Relationship between salvage strategy and outcomes in patients with CBF or NPM1-mutated AML and first molecular relapse

Abstract Introduction: Patients with Core-Binding Factor (CBF) and NPM1-mutated acute myeloid leukemia (AML) can be monitored by quantitative PCR after having achieved complete remission (CR) to detect molecular relapse with a growing interest in the use of preemptive therapy to prevent overt morphologic relapse. The best salvage strategy, including lower-intensity venetoclax (VEN)-containing regimens and allogeneic hematopoietic cell transplantation (HCT), is however unclear. In this study, we retrospectively analyzed the outcome of CBF or NPM1-mutated AML patients who received preemptive therapy for molecular relapse according to the salvage strategy used. Methods: All CBF and NPM1-mutated AML adults, who had molecular relapse after first-line therapy including induction therapy with daunorubicin/idarubicin and cytarabine followed by standard chemotherapy consolidation, without allogeneic HCT in first CR, were included. Molecular relapse was defined according to ELN recommendations (Heuser M, Blood 2021) as a conversion from MRD negativity to positivity or an increase in MRD copy numbers >1 log between two positive samples. MRD response included patients with MRD negativity and MRD reduction ≥1 log. Results: Between 2010 and 2023, we included 121 adults from 12 centers with CBF (n=28) and NPM1-mutated (n=93) AML who had molecular relapse. The different salvage strategies included upfront allogeneic HCT (n=19), intensive chemotherapy (IC; n=21), venetoclax and azacitidine combination (VEN-AZA; n=70), and other strategies (n=11; including AZA, gemtuzumab ozogamicin, selective inhibitors). Patients receiving upfront allogeneic HCT had lower transcript levels at molecular relapse (median, 1% vs. 4%, 17%, and 14% for IC, VEN-AZA, and other in bone marrow, respectively, P=0.033 and 0% vs. 1%, 2%, and 3% in peripheral blood, respectively, P=0.036) and a longer time to initiation of salvage therapy (72 days vs. 50, 48, and 60 days, respectively, P=0.01) while patients receiving VEN-AZA were older (58 years vs. 51,41, and 44 years for upfront allo, IC, and other, respectively, P<0.001) and more likely to have NPM1-mutated AML (94% vs. 37%, 48%, and 91%, respectively, P<0.001). MRD response was higher in patients undergoing IC (89% [72% MRD negativity, 17% MRD reduction] vs. 72% for VEN-AZA [33% MRD negativity, 39% MRD reduction] vs. 17% for other [17% MRD negativity], P=0.002). With a median follow-up of 3.38 years (interquartile [IQR]: 2.39-4.95) in survivors, there were 27 deaths contributing to the estimates for overall survival (OS). At three years, OS was not statistically different between the four groups (84% [69-100%] for upfront allo vs. 81% [66-100%] for IC vs. 79% [69-90%] for VEN-AZA vs. 64% [41-100%] for other, P=0.31). After multivariable adjustment, age at molecular relapse (hazard ratio [HR]=1.04 [1.00-1.08], P=0.049), male gender (HR=1.37 [1.12-3.33], P=0.007), and transcript levels in blood at initiation of salvage (HR=1.01 [1.00-1.01], P<0.001) were associated with OS. Allogeneic HCT was received by 98 patients (81%) with a cumulative incidence of allogeneic HCT at 12 weeks of 100% for upfront allo, 71% [51-92%] for IC, 73% [50-97%] for VEN-AZA, and 82% [55-100%] for other (P<0.001). In a 190-day landmark analysis, when 75% of transplant recipients had undergone the procedure, three-year OS was statistically significantly higher in those who received allogeneic HCT (82% [73-93%] vs. 70% [57-86%] for those who did not receive HCT at 190 days, P=0.041). In patients who received allogeneic HCT, type of salvage therapy was not statistically associated with post-HCT relapse (11% [0-25%] for upfront allo vs. 13% [0-31%] for IC vs. 19% [7-32%] for VEN-AZA vs. 10% [0-30%] for others, P=0.77), relapse-free survival (90% [77-100%] vs. 80% [62-80%] vs. 71% [58-86%] vs. 60% [36-100%], P=0.2), or OS (90% [77-100%] vs. 86% [70-100%] vs. 86% [77-96%] vs. 70% [47-100%], P=0.42) but with non-relapse mortality (0% vs. 7% [0-15%] vs. 10% [2-18%] vs. 30% [0-60%], P=0.043) at three years. Conclusion: Our data suggests that different salvage strategies are associated with similar outcomes with a similar proportion of patients receiving allogeneic HCT. Allogeneic HCT is associated with increased OS in the context of molecular relapse. Upfront allogeneic HCT may be a valuable option in patients with low transcript levels at molecular relapse and an available donor. Either IC or VEN-AZA can be used as bridge to transplant in other patients.