Malignant stem cells in childhood ALL: the debate continues!

To the editor:

Recently, Cox et al¹  described that only acute lymphoblastic leukemia (ALL) blasts with a CD19⁻CD133⁺ immune-phenotype are able to propagate human leukemia in immune-deficient nonobese diabetic severe combined immunodeficiency (NOD/scid) mice. We welcome this contribution to the rapidly evolving debate on leukemic stem cells.

Published data on ALL stem cells are conflicting. While some groups have detected self-renewing blasts as being restricted to the primitive CD19⁻ or CD38⁻ stem cell compartment,^1-3  others found repopulating cells only in the more mature CD19⁺ lymphoid compartment⁴  or within aberrant CD19⁺ lymphoid cells that lack CD38 expression.⁵  To make the picture even more confusing, we have recently shown that a wide variety of blasts at different stages of maturation (CD34⁺ and CD34⁻, CD19⁻ and CD19⁺, CD20⁻ and CD20⁺) are able to propagate leukemia in natural killer (NK) cell–depleted NOD/scid and NOD/scid gamma (NSG) mice after intrafemoral injection.⁶  Unlike the mechanism suggested by Cox et al,¹  our data cannot be explained by prior passage of the human cells in the mice as primary sorted cells show the same engraftment pattern (Table 1). Although this may have been an obvious concern, we believe that we are not creating in vivo cell lines with an altered stem cell hierarchy.

Similar to our own results, Kong et al⁷  have recently shown that both CD34⁺CD19⁺CD38⁻ and CD34⁺CD19⁺CD38⁺ blasts in infant ALL/t(4,11) are able to transfer leukemia onto NSG mice. Most importantly, work out of Civin's laboratory as presented at the 2008 Annual Meeting of the American Society of Hematology showed that, after intravenous injection, as few as 10 unselected ALL cells were sufficient to reinitiate the leukemia in NSG mice.⁸ 

How can it be explained that many groups seem to come up with a different phenotype for ALL stem cells? It is interesting to note that studies identifying rare leukemic stem cells with an immature immune phenotype use more conventional NOD/scid mice as recipients, whereas studies observing leukemia-initiating potential in a wider variety of populations use either NK cell–depleted NOD/scid mice or one of the newer genetically engineered NOD/scid mouse strains with deletion of the IL2 receptor gamma chain (eg, NSG mice). Work from Bonnet's laboratory has elegantly demonstrated that conventional NOD/scid mice can clear human cells coated with certain antibodies, such as anti-CD38.⁹  Consequently, these models may not only underestimate the frequency of human stem cells but also wrongly imply a hierarchy between different populations. Similar observations have recently been made in several solid tumor models (reviewed in Dick¹⁰ ): while earlier xenograft assays had suggested the presence of only rare cancer stem cells, recent results using NSG mice showed stem cell frequencies as high as 1 in 4 tumor cells.

Much of the confusion could therefore be around methodology. It will be paramount to standardize the xenotransplantation approaches and to use optimized mouse models that offer the least hostile microenvironment for human cells.¹⁰  With these newer and more sensitive models, we may be surprised to find that many more populations have stem cell properties then previously thought. We look forward to continuing this debate.