Leukemia-Initiating Cells Are Frequent in Very High Risk Childhood Acute Lymphoblastic Leukemia and Give Rise to Relatively Stable Phenotypes in Immunodeficient Mice.

Abstract 86

High-risk acute lymphoblastic leukemia (ALL) has been proposed to arise from a limited immature leukemia-initiating cell (LIC) compartment that may confer treatment resistance. Recent reports challenge this view. Using syngeneic leukemia mouse models, higher frequencies of LIC were detected. Moreover, sorted ALL subpopulations with different degree of lineage maturation could reconstitute the same leukemia phenotype effectively in ALL xenograft models. We have established a xenotransplantation model of primary B-cell precursor ALL cells in immunodeficient NSG mice (NOD/SCID strain with additional IL-2 receptor common gamma chain deletion), starting from carefully selected very high-risk (VHR) and standard-risk (SR) patients as defined by minimal residual disease (MRD)-based risk stratification criteria on the ALL-BFM-2000 treatment protocol. Median time to engraftment in NSG mice was similar for 11 VHR and 8 SR patients with 8 (range 6–18) weeks for VHR patients and 12 (range 4–32) weeks for SR patients. The comparison of the immunophenotype of ALL cells at diagnosis and after up to three passages in NSG mice revealed a concordance of 90% using 8 cell surface markers that are most commonly included in the diagnostic flow cytometry panels, with no differences between passage rounds. Discordant observations included loss of CD22 expression in 6/19 cases and low levels of CD13 co-expression in 4/19 cases in xenografted samples. To get insight into the genetic stability upon serial passages of ALL cells in NSG mice, we performed Affymetrix Genome-Wide Human SNP Array 6.0 analyses and transcriptional profiling to compare copy number alterations (CNAs) in diagnostic ALL samples and after serial transplantation in NSG mice. Three VHR ALL cases having data from two passages in NSG available were included in a pilot analysis, comparing leukemic DNA to the patients' constitutional DNA. A small number of focal CNAs (>100-<1000 Kb) were detected, most of them previously reported in larger studies. In all cases, CNAs corresponding to expected immunoglobulin and/or T-cell receptor gene rearrangements were seen. In one case, 5 out of 12 CNAs and 2 copy-neutral losses of heterozygosity were maintained after two passages in mice. In a second case with high hyperdiploidy, 15 out of 17 CNAs were stable. In both samples, small deletions in 9p21.3 were maintained. A third case with a t(17;19) translocation lost all 3 CNAs present at diagnosis, but acquired a small deletion in 9p21.3. Between 1 and 6 CNAs were acquired after two passages in mice. An analysis of copy number ratios suggested that selection of subclones had occured. To characterize the LIC compartment of ALL cells in VHR ALL we performed limiting dilution experiments by orthotopic intrafemoral xenotransplantation of primary ALL cells in NSG mice. Hundred thousand unsorted ALL cells generated leukemia in NSG mice in 5/5 cases, and 100 cells were sufficient for engraftment (in 4/5 cases) without conditioning, despite the xenograft barrier. Secondary transplantations demonstrated conserved self-renewal properties. Collectively, our data indicate that this xenograft model of treatment-resistant ALL is remarkably stable with successive passages in mice. This approach will constitute a powerful model to scrutinize clonal evolution of cytogenetic lesions in resistant leukemia and to evaluate new therapeutic options in the fraction of patients that is expected to be recruited in future phase I/II studies. Furthermore, our data indicate that a large proportion of the ALL cell population can recapitulate the leukemia phenotype and thus retain LIC properties. These data also support the use of unsorted populations for therapeutic modeling in NSG mice.