PS1060 CO‐OCCURRENCE OF DNMT3A‐R882, FLT3 AND NPM1 MUTATIONS IDENTIFIES A SUBSET OF AML PATIENTS WITH ADVERSE PROGNOSIS

Abstract

Background:DNMT3A mutations are found in approximately 22% of adult acute myeloid leukemia (AML) cases, occurring more commonly in a hotspot at codon 882. These mutations are an early event in leukemogenesis, causing impaired DNA methylation and aberrant gene expression. Previous studies have shown that isolated DNMT3A mutations may cause limited clonal hematopoiesis in healthy individuals, however, additional mutations in genes such as FLT3 or NPM1 are required for leukemic transformation. Despite its well‐recognized role in AML development, little is known about the prognostic repercussions of the interaction between these three genes.Aims:Here, we aimed to investigate i) the clinical and biological differences between #R882 (hotspot) and non‐R882 mutations in AML and ii) the impact of the synergism between DNMT3a, FLT3 and NPM1 mutations on patient survival.Methods:218 non‐selected adult AML patients (non‐AML M3) followed in a reference center in northeast Brazil were enrolled. Additionally, 180 patients from The Cancer Genome Atlas (TCGA) databank were evaluated as an independent cohort. Brazilian patients were treated according to the 7+3 chemotherapy protocol followed by consolidation with high doses of cytarabine, while patients from the TCGA cohort were treated according to the National Comprehensive Cancer Network recommendations. Screening for DNMT3a mutations was performed by direct Sanger Sequencing.Results:DNMT3A R882 mutations were associated with FAB monocytic AML subtypes (M4 and M5), higher leukocyte counts, normal karyotype and FLT3 and NPM1 mutations (more evident for FLT3/NPM1 double‐positivity), while non‐R882 mutations were associated with NRAS and IDH2 mutations. Analysis of TCGA patient's transcriptome data revealed that patients harboring concomitant DNMT3a, FLT3 and NPM1 mutations had different patterns of gene expression compared to their counterparts with isolated mutations. Next, to address whether the association between DNMT3a R882, FLT3 and NPM1 mutations is clinically relevant, we created a composite variable grouping patients harboring DNMT3a R882 mutations with at least one mutation in FLT3 or NPM1 (here referred to as R882/FLT3/NPM1). Strikingly, patients assigned to the R882/FLT3/NPM1 group had significantly poorer overall (Brazilian cohort: HR: 2.44, 95% CI: 1.35–4.44; P = 0.002 and TCGA cohort: HR: 1.83, 95% CI: 1.09–3.06; P = 0.022) and disease‐free survival rates (Brazilian cohort: HR: 1.94, 95% CI: 1.08–5.92; P = 0.027 and TCGA cohort: HR: 2.06, 95% CI: 1.11–3.81; P = 0.021) in both cohorts. Similar results were observed when analysis was restricted to the normal‐karyotype group. On the other hand, individual analysis of DNMT3a (or #R882 only) mutations showed only a tendency for shorter survival that did not reached statistical significance in both cohorts. Proportional hazards modeling showed that R882/FLT3/NPM1, but not single mutational status, remained independently predictive of poor outcomes (Brazilian cohort: HR: 1.97, 95% CI: 1.06–3.84; P = 0.048 and TCGA cohort: HR: 2.14, 95% CI: 1.24–3.68; P = 0.006).Summary/Conclusion:AML patients with R882 mutations have distinct clinical and molecular signatures from those with non‐R882. At the transcriptome level, triple positive‐mutated patients (DNMT3a‐R882, FLT3 and NPM1) cluster differently from the single positives ones, suggesting functional interaction between these mutations. By transposing these findings into a composite variable, we show that the synergism between DNMT3a, FLT3 and NPM1 mutations identifies a subset of AML with a particularly adverse prognosis.