In Acute Lymphoblastic Leukaemia, Stemness Is Frequent and Ubiquitous

Abstract 92

Research on cancer stem cells, cells that self-renew and reconstitute the full phenotype of the original malignancy, has yielded controversial results regarding their frequency and identity for many cancers. The hierarchical stem cell model has been well established in some malignancies such as acute myeloid leukemia and states that only rare, immunophenotypically immature blasts harbor stem cell activity, resembling a normal physiological hierarchy. The opposing stochastic model proposes that stemness in cancer cells is supported by extrinsic stimuli and that a substantial fraction of malignant cells have this potential. Continued optimization of in vivo xenotransplantation modeling recently caused a paradigm shift for some cancers, for example in malignant melanoma where stem cell activity was found in as many as 1 in 4 cells. For acute lymphoblastic leukemia (ALL) we and others previously challenged the hierarchical model by demonstrating that both immature and more mature leukemic blasts contain self-renewal properties (Cancer Cell 2008, 14(1), p47-58). In this study we address the frequency of leukemic stem cells in the bulk leukemia and also, more specifically, in subpopulations of different blast maturity by using unsorted and highly purified flow sorted cell fractions. Primary patient material as well as leukemic blasts harvested from engrafted mouse bone marrow (secondary and tertiary material) were sorted for their CD10, CD20 or CD34 expression followed by orthotopic intrafemoral transplantation into severely immunocompromised NOD/scid IL2Rγnull (NSG) mice. Engraftment of transplanted CD19+CD10low and CD19+CD10high, CD19+CD20low and CD19+CD20high and CD19+CD34low and CD19+CD34high blast populations was monitored by 5 color flow cytometry using material from consecutive bone marrow punctures, final bone marrow harvests and/or single cell suspensions from spleens. Primary ALL samples from 15 high risk (BCR/ABL positive (n=8), BCR/ABL like ALL (n=2), high hyperdiploid/MRD positive (n=2), MRD positive (n=1), MLL/AF4 (n=2)), 3 intermediate risk (high WBC/MRD negative (n=2), age >10 years (n=1)) and 3 standard risk (n=3) patients were included. Cells sorted into CD19+CD10low and CD19+CD10high fractions were transplanted from primary patient material (n=4, HR; n=1, SR) and from secondary samples (n=4, HR; n=1; IR) with cells from one HR patient used at limiting dilutions. As few as 100 sorted cells of either fraction were sufficient to repopulate the leukemia. CD19+CD20high and CD19+CD20 low fractions from primary (n=7, HR; n=1, IR), secondary (n=5, HR; n=1, IR) and tertiary material (n=2, HR; n=1, IR) engrafted NSG mice. Limiting dilutions were performed on secondary (n=4, HR) and tertiary material (n=2, HR). Cell numbers required for engraftment varied between leukemias with as few as 100 cells being sufficient to cause engraftment. Limiting dilution experiments using CD19+CD34high and CD19+CD34low fractions from secondary (n=1, HR) and tertiary (n=1, HR) material yielded engraftment with as few as 10 CD19+CD34high and 100 CD19+CD34low cells. Similarly, unsorted primary (n=11, HR; n=2, IR), secondary (n=2, HR) and tertiary material (n=1, HR) required as few as 10 cells for leukemic reconstitution. Taken together, both unsorted and sorted blasts of all immunophenotypes and transplanted with low numbers were able to reconstitute the complete original phenotype of the patient leukemia. All limiting dilutions were transplanted down to 10 cells per mouse and those mice not engrafted yet are still under observation. Furthermore, the ability to self-renew was demonstrated by serial transplantation. Finally, we compared expression of self-renewal associated genes (BMI1, EZH2, HMGA2, MEIS1, TERT) in CD19+CD34low and CD19+CD34high fractions of 5 HR and 1 SR samples with that in cord blood. Interestingly, expression of these genes was not dependent on the CD34 status of the leukemic cells, whereas HMGA2, MEIS1 and TERT were upregulated in CD34+ cord blood cells. In summary we provide strong evidence for the stochastic cancer stem cell model in B precursor ALL by demonstrating that (i) a broad spectrum of blast immunophenotypes exhibit stem cell characteristics and (ii) that this stemness is highly frequent among ALL cells.