GENOMIC AND DRUG RESPONSE PROFILING OF FATAL TCF3-HLF-POSITIVE PEDIATRIC ACUTE LYMPHOBLASTIC LEUKEMIA IDENTIFIES RECURRENT MUTATION PATTERNS AND NOVEL THERAPEUTIC OPTIONS

Background
The translocation t(17;19)(q22;p13) results in the fusion of the transactivating domain of the B cell transcription factor TCF3 and the DNA binding domain of the hepatic leukemia factor HLF. TCF3-HLF defines a rare subtype of ALL that is currently incurable. In contrast, the fusion of the same portion of TCF3 to the PBX1 gene (t(1;19)) results in a leukemia subtype with a favorable prognosis. TCF3-HLF ALL constitutes a paradigm of treatment refractory leukemia in which the oncogenic fusion protein drives programs that are fundamental for the malignant process. However genetic engineering experiments in mice did not recapitulate disease, suggesting that the cellular context in which the translocation occurs is critical for transformation. 

Aims
We explored the genetic landscape of ten TCF3-HLF positive ALL cases in comparison to TCF3-PBX1 positive ALL using primary diagnostic samples and corresponding patient derived leukemia xenografts in order to analyze whether TCF3-HLF is associated with characteristic patterns of genomic lesions and to establish a well characterized humanized mouse model for functional investigations.

Methods
We used an integrated multi-OMICs approach to identify mutation and gene expression patterns in TCF3-HLF- and TCF3-PBX1-positiveALL. To compare drug response profiles of treatment resistant TCF3-HLF-positive and responsive TCF3-PBX1-positive ALL we established a xenograft model of patient derived leukemia. At least two biological replicates per case were xenografted in NOD/SCID/IL2rγ^null (NSG) mice. Drug response profiles to a customized library of >100 therapeutic compounds were obtained and in vivo validation studies were performed using this model.

Results
In TCF3-HLF-positive ALL, intragenic deletions in the lymphoid transcription factor PAX5 or somatic mutations in the non-translocated allele of TCF3 (acting upstream of PAX5), but not in any of the IFKZ family members, were recurrent and frequently observed in conjunction with RAS pathway aberrations. The pattern of co-occurring mutations was not overlapping in TCF3-PBX1 ALL, supporting the notion that completely different genetic interactions may be required for disease pathogenesis. The profile of mutations was maintained in the corresponding xenografts. Sequence features at the translocation breakpoints suggested RAG mediated recombination and an already lymphoid-committed cell of origin, but the TCF3-HLF-positive ALL transcriptome was enriched for stem cell and myeloid features, consistent with reprogramming towards a hybrid, more drug-resistant hematopoietic state. This gene expression signature was maintained in matched patient-derived xenografts. TCF3-HLF-positive ALL revealed a distinct drug-response profile with resistance to some agents commonly used in ALL therapy, but sensitivity towards glucocorticoids and some therapeutic agents in clinical development. Striking on-target sensitivity was achieved with the BCL2-specific inhibitor ABT-199 indicating BCL2-dependency of TCF3-HLF-positive ALL. Our results show, that ABT-199 acts synergistically with conventional chemotherapeutic agents, including dexamethasone and vincristine, providing a rationale for experimental therapy.

Summary
The molecular portrait of TCF3-HLF-positive ALL indicates that the initiating event occurs in cells committed to the lymphoid lineage and that leukemogenesis requires a restricted pattern of additional genetic lesions. Maintenance of dominant genomic and transcriptomic patterns of both TCF3-translocated subtypes in the corresponding xenografts provides the basis for systematic functional investigations. Integration of drug response profiling revealed recurrent patterns of resistance and new options for the treatment of this deadly disease.

Keyword(s): Acute lymphoblastic leukemia, Drug resistance, Drug sensitivity, Xenotransplantation

www.eha.org