DNMT3A Mutations in Acute Myeloid Leukemia

Abstract SCI-31

DNMT3A is one of the two human de novo methyltransferases; both are essential for methylating specific cytosine residues, a process that is critical for regulating gene expression during early development. In 2010, recurring mutations in DNMT3A were discovered in patients with AML (1, 2); mutations were not found in the other de novo methyltransferase (DNMT3B) or in the binding partner for both (DNMT3L). After discovering the first mutation by whole genome sequencing, we sequenced all of the exons of DNMT3A in 281 additional AML patients and found a total of 62 mutations (22% of all cases) in the gene that were predicted to affect translation. Eighteen different missense mutations were found, the most common of which was at amino acid R882 (37 patients). We identified six frameshift, six nonsense, and three splice site mutations, and also a 1.5 MB deletion that included the entire DNMT3A gene. DNMT3A mutations are highly enriched in patients with intermediate risk cytogenetics (56 of 166 patients, 33.7%) and were absent in 79 patients with favorable risk cytogenetics (p < 0.001 for both). In our series, DNMT3A mutations were independently associated with poor outcomes in multivariate analyses. Since the initial reports, DNMT3A mutations have been identified in ∼5%–10% of patients with myelodysplastic syndromes (3, 4), in 10%–20% of patients with myeloproliferative neoplasms and secondary AML (5), and in pediatric AML cases, but at much lower frequencies than adult AML (0%–1%) (6, 7). The adverse clinical outcomes associated with DNMT3A mutations have been confirmed by Yan et al (8) and by Thol et al (9). DNMT3A mutations are therefore among the most common in AML and other myeloid malignancies, and may be important for identifying patients with adverse outcomes. However, the mechanisms by which DNMT3A mutations influence AML pathogenesis are not yet clear. In our original study, we did not identify a clear gene expression signature associated with DNMT3A mutations, nor were we able to identify global or focal alterations in DNA methylation patterns that are clearly caused by the mutations. Yan et al (8) reported that DNMT3A mutations were associated with increased expression and hypomethylation of several HOXB genes, but this could not be validated with our data. Many important questions about DNMT3A mutations remain to be answered, including: 1) What are the relationships between DNMT3A mutations and other mutations that affect outcomes?, 2) Do DNMT3A mutations predict responses to hypomethylating agents or other regimens?, 3) Are the recurring mutations at R882 associated with a gain-of-function activity or do they act as dominant negatives?, and 4) Do the missense mutations alter methylase activity, DNA binding and/or methylation specificity, protein/protein interactions, or other (as yet unknown) functions of DNMT3A? Although the mechanism of action of DNMT3A mutations is currently unknown, Challen et al (10) noted that loss of Dnmt3a in the hematopoietic cells of mice caused a dramatic expansion of stem cells upon serial transplantation. These results strongly suggest that loss-of-function mutations in DNMT3A may affect the self-renewal properties of hematopoietic stem cells, which may be relevant for AML pathogenesis.