Background:Differentiating agents (ATRA/ATO) with or without anthracycline‐based chemotherapy have led to significant improvements in APL. We acknowledge that early death remain as the unsolved issue in this disease, especially when we analyse real‐life population cohorts (this cause of failure is underreported in clinical trials/protocols).Aims:To review the incidence or early death in APL in a large real life setting, as well as in patients who were eligible for protocols. To identify patterns of early death and risk factorsMethods:Between 1996 and 2017, consecutive adult and pediatric patients were reported to the PETHEMA registry data‐base (120 Institutions from Spain (PETHEMA), Poland (PALG), The Netherlands (HOVON), Uruguay, Argentina (GATLA), and Colombia). All patients with suspicion of APL had to be reported. Exclusion criteria for LPA96/99/2005 and 2012 PETHEMA trials were: unfit for chemotherapy/ECOG 4; death before starting ATRA; lack of genetic diagnosis, protocol violation; and secondary APL (sAPL). Induction therapy consisted of AIDA schedule (3 days IDA for older patients). Supportive measures defined per protocol. Early death (ED) defined as induction death plus death before starting ATRA.Results:Overall 2421 patients were registered, of whom 1962 (81%) were considered eligible for PETHEMA trial (exclusion because of sAPL (n = 189), unfit or ECOG4 (n = 154), no genetic diagnosis (n = 32) and protocol violation (n = 77)). ED occurred in 352 patients (14.6%) (hemorrhage n = 170, 7%, infection n = 77, 3%, DS n = 46, 2%, Thromboischemic events n = 31, 1.4%, and other/unknown n = 28, 1.3%). The rate of ED was similar among the 4 consecutive protocols (p = .98), with a non‐significant decrease of hemorrhagic death form the LPA96 trial (9%) to the later LPA2012 trial (5%). ED rate was significantly lower among patients included in trial (8.6%) vs non eligible (22.2% among sAPL, 77.9% in unfit/ECOG4, 40.6% among non‐genetic diagnosis), p < .001. In all subsets hemorrhage was the main cause of ED, except sAPL (8% hemorrhagic ED and 8% infectious ED rate). Univariet analysis showed the following risk factors for ED: age>60 yo, male, ECOG>1, sAPL, high‐risk, fever, platelet count <40 x 109/L, circulating blasts >30 x 109/L, M3variant, BCR3, FLT3‐ITD, prolonged PT, prolonged APTT, any bleeding at presentation, thrombosis at presentation, BMI>25, elevated creatinine, uric acid, bilirubin, LDH, GOT>50 UI/L, glucose >150 mg/dL, albumin <3.5 gr/dL, CD34neg, CD56+, and CD2+ (all p < .05). The multivariate analysis, performed in 1854 patients showed that age (RR 1.03 per year), WBC >10x109/L (RR 2.1), ECOG (RR 1.9 per unit), elevated LDH (RR 1.7), elevated creatinine (RR 4.2), circulating blasts (RR 1.009 per 1x109/L), and sAPL (RR 1.7), were independent risk factors for ED. The Figure 1 shows the cause of death by timing since diagnosis date. CNS was the more frequent site of lethal bleeding (72%, vs 24% pulmonary). In addition, we identified specific risk factors according to the main cause of ED. The time from diagnosis to starting ATRA (median 0 day) had no impact on ED rate.Summary/Conclusion:ED rate in this real‐life cohort was 14.6%, higher compared with 8.6% in selected patients for PETHEMA trials. Hemorrhage, but also infection, DS and thrombosis remain challenging issues to solve.image