Failure to eradicate cancer may be as fundamental as a misidentification of the target. Current therapies succeed at eliminating bulky disease and proliferating cells, but miss a tumor reservoir that leads to disease recurrence and metastasis. Advances in the understanding of normal tissue development and repair provide a basis for revisiting oncogenesis, tumor heterogeneity, and drug resistance. In order to test the hypothesis that tumor stem cells, small resting cells with the stem cell-like properties of self-renewal, protection, and apoptosis resistance, exist as rare populations in heterogeneous tumors, we used flow cytometry to simultaneously measure tumor and stem cell markers, and constitutive multiple drug resistance transporter expression and activity (side population) in cancer specimens (breast, ovarian, lung, colon, renal, prostate). We concentrated on pleural effusions and ascites, where analysis is not confounded by the presence of normal tissue stem cells. In all cases (n=28) we have succeeded in detecting cytokeratin+, ESA+, CD45-, MDR+ tumor cells of small resting morphology. Building on the findings of the Clarke laboratory, we have shown that these cells are contained within the tumorigenic CD44+/CD24 negative or dim population. Small resting tumor cells are cytokeratin dim, exclusively display P-gp (ABCC1) activity (rhodamine 123 transport) and also express the classical stem cell markers c-kit (CD117), Thy-1 (CD90) and the MDR transporter ABCG2 (BCRP1). This population represents the best candidate for the resting tumor stem cell and is present in pleural effusions at a frequency of ∼1/100,000 cells. In ovarian and breast cancer we have isolated this subset and demonstrated long-term in vitro clonogenicity at frequencies < 1/30 cells. A subset of larger, more granular CD44+/CD24dim tumor cells does not display P-gp activity, but is Thy-1+, ABCG2+, and c-kit+. These cells represent a candidate for the tumor analog of transient amplifying progenitor cells. A subset of these cells is CXCR4+ (stromal derived factor receptor), a known marker of invasive cancer cells. In tumor cells, CD133, a progenitor cell marker, is progressively upregulated as tumor cells differentiate from ABCG2+, Thy-1 bright to the Thy-1-, ABCG2- phenotype. We detected stem cell- and progenitor cell-like populations in all samples, including primary untreated cancer, indicating that the MDR phenotype is constitutive and not treatment induced. These data strongly suggest the existence of tumor counterparts to normal tissue stem and progenitor cells in a wide variety of malignancies. Current therapies target differentiated and proliferating tumor cells, while avoiding irreversible damage to normal tissue stem cells. Given the stem cell-like properties of small resting MDR+ tumor cells, they may be inadvertently spared as well. We are currently testing the in vivo tumorigenic potential of purified tumor stem cells in SCID/NOD mice.

[Proc Amer Assoc Cancer Res, Volume 46, 2005]