes Research (CanVECTOR) Network.

SOURCE: Lazo-Langner A et al. EHA 2019, Abstract S1642.

Next Article: