NUP98/JARID1A Is a New Recurrent Genetic Abnormality in Pediatric Acute Megakaryoblastic Leukemia with a Distinct HOX-Gene Expression Pattern

Cure rates in pediatric AML are currently in the 60–70% range despite treatment with intensive chemotherapy. To improve prognosis new treatment targets need to be identified, hence there is a need to better understand the underlying biology. It is hypothesized that AML results from at least two types of mutations which non-randomly collaborate in leukemogenesis. The type-I aberrations confer a proliferative advantage, type-II mutations lead to impairment of hematopoietic differentiation (Kelly et al, 2002). We recently described NUP98/NSD1 as recurrent event in cytogenetically normal AML (Hollink et al, 2011). Patients with NUP98/NSD1 had dismal outcome, and a stem-cell phenotype characterized by overexpression of homeobox (HOX) A and –B genes. Using split-signal FISH on 122 pediatric AML cases without driving oncogenic mutation, 26 NUP98- rearranged cases were identified, including 1 patient with acute megakaryoblastic leukemia (AMKL). We previously reported a patient with fusion of JARID1A, located on chromosome 12p13, to NUP98, located on chromosome 11p15, in a non-Down Syndrome (DS) AMKL case (Van Zutven et al, 2006). Therefore, a large series of non-DS AMKL patients was screened for NUP98/JARID1A and for other abnormalities, including the novel CBFA2T3/GLIS2 translocation (Gruber et al, ASH2011; #757).

Samples from 105 pediatric non-DS AMKL cases, diagnosed between 1998 and 2011, were obtained from the DCOG, the AML-BFM SG, the Saint-Louis Hospital in Paris, and the COG. AMKL is more common in DS patients, therefore we also screened a series of DS AMKL (n=16). Centrally reviewed clinical and cell-biological data were provided by these study groups. Translocation of NUP98/JARID1A, MLL-rearrangements, RBM15/MKL1, and CBFA2T3/GLIS2 were identified using RT-PCR, as well as molecular characterization including hospots for the following mutations: FLT3, KIT, RAS, PTPN11, NPM1, WT1, and CEBPA. HOXA and –B expression levels were analyzed using gene expression profiling (Affymetrix) in 274 pediatric AML patients (Balgobind et al, 2011) including 9 AMKL patients, and validated with quantitative real-time PCR (n=37).

NUP98/JARID1A translocations were identified in 11 patients (11%). Four other patients had a NUP98- aberration with unknown translocation partner based on split signal FISH. We identified 16/105 patients with RBM15/MKL1, 13/105 with CBFA2T3/GLIS2 translocation, and 13/96 harbouring an MLL-rearrangement. Hence, specific non-random abnormalities could be defined in 61% of pediatric AMKL cases. Only 3/45 cases harboured a type-I mutation, all localized in the RAS gene. Comparing NUP98/JARID1A positive cases with negative cases in pediatric AMKL, no significant differences in patient characteristics including sex, age, and white blood cell count (WBC) were found. Considering prognosis, 5-year pEFS (22±14% vs. 36±6%, p=0.50) did not differ significantly from all other AMKL patients, nor did the cumulative incidence of relapse (56±19% vs. 54±7% p=0.9). CBFA2T3/GLIS2 translocated patients also did not differ from other AMKL patients (pEFS 19±16% vs. 36±6%, p=0.63). However, 5-year pEFS for RBM15/MKL1 translocated patients was significantly better (73±13% vs. 28±6%, p=0.043), but not in multivariate analysis adjusted for age and WBC. Gene expression analysis showed significantly higher HOXA5/A9/A10 and HOXB2/B3/B4/B5/B6 expression in NUP98/JARID1A compared to other pediatric AML cases. We did not identify any NUP98/JARID1A cases in the 16 DS AMKL patients.

NUP98/JARID1A is a recurrent cryptic translocation in approximately 11% of pediatric AMKL cases. In 61% of all AMKL cases a type-II mutation could now be identified. Similar to NUP98-NSD1 a stem-cell phenotype was detected with persistent HOXAB-gene expression. Although NUP98/JARID1A did not influence prognosis, outcome in pediatric AMKL is unsatisfactory. NUP98 is known to recruit CREBBP/p300 resulting in histone acetylation, and transcriptional activation of HOX genes (Wang et al, 2007), suggesting that histone acetyltransferase inhibitors may be active. Moreover, JARID1A is unable to demethylate H3K4me2/3, which also results in sustained up regulation of HOX genes. This may provide options for targeted therapy.

No relevant conflicts of interest to declare.