DNMT3A mutations in De Novo Myelodysplastic Syndrome: Distinct Clinico-Biological Features and Prognostic Relevance

Mutations of the DNMT3A gene, which encodes the enzyme DNA methyltransferase 3A, were identified in patients with myeloid malignancies and are associated with poor prognosis in primary AML patients. However, the clinical and prognostic implications of these mutations in myelodysplastic syndrome (MDS) remain to be determined.

A total of 328 de novo MDS patients diagnosed according to French-American-British (FAB) criteria at the National Taiwan University Hospital who had cryopreserved bone marrow cells for study were recruited into mutational analyses. Mutations in DNMT3A gene at exon 2–23 were analyzed by polymerase chain reaction and direct sequencing. The results were correlated with clinical features, cytogenetics, gene mutations and treatment outcomes.

Among the 328 patients, 115 patients (35.0%) had refractory anemia (RA), 19 (5.8%) had RA with ring sideroblasts (RARS), 122 (37.2%) had RA with excess blasts (RAEB), 35 (10.7%) had RAEB in transformation (RAEBT), and 37 (11.3 %) had chronic myelomonocytic leukemia (CMMoL). DNMT3A mutations at 20 different positions were identified in 33 patients, including thirteen missense mutations, two nonsense mutations and five frame-shift mutations. Among these 33 patients, 31 had single mutation of DNMT3A, and the other 2 patients had double mutations. The most common mutation was R882H (n = 8), followed by R882C (n = 7), Y735C (n=2), and R720H (n=2). All other mutations were detected in only one patient each. Totally, DNMT3A mutations were identified in 33 (10.1%) of 328 patients diagnosed according to the FAB classification and in 25 (9.8%) of 256 diagnosed according to 2008 WHO classification.

DNMT3A-mutated patients were older (median age, 74 years vs. 66 years, P=0.048) and had higher platelet counts at diagnosis than DNMT3A-wild patients (median, 123.5×10³/μL vs. 73 ×10³/μL, P=0.016). According to FAB classification, patients with RARS had the highest incidence (26.3%) of DNMT3A mutations, followed by RAEBT (14.3%), RAEB (11.5%), and CMMoL (8.1%), whereas those with RA had the lowest incidence (5.2%, P=0.035). Chromosome data were available in 308 patients (93.9%) at diagnosis and clonal chromosomal abnormalities were detected in 155 patients (50.3%). There was no difference in the distribution of 2008 WHO classification, karyotype and international prognostic scoring system (IPSS) between patients with and without DNMT3A mutations.

To investigate the association of gene mutations in the pathogenesis of MDS, a mutational screening of 10 other genes was also performed. Among the 33 patients with DNMT3A mutations, 16 patients (48.5%) showed additional molecular abnormalities at diagnosis, including seven with concurrent IDH1/IDH2 mutations, seven ASXL1 mutations, five AML1/RUNX1 mutations, two MLL-PTD, two RAS mutations and one JAK2 mutation. Eight of these 16 patients (50%) had two other concurrent mutations, and the others had one additional mutation. It's clear that DNMT3A mutation was closely interacted with IDH mutation in MDS (IDH mutation occurring in 21.2% of DNMT3A-mutated patients vs. 3.4% in DNMT3A-wild ones, P=0.001).

With a median follow-up of 57.6 months (range, 0.1–250.7 months), there was no significant difference in overall survival (OS) between patients with and without DNMT3A mutation by either FAB or 2008 WHO classifications (median, 22.5 months vs. 30.9 months, P=0.669 and 25.2 months vs. 34.9 months, P=0.538, respectively) as well as in the rate of acute transformation. However, among the subgroup of patients with RA by FAB classification or refractory cytopenia with unilineage dysplasia by 2008 WHO classification, DNMT3A-mutated patients had significantly shorter OS than DNMT3A-wild patients (median, 28.3 months vs. 39.8 months, P<0.001 and 23.8 months vs. 40.3 months, P=0.026, respectively). Further, DNMT3A mutation is an independent poor prognostic factor in these two subgroups.

Our findings provided evidence that DNMT3A mutations could be detected in a substantial portion of de novo MDS patients. DNMT3A mutations are associated with distinct clinical and biological features and poor prognosis in selected groups of patients.

No relevant conflicts of interest to declare.