Previous studies have suggested that ?-catenin might play an important role in the progression of CML to myeloid blast crisis. It was recently reported that while wild-type mice transduced with p210-infected mouse bone marrow readily develop an acute myeloid disease resembling CML blast crisis,51
mice lacking ?-catenin in the hematopoietic compartment rarely develop myeloid disease, instead generally succumbing to acute lymphoid leukemias.52
Thus, ?-catenin appears to facilitate the progression of advanced myeloproliferative disease. Furthermore, a recent study that used microarray analysis to characterize the changes associated with CML progression found that the ?-catenin pathway was one of the pathways most deregulated in blast crisis patients.53
It is tempting to speculate that the involvement of ?-catenin in the progression of CML is due to its role in conferring self-renewal properties to the GMP population during the advanced stages of the disease.
The available data strongly suggest that there are some cancers that follow the cancer stem cell hypothesis and some that do not. In addition, there may be cancers in which the types of cells capable of initiating tumorigenesis change as the disease progresses. Especially for cancers that initially follow the cancer stem cell hypothesis, where only a fraction of the cells are tumor-initiating, it seems reasonable to expect that the selection pressure for cells with a growth or survival advantage may give rise to additional populations of tumor-initiating cells. Even limiting the discussion to leukemia, one can find tumors that fall into each of the above groups.
As discussed earlier, most acute myeloid leukemias (AML) appear to adhere to the cancer stem cell hypothesis.1,15
In contrast, it has been shown that for certain types of high-risk childhood acute lymphoblastic leukemia (ALL) all sorted populations of cells are able to induce tumorigenesis in immunocompromised mice.54
Importantly, the tumors that arise from these sorted populations all recapitulate the heterogeneity of the original tumor. This demonstrates the remarkable plasticity of these tumor cells and convincingly shows that these tumors do not follow the cancer stem cell hypothesis. The studies described in this review, on the other hand, demonstrate that CML is a malignancy in which the tumor-initiating population of cells evolves as the disease progresses. In chronic phase, the tumor-initiating cell is the Ph+
HSC, which functions as a cancer stem cell. As the disease progresses and additional genetic mutations occur, the GMP population acquires self-renewal potential and the ability to function as a second CML stem cell (Fig. 3
). The specific genetic alterations that endow GMP with the ability to self-renew remain largely unknown, although a recent study suggests that, at least in some patients, self-renewal may result from missplicing of GSK3-? and a resultant increase in ?-catenin activity.42
While progression of CML leads to the evolution of a second cancer stem cell, it also seems possible that a cancer initially following the cancer stem cell hypothesis would progress to the point where cells of the bulk tumor acquire self-renewal. Thus, the more advanced disease may no longer follow the cancer stem cell hypothesis. In this regard, studies of Ph+
ALL are quite interesting. Studies using xenotransplantation of human Ph+
ALL cells have identified a rare population of committed progenitors that function as CSCs.55,56
However, a recent mouse model of Ph+
ALL generated in an Arf?/?
background demonstrated that nearly all cells in this model are able to initiate tumorigenesis in syngeneic transplants.57
This may indicate that events such as ARF loss can confer tumor-initiating properties to the bulk tumor. As loss of the ARF locus is frequently observed in Ph+
it would be interesting to determine if tumors from such patients contain a higher percentage of LICs than those from patients with an intact ARF locus.
Studies such as these on Ph+ ALL, as well as the aforementioned studies on CML, underscore the need to consider the stage of the disease when characterizing cancer stem cells as the tumor-initiating population of cells may evolve as the disease progresses.
Most conventional cancer therapies target the bulk tumor. The most important implication of the cancer stem cell model is that treatment of the bulk tumor may be insufficient to cure disease if a cancer stem cell population is spared. Nowhere has this been manifest more clearly than in the case of CML, where treatment with imatinib is incredibly effective in eliminating the bulk tumor, but fails to eradicate all leukemic stem cells. The failure of imatinib to eradicate these cells is likely due to an inability to eliminate the quiescent fraction of Ph+
Thus, therapeutics that target this population are of significant interest.
The observation that GMPs can function as cancer stem cells in blast crisis, where imatinib is largely ineffective, raises the intriguing question of whether these cells might also be intrinsically insensitive to imatinib or whether they might acquire resistance during their evolution to cancer stem cells. In this regard, QPCR analysis of chronic phase patients with major molecular responses to imatinib rarely detect BCR-ABL transcripts in the GMP population, suggesting Ph+
GMP are not intrinsically resistant to imatinib.28
However, we have observed that GMPs in a murine model of CML are capable of proliferating in the presence of imatinib, suggesting that blast crisis GMP may be resistant to the drug (Minami et al. manuscript in preparation). Thus, additional studies will be needed to determine if leukemic GMP contribute to the imatinib-insensitivity of blast crisis CML.
While the differences between CSCs and cells of the bulk tumor may prevent CSCs from being eliminated by therapies that target the bulk tumor, these differences may also provide unique therapeutic targets. Therefore, the identification of cancer stem cells may open the door to new targeted therapies as the differences between the cancer stem cell, the bulk tumor, and normal cells are realized. The observation that the leukemic GMPs in CML blast crisis largely depend on the ?-catenin pathway for self-renewal point to this pathway as one attractive therapeutic target. Future studies with purified populations of HSCs and GMPs from patients with CML will be essential to identifying additional differences amenable to therapeutic intervention.