Infant leukemia, defined as leukemia before 1 year of age, is an aggressive set of diseases with dramatically poorer outcome when compared to childhood leukemia. A particularly intractable subset of infant leukemia patients, who are prone to relapse, are those with B-cell acute lymphoblastic leukemia (B-ALL) harboring a chromosomal translocations involving the epigenetic modifier gene mixed lineage leukemia (MLL), referred hereafter as MLL rearranged (MLL-r) B-ALL. Rearrangements in MLL occur in 70-80% of all infant B-ALL patients (compared to 5% in pediatric leukemia) and this cohort is characterized by a record setting dearth of cooperating lesions beyond MLL-r.

To better understand the mechanism of relapse in this patient cohort we employed RNA sequencing and deep mass spectrometry based proteomics to characterize infant MLL-R B-ALL patient derived xenografts. Five infant MLL-r B-ALL patients with matched diagnosis and relapse primary sample were individually injected into NSG immunocompromised mice. For all patients >95% human engraftment purity was achieved in mouse bone marrow and spleen. Fresh cells were submitted for RNA sequencing and nuclear fractionated as well as whole cell lysates were generated for mass spec protein analysis. Protein and RNA quantitation at relapse and diagnosis across all patients was achieved on molecules that correspond to about 8,000 genes for protein and 17,000 genes for RNA.

Hierarchal clustering as well as multi-dimensional scaling analysis of RNA sequencing data revealed a closer intra-patient relationship at the two treatment time points (diagnosis and relapse) than an inter-patient relationship at diagnosis or at relapse. A common relapse signature was simply non-existent according to the data, however this could be attributed to the small sample size. The integration of proteomic and transcriptomic data revealed a strong correlation of relapse/diagnosis fold changes at the level of RNA and protein. Intriguingly, a plasticity at relapse was observed in key developmental and differentiation marker genes associated with the myeloid, T, and B cell lineages. For example, in Patient 1 up-regulation at the level of protein and RNA of myeloid/acute myeloid leukemia (AML) associated genes; RUNX2, MPO, and CD33, and down-regulation of T and B cell associated genes; CD1A, CD2, and BCL6 where observed at relapse. Patient 2 showed an RNA/protein up-regulation of myeloid and AML markers CEBPB and MPO while a concurrent down-regulation of the Ikaros family member Helios, B/T cell attenuator BTLA, and the oncogene ETV6 at relapse. Patient 3 showed an RNA/protein up-regulation of Helios, the oft mutated in B-ALL; ARID5B, and the pre-B-cell leukemia associated transcription factor PBX3 at relapse. Patient 1 and 2 appear to show signs that myeloid programs are being greater utilized at relapse while in Patient 3 seems to intensify its commitment to the B-cell lineage. All patients showed some degree of lineage marker infidelity at relapse and interestingly, this plasticity was not reflected in the traditional flow cytometric markers used to differentiate AML from B-ALL suggesting that this lineage infidelity might be more common than previously thought.

Despite the mutational uniformity of infant MLL-R B-ALL genomes it is surprising that such a diversity of RNA expression and subsequent protein levels is observed. This data suggests that MLL fusion proteins have great flexibility to control of the transcriptional regulation/protein expression and could in part help explain why these patients do not rely upon newly acquired mutations to evade treatment protocols. Furthermore, lineage switching in leukemia is rarely observed but as this data suggests that this is more common in MLL-r B-ALL than previously thought.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.

Sign in via your Institution