Background: Early bone marrow (BM) evaluation on the day (D) 14-22 has been widely used to predict the likelihood of attaining complete remission (CR) in acute myeloid leukemia (AML) patients (pts) undergoing either single or double induction chemotherapy (IC). Recently, a German SAL group showed that in pts with BM blast clearance <5% after initial IC, a second induction did not lead to an overall survival (OS) benefit. While assessment of measurable residual disease (MRD) by flow cytometry (FC) after two cycles of intensive therapy and at the end of consolidation provide independent prognostic information, data on the value of earlier FC-MRD assessment during aplasia are limited. Aims: The aim of this analysis was to evaluate the independent prognostic impact of D14 BM MRD (MRD-0) compared to established risk factors. Methods: PALG-AML1/2016 study is an open-label multicenter prospective randomized phase III trial. Pts 18-60 years with newly diagnosed AML, normal cardiac and organ function are randomized 1:1 to DA-90 vs. DAC Induction. Pts with >10% blasts in D14 BM evaluation by cytology are given early second IC (DA-45 and DAC, respectively). Pts with CR/CRi receive further post-remission therapy adjusted to predefined risk groups. Serial samples were collected for FC-MRD at D14 (MRD-0), at the time of CR/CRi after one or two induction cycles (MRD-1), and after each consolidation cycle (MRD-2, -3, -4). The predictive power of MRD level and D14 blast % after induction was evaluated by receiver operating characteristics (ROC) and area under the curve (AUC) analysis. Multivariable logistic regression models were developed using significant variables from univariate analyses, including MRD and blast count. Results: To date, 374 pts were evaluable at D14, and 270 cCR (CR+CRi) pts had data on MRD1. A cut-off of D14 blasts <5% had the highest accuracy in predicting cCR, with AUC=0.63 (p=0.0002). Overall cCR rate was 80.5%, with 72.5% in pts with D14 BM blasts <5% vs. 45.1% in pts with blasts ≥5% (p<0.0001). However, D14 BM blast % was not significant in predicting FC MRD-negative cCR (MRD-neg). Using ROC analyses, optimal MRD-0 values were determined for predicting cCR and cCR MRD-neg. MRD-0 <0.3% (AUC=0.66, p<0.0001) and MRD-0 <0.11% (AUC 0.69, p<0.0001) had the highest accuracies for predicting cCR and cCR MRD-neg, respectively. MRD-0 value of <0.3% increased the probability of achieving cCR with OR 5.94 (p<0.0001) in univariate analysis. The cCR rates in pts with MRD-0 <0.3% and MRD-0 ≥0.3% were 84.2% vs 47.2% respectively (p<0.001). When the same MRD-0 cut-off was analyzed in pts with 5-10% blasts at D14, it remained predictive for cCR (86.7% vs. 33.3% respectively for pts with MRD-0 <0.3% and MRD-0 >≥0.3%; p=0.002). The overall rate of cCR MRD-neg after induction was 37.1%. Pts with MRD-0 <0.11% had higher rate of cCR MRD-neg than those with MRD ≥0.11% (42.5% vs. 15.8%; p<0.0001 respectively). Interestingly, MRD-0 <0.11% was highly predictive of cCR MRD-neg achievement in pts with 5-10% blasts in D14 BM (75% vs. 13% respectively for pts with MRD-0 <0.11% vs. MRD-0 >≥0.11%; p=0.003). In multivariate analyses, only MRD-0 <0.3% and poor-risk ELN category independently predicted cCR. Similarly, MRD-0 <0.11% significantly increased (OR 2.79, 95%CI: 1.39-5.59, p=0.0038), whereas poor-risk ELN category decreased the probability of CR MRD-neg (Table 1). Median EFS censored at SCT for the whole cohort was 22.3 months (mo). D14 BM blasts cut-off >5% was associated with shorter EFS (HR=1.77, p=0.006) and OS (HR=2.59, p<0.0001). Summary/Conclusion: Our results suggest that FC D14 BM assessment has an independent predictive value for achievement of cCR and cCR MRD-neg and potential clinical utility in selecting patients who might benefit from second induction.Keywords: AML, MRD, Survival, Flow cytometry