An effective early death scoring system for predicting early death risk in de novo acute promyelocytic leukemia

The Sanz risk, which was originally used to predict the risk of acute promyelocytic leukemia (APL) relapse, is a recognized method to predict the prognosis of APL. About 570 de novo APL patients admitted to our center were randomly divided into a training cohort (N = 344) and validation cohort (N = 226). Multivariate analysis of training cohort demonstrated that age >52 (OR = 5.170, p = .002), white blood cell count >10 x 10(9)/L (OR = 9.062, p < .001), PLT count <= 10 x 10(9)/L (OR = 4.254, p < .001), and LDH level >500 U/L (OR= 3.002, p = .046) were independent risk factors for early death. A risk score (age >52: 1.5 points; WBC >10 x 10(9)/L: 2 points; PLT <= 10 x 10(9)/L: 1 point; LDH >500 U/L: 1 point) was used to predict early death risk. The model shows a better predictive power of early death in training cohort and validation cohort compared with Sanz risk stratification.