Venetoclax and azacitidine for molecular relapse after intensive chemotherapy in NPM1 or CBF AML: a FILO study

Dear Editor,

In acute myeloid leukemia (AML), molecular response assessment and sequential follow-up by real-time quantitative PCR (RT-qPCR) during remission are routinely used in patients with NPM1 mutations (NPM1^mut), RUNX1::RUNX1T1, or CBFB::MYH11 transcripts [1, 2]. Even though patients who achieve complete molecular response have a better prognostic and can be cured without transplantation, they should be closely monitored according to ELN MRD guidelines since up to 30% of them may present MRD relapse (MRD_Rel) [3]. Recent retrospective studies showed that NPM1^mut or CBF-AML patients who received preemptive therapy at time of MRD_Rel had a better overall survival (OS) than those treated for morphologic relapse [4, 5]. However, there is no consensus on the best treatment approach in this situation. Preliminary studies have shown promising results with venetoclax–based low intensity therapies for both molecular failure and MRD_Rel in NPM1^mut and CBFB::MYH11 AML [6,7,8].

Patients with NPM1^mut or CBF-AML who received VEN-AZA for MRD_Rel between February 2020 and October 2024 were retrospectively identified from 10 French Innovative Leukemia Organization (FILO) centers. MRD monitoring by RT-qPCR in blood (PB) or bone marrow (BM) samples (RUNX1::RUNXT1, CBFB::MYH1 or NPM1 mutations) during first-line chemotherapy and follow-up is standard-of-care since the CBF-2006 and BIG-1 trials [9, 10]. All MRD_Rel were confirmed on a second sample (day of first sample was used to define the date of MRD_Rel). Inclusion criteria were: age ≥18 years, first morphologic CR with negative MRD or MRD-LL ( <2%) during first line chemotherapy, no previous allo-HSCT, MRD_Rel according to ELN criteria [2, 11], at least 1 cycle of VEN-AZA for MRD_Rel. Patients received off-label venetoclax 400 mg/d (d1-7, d1-14, d1-21 or d1-28 according to centers) without ramp-up and azacitidine 75 mg/m²/d subcutaneously (d1-7 or d1-5, d8-9).

Assessment of MRD responses was performed locally and response definitions followed those published by Jimenez-Chillon et al. : MRD negativity, MRD reduction for reduction ≥1log₁₀ from pre-treatment value, MRD progression for increase ≥1log₁₀ from pre-treatment value, and stable MRD for patients not meeting any of these previous criteria [7].

Morphologic relapse was defined according to ELN criteria. OS was measured from the date of MRD_Rel to the date of death from any cause. Relapse-free Survival (RFS) was measured from the date of MRD_Rel, to the date of morphologic relapse or death of any cause. RFS_MRD was measured from the date of MRD_Rel to the date of morphologic relapse, second MRD_Rel (for patients achieving MRD negativity or MRD reduction after VEN-AZA) or death from any cause. In accordance with the Declaration of Helsinki, the study was approved by the research ethics committee at Toulouse University Hospital. Because of the retrospective nature of our study, informed consent was waived according to national regulations.

Seventy patients were included (Supplementary Table 1). Most patients had NPM1 mutations and only four patients had CBF-AML. All patients achieved best molecular response of CR_MRD- (n = 44) or CR_MRD-LL (n = 26) during first-line treatment according to the ELN criteria [11]. The median time between diagnosis and MRD_Rel was 10 months (IQR 8-15; min-max 2-51).

The median time between MRD_Rel and first VEN-AZA cycle was 48 days (IQR 32-71; min-max 6-526). Most patients (77%) received the first cycle as outpatient (Supplementary Table 2). Fifteen patients (21%) received posaconazole prophylaxis and 28 (42%) patients received G-CSF. Patients received a median of 2 cycles (IQR 2-3.8; min-max 1-28). During cycle 1, the duration of VEN treatment was 7, 14, 21, and 28 days in 2 (3%), 30 (43%), 15 (21.5%) and 22 (31.5%) patients, respectively. Six (9%) and seven (10%) patients had red blood cell or platelet transfusions. Grade 3-4 neutropenia was observed in 36 patients (52%) but only 11 patients (16%) presented febrile neutropenia. Neutropenia was not associated with duration of VEN treatment (Supplementary Table 2). There was no early death at day-60.

Among the 64 patients evaluated in PB after one or two VEN-AZA cycles, 27 (42%), 16 (25%) and 14 (22%) achieved MRD negativity, MRD reduction or stable MRD respectively. Of the 44 patients evaluated in BM after one or two VEN-AZA cycles, 12 (27%), 20 (46%), and 5 (11%) achieved MRD negativity, MRD reduction or stable MRD respectively (Table 1). Response rates were consistent between NPM1^mut patients with (n = 31) or without (n = 35) FLT3-ITD or N/KRAS mutations (Table 1). At last news, 58 patients (83%) were still in morphologic CR including 52 patients (74%) with negative MRD. With a median follow-up from MRD_Rel of 22.5 months (IQR 14.25-28), median OS was not reached with 1y- and 2y-OS of 95% and 83%, respectively (Fig. 1A). Median RFS and RFS_MRD were not reached, with 1y- and 2y-RFS of 82% and 71%, and 1y- and 2y-RFS_MRD of 73% and 62% (Supplementary Figure 1).

Table 1 MRD response after C1 + /− C2.

Fig. 1: Survival outcomes.

Overall survival since molecular relapse in the whole cohort (A), in transplanted (B) and non-transplanted patients (C).

Morphologic relapse was documented in 14 patients (20%) including 2 pre-transplant, 3 post-transplant, and 9 in non-transplanted patients. The two patients with pre-transplant relapses were salvaged with etoposide-amsacrin or gilteritinib, then transplanted and achieved negative MRD. Targeted NGS analysis on available paired samples at diagnosis and morphologic relapse identified one patient with NPM1^mut loss and two patients with emergence of BAX mutations. We also observed two patients with FLT3-ITD loss or acquisition, and three patients with FTL3-TKD loss (Supplementary Table 3).

Fifty-four patients (77%) proceeded to allo-HSCT after a median of two VEN-AZA cycles (IQR 1-3). The median time between MRD_Rel or first VEN-AZA cycle and allo-HSCT was 138.5 days (IQR 112.3-189.5) and 94 days (IQR 66-109.8), respectively. Of the 48 evaluated patients, 34 (71%) achieved response in PB before transplantation, including 27 (56%) and 7 (15%) with MRD negativity or MRD reduction respectively. Most patients (n = 36, 67%) received reduced-intensity conditioning regimen mainly from unrelated donors (Supplementary Table 4). At last news, 47 were alive, including 46 (98%) with morphologic CR and 41 (87%) with negative MRD whereas 7 patients died including 4 deaths while in CR because of infection and/or graft versus host disease, and 3 after morphologic relapse. With a median follow up of 23.5 months (IQR 18-31.5), median OS was not reached with 1-y and 2y-OS of 96% and 86%, respectively (Fig. 1B). The median RFS and RFS_MRD were not reached. 1-y and 2y-RFS was 88% and 81%, and 1-y and 2y-RFS_MRD was 83% and 74% respectively (Supplementary Figure 1).

The 16 patients who did not proceed to allo-HSCT received a median of 9 VEN-AZA cycles (IQR 7-14). The reasons for not proceeding to allo-HSCT were comorbidities or age (n = 11), donor availability (n = 2), refusal (n = 1), or disease progression (n = 2). During follow-up, 9/16 patients relapsed, and 7/16 were still in morphologic remission, including 6 (44%) with negative MRD. Two patients discontinued VEN-AZA after 8 and 14 cycles and maintained negative MRD with treatment-free survival of 29 and 9 months, respectively. With a median follow up of 13.5 months (IQR 8.8-25.5), median OS was not reached with 1-y and 2y-OS of 93% and 71%, respectively (Fig. 1C). The median RFS was 14 months with 1-y and 2y-RFS was 61% and 38%, respectively. The median RFS_MRD was 12 months with 1-y and 2 y RFS_MRD was 41% and 27% respectively (Supplementary Figure 1).

VEN-AZA treatment for molecular relapse is safe and effective in patients with NPM1 mutations. As patients with molecular relapse have no disease-related symptoms, it was expected that general and hematological toxicity would be less pronounced than in the active phase of the disease [12]. Most patients achieved a second molecular response quickly and were therefore able to undergo transplantation in the best possible conditions (i.e., with optimal response and limited sequelae from salvage treatment toxicity).

This is the largest study evaluating VEN-AZA in the specific situation of ELN-defined molecular relapse during first line therapy in non-transplanted patients. Other recent studies have evaluated venetoclax-based combinations in oligoblastic relapses or molecular failure including molecular relapse, progression, or persistence. In the VALDAC prospective trial [12], 48 patients were enrolled including 22 oligoblastic relapse and 26 MRD_Rel, treated with low dose cytarabine (LDAC) and venetoclax. In the MRD relapse cohort, the rate of negative MRD was 55% in the 20 NPM1^mut patients. In the retrospective study conducted by Jimenez-Chillon et al. [7]. 79 NPM1^mut AML patients were treated with VEN-AZA or VEN-LDAC for molecular failure, including 34/43 patients (79%) with molecular relapse (i.e. conversion from MRD negativity to positivity confirmed on a second sample) who achieved negative MRD In these two studies, 2-year OS was 63% [12] and 67% [7] which, combined with our result (2-year OS, 82%), compares favorably with the outcome of NPM1^mut AML patients treated with intensive chemotherapy for morphologic relapse [12, 13].

Our study has several limitations due to its retrospective, non-comparative nature, and the limited number of patients, especially CBF-AML patients. VEN was used off-label with a heterogeneous treatment duration and the timing of molecular evaluation was not pre-specified. Finally, patients who may have progressed rapidly from molecular to morphologic relapse were not included, so we were unable to assess the proportion of patients in this case. Nevertheless, our study contributes to provide benchmark data in a challenging clinical situation that needs to be prospectively studied.

In conclusion, VEN-AZA with its favorable efficacy/toxicity ratio appears to be a relevant therapeutic option for NPM1^mut patients in first molecular relapse. Furthermore, our study illustrates a recent change in practice and supports the creation of new ELN 2022 evaluation criteria that now include molecular relapse as an event to be considered in the calculation of EFS and RFS.