DNMT3A Mutations Independently Predict Poor Outcome in Older AML Patients: A SWOG Report,

Mutations of the DNA methyltransferase 3A (DNMT3A) gene, which encodes DNA cytosine-5-methyltransferase 3A, have recently been reported to have prognostic implications in adult acute myeloid leukemia (AML). We studied the incidence and prognostic impact of these mutations, in the context of other prognostic markers, in older patients with de novo AML.

Diagnostic specimens were available for 268 patients who were registered on SWOG trials S-9031 and S-9333 (both of which enrolled patients > 55 years), and S-9500 (which enrolled patients from 18–55 years). The median age of patients was 63 years (range, 18 to 88 years). The samples were analyzed for the presence of DNMT3A mutations via direct sequencing of exon 23. DNMT3A mutation status was correlated with clinical characteristics, other genetic risk factors (cytogenetics and mutations in NPM1, FLT3-ITD, Kit, CEBPA and IDH1/2) and outcome.

DNMT3A mutations were found in 50 patients (19%). A total of 9 different mutations were identified (R882H, R882C, R882S, L901R, C911Y, N879D, P904L, R887I and W893S). The codon most frequently mutated was R882 (n=44). There were no statistically significant differences in the median age, WBC, blast or platelet count at diagnosis between mutated DNMT3A and wild type (WT). The DNMT3A mutations occurred most commonly in cytogenetically normal (CN) patients and did not occur in those with core binding factor abnormalities. DNMT3A mutations were more common in NPM1+ patients (67% vs 27%, p<0.01) and were somewhat more common in FLT-ITD+ patients (26% vs 16%, p=0.059) patients. There were no differences in the incidence of CEBPA, KIT, or IDH1/2 mutations among patients with DNMT3A mutations as compared to DNMT3A WT. When adjusting for age, PS, WBC, blast count, FLT3/NPM1 status, and cytogentics, the presence of DNMT3A mutations independently predicted worse OS (20% vs 29% at 2 years, HR=2.3, 95% CI: 1.4–3.9, p=0.002) and worse EFS (14% vs 23% at 2 years, HR=2,1, 95% CI: 1.3–3.3, p=0.003). In CN-AML, 33% of the patients (p<0.001) harbored DNMT3A mutations. In this patient subgroup, multivariate analysis revealed that the DNMT3A mutations predicted a shorter OS (23% vs 40%, HR=1.8, 95% CI: 1.1–3.0, p=0.03) and EFS (17% vs 32%, HR=2.0, 95% CI: 1.2–3.2, p=0.01). Among the high-risk NPM1/FLT3-ITD subgroup (FLT3-ITD+ regardless of NPM status, and FLT3-ITD-/NPM1-) OS was significantly worse in the DNMT3A+ patients (12% vs 30%, HR=2.1, 95% CI=1.3–3.4, p=0.002) as was the EFS (12% vs 22%, HR=1.9, 95% CI= 1.2–2.9, p=0.006). Conversely, in the FLT3-ITD-/NPM1+ favorable subgroup, DNMT3A+ patients trended towards improved RFS (71% vs. 47%, HR=0.2, 95% CI=0.1–1.1, p=0.063), although there was no difference in OS (40% vs. 40%, HR=0.9, CI=0.4–2, p=0.82). When restricting analysis to elderly patients, DNMT3A+ patients older than 60 years had significantly worse EFS (0% vs 14%, HR=1.6, 95% CI: 1.0–2.3, p=0.04) and trended towards worse OS (10% vs 18%, HR=1.8, 95% CI: 1–1.2, p=0.08), although this did not reach statistical significance.

The high prevalence of DNMT3A mutations and its independent negative impact on OS and EFS identifies DNMT3A as a potentially important molecular marker for patients with AML, especially those with CN-AML. Within the high-risk FLT3-ITD/ NPM1 group, DNMT3A mutations identify a subset of patients with especially poor prognosis. Our results are in keeping with the recent published studies on the prognostic impact of DNMT3A in patients with AML and add valuable information for risk stratification of older AML patients. DNMT3A mutation status should be evaluated prospectively in future adult AML clinical trials.

No relevant conflicts of interest to declare.