Abstract 2506


The development of massively parallel sequencing technologies has led to the identification of somatic DNA methyltransferase 3A (DNMT3A) gene mutations in acute myeloid leukemia (AML), with the highest frequency being found in cytogenetically normal (CN) AML. DNMT3A mutations have been suggested to predict poor clinical outcome in AML, but only few data are available on their prognostic significance within CN-AML. The aim of this study was to determine the frequency, the main associated features, and the prognostic significance of DNMT3A mutations in CN-AML.

Patients and methods:

This retrospective study was performed in 123 young adult patients (16–60 years) with previously untreated primary CN-AML and enrolled on two concomitant protocols of the Acute French Leukemia Association (ALFA), the ALFA-9801 and ALFA-9802 trials. DNMT3A mutations were screened on genomic DNA by PCR and direct Sanger sequencing. We focused our screening on the 3 conserved domains of DNMT3A (the proline-tryptophane-tryptophane-proline (PWWP) domain, the ADD-type zinc finger domain, and the C5-methyltransferase domain), corresponding to exons 8–9 and 11–23. The patients were also assessed for the presence of FLT3 internal tandem duplication (FLT3-ITD), FLT3 tyrosine kinase domain (FLT3-TKD), NPM1, CEBPA, WT1, IDH1, and IDH2 mutations.


Thirty-eight DNMT3A mutations were identified in 36 of the 123 (29%) patients. These alterations consisted of 36 nucleotide substitutions and 2 frameshift deletions. Thirty out of 36 (83%) nucleotide substitutions affected the amino acid residue R882 (R882H, n = 21; R882C, n = 7; R882P, n = 2), 5 represented other missense alterations, and 1 was a nonsense mutation. Two patients exhibited 2 heterozygous missense mutations in different exons, and one patient had a homozygous missense mutation. DNMT3A mutated and wild-type cases did not differ in terms of age, gender, and white blood cell (WBC) count at presentation. DNMT3A mutations were strongly associated with the French-American-British (FAB) subtypes M4/M5 (P =.0002) and the presence of NPM1 mutations (P =.0006), and tended to often co-occur with IDH1R132 mutations (P = .09). In the entire cohort, complete remission rate was found lower in DNMT3A mutated patients than in DNMT3A wild-type patients, but without reaching statistical significance (80% vs 90%, P =.24). DNMT3A mutated patients had a shorter event-free survival (5-year EFS: 13% vs 32%, P =.02) and overall survival (5-year OS: 23% vs 45%, P =.02) compared to DNMT3A wild-type patients. We next performed subgroup analysis according to the NPM1/FLT3-ITD genotypes. In patients with the non-favorable genotypes, that is NPM1 mutated/FLT3-ITD positive, NPM1 wild-type/FLT3-ITD positive, NPM1 wild-type/FLT3-ITD negative (n = 86), 18 (21%) had a concomitant DNMT3A mutation. In this high-risk subgroup of CN-AML, DNMT3A mutations conferred a worse clinical outcome (5-year EFS: 0% vs 23%, P =.02; 5-year OS: 14% vs 37%, P =.06). In patients with the favorable genotype NPM1 mutated/FLT3-ITD negative (n = 37), 18 (49%) were found to display a concomitant DNMT3A mutation. Within this favorable subgroup, patients carrying a DNMT3A mutation had a significantly inferior EFS and OS compared to DNMT3A wild-type patients (5-year EFS: 25% vs 65%, P =.01; 5-year OS: 29% vs 76%, P =.02). Furthermore, in multivariate analysis including age, WBC count, NPM1/FLT3-ITD genotypes, and DNMT3A mutational status, the presence of a DNMT3A mutation remained an independent adverse prognostic factor for EFS (hazard ratio = 2.29; 95% CI, 1.42 to 3.70; P =.0007) and OS (hazard ratio = 2.34; 95% CI, 1.37 to 4.00; P =.002).


DNMT3A mutations are one of the most common gene mutations in CN-AML and independently predict poor clinical outcome. Testing for DNMT3A mutations could help further improve risk stratification in CN-AML.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.