Cancer Stem Cells: A Review of Potential Clinical Applications

All specialized cells of the human body derive from tissue-specific stem cells, which are defined by their capacity to self-renew and to differentiate. The original hypothesis of CSCs was based on the clonal theory of cancer initiation and progression.¹  This hypothesis states that any normal stem cell, upon acquiring a mutation giving it selective growth advantage, gives rise to a neoplastic clone of homogenous neoplastic cells. The clone, upon expansion and acquisition of additional mutations, expands as a tumor. This model, which is called classical or stochastic, assumes that all cells within a given tumor have the same tumorigenic potential.

A more-recent hypothesis, the so-called hierarchical model, suggests that any given tumor consists of a heterogeneous population of cells, with only a small proportion of them being CSCs.²  However, that small, self-renewing population of CSCs is responsible for tumor initiation and growth maintenance. Those CSCs could be tissue stem cells or a more differentiated progeny, which acquired self-renewal capacity.^3–5  There are 2 mechanisms that could mediate the transformation of normal stem cells to CSCs:

• • 

  Early progenitor cells can gain mutations which gives them self-renewal capacity⁶ 

• • 

  Fully differentiated cells or cells in the late progenitor stage can become de-differentiated to acquire the properties of stem cells.

Two major types of mutations are involved in the above processes: either activation of oncogenes or inactivation of tumor suppressor genes. Additionally, the tumor microenvironment, that is, the CSC niche, and cytokine loops play essential roles in the maintenance of CSCs and in tumor growth and development.^7,8  Using a model of myeloma, Feng et al⁹  demonstrated that stromal cells of the tumor microenvironment create a growth advantage for CSCs. In that study,⁹  the proliferative capacity of multiple myeloma stem cells was stronger when grown in the presence of stromal cells from patients with myeloma than it was in the presence of stromal cells from normal control bone marrow.

The confirmation of CSC theory resulted from in vivo experiments where human tumor cells were xenotransplanted into immunodeficient mice, and the subsequent tumor development was histologically analyzed using an in vitro “sphere-forming” assay.^10,11  Initially, CSCs were considered very rare in human tumors because many tumor cells had to be transplanted into mice to grow the tumors. However, that could have been due to the difference in the human and murine microenvironments because mice-to-mice transplant of even a few leukemic cells led to tumor establishment in recipient mice. In vivo, serial-dilution methods are based on a capacity of normal stem cells or CSCs to reconstitute recipient's bone marrow when transplanted into immunodeficient mice. Once infused, they compete against a predefined number of host murine recipient cells. Different dilutions are performed by using varying ratios of donor and recipient cells. Serial transplantation of these various dilutions allows for a more-precise detection of the highest yield of immature stem cells. This method allows CSC detection on a single-cell level. The serial transplantation technique is when CSCs are transplanted into serial, sequential recipients to sustain tumorigenesis. It allows detection of the most-immature cells in the CSC hierarchy, thus being the most stringent in vivo assay. Even though the above techniques are very sensitive and are still considered the gold standard in CSC detection, they are dependent on host factors, such as microenvironment and bone marrow reconstitution as well as on the timing of posttransplant analyses.

The animal studies outlined above are relatively expensive and require considerable time and effort. Therefore, numerous in vitro assays have been developed for CSC identification, which include the colony assay, the sphere assay, the “side population” (SP) assay by Hoechst labeling,¹²  staining for CSC surface antigens (such as ABCG2 and other markers), aldehyde dehydrogenase (ALDH) activity assay,¹³  and label-retaining cell assay using PKH (Paul Karl Horan) dyes. These methods are summarized below.

The colony-forming cell assay is an in vitro assay used in the study of stem cells. It assesses the ability of progenitor cells, in a semisolid media or methylcellulose-based culture media, to proliferate and differentiate into colonies in response to cytokine stimulation. The colonies formed can then be quantitated and characterized according to their unique morphology. This method can be used to quantitate CSC and assess their proliferation capacity when in vivo assays cannot be used. However, reliability of this assay is controversial, and it is subject to interlaboratory variation because of varying feeder layers and specific culture conditions.

The microsphere assay,¹⁰  initially developed in the 1960s as a “neurosphere” assay, assesses the ability of neural stem cells from mammalian adult brain subventricular neural cells to grow in serum-free medium in nonadherent conditions, forming sphere-shaped cell aggregates. Since then, it has been used to isolate stem cells from many other organs, such as breast, heart, pancreas, and prostate. Before plating, the cells are isolated from tissue by laser microdissection with subsequent enzymatic digestion and are then analyzed with flow cytometry for the expression of certain stem cell markers and purity. They are then plated and grown in nonadherent conditions in serum-free cell culture medium. Cell density is the most critical parameter affecting the yield of clonal cells. However, despite the simplicity of this method, it has some disadvantages. Sphere assays may not detect quiescent CSC. Furthermore, enzymatic digestion, used in the early steps, can alter surface markers and corroborate subsequent sorting by fluorescence-activated cell sorters. In addition, culture media with growth factor differentiation may cause a potential differentiation bias.

The SP assay measures upregulation of P-glycoprotein by CSCs, which leads to active transport of some dyes, such as Hoechst or rhodamine, out of the cells. This phenomenon led to the development of the so-called side population flow cytometry–based isolation technique.^12,14  This technique is being used for isolation of multiple myeloma (MM) stem cells in our laboratory (Figure 1, A through F). Briefly, MM cells are incubated at 37°C, for 90 minutes, in the presence of 2 μg/mL of Hoechst 33342 dye. Subsequently, the cells are centrifuged at 4°C and resuspended in ice-cold, phosphate-buffered saline. Before flow cytometry analysis, propidium iodide is added to exclude dead cells. For flow cytometry analysis, Hoechst 33342 dye is excited at 355 nm, and its emission is detected using 440/30 nm (blue) and 670/40 nm (red) filter systems. Side population cells (which are presumed to be MM CSC), because of dye efflux, are negative for both Hoechst red and Hoechst blue, whereas most tumor cells retain the dye and are, therefore, classified as non-SP. Even though the SP method of CSC isolation is relatively easy, strict procedures must be followed because the results can vary depending on stain and culture conditions (ie, length of time from tumor cell culture to staining, dye concentration, incubation time, and even interval times between shaking during the incubation).¹⁵  In addition, if there are few SP cells, their analysis and sorting by flow cytometry is limited.

Expression of the ABCG2 protein, which is a member of the adenosine triphosphate–binding cassette transporter, might be used as another way of identifying CSCs.^16,17  Expression of ABCG2 was originally described in breast cancer cells, and, more recently, it was reported that SP cells of MM cell lines express higher levels of ABCG2 mRNA than do non-SP cells and that they have increased sensitivity to lenalidomide.¹⁸  Antibody to ABCG2 for direct cell staining is available for use in flow cytometry assays; however, additional studies are needed to determine whether ABCG2 should be used as a surrogate marker for detection and selection of clonogenic and self-replicating CSCs.

High levels of ALDH activity have recently been identified as a characteristic of CSCs, and the Aldefluor flow cytometric assay has been widely used for isolation and study of CSCs in different types of cancer, such as leukemia and prostate and cervical cancer.^13,19,20  In this assay, the active reagent boron-dipyrromethene (BODIPY)-aminoacetaldehyde is added to the cells; BODIPY-aminoacetaldehyde is then converted by ALDH to the fluorescent BODIPY-aminoacetate. The ALDH inhibitor is used as a negative control. Some studies have also included polymerase chain reaction and Western blot for measuring the ALDH gene and ALDH protein expression levels, respectively. Aldehyde dehydrogenase activity is significantly higher in the clonogenic cell population than it is in nonstem tumor cells. Recently, immunohistochemical staining for ALDH has been used in detecting CSCs in various types of cancer, such as cervical, endometrial, pancreatic, breast, and hepatocellular adenocarcinoma.^21–24  Like any other technique, the ALDH assay has some limitations. Aldehyde dehydrogenase activity is affected by chemotherapy. Numerous tissue-specific isoforms limit its detection by immunohistochemistry.

The PKH26 labeling method has been used to identify CSCs by demonstrating their asymmetric division.²⁵  The explanation for the phenomenon underlying this identification method is that the PKH26 dye irreversibly binds to the lipid bilayers on cell membranes and is equally distributed among daughter cells during each subsequent cell division. Therefore, when a CSC goes through asymmetric division, the non-CSC daughter cells will divide rapidly, diluting the dye during cell divisions and, eventually, losing it altogether. In contrast, CSCs are more quiescent and maintain the PRH26 dye. Figure 2, A and B (C. C. Chang, MD, PhD, unpublished data, May 2011) demonstrates asymmetric division of multiple myeloma Roswell Park Memorial Institute, Buffalo, NY (RPMI) 8226 cells stained with PKH26 dye.

Cancer stem cells can also be identified by expression of specific surface markers, which can, in some instances, be tissue specific. Examples of such markers include CD44, CD24, and CD133.^12,26,27  When sorted by flow cytometry, tumor cells expressing the above molecules were more tumorigenic when transplanted into immunodeficient mice than were all of the other types of cells in the tumor. Examples of CSC markers in different types of human cancers include the following (Table): CD24 and CD44 in breast and pancreatic adenocarcinomas; CD44, CD133, EpCAM, and CD166 in colorectal cancer; CD44 and CD133 in prostate adenocarcinoma; and CD133 and EpCAM-hepatocellular carcinoma, CD34, and CD38 in acute myeloid leukemia.

Resistance of tumors to the broad armamentarium of cancer treatments may largely be due to the presence of residual CSCs, which are very hard to eradicate. The presence of circulating tumor cells in patients with cancer correlates with poor prognosis.^28–30  Studies have demonstrated that, in patients with breast cancer, circulating tumor cells expressed a stem cell–like phenotype (CD44⁺, ALDH-1⁺) and were in a nonproliferating state.³⁰  A recent study by Gerber et al³¹  demonstrated, for the first time, that the presence of CSCs (in acute myeloid leukemia) was correlated with clinical outcome and suggested that those cells may be responsible for relapse. They showed that, in the setting of minimal residual disease, remaining leukemic cells were enriched for a particular leukemic stem cell phenotype (CD34⁺, CD38⁻, ALDH^int). Those cells engrafted into immunodeficient mice carried the cytogenetic aberration present at initial diagnosis. The size of that cell population decreased during overt leukemia relapse and was replaced by a more-differentiated leukemic progeny. Furthermore, presence of those cells during complete remission was highly predictive of subsequent acute myeloid leukemia relapse, thus supporting their clinical relevance.

Using an in vivo ovarian cancer model, Kusumbe et al²⁴  demonstrated that, upon exposure to chemotherapeutic drugs, a minor aneuploidy-quiescent, stem cell–like population of tumor cells can reacquire proliferative capacity and, at the same time, can be refractory to chemotherapy. Therefore, these cells could be responsible for the selection of drug-resistant clones and the eventual development of multidrug resistance. There are several mechanisms that mediate CSC resistance to chemotherapy and radiation.³  They include the quiescent nature of CSCs, their presence in hypoxic niches, the upregulation of DNA damage response, and the increase in CSCs of drug efflux.

Alternative therapeutic methods of selectively targeting CSCs are currently in development. Interruption of key signaling pathways (eg, Wnt, hedgehog, and Notch pathways) could be one such approach. On the other hand, application of epigenetic manipulations (such as histone deacetylase inhibitors and DNA methyltransferase inhibitors) could lead to reexpression of tumor suppressor genes.

Another attractive approach is immunotherapy targeting CSCs. Foster et al^32,33  generated cytotoxic lymphocytes directed against a stem cell–like SP of cells in chronic lymphocytic leukemia. The same group similarly demonstrated the potential cytotoxicity of cytotoxic lymphocytes against SP cells in Hodgkin lymphoma.³⁴ 

The trophoblastic theory of cancer suggested by John Beard in 1906 states that all cancers originate from germ cells.³⁵  Subsequent studies discovered a specific group of antigens that are present only in germ cells, trophoblasts, and cancer cells.³⁶  These antigens have been called cancer/testis antigens (CTAs). Since then, 130 CTAs have been discovered, which can be grouped into 83 gene families, separated into 12 sequence homology–based groups.³⁷  The CTAs localized to the X chromosome comprise 44 multigene families and are highly immunogenic. In contrast, immunogenicity of CTAs localized on autosomal chromosomes has not yet been proven.^36,37 

Mechanisms involved in regulation of CTA expression include epigenetic as well as nonepigenetic ones.³⁷  Hypermethylation (both global and promoter-specific) suppresses CTA activity, whereas demethylating agents (eg, decitabine) lead to reinduction of CTA expression. Histone deacetylases repress CTAs, whereas histone deacetylase inhibitors induce the opposite effect. Nonepigenetic mechanisms seem to involve sequence-specific transcription factors, as well as signal transduction pathways, including activated tyrosine kinases.

There have been several studies that demonstrated an association between CTA expression and poor prognosis in patients with solid tumors, such as head and neck cancer,³⁸  lung cancer,^39–42  and cholangiocarcinoma,⁴³  as well as hematopoietic malignancies, such as MM.^44,45  Therefore, studies have explored the potential of using CTAs as immunotherapy targets. A CTA peptide vaccination of patients with advanced esophageal cancer showed objective clinical responses.⁴⁶  Quintarelli et al^47,48  demonstrated PRAME-specific, cytotoxic lymphocyte cytotoxicity against chronic myelogenous leukemia cells. Goodyear et al⁴⁹  demonstrated the presence of CD8⁺, CTA-specific, cytotoxic lymphocytes in patients with MM, which, although the lymphocytes were unable to control tumor cell growth in vivo, lysed MM cells in vitro. Furthermore, patients with evidence of a CTA-specific immune response had a 53% reduction in mortality during a median follow-up of 4 years. Another study⁵⁰  demonstrated significant CTA-specific immune responses in a cohort of 41 patients who underwent allogeneic stem cell transplantation for the management of acute myeloid leukemia or MM. In one study,⁵¹  immunization of a healthy donor with defined CTA protein led to generation of an antimyeloma response in MAGE-A3⁺ MM. The donor was immunized with MAGE-A3 protein, and primed donor cells collected after immunization were transferred to her identical twin after syngeneic stem cell transplant, followed by repeated patient immunizations. Persistent, cytotoxic, lymphocyte-mediated immune response was obtained, and the patient remains in remission more than 2 years after transplant.

Besides expression in germ cells and various tumors, CTAs have recently been shown to be present in normal mesenchymal stem cells and to regulate self-renewal and epithelial-to-mesenchymal transition, a crucial process in the development of metastases, indicating that CTAs may play a role in the “stemness” of various stem cells.^52–54  Therefore, CTAs could possibly be upregulated by CSCs, such as the stem cell–like SP cells of MM, and serve as targets for immunotherapies focusing on CSCs. Our laboratory has been investigating this hypothesis. We have recently demonstrated increased CTA expression by SP cells of MM (Figure 3; C. C. Chang, MD, PhD, unpublished data, Month 2011). We have also shown that treatment of tumor cells with demethylating agents led to upregulation of PRAME expression (C. C. Chang, MD, PhD, unpublished data, November 2011). As mentioned above, studies by Shafer et al³⁴  and Foster et al^32,33  demonstrated selective susceptibility of SP cells in chronic lymphocytic leukemia and Hodgkin lymphoma to cytotoxic T cells.

In summary, CSCs comprise a minor subpopulation of tumor cells that are relatively quiescent and self-renewing and responsible for drug resistance, minimal residual disease, and cancer recurrence. In vivo assays of their detection, even though still considered the gold standard, are cumbersome, nonquantitative, and highly dependent on multiple factors, such as reconstitution in a murine microenvironment. In vitro assays, such as the efflux-based SP assay, ALDH activity assay, and cell surface marker assays, have proven to be reproducible and straightforward methods of CSC detection. Future studies of CSCs and expression of CTAs, as well as other tumor-specific antigens, may not only shed additional light on the biology of cancer but also may lead to the development of synergistic strategies in battling minimal residual disease, such as immunotherapeutic approaches targeting CTAs.⁵⁵