Measurable Residual Disease (MRD) Determinants, Kinetics and Its Impact on Survival in Patients Treated with Azacitidine and Venetoclax for Acute Myeloid Leukemia in Frontline Setting : A Multicentric Study from French Auraml Group

In intensively treated acute myeloid leukemia (AML) patients, measurable residual disease (MRD) has become a critical standardized endpoint which is highly predictive of relapse-free (RFS) and overall (OS)survival. Azacitidine and venetoclax (AZA/VEN) combination regimen recently emerged as a new standard of care in frontline unfit AML patients. However, MRD negativity determinants and its impact on survival remains elusive in AZA/VEN treated patients. Identifying patients with favorable outcomes might be the corner stone for future de-escalation strategy. We aimed in this retrospective study to evaluate the predictive value of MRD negativity and its kinetics on survival on a large cohort of AML patients treated upfront with AZA/VEN.

Patients from the VENAURA registry (n=585) with the following criteria were included in this study : patients with AML according to WHO 2022 (including 10-20% of marrow blasts), receiving AZA/VEN in frontline setting. The VENAURA registry retrospectively collates AZA/VEN data from 12 different French centers in the Auvergne Rhône Alpes (AURA) region, between January 2019 and February 2024. Composite complete remission (CRc) was defined as in VIALE-A. MRD negativity was defined as ≤ 10^-3 by flow cytometry (on bone marrow), ≤ 10^-3 for WT1 by RT-qPCR and/or ≤ 10^-4 for NPM1 by RT-qPCR. Optimal, stable and progressive MRD kinetics by FCM and/or RT-qPCR MRD was defined as MRD negativity or conversion to MRD negativity between two time points, detectable MRD between two time points and conversion from MRD negativity to MRD positivity, independent of method. Risk groups were defined according to ELN 2022 and molecular prognostic risk score (mPRS) as previously reported.

Overall, 268 patients (de novo AML=78, MRC-AML= 168, myeloid sarcoma=2, MDS/AML=20) received frontline AZA/VEN. With a median follow-up (FU) of 10.3 months (0-54.2), CRc rate of the entire cohort was 70.5% (80.1% at cycle 1, 19.9% later) and median overall survival was 14.7 months (m). NPM1 (OR=3.74, p=0.014) and IDH1/2 (OR=2.08, p=0.04) mutations were associated with higher rates of CR_MRDneg conversely to complex karyotype (OR=0.31, p=0.04) and/or TP53 mutations (OR=0.4, p=0.05). In multivariate analysis, MRD negativity determinants during the 6 first cycles was neutrophil counts at AZA/VEN initiation, while NPM1IDH/TP53 mutations, complex karyotype, and VEN dosage were not associated with MRD response. No other mutation was predictive of MRD negativity. ELN 2022 risk groups were not predictive for survival while mPRS median OS was 19.3, 13.8 and 7.3m for favorable, intermediate and adverse risk groups (p<0.001). Reaching MRD negativity by FCM during the first 6 cycles was associated with improved OS (31.3 vs 14.2 m, p=0.001) and RFS (21.7 vs 8.2 m, p<0.001). As FCM, NPM1_mut MRD negativity was associated with improved OS and RFS while WT1 MRD assessment was not. Median OS of MRD_neg in favorable and adverse mPRS patients were similar. Conversely, median OS for MRD_pos was 14.7 and 9.1 m in favorable and adverse mPRS patients (p=0.23). However, MRD by FCM and/or RT-qPCR was not predictive of OS and RFS in intermediate mPRS patients. In multivariate analysis, only CR_MRDneg (HR=4.35, p=0.001) and blasts monocytic bias (HR=3.08, p=0.038) were independently associated with survival while mPRS or other genotypes were not. Seventy-two patients were evaluable at two different time points during FU. Optimal MRD (MRD_optimal) response was associated with a favorable outcome (mRFS : 40.7m), while stable and progressive MRD (MRD_partial) kinetics identify patients with poor outcomes (12.1 and 8.2m respectively (p<0.001)). In responding patients, 61 stopped VEN and pursued AZA, mainly due to hematological toxicity. In MRD_optimal patients that stopped VEN, RFS probability at 24m was 84.6% while all MRD_partial patients that stopped VEN ultimately relapsed within a median time of 7.7m.

To the best of our knowledge, this is one of the largest studies reporting MRD predictive value on survival in AML patients treated with AZA/VEN. Our results suggest that deep responses within the first 6 cycles alleviate mPRS prognostic impact in favorable and even adverse risk groups. However, in intermediate mPRS risk patients, MRD may not represent a good surrogate marker for survival. Thus, optimizing MRD response might represent a new therapeutic goal to further improve outcomes in AZA/VEN treated patients.