Most acute myeloid leukemia progenitor cells with long-term proliferative ability in vitro and in vivo have the phenotype CD34(+)/CD71(-)/HLA-DR-

Abstract

Acute myeloid leukemia (AML) occurs as the result of malignant transformation in a hematopoietic progenitor cell, which proliferates to form an accumulation of AML blasts. Only a minority of these AML cells are capable of proliferation in vitro, suggesting that AML cells may be organized in a hierarchy, with only the most primitive of these cells capable of maintaining the leukemic clone. To further investigate this hypothesis, we have evaluated a strategy for purifying these primitive cells based on surface antigen expression. As an in vitro endpoint, we have determined the phenotype of AML progenitor cells which are capable of producing AML colony-forming cells (CFU) for up to 8 weeks in suspension culture (SC) and compared the phenotype with that of cells which reproduce AML in nonobese diabetic/severe combined immunodeficiency (NOD/SCID) mice. AML cells were fluorescence-activated cell sorted (FACS) for coexpression of CD34 and CD71, CD38, and/or HLA-DR and the subfractions were assayed in vitro and in vivo at various cell doses to estimate purification. While the majority of primary AML CFU lacked expression of CD34, most cells capable of producing CFU after 2 to 8 weeks in SC were CD34(+)/CD71(-). HLA-DR expression was heterogeneous on cells producing CFU after 2 to 4 weeks. However, after 6 to 8 weeks in SC, the majority of CFU were derived from CD34(+)/HLA-DR- cells. Similarly, the majority of cells capable of long-term CFU production from SC were CD34(+)/CD38(-). Most cells that were capable of engrafting NOD/SCID mice were also CD34(+)/CD71(-) and CD34(+)/HLA-DR-. Engraftment was not achieved with CD34(+)/CD71(+) or HLA-DR+ subfractions, however, in two patients, both the CD34(+) and CD34(-) subfractions were capable of engrafting the NOD/SCID mice. A three-color sorting strategy combining these antigens allowed approximately a 2-log purification of these NOD/SCID leukemia initiating cells, with engraftment achieved using as few as 400 cells in one experiment. Phenotyping studies suggest even higher purification could be achieved by combining lack of CD38 expression with the CD34(+)/CD71(-) or CD34(+)/HLA DR- phenotype. These results suggest that most AML cells capable of long-term proliferation in vitro and in vivo share the CD34(+)/CD71(-)/HLA-DR- phenotype with normal stem cells. Our data suggests that in this group of patients the leukemic transformation has occurred in a primitive progenitor, as defined by phenotype, with some degree of subsequent differentiation as defined by functional assays.