Impact of DNMT3a Status on Post Induction NPM1 MRD Predictive Value on Survival in Elderly AML Patients Treated Intensively

Minimal residual disease is now a powerfull surrogate marker to assess response to chemotherapy in acute myeloid leukemia (AML). In younger adults, NPM1 MRD has recently demonstrated to be a favorable predictive marker for EFS and OS independently of fms-like tyrosine kinase-3 internal tandem duplications (FLT3-ITD) status. However, there is very few datas regarding predictive value of NPM1 MRD in elderly patients treated with intensive chemotherapy. Moreover, numerous studies have suggested the negative impact of DNMT3a mutation in NPM1 AML patients, especially in those with concurrent FLT3-ITD mutation. In this study, we aimed to investigate the impact of DNMT3a status on post induction NPM1 MRD1 predictive value for survival in a retrospective cohort of AML patients aged over 60 years old treated intensively.

A total of 138 patients treated for NPM1 mutated AML in two French institutions (Lyon, Lille) were analyzed retrospectively. Median age of the entire cohort was 66.1 years old (range 60-78.2). An FLT3-ITD mutation was evidenced in 52 of 138 patients (37.6%) with a median FLT3-ITD AR of 0.53 (range, 0.05-3). With a median follow-up of 19.61 months (0.07-128.4), the overall CR rate was 89.9% with no influence of DNMT3a or FLT3 mutational status on the probability of CR. In this elderly cohort of NPM1_mut patients, a 4log reduction of NPM1 bone marrow (BM) MRD1 was associated with better outcome (median OS: NR vs 13.4 months, HR=0.35, p<0.01)(Figure A). Overall, DNMT3 status did not influence the probability of having a ≥ 4log MRD1 reduction after induction. However, only 9/44 (20.4%) FLT3-ITD patients reached ≥ 4log MRD1 reduction whereas 38/80 FLT3_wt (47.5%) were good molecular responders (p<0.001). FLT3-ITD mutated patients who achieved a 4log reduction had a superior outcome compared to those who did not (HR=0.34; 95% CI, 0.16 to 0.70; P <0.001). Similarly, NPM1_mut FLT3_wt patients with a 4log reduction in NPM1 BM-MRD1 had a longer OS (3-year OS, 68.1%; 95% CI, 48.8 to 82.9) than those without good molecular response (3-year OS, 46.5%; 95% CI, 30.2 to 61.7)(Figure B). DNMT3a negative patients who achieved a 4log reduction had a superior outcome to those who did not reached at least a 4log reduction (HR=0.23; 95% CI, 0.07 to 0.72; P <0.001). However, postinduction NPM1 MRD1 reduction was not predictive of OS and leukemia free survival (LFS) in DNMT3a_mut patients. DNMT3a_mut patients has a very poor LFS which was even worst in poor NPM1 MRD1 responders compared to those who reached at least 4log reduction (median LFS: 8.3 months vs 17.4 months, HR = 0.48, 95% CI, 0.25-0.91, p=0.023)(Figure C). In multivariate analysis, only DNMT3a mutational status and a 4-log reduction in NPM1 BM-MRD were significantly associated with survival. Based on these results, we identified among NPM1 positive patients 3 groups with distinct prognosis, based on FLT3-ITD, DNMT3a status and NPM1 BM-MRD post induction response (NPM1 scoring system)(Figure D). When compared to ELN 2017 intermediate risk group (AUC=0.695), NPM1 scoring system (NPM1 SS) was more accurate for OS prediction in patients within intermediate (AUC=0.833) and unfavorable (AUC=0.863) NPM1 SS risk group. However, there was no significant difference in AUC between NPM1 SS favorable and ELN 2017 favorable risk group.

These results confirm that post-induction NPM1 MRD1 is a reliable tool to assess disease outcome in elderly AML patients. However, the presence of DNMT3a also identify a subgroup of patients at very high risk of relapase, despite good molecular responses. As hematopoietic stem cell transplantation (HSCT) might improve OS in elderly patients, DNMT3a positive AML elderly patients should be considered for HSCT or post induction maintenance strategies, even within the favorable ELN risk group.