Abstract C009: Zebrafish cancer models predict clinical outcomes in human acute lymphoblastic leukemia

Abstract

The complex heterogeneity of acute lymphoblastic leukemia (ALL) often predicts poor prognosis, high morbidity, and drug resistance. Yet, the molecular drivers that initiate aggressive ALL have yet to be fully elucidated, especially in the context of the 38 subtypes described to date. Here, we used a large-scale F0 transgenic screen in zebrafish to identify synergistic combinations of 64 putative oncogenes that induce T- or B-cell ALL. Notably, the proto-oncogene SET collaborates with both mutationally activated notch1 and IL7-receptor to initiate a wide array of leukemias and is broadly and highly expressed across all human T- and B-ALL, suggesting important roles in leukemia initiation. Gene expression and non-negative matrix factorization (NMF) analysis uncovered highly expressed gene programs found across leukemia subtypes, both in zebrafish and human ALL. In total, seven transcriptional programs enriched in zebrafish leukemia independently predicted overall and event-free survival in human T-ALL. Multivariable analysis using combined zebrafish transcriptional programs significantly improved risk stratification beyond established clinical diagnostic variables and minimal residual disease after treatment. This effect was seen both in aggregate and within defined molecular subtypes, including identifying low-risk Early T-cell Progenitor (ETP) patients and stratifying the most common Double Positive-like T-ALLs into good and poor prognostic groups. Taken together, our zebrafish screening approach is a powerful tool for comparative genomic studies to identify novel oncogenic drivers and genetic synergies that induce leukemia with translational application to human ALL.