EHA Library - The official digital education library of European Hematology Association (EHA)

DNMT3A is frequently mutated in AML and associated with a poor overall (OS) and relapse-free survival (RFS). The presence of DNMT3A mutations (DNMT3A^mut) in early preleukemic stem cells and in apparently healthy elderly highlight its pathogenic role as a founder mutation and as pre-disposing event in leukemia. Within our clinical AMLSG trials we addressed the question if MRD monitoring in DNMT3A^mut patients (pts) has clinical impact and delivers further information for risk-adapted therapy and clonal hematopoiesis.

To monitor MRD for the most common DNMT3A^mut (DNMT3A^mut-R882H, n=111 and -R882C, n=48) in a large cohort of AML pts entered on three AMLSG treatment trials [AML HD98A (n=14; NCT00146120), AMLSG 07-04 (n=87; NCT00151242), AMLSG 09-09 (n=58; NCT00893399)].

DNMT3A^mut MRD monitoring was performed using a cDNA-based RQ-PCR-assay by TaqMan technology with a sensitivity between 10^-3 and 10^-4. MRD levels have been reported as normalized values of DNMT3A^mut transcripts per 10⁴ ABL1 transcripts (DNMT3A^mut/10⁴ ABL1).

In total, 1,168 samples [bone marrow (BM), n=615; peripheral blood (PB), n=553] from 159 DNMT3A^mut pts were analysed [diagnosis, n=256; during therapy, n=719; follow-up, n=193]. Median BM DNMT3A^mut transcript level (TL) at the time of diagnosis was 12690 (range, 0-54280); TL were not associated with known clinical characteristics or mutations in NPM1 or FLT3; there was no impact on OS and RFS. DNMT3A^mut TL during therapy were significantly higher in BM after induction I, consolidation I and II (p=0.01; p=0.0003; p=0.01) than in PB. After double induction therapy (DI) there was no difference in the median of TL between 102 pts in complete remission (CR) and 12 pts not in CR (12694 and 11309, respectively). There was no prognostic impact of BM DNMT3A^mut TL as log 10 transformed continuous variable during therapy with regard to death and relapse. Strikingly, the greatest TL reduction was seen after induction I (one log) whereas subsequent therapies did not significantly influence TL. Next, we evaluated the impact of DNMT3A^mut MRD monitoring for the clinical endpoints OS, cumulative incidence of relapse (CIR) and remission duration (RD) after DI and after end of therapy (ET). After DI and ET, only 7/75 and 4/63 BM samples became MRD negative. At these two time-points MRD positivity did not significantly impact OS (p=0.89; p=0.73), CIR (p=0.74; p=0.29) and RD (p=0.61; p=0.30). Then we investigated the MRD DNMT3A^mut log10-reduction (compared to level at diagnosis) using the median as a cut-off after DI and ET. Again, there was no significant correlation for pts with a higher TL reduction compared with a lower TL reduction for OS and RD after DI and ET (p=0.87; p=0.38; p=0.67; p=0.57, respectively). Finally, we evaluated the BM DNMT3A^mut TL as 4 increasing equally sized intervals according to the quartiles of the distribution. There was no prognostic impact after DI on OS and RD (p=0.24; p=0.85) and ET (p=0.38; p=0.44). To increase the number of samples, we also performed the analyses for BM and PB samples, but again, there was no significant impact on prognosis.

In our study neither the absolute BM DNMT3A^mut TL at the time of diagnosis nor the reduction of TL or the increasing quartiles after DI and ET showed prognostic impact. In contrast to AML with NPM1 mutation or associated gene fusions, most pts had persistent DNMT3A^mut TL supporting the role of persistent clonal hematopoiesis. Serial investigation of DNMT3A^mut and its concurrent mutations will provide further insights into the clonal evolution of AML.