Acute myeloid leukemia with translocation t(3;5): new molecular insights

AML with translocation t(3;5) belongs to the “AML with myelodysplasia-related changes” defined in the 2008 WHO classification.1 The incidence of this balanced abnormality is less than 0.5% of AML.2 The identified breakpoints occur at 3q25.1 on chromosome 3 and at 5q34 on chromosome 5, where the nucleolar phosphoprotein nucleophosmin 1 (NPM1) is located.3 At locus 3q25.1, Yoneda-Kato et al. identified a new gene, myeloid/myelodysplastic leukemia factor 1 (MLF1), and thus highlighted the fusion transcript NPM1-MLF1.4 The physiological role of MLF1 has not been well characterized. In hematologic diseases, MLF1 was found to be overexpressed in more than 25% of myelodysplastic syndrome (MDS) in transformation phase and MDS-associated AML.5 NPM1, the partner gene of MLF1 in the t(3;5)(q25.1;q34) translocation, is better known for being affected by a 4 bp insertion in exon 12 that occurs in 30-35% of all AML cases.6

To better characterize AML with NPM1-MLF1, we report morphological, immunophenotypic, cytogenetic features, and the first description of gene mutations analysis and gene expression profiling (GEP) in this cytogenetic entity.

This study included 7 cases diagnosed between 2002 and 2011 in France; case n. 1 has been reported previously.7 The main clinical and biological characteristics of the patients studied are shown in Table 1. Molecular analysis was performed on cryo-preserved bone marrow mononuclear cells. The RT-PCR for the detection of NPM1-MLF1 fusion transcript was carried out using the primers described by Yoneda-Kato et al.4 The screening for mutations in NPM1, FLT3, CEBPA, WT1, IDH1/2, DNMT3A was performed as previously reported.8 GEP was performed in 4 patients (UPN 2, 3, 6 and 7), according to standard protocol with Genechip Affymetrix HG 133 plus 2.0 array.9

The analysis of the Acute Leukemia French Association (ALFA) trials database showed a very low incidence (3 of 1333; 0.23%) of AML with NPM1-MLF1 among adult AML patients with available karyotype. This incidence is consistent with that reported by Grimwade et al. (<0.5%).10 The mean age of patients with NPM1-MLF1 was 24 years. Cytological characteristics showed that a 3-lineage dysplasia was present in the great majority of cases, both in peripheral blood and bone marrow smears. Dysmegakaryopoiesis was constantly observed, dysgranulopoiesis was associated with peroxidase deficiency in 3 cases, and dyserythropoiesis occurred in 5 of 7 cases. Flow cytometry analysis showed that blast cells were positive for the myeloid antigens CD117, CD13, and CD33. CD34 was negative in 6 of 7 patients, as frequently observed in NPM1 mutated AML.11 Cytogenetic analysis showed that the translocation t(3;5)(q25.1;q34) was present as the sole abnormality at diagnosis in all cases. The presence of the chimeric fusion transcript NPM1-MLF1 was confirmed in each case. Furthermore, all 7 cases harbored WT1 overexpression and at least one mutation (WT1 exon 7, FLT3-ITD and/or IDH2R140 mutation) was identified in the 4 adult cases (Table 1). In contrast, no mutation was found in the 3 pediatric cases. Considering age at AML diagnosis, associated gene mutations in NPM1-MLF1 positive AML seem to be similar to those in NPM1 mutated AML. DNMT3A mutations, that occur in 50-60% of NPM1 mutated AML, were not found in our cohort. GEP of AML with NPM1-MLF1 was performed searching for a specific signature associated with this translocation. In order to do this, we compared 4 patients with NPM1-MLF1 to a reference group consisting of 436 patients with AML without t(3;5). Only two probes targeting MLF1 were significantly overexpressed (P<10) (Figure 1A). In a second step, we performed principal component analysis using the same patient cohort and a set of 189 probes defining a robust NPM1 mutated signature (Online Supplementary Table S1). This analysis revealed that patients with AML with NPM1-MLF1 co-segregated with the group of AML with NPM1 mutations (Figure 1B). Thus, our results suggest that NPM1-MLF1 positive AML and NPM1 mutated AML may share common signaling pathways critical for leukemogenesis.

Table 1.Clinico-biological characteristics of the 7 AML patients with t(3;5)(q25.1;q34).

Figure 1.Results of gene expression profiling. (A) Volcano plot representing probes with statistically significant changes in gene expression among 189 probes used in 126 AML NPM1 mutated without FLT3-ITD as control group and 4 AML with t(3;5)(q25.1;q34). The arrows indicate MLF1 probes. The horizontal line and the vertical lines show respectively P=0.001 and fold-change >3. (B) Principal component analysis representing 436 patients with AML separated on a robust NPM1 mutated signature and 4 AML with t(3;5)(q25.1;q34). Blue squares represent NPM1 wild-type patients; green triangles represent NPM1 mutated patients; red circles represent patients with t(3;5)(q25.1;q34).

Regarding clinical outcome, all patients achieved complete remission (Table 1). Four out of 7 patients died, 2 from a cause other than leukemia. The 3 patients alive are in first complete remission, and minimal residual disease based on WT1 expression and NPM1-MLF1 transcript still remains negative at last follow up. Overall, the clinical outcome of the 7 patients studied appears consistent with the intermediate prognosis recently reassessed from poor prognosis by Grimwade et al.10

In conclusion, the comparison of AML with NPM1-MLF1 and AML with NPM1 mutations showed similar immunophenotypical and molecular features, including gene mutation patterns and GEP. Our findings suggest that these two types of AML may share common signaling pathways critical for leukemogenesis. However, this hypothesis needs to be confirmed in larger patient cohorts.