Characterization of a hierarchy in human acute myeloid leukemia progenitor cells.

Despite the usual uniform and primitive appearance of cells derived from the leukemic clone in most patients with acute myeloid leukemia (AML), there is considerable heterogeneity among leukemic blasts, particularly with respect to their capacity to proliferate and/or self renew. We have assessed whether these differences in proliferative potential are correlated with the phenotypic changes that characterize normal hematopoiesis, which might suggest an analogous hierarchy of AML progenitors. We have used the ability of primitive AML cells to persist or produce blast colony forming cells (CFU-blast) detected after 2 to 8 weeks in the presence of growth factors in suspension cultures (SC) termed SC-initiating cells (IC), or with stroma in long-term cultures (LTC-IC) as a quantitative assay for a cell that may have primitive characteristics. This SC assay is linear, cell concentration independent, and the frequency of SC-IC by limiting dilution analysis is lower than primary CFU-blast. The average output of CFU-blast after 2 to 8 weeks by individual SC-IC varied between 2 and more than 100 in individual patients. Leukemic blasts were sorted based on their expression of antigens previously found useful to characterize normal progenitor differentiation, and analyzed for the percentage of CFU-blast SC-IC, and leukemic LTC-IC within each fraction. All of these progenitor types were heterogeneous in their expression of CD45RA and CD33, but expressed uniformly low levels of CD15 and differed from normal primitive progenitors in their high expression of HLA-DR. CFU-blast had a significantly higher expression of CD71 and CD38 as compared with SC-IC or leukemic LTC-IC. In patients with CD34+ blasts, the majority of their SC-IC at 4 weeks were CD34+/CD38-; however, patients with CD34- blasts had at least some CD34- progenitors. These results show that while heterogeneity exists between patients, it is possible to physically separate subpopulations of AML cells with different proliferative potentials. It also provides some support for the concept that quantitation of leukemic cells capable of producing CFU-blast for 4 weeks or more in vitro measures a less frequent leukemic progenitor with higher proliferative potential that may be the only relevant cell for maintaining the leukemic clone in vivo.