The cryptic CBA2T3-GLIS2 fusion generated by the inv(16)(p13.3q24.3) was initially identified in megakaryocytic leukemia and later implicated in other acute myeloid leukemia (AML) subtypes. Presence of this fusion may lead to altered expression of the potentially targetable sonic hedgehog and bone morphogenic protein pathways. We determined the prevalence of CBA2T3-GLIS2 in children treated on COG AAML03P1 and AAML0531 protocols, which collectively enrolled 1361 eligible children, adolescents, and young adults with de novo AML, and correlated the presence of this fusion with patient demographics, laboratory features, and clinical outcomes. We also determined the prevalence and clinical implications of CBA2T3-GLIS2 in 71 children with FAB M7 AML treated on 4 consecutive COG AML trials.

Of the 1042 diagnostic samples available and tested for CBFA2T3-GLIS2, 45 (4.3%) were positive for the fusion. Fusion-positive patients were significantly younger than fusion-negative patients (2.1 vs. 10.3 years; P<0.001). CBFA2T3-GLIS2 was most prevalent in the youngest patients (10.6% for 0 to <2 year olds and 8.6% for 2 to <5 year olds) [Figure A]. In contrast, no fusion transcripts were identified in 299 unselected adult patients. All FAB subtypes were represented in fusion-positive patients. Overall, FAB M5 and M7 were equally prevalent in fusion-positive patients, and each subtype accounted for 20% of cases [Figure B]. There was a preponderance of MLL rearrangements (N=7) in fusion-positive patients, and 3 more patients had either t(8;21) or inv(16), and 25.6% (N=11) without karyotypic alterations (CN-AML). None of the fusion-positive patients had the t(7;12) or 12p abnormality. There were few common AML-associated mutations: 1 patient had FLT3ITD and 1 had the WT1 mutation (no NPM1 or biallelic CEBPA mutations were identified).

Rates of morphologic complete remission (CR) at the end of induction course 1 were similar for fusion-positive and -negative patients (68.9% vs. 77.7%; P=0.17). However, fusion-positive patients were more likely to have minimal residual disease (MRD) by flow cytometry at this time point (50% vs. 28.9%; P=0.006) with a correspondingly higher relapse rate (RR) from remission of 58% vs. 35% (p=0.005). Disease-free survival for those with and without fusion was 42% vs. 58% (p=0.060)

In a subset analysis of 193 patients with CN-AML, the prevalence of CBFA2T3-GLIS2 was 4.7%. Fusion-positive patients were younger than fusion-negative patients (1.6 vs. 13.1 years; P<0.001) and more likely to have MRD at the end of induction (85.7% vs. 40.8%; P=0.043). CN-AML fusion-positive patients had significantly worse 5-year OS (36% vs. 67% P=0.025) and EFS (18% vs. 51%, P=0.017) than fusion-negative patients. All fusion-positive patients in CR had a higher 5-year RR than fusion-negative patients (88% vs. 33%; P<0.001), with a corresponding disease-free survival (DFS) of 13% vs. 59% (P<0.001).

Implications of CBFA2T3-GLIS2 were evaluated in 71 cases of children with FAB M7 AML where the fusion was identified in 12 patients (17%). Fusion-positive FAB M7 patients had significantly lower CR rate than fusion-negative patients (33.3% vs. 77.6%, P=0.005) and all were MRD positive at the end of induction (100% vs. 30%, P=0.001). All FAB M7 fusion-positive patients relapsed (100% vs. 36%, P=0.007), with a DFS of 0% vs. 60% (p=0.013).

As CBFA2T3-GLIS2 was most common in younger patients, we compared clinical implications in children <2 years of age. Although CR rates in fusion-positive and -negative patients were similar (68.2% vs. 74.9%; P=0.50), fusion-positive patients were more likely than fusion-negative patients to have MRD at the end of course 1 (63.2% vs. 24.8%, P=0.006). Five-year RR rates in fusion-positive and -negative patients were 71% vs. 39% (P=0.012), with corresponding DFS of 29% vs. 56% (P=0.030).

This study provides a comprehensive evaluation of incidence and prognostic implications of the CBFA2T3-GLIS2 fusion in pediatric de novo AML. Along with MLL rearrangements, 12p abnormalities, and the recently described NUP98-JARID1A fusion, this cryptic inversion 16, as defined by presence of the CBFA2T3-GLIS2 fusion, represents a distinct, recurrent chromosomal abnormality associated with poor prognosis in infant AML and is a potential therapeutic target.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.

Sign in via your Institution