Post-induction minimal residual disease assessment by flow cytometry and molecular biology enables prognostic stratification of patients with NPM1-mutated Acute Myeloid Leukemia: A dataml registry study.

Abstract IntroductionMRD assessment in AML provides a quantitative approach to define a deeper remission status and refine the risk assessment of postremission relapse. Currently, the two most extensively validated methodologies are multiparameter flow cytometry-based MRD (MFC-MRD) and molecular MRD (Mol-MRD) assessed by quantitative PCR (qPCR). The latest update of the ELN recommendations on MRD includes new technical guidance for standardized MFC-MRD and Mol-MRD analyses, MRD thresholds, definitions of MRD response, and recommendations for clinical application. In this context, we evaluate the clinical utility of MRD assessment using MFC and qPCR in patients with NPM1-mutated AML.MethodsBetween January 1, 2015, and September 30, 2024, 506 adult patients (≥18 years) with NPM1-mutated AML were included in the DATAML registry. Immunophenotyping at diagnosis was available for 383 patients from Toulouse University Hospital and 123 patients from Bordeaux University Hospital. Among them, 306 patients received a standard intensive chemotherapy regimen. 270 patients achieved complete remission (CR/CRi), and 194 of these were monitored for post-induction bone marrow minimal residual disease (MRD1) using both MFC-MRD and qPCR techniques.MFC analysis of leukemic cells was performed on fresh bone marrow samples. For leukemic bulk analysis, the antibody panel included CD34, CD13, CD38, CD7, CD33, CD56, CD117, HLA-DR, and CD45, as recommended by ELN. MFC-MRD data were analyzed using the FlowSOM algorithm (Vial et al., Cancers, 2021). The threshold for MRD1 positivity was defined as ≥0.1% for MFC-MRD and ≥2% for low-level (LL) Mol-MRD.Results194 NPM1-mutated AML patients in first CR/CRi after induction chemotherapy were included. Main co-mutations were: DNMT3A (38%), FLT3-ITD (34%) IDH1 (18%), IDH2 (14%) and FLT3-TKD (14%). 31 patiens (16%) had NPM1/DNMT3A/FLT3-ITD mutations.The median follow-up (FU) was 40 months. In patients with positive MFC-MRD, 1-y and 3-y OS was 83% and 67%, vs 93% and 82% in patients with negative MFC-MRD (Hazard Ratio HR: 2.39, p=0.0056). The 3-y cumulative incidence of relapse estimation (CIR) was 55% for MFC-MRD positive patients vs 30% for MFC-MRD negative patients (HR: 3.17, p0.0001).In patients with LL Mol-MRD, 1-y and 3-y OS rates were 83% and 59% vs 93% and 85% in patients with negative Mol-MRD (HR: 3.52, p=0.0001). The 3-y CIR estimation was 58% in LL Mol-MRD positive patients vs 30% in Mol-MRD negative patients (HR: 3.01, p0.0001).The combination of Mol-MRD and MFC-MRD identified four distinct patient groups: double negative (-/-; n=74, 38%), MFC-MRD positive only (-/+; n=49, 25%), Mol-MRD positive only (+/-; n=41, 21%), and double positive (+/+; n=30, 15%). At 3 years, OS estimates were 90%, 77%, 72% and 48% respectively in these 4 subgroups (p0.0001). CIR estimates were 18%, 45%, 49% and 71% respectively (p0.0001). In multivariate analysis, both LL Mol-MRD positivity and MFC-MRD positivity were independently and significantly associated with inferior OS (HR: 3.59, p0.001; HR: 2.16, p=0.014, respectively) and increased CIR (HR: 2.82, p0.001; HR: 2.30, p=0.001, respectively) whereas DNMT3Am, FLT3-ITD or allo-HSCT did not retain statistical significance.ConclusionIn NPM1 mutated AML, where transcript levels typically decline rapidly after induction, FlowSOM assisted MFC-MRD yields prognostic information that is complementary to qPCR. Combining both assays at post-induction (MRD1) delineates four clinically distinct groups, ranging from a double-negative cohort with excellent outcomes (3-year OS=90%, CIR 18%) to a double-positive cohort at very high risk (3-year OS=48%, CIR 71%). MFC-MRD positivity at MRD1 remained independently associated with inferior OS and higher relapse, even after adjusting for molecular MRD. Although larger prospective studies are needed, these findings strongly suggest that MFC-MRD may not be redundant in NPM1-mutated AML. Notably, when integrated with qPCR, it could improve the accuracy of predicting early relapse. These results support the incorporation of dual-modality MRD monitoring into prospective testing.