Oncogenic subtypes in childhood B-cell precursor acute lymphoblastic leukemia (BCP-ALL) are used for risk stratification. However, a significant number of BCP-ALL patients are still genetically unassigned. Using array-comparative genomic hybridization in a selected BCP-ALL cohort, we characterized a recurrent V(D)J-mediated intragenic deletion of the ERG gene (ERG(del)). A breakpoint-specific PCR assay was designed and used to screen an independent non-selected cohort of 897 children aged 1-17 years treated for BCP-ALL in the EORTC-CLG 58951 trial. ERG(del) was found in 29/897 patients (3.2%) and was mutually exclusive of known classifying genetic lesions, suggesting that it characterized a distinct leukemia entity. ERG(del) was associated with higher age (median 7.0 vs. 4.0 years, P=0.004), aberrant CD2 expression (43.5% vs. 3.7%, P<0.001) and frequent IKZF1 Δ4-7 deletions (37.9% vs. 5.3%, P<0.001). However, ERG(del) patients had a very good outcome, with an 8-year event-free survival (8-y EFS) and an 8-year overall survival of 86.4% and 95.6%, respectively, suggesting that the IKZF1 deletion had no impact on prognosis in this genetic subtype. Accordingly, within patients with an IKZF1 Δ4-7 deletion, those with ERG(del) had a better outcome (8-y EFS: 85.7% vs. 51.3%; hazard ratio: 0.16; 95% confidence interval: 0.02-1.20; P=0.04). These findings have implications for further stratification including IKZF1 status.

No abstract

121 The genetic landscape of B-cell precursor acute lymphoblastic leukemia (BCP-ALL) in children above 10 years and adolescents remains poorly defined. Specifically, more than half of these patients have none of the cytogenetic abnormalities that define oncogenic subtypes and underlie risk stratification. To uncover new genetic abnormalities in these unassigned cases, we studied 85 BCP-ALL from patients aged 10 to 17 diagnosed at St-Louis hospital (Paris, France), for which the main classifying genetic lesions were assessed (i.e. high hyperdiploidy, t(12;21)/ETV6-RUNX1, t(1;19)/TCF3-PBX1, t(9;22)/BCR-ABL1, iAMP21, MLL translocations, low hypodiploidy, and near haploidy). Fifty of these BCP-ALL presented no classifying genetic lesions. Paired leukemic and remission samples could be analysed by high density array-CGH (Agilent 1M arrays) in 17 of these unassigned cases. We focused on acquired, focal, and recurrent copy-number abnormalities. A mono-allelic intragenic deletion of the ETS-related Gene (ERG) was found in 3 cases. ERG belongs to the ETS family of transcription factors and is implicated in chromosomal translocations associated with several cancer types including acute myeloid leukemia. The possibility of a cryptic unbalanced translocation was ruled out in the 3 cases by FISH analysis. The deletions encompassed exons 3 to 7, or 3 to 9, and the breakpoints were tightly clustered. Based on the breakpoint sequences we designed a PCR assay that allowed us to screen ERG intragenic deletions in the whole cohort. ERG deletion was identified in 9 additional cases, none of them having any of the known classifying genetic lesions, bringing up to 25% (12 out of 50) the frequency of ERG deletion in unassigned BCP-ALL of children older than 10. These results suggested that ERG deletion characterized a novel oncogenic subtype of BCP-ALL. Of note, these results were consistent with independent data of Harvey et al. (2010) that reported ERG deletions in a distinct gene-expression cluster. To confirm and extend these findings in the whole population of paediatric BCP-ALL, we used our breakpoint-specific PCR assay to screen ERG deletions in an independent cohort of 822 unselected patients aged 1 to 17, enrolled in the EORTC 58951 trial. ERG deletion was identified in 31/822 (3.7%) patients. Again, none of them had another known classifying genetic lesion, confirming that ERG deletion characterizes a distinct oncogenic subtype. Patients with ERG deletion were significantly older compared to other patients (median 7.0 vs 4.0, p=0.002), but they had similar white blood counts at diagnosis. They had a favourable outcome, with a 8-year event free survival (EFS) of 82.4% and overall survival (OS) of 96.0%, which is similar to EFS of 83.4% and OS of 91.6% obtained for patients having no very high risk initial features (i.e. no t(9;22)/BCR-ABL1, MLL rearrangement or haploidy/low hypodiploidy). IKZF1 deletion is a cooperative genetic lesion that has been recently shown to be associated with a poor outcome in BCP-ALL. Remarkably, the incidence of IKZF1 deletions in patients with ERG deletion was significantly higher than in other BCR-ABL1-negative patients, especially when considering the IKZF1 intragenic deletion Δ4-7 (10/31, 32.3% vs 34/744, 4.6%, P<0.001), and this regardless of age. Surprisingly, IKZF1 deletion had no impact on the prognosis of ERG deleted patients. Indeed, patients combining ERG and IKZF1 Δ4-7 deletions had a better outcome than other BCR-ABL1-negative patients with IKZF1 deletions (8-year EFS 83.3% vs 53.0%, hazard ratio (HR) 0.19, 95% CI 0.02–1.41; p=0.069). Altogether, we have identified a novel oncogenic subtype of BCP-ALL characterized by ERG deletion. This subtype is frequently associated with IKZF1 deletions, suggesting a preferred oncogenic cooperation. Importantly, despite having older age and frequent IKZF1 deletions, which are factors usually predictive of a poor prognosis, patients with ERG deletion have a favourable outcome. Therefore, this genetic abnormality may be systematically assessed as part of the diagnostic work-up of BCP-ALL and taken into account when considering treatment stratification. Disclosures: No relevant conflicts of interest to declare.