Abstract 880

Background

The prognosis of childhood B cell precursor acute lymphoblastic leukemia (BCP-ALL) has improved enormously due to risk-adapted therapeutic stratification of patients. Still, most relapses occur in the group that is supposed to have an intermediate relapse risk. To overcome this problem, new prognostic markers are needed. Recently, new high-risk molecular features were identified in BCP-ALL: the BCR-ABL1-like signature, IKZF1 deletions and high CRLF2 expression (CRLF2-high). These features can occur together; however, whether they confer independent prognostic value is currently unknown. In this study we determined the frequency, co-occurrence and prognostic relevance of these new molecular features in four independent cohorts of children with ALL.

Methods

BCR-ABL1-like gene signature (110-gene probe classifier), IKZF1 deletions (Multiplex Ligation-dependent Probe Amplification) and CRLF2 expression levels (Affymetrix U133 plus 2.0) were determined in leukemic cells (>90% blasts) from 1128 children with newly diagnosed ALL enrolled in three consecutive Dutch Childhood Oncology Group trials (DCOG ALL-8/-9/-10) and in the German COALL-97/03 trial. The cumulative incidence of relapse (CIR) was calculated using death as a competing event.

Results

BCR-ABL1-like, IKZF1-deleted and CRLF2-high cases were found in 16%, 17% and 10% of the BCP-ALL cases, respectively. Collectively, these three features constitute 34% of BCP-ALL cases for which 4% of those had all three, 29% had any two and 67% had only one these features. The BCR-ABL1-like signature and IKZF1-deletions co-occurred the most frequent of these three features (22%). The BCR-ABL1-like signature was only found in B-other cases negative for the genetic abnormalities ETV6-RUNX1, BCR-ABL1, hyperdiploid, TCF3- and MLL-rearrangement, whereas IKZF1-deleted and CRLF2-high cases were also found in other subtypes of BCP-ALL. The frequency of IKZF1- deletions was highest in BCR-ABL1- positive (70%) and BCR-ABL1-like (40%) cases. The BCR-ABL1-like signature, IKZF1 deletions and CLRF2-high were associated with prognosis in four, three and one out of 4 studied treatment protocols, respectively. Protocol-stratified analysis revealed that five-year CIR was higher in BCR-ABL1-like (32%, p<0.001) and IKZF1- deleted (34%, p<0.001) compared to other BCP-ALL cases (11% and 13%, respectively), whereas no unfavorable CIR was associated with the CRLF2-high feature. IKZF1- deleted cases also were at higher risk of relapse in the hyperdipoid (5-year CIR 34%, p=0.04) and non-BCR-ABL1- like B-other group (5-year CIR 40%, p=0.02) compared to 9% for non-deleted BCP-ALL cases whereas IKZF1-deletions had no (additive) prognostic value in ETV6-RUNX1- positive or BCR-ABL1-like cases. Cases with only the BCR-ABL1-like signature (HR 5.3) or only an IKZF1 deletion (HR 4.4) as well as those cases positive for both features (HR 3.7) were at higher risk to develop an adverse event than those cases negative for both features (p<0.001). Together these two features accounted for 62.9% of all relapses in BCP-ALL whereas no relapse occurred in the cases with high CRLF2 expression as single feature. The adverse prognosis associated with BCR-ABL1-like signature and/or IKZF1-deletions was independent of other risk factors including age and white blood cell count. Moreover, the BCR-ABL1-like signature (HR=3.7 p=0.026) and IKZF1- deletions (HR 2.7, p=0.043) were predictive of an unfavorable prognosis in the group of BCP-ALL cases with intermediate minimal residual disease levels at day 79 of treatment.

Conclusions

High CRLF2 expression is the least predictive, whereas the BCR-ABL1-like signature and IKZF1- deletions –both individually and combined- are strong independent prognostic factors and have additional prognostic value on top of intermediate minimal residual disease levels in children with BCP-ALL. These findings demonstrate that these two molecular features are important clinically for identifying high-risk patients who may require more intensive therapy.

Disclosures:

No relevant conflicts of interest to declare.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution