Predicting Bleeding in AML-Associated DIC: Limitations of the ISTH Score and a Modified Approach

Abstract

Background and Objectives: Bleeding is a major cause of mortality in cases of acute myeloid leukemia (AML)-associated disseminated intravascular coagulation (DIC). The predictive power of the standard International Society on Thrombosis and Hemostasis (ISTH) score for bleeding in patients diagnosed with AML is limited. This study aimed to evaluate the performance of the standard ISTH score and modified versions in predicting bleeding among acute promyelocytic leukemia (APL, M3) and non-APL AML subgroups. Methods: This single-center, retrospective study included 190 AML patients (61 APL, 129 non-APL). The predictive power of the original ISTH score and eight different modified scores-incorporating parameters such as lactate dehydrogenase (LDH) and genetic positivity-for DIC-related bleeding was assessed using receiver operating characteristic (ROC) analysis. Results: In the APL group, the original ISTH score was statistically significant in predicting DIC-related bleeding (AUC = 0.727), but the modifications did not improve performance. In the non-APL AML group, the original score did not predict bleeding (AUC = 0.632, p = 0.079). However, a modified ISTH score excluding D-dimer and including LDH (>= 800 mg/dL) and genetic positivity significantly improved prediction (AUC = 0.710, p = 0.005). This modification increased specificity from 48.2% to 60.7% and sensitivity from 76.5% to 82.4%. Conclusions: A subtype-specific approach is required to predict bleeding risk in AML-associated DIC. Modified ISTH scores remain suboptimal for APL; however, a modified score incorporating LDH and genetic status represents a promising tool to identify non-APL AML patients at risk of bleeding and warrants prospective validation.