NUP98-KDM5A Fusion Gene Is Not Exclusive to Acute Megakaryoblastic Leukemia and Is an Independent Predictor of Poor Prognosis

NUP98 rearrangements are recurrent genetic events in pediatric acute myeloid leukemia (AML). We identified NPU98-NSD1 as a recurrent poor risk abnormality (Hollink et al, Blood 2011). Another recurrent fusion is NUP98-KDM5A, which was found in pediatric acute megakaryocytic leukemia (AMKL) and was associated with poor clinical outcome (de Rooij et al, Nature Genetics 2017). To date, it is unclear whether NUP98-KDM5A exclusively occurs in AMKL or also in other FAB-subtypes of childhood AML. The aim of this study was to determine the incidence of NUP98-KDM5A in large pediatric AML cohorts, and to define the biological and clinical characteristics of this subgroup.

The study comprised a COG and a European cohort of patients, and included 997 patients from COG trials 03p1 and 0531 (Gamis et al, Blood 2013) and 354 European pediatric AML patients, previously described by Balgobind et al (Haematologica 2011), NUP98-KDM5A rearrangements were detected either by paired-end RNA sequencing or RT-PCR. Patients with FAB M3 AML were excluded from this study. For comparison of clinical characteristics between NUP98-KDM5A positive and negative patients, the χ-squared and the Fisher's exact test were used. Event-free survival (EFS) and overall survival (OS) were estimated using the Kaplan-Meier method; the Cox proportional-hazard model was used for multivariate analysis. For this analysis, age, white blood cell count (WBC), cytogenetic risk group, allogeneic hematopoietic stem cell transplant (HSCT) as time dependent variable and study group (COG vs European cohort) were used as co-variables.

NUP98-KDM5A rearrangements were detected in 31 out of 1351 patients (2.3%). There was no significant difference in prevalence between the COG and European cohorts. 41% of the NUP98/KDM5A+ patients were male, compared to 53% of the NUP98-KDM5A- patients (p=0.24). The median age of NUP98-KDM5A patients was significantly lower as compared to patients without this fusion gene (5.8 vs 9.4 yrs; p=0.03). There was no difference in WBC between NUP98-KDM5A positive and negative patients (13.1*10⁹/l vs.25.2*10⁹/l, respectively, p=0.13). NUP98-KDM5A occurred in all FAB types, with FAB M7 being most frequent (33% vs 8% in NUP98-KDM5A - patients; p<0.001).

The 5-year EFS of NUP98-KDM5A + patients was 36.8 ± 18.8%, compared to 48.1 ± 2.8% in the NUP98-KDM5A - patients (p=0.12). The 5-year OS of NUP98-KDM5A + patients was 41.1 ± 19.7%, a value significantly lower compared to the NUP98-KDM5A - patients (63.9 ± 2.8%; p=0.01), this finding suggesting that NUP98-KDM5A+ patients are difficult to rescue when they relapse. In multivariate analysis, NUP98-KDM5A was an independent risk factor for OS, but not for EFS, with hazard ratios of 1.76 (p=0.045) and 1.58 (p=0.083), respectively. Non-M7 (n=19) patients had lower survival rates with EFS of 26.4 ± 21.7% and OS of 30.0 ± 22.3% compared to M7 (n=9) patients (EFS 55.6 ± 33.1% and OS 64.8 ± 33.1%). These differences were not statistically significant.

This study shows that NUP98-KDM5A is an aberration that occurs in all FAB types and not just in AMKL. NUP98-KDM5A is an independent risk factor for OS but not for EFS and identifies a poor prognostic subgroup. Although there was a big difference in survival between M7 and non-M7 NUP98-KDM5A rearranged patients, this was not statistically significant, probably because of the low number of patients in the two cohorts. Overall, these data suggest to screen for the presence of NUP98-KDM5A rearrangements all children with non-APL AML irrespectively of their AML FAB subtype, and to stratify these patients in the high risk group.