Platelet function tests predict bleeding in patients with acute myeloid leukemia and thrombocytopenia.

Bleeding is frequent among patients with thrombocytopenia. We studied whether in vitro platelet function predicts future bleeding in patients with acute myeloid leukemia (AML), and thrombocytopenia. Adult AML patients with platelet count <5.0 × 1010 /L were included. Detailed bleeding history and blood samples were collected at inclusion and every 7 days. We analyzed hematology and coagulation parameters. With flow cytometry we evaluated platelet activation (activated GPIIb/IIIa, P-selectin, and CD63 expression), and platelet aggregation. Agonists were thrombin-receptor activating peptide (TRAP) and collagen-related peptide (CRP-XL). During 18 months, sixty participants were enrolled and followed for a total of 114 weeks. Bleeding occurred in 52 weeks, and was not associated with clinical, hematology or coagulation parameters. Predictors of bleeding were assessed using multivariate logistic regression, adjusted for platelet count, sex, and age. Bleeding history and receiver operating curves were compared using c-index. Reduced TRAP-induced platelet aggregation had Odds ratio 3.00 (95% confidence interval [CI] 1.38-6.60). Reduced CRP-XL-induced platelet aggregation had Odds ratio 4.00 (95%-CI 1.70-9.20) for bleeding. Overall, C-index was 0.71 for the models including platelet aggregation results, 0.72 for activated GPIIb/IIIa-positive platelets after stimulation with TRAP, 0.68 for percent P-selectin positive platelets with TRAP and 0.63 for the platelet count. Among patients receiving no platelet transfusion, C-index was 0.83-0.91 for bleeding; highest for models using platelet aggregation. Change in platelet aggregation did not correlate with the number of platelet concentrates administered. In conclusion, platelet aggregation and platelet activation results predict bleeding better than platelet count alone, among AML patients with thrombocytopenia.