Encyclopedia of Cancer (2016)
Cancer stem(-like) cells are those cells that possess the capacity for self-renewal and for causing the heterogeneous lineages of cancer cells that comprise the tumor.
The definition follows a consensus at a workshop on cancer stem(-like) cells (CSC) organized by the American Association for Cancer Research (AACR). There is considerable debate and some controversy on the CSC concept, so that a consensus definition is required. The importance of the debate is proportional to its relevance to the change in our perception of cancer, intrinsic to the CSC paradigm, implying that not all cancer cells are equal but that only a small fraction of them is endowed with the properties of perpetuating the disease. This hierarchical model has not only important biological consequences but also relevant therapeutic implications, as we discuss in this essay.
The CSC paradigm fits in a model of cancer as a caricature of an organ that is already present in the literature as suggested by data published 30–40 years ago. In particular, Hamburger and Salmon established growth conditions for cancer cells in soft agar medium and found that tumor stem cell colonies, arising from different types of cancer with 0.001–0.1 % efficiency, had differing growth characteristics and colony morphology. Studies by Dick and coworkers in the 1990s showed that in several forms of acute myeloid leukemia (AML) cells that could engraft in immunodeficient mice are restricted to a minority subpopulation defined as [CD34+/CD38neg]: these cells, therefore, shared a cell surface phenotype with normal human primitive hematopoietic progenitors, suggesting that they may have originated from normal stem cells rather than from committed progenitors. Also of interest was the observation that leukemic cells engrafted in NOD-SCID mice (nonobese diabetic-severe combined immunodeficiency: an immunodeficient mouse strain characterized by lack of B, T, and NK lymphocytes) showed similar phenotypic heterogeneity to the original donor: thus, [CD34+, CD38neg] retain the differentiating capacity necessary to give rise to CD38+ and Lin+ cells (lineage positive).
Table 1. Molecular markers of cancer stem-like cells
The presence of CSC has also been demonstrated in chronic myeloid leukemia (CML). This disease has a chronic phase and a terminal stage; the blast crisis and molecular events underlying this evolution are not completely understood. In the chronic phase, the chromosomal translocation t(9:22)BCR-ABL, a diagnostic marker of CML, can be detected in most circulating mature lineages. In the blast crisis, however, highly undifferentiated BCR-ABL+ cells accumulate in the blood. In particular, an expansion of granulocyte-macrophage progenitors (GMP) is present in blast cells, showing aberrant acquisition of self-renewal properties and nuclear expression (i.e., activation) of beta-catenin, a key, positive regulator of stem cell self-renewal. These observations imply that during progression of CML the GMP subfraction of leukemic progenitors acquire stem cell characteristics. Thus, the functional hierarchy of CSC can be modified during the natural history of this tumor as a result of its progression.
The requirement for a periodical renovation is not only present in blood but also in the skin and epithelia of the respiratory, gastrointestinal, reproductive, and genitourinary systems. Other tissues like brain, previously considered as exclusively post-mitotic, contain stem cells that can be mobilized and activated under conditions of stress, such as hypoxia. Thus the CSC model could also be applied to solid tumors, and a series of recent papers report data supporting the identification of a stem cell population in different cancers (see Table 1). Initial data were gained in breast cancer where a small population of cells with a CD44+/CD24neg-low phenotype appears exclusively capable of tumor initiation.
The most malignant of brain tumors, glioblastoma multiforme (GBM), was also found to contain a fraction of neoplastic cells identified and selected on the basis of CD133 expression. Not only could CD133+ cells self-renew and differentiate into different neural lineages but also, in vivo, only the CD133+ cells were able to reinitiate malignant gliomas with phenotype similar to the original tumor.
The CSC paradigm may also help explain intratumor heterogeneity, a frequent finding in most cancers: heterogeneity could be consequent to functional diversity of cells at different states of differentiation. On the other hand, the patterns of tumor heterogeneity and gene expression profiles can be highly similar in the original tumor and in distant metastasis.
It is conceivable that the existence of cancer stem cells may provide novel therapeutic targets of increased effectiveness in contrasting or even eliminating cancer. Brain tumors have provided a highly fertile ground to start verifying this hypothesis, as outlined in Table 2. Data are piling up indicating that CD133+ GBM CSC are highly proangiogenic, because of the high levels of VEGF expression, and have greater resistance to chemotherapy and radiotherapy. As a consequence, specific therapeutic strategies can be attempted and combined to overcome CSC. Upon radiotherapy CD133+ GBM CSC activates checkpoint kinases 1 and 2 and repair mechanisms more effectively than CD133neg cells. Resistance to chemotherapy can be linked to an intriguing aspect of the CSC phenotype, the side population (SP) phenotype. SP cells have the ability to extrude the DNA binding dye Hoechst 33342 via the drug transporter BCRP1/ABCG2. Interestingly, the BCRP1/ABCG2 pump can also effectively extrude chemotherapeutic drugs such as mitoxantrone.
Also related, although of less immediate relevance in the clinical setting, are the observations reported by Pellegatta et al. using glioma neurospheres as a target for dendritic cell (DC, the most potent of antigen presenting cells) immunotherapy. Normal neural stem cells may growas neurospheres (NS) in the absence of serum and in the presence of two critical growth factors, EGF and bFGF. NS are enriched in neural stem cells but also contain partially committed progenitors as well as a differentiated progeny. Oncospheres with similar characteristics were obtained from GBM but also from other solid tumors like breast or colon carcinomas. Pellegatta et al. set up a murine model showing that DC loaded with GBM NS are much more effective in protecting mice against the GBM challenge than DC loaded with GBM cells where CSC are poorly represented. Thus, CSC targeting by immunotherapy is feasible and highly effective, opening new scenarios for clinical immunotherapy and supporting the idea that CSC are at the heart of malignant growth. Also of interest is the observation by Piccirillo et al. that treating GBM CSC with the differentiating factor BMP4 can block growth in vitro and avoid tumor formation in the majority of mice in vivo.
Table 2. Therapeutic potential of cancer stem-like cells
Given the increasing number of observations supporting the CSC paradigm in different tumors, it is expected that more therapeutically relevant observations will be proposed in the near future.
Together with therapeutic and clinical implications, the CSC concept seems to have important consequences for our understanding of tumor biology. Modern genetics and molecular biology have given a definition of cancer as a genetic disease in which a growing burden of mutations leads to a progressively more aggressive and ultimately lethal phenotype (Fig. 1). A Darwinian selection for these mutations, privileging those that can confer resistance to different challenges, like hypoxic stress or immune attack, appears to be the most plausible rationale for making sense of this evolutionary catastrophe. The hierarchical CSC model seems to introduce an element of rigidity in this highly flexible scenario, implying that only cells endowed with stem cell properties can afford tumor perpetuation (Fig. 1). Are these two models different or are they compatible? A convincing answer to this tough question will undoubtedly require a lot of robust science in the time to come but comments can be given on the basis of data that are already available. One important issue that the CSC model addresses is that of the cell of origin for cancer(s): stem cells, because they are long-lived and self-renewing, are excellent candidates to play the “cell of origin” role. A stem cell hosting a critical mutation could be quiescent for years and then be engaged in a repair response requiring mobilization and proliferation. For example, hypoxic stress may activate the CXCR4 pathway that not only attracts stem cells but may also favor their proliferation, thus being the spark initiating the cancer fire. However, an initiating mutation could also arise in a more committed progenitor (see the integrated model in Fig. 1): acquisition of a stem-like phenotype could in this context be the consequence of environmental challenges; in vivo, for instance, hypoxia could play an important role in dedifferentiation; in vitro, the modification of growth factors could have similar consequences. Epigenetic changes could play important roles in mediating rapid and genome-wide changes that can substitute for genetic mutations and lead to dedifferentiation.
In the Darwinian model, different mutations (M1 through M4) accumulate during evolution and confer heterogeneity.
In the Hierachical model, tumor arises in a stem cell, thus becoming a cancer stem cell (CSC): heterogeneity is conferred by asymmetrical divisions creating different types of cancer cells (CC1 through 4).
In the Integrated model, a first mutation (M1) can arise in a progenitor or even a committed cell. During progression, though, external stimuli may give rise to a cancer stem cell that through asymmetric division will create other CSC as well as more differentiated tumor cells.