CBFA2T3-GLIS2 fusion transcript is a novel common feature in pediatric, cytogenetically normal AML, not restricted to FAB M7 subtype

Pediatric acute myeloid leukemia (AML) is a molecularly heterogeneous disease that arises from genetic alterations of pathways that regulate self-renewal and myeloid differentiation. While the majority of patients carry recurrent chromosomal translocations, almost 20% of childhood AMLs do not show any recognizable cytogenetic alteration and are defined as cytogenetically normal AML (CN-AML).¹  Many genetic abnormalities have been identified in AML with normal karyotype, with the most frequent affecting genes such as NPM1, FLT3, CEBPA, and WT1.^1-5 

Genome-wide analyses have been used with the aim of determining the full array of genetic lesions of CN-AML. Recent studies have provided new insight into the molecular genetics and biology of AML, confirming both the complexity and the heterogeneity of this disease.⁶  Novel lesions such as mutations in IDH1 and DNMT3A have been identified.^7,8  However, these alterations are rare in pediatric AML, with IDH1/IDH2 accounting for 2% to 4% of cases^9,10  and DNMT3A not even being found mutated in childhood AML.¹¹  Recently, 2 studies identified a novel recurrent translocation involving CBFA2T3 and GLIS2 in about 30% of children with non-Down syndrome acute megakaryoblastic leukemia (non-DS AMKL, AML French–American–British [FAB] M7).^12,13  Nevertheless, there are many children with CN-AML in whom no genetic abnormality has been detected. The identification of the different genetic profiles characterizing this subgroup is a primary objective to be pursued. To this end, we performed whole-transcriptome massively parallel sequencing of 7 cases of pediatric CN-AML with the aim of identifying recurrent somatic mutations or genomic rearrangements. Subsequently, we validated our findings in a larger cohort of 230 pediatric CN-AML patients.

Patients analyzed either in the parallel sequencing screening or in the validation cohort are children with newly diagnosed de novo AML other than promyelocytic leukemia. The patients are enrolled in the Associazione Italiana Ematologia Oncologia Pediatrica 2002/01 Protocol, which was approved by the institutional review board of the Sant’ Orsola-Malpighi Hospital, Bologna, Italy.¹⁴  Patients gave informed consent in accordance with the Declaration of Helsinki. FAB morphological classification and immunophenotypic analysis were centrally established at the laboratory of Pediatric Hematology of the University Hospital in Padova, Italy. Chromosome analysis was performed on bone marrow (BM) aspirates using standard laboratory procedures. Karyotypes were reported according to the International System for Human Cytogenetic Nomenclature.¹⁵  For fluorescence in situ hybridization, an MLL locus-specific dual-color probe for 11q23 (Abbott-Vysis, Downers Grove, IL) was used. All CN patients were also negative for known recurrent genetic abnormalities involving MLL, CBFB, NPM1, and FLT3 genes.

Total RNA was extracted from BM leukemia cells of CN-AML patients by TRIzol, following the manufacturer’s protocol (Invitrogen, Karlsruhe, Germany); 250 to 1000 ng of total RNA was used for the synthesis of cDNA libraries with TruSeq RNA Sample Prep Kit v2 (Illumina, San Diego, CA) according to the manufacturer’s recommendations. Sequencing by synthesis was performed on HiScanSQ sequencer (Illumina) at 75bp in paired-end mode. Reads were aligned with TopHat2/BowTie2¹⁶  to the reference human genome hg19/GRCh37. deFuse¹⁷  and ChimeraScan¹⁸  packages were used to detect chimeric transcripts from RNA-seq data.

Total RNA was extracted from BM leukemia cells of all samples using TRIzol. CBFA2T3-GLIS2 fusion transcript was detected with reverse-transcription polymerase chain reaction (RT-PCR) and sequenced with the BigDye terminator v3.1 Cycle Sequencing kit (PE Applied Biosystems, Foster City, CA) on an Applied Biosystems 310 analyzer. The following primers were used: forward 5′-CGAAGGGCCTCAGCTAGACGT-3′, reverse 5′-AGCCACTGCGCTATTTGGAT-3′.

Whole-transcriptome massively parallel sequencing in the 7 children with CN-AML yielded an average of 88.3 million mapped reads per patient, thus reaching an average coverage of 36×. Single nucleotide variant calling confirmed the absence of mutations in genes such as NPM1, CEBPA, FLT3, MLL, IDH1, IDH2, and DNMT3A.

Two algorithms, ChimeraScan¹⁷  and deFuse,¹⁸  were used to identify new putative fusion transcripts. In 3 of 7 patients (1, 2, 17; Table 1) we identified a chimeric transcript involving CBFA2T3 and GLIS2, resulting from a cryptic inversion of the telomeric region of chromosome 16 that fuses the 5′ portion of CBFA2T3 in frame with the 3′ region of GLIS2. CBFA2T3, also known as MTG16 or ETO2, is a member of the myeloid translocation gene family that is fused to AML1 in the t(16;21) (q24;q22) translocation that is identified in therapy-related AML.^19,20 GLIS2 is a transcription factor of the GLI-similar Krüppel-like zinc finger protein family that acts as an inhibitor of the hedgehog signaling pathway.²¹  This fusion was the only recurrent one, and its identification was supported by an average of 42 span and 153 split reads. Two patients showed the fusion between exon 11 of CBFA2T3 and exon 3 of GLIS2, while the third showed CBFA2T3 exon 10 fused to exon 2 of GLIS2 (Figure 1A). These data were also confirmed by RT-PCR analysis and Sanger sequencing (Figure 1B). Interestingly, we identified 1 novel breakpoint for the CBFA2T3-GLIS2 fusion transcript (CBFA2T3-ex10/ex2-GLIS2), which is different from breakpoints already described in the 2 recent reports on non-DS AMKL.^12,13  Taken together, these data suggest that the cryptic inv(16) generates a CBFA2T3-GLIS2 fusion gene that can be truncated at different positions.

To assess the prevalence of CBFA2T3-GLIS2 fusion, we then examined a validation cohort of 230 children with newly diagnosed de novo CN-AML, also negative for known recurrent genetic abnormalities involving MLL, CBFB, NPM1, and FLT3 genes. Globally, the CBFA2T3-GLIS2 rearrangement was detected in 20 of 237 cases (8.4%) with CN-AML. RT-PCR analysis and Sanger sequencing confirmed that all positive cases in the validation cohort carried the CBFA2T3 exon 11-GLIS2 exon 3 fusion. Fifty percent (N = 10) of the positive patients belonged to the M7 FAB subgroup, while the remaining patients (N = 10) were distributed among the other FAB classes (see Table 1). These results indicate that the CBFA2T3-GLIS2 fusion transcript, recently described as a distinctive feature of pediatric non-DS AMKL,^12,13  should be more broadly considered as a genetic abnormality that is shared with other FAB subgroups of pediatric CN-AML.

We evaluated whether the presence of CBFA2T3-GLIS2 fusion product influences patients outcome. The 5-year event-free survival (EFS) of the 20 patients with the CBFA2T3-GLIS2 fusion gene was significantly worse than that of the 217 pediatric CN-AML patients not harboring the translocation (27.4%, standard error [SE] 10.5 vs 59.6%, SE 3.6; P = .01; Figure 1C). We also stratified the patients with respect to FAB subgroups (M7 vs non-M7; Figure 1D). The 5-year EFS of FAB M7 patients with or without the CBFA2T3-GLIS2 fusion gene was significantly different: 26.6% (SE 15.0) and 60.7% (SE 8.3; P = .04), respectively. Similar results were obtained in non-M7 patients; the 5-year EFS of patients assigned to other FAB categories with or without the CBFA2T3-GLIS2 fusion transcript was 30.0% (SE 14.4) and 59.4% (SE 3.5; P = .04), respectively. No statistically significant difference in EFS of non-M7 and M7 patients harboring the CBFA2T3-GLIS2 fusion transcript (30.0%, SE 14.4 vs 26.6%, SE 15.0; P = .91) was found, suggesting that FAB classification does not interact with theCBFA2T3-GLIS2 fusion product in influencing outcome. Taken together, these data indicate that the CBFA2T3-GLIS2 fusion transcript is a novel common feature in pediatric CN-AML that is not restricted to the FAB M7 subtype, predicting poor outcome.

This work was supported in part by grants from Fondazione Umberto Veronesi, Milan (R.M.); by Fondazione Istituto di Ricerca Pediatrica-Città della Speranza Padova (M.P., E.M., G.B.); and by Associazione Italiana Ricerca sul Cancro Special Project 5 × mille (F.L.).

Contribution: R.M. and M.P. equally coordinated the work, analyzed data, performed statistical analyses, and wrote the paper; M.T. and A.A. performed the whole-transcriptome massively parallel sequencing; V.I. and R.C. performed bioinformatics analyses; E.M. performed mutation analyses on the validation cohort; and A.P., G.B., and F.L. designed and supervised the research. A.P. and F.L. equally contributed to the critical revision and writing of the manuscript. All authors read and approved the final version of the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Andrea Pession, Paediatric Oncology and Haematology Unit, Lalla Seràgnoli, Sant’Orsola-Malpighi Hospital, Via Massarenti 11, 40137 Bologna, Italy; e-mail: andrea.pession@unibo.it.