Cancer Stem Cell Hypothesis: Proceed with Caution

Where the Cancer Stem Cell Hypothesis Comes From

Researchers first proposed the cancer stem cell hypothesis in 1937, to explain why tumors are so heterogeneous (Clevers, 2011). In the mid-1990s, as stem cell biology took off, John Dick revived the idea.

He showed that some leukemia cell subpopulations have stem cell-like properties (Clevers, 2011). Scientists called these subpopulations “cancer stem cells” — though many now prefer “tumor-initiating cells” — because they can form tumors and appear to self-renew, much like adult stem cells do.

Researchers have since found cancer stem cell populations in tumors of the brain, pancreas, ovary, colon, and liver, as well as in leukemia (Magee et al, 2012). No one has proven tumor-initiating cells exist in every tumor type (Magee et al, 2012). Still, studying cancer stem cells matters because of what they could mean for cancer therapy.

Why Cancer Stem Cells Matter for Therapy

In many tumors, certain cancer cell subpopulations seem to resist chemotherapy and radiation (Clevers, 2011; Magee et al, 2012). This is a problem: if therapy fails to target cancer stem cells, the results can be counterproductive.

• Computer simulations show that therapy aimed only at non-tumorigenic cancer cells would leave behind — and enrich for — the tumorigenic, tumor-initiating cells. This could make many cancers more aggressive (Vermeulen et al, 2012).
• This may explain why some cancers grow more aggressive after treatment with current therapies.
• Tumor heterogeneity makes this worse, since different cancer cell types respond differently to the same therapy (Vermeulen et al, 2012) — making it hard to design one ideal treatment.

Competing Models of Tumor Heterogeneity

The cancer stem cell hypothesis isn’t the only explanation for tumor heterogeneity. Three models are commonly discussed:

• Stochastic model: Random genetic and epigenetic variation drives heterogeneity. Selection favors the hardiest subpopulations, and clonal evolution causes uneven growth within a tumor (Magee et al, 2012).
• Microenvironment model: Heterogeneity comes from external factors in the tumor microenvironment. Cells near structures like blood vessels form a niche that temporarily or permanently changes nearby tumor cells (Magee et al, 2012).
• Cancer stem cell model: A distinct subset of the tumor is tumorigenic. These cells can self-renew (making more tumorigenic cells) or differentiate into the tumor’s non-tumorigenic bulk (Magee et al, 2012).

Increasingly, researchers favor combining these models rather than picking just one (Clevers, 2011; Magee et al, 2012). If you’re studying a potentially tumorigenic population in a tumor model, you should account for all three factors.

For example, studying ovarian cancer stem cells in vitro alone misses the microenvironment entirely, which can skew your results.

Methods for Studying Cancer Stem Cells — and Their Limits

Common approaches include:

• Isolating specific surface marker phenotypes.
• Using cultures thought to favor cancer stem cell clonogenicity, like sphere-forming cultures.
• Serial transplantation of specific populations into immunocompromised mice, to test tumorigenicity.
• Microscopic analysis of tumor heterogeneity using markers.

Each method has real limits to keep in mind:

• The cancer stem cell phenotype may only show up in certain patient samples or at certain ages — it may be context-specific (Magee et al, 2012).
• It’s still unknown whether non-tumorigenic cells can become tumorigenic cancer stem cells, either spontaneously or through de-differentiation.
• This flexibility (plasticity) raises doubts about isolating one population and later claiming it explains tumor heterogeneity. There’s often no data linking the original isolated population to what shows up in later studies.

The bottom line: don’t rely on just one assay. Combine multiple assays for rigorous tumor-initiating cell research.

For instance, when studying different brain tumor populations, keep each method’s limits in mind. Confirm your findings with both a well-formed sphere formation assay and an in vivo limited-dilution tumorigenicity model. Even then, the number and behavior of cancer stem cells can vary a lot between patients — so test each case rigorously before applying general concepts to treatment.

Things to Keep in Mind When Studying Cancer Stem Cells

• Certain cell surface markers are correlated with a cancer stem cell phenotype (though this remains debated):
  • Glioma: CD133, SSEA1, CD49f, Musashi-1, and Nestin
  • Breast: BMI-1, CD44, CD24, CD49f, ALDHA1, and EpCAM
  • Lung: ALDHA1, CD90, CD117, and EpCAM
• Researchers also traditionally use upregulation of stem cell-associated genes — such as Nestin, Oct4, Sox2, Nanog, Mushashi1, Notch1, and Notch4 — to identify cancer stem cell subpopulations.
• Multiple primary tumors generally make better specimens than immortalized cancer cell lines, which build up mutations over repeated passages that can distort the true phenotype of tumor-initiating cells.
• Many cancer stem cell labs agree that lineage tracing and side-by-side fate mapping of tumor subpopulations is essential for proper studies.
• Single-cell serial transplantation into immunocompromised mice is a solid assay for testing cancer stem cell phenotype, if feasible in your system. Note that minor immunoediting can still happen in immunocompromised mice — using a syngeneic mouse line can limit this if you’re working with murine specimens.
• Consider quiescent (dormant) cancer stem cells:
  • You can test for these with a western blot on your cell population of interest, comparing stem cell-associated proteins (Sox2, Nestin, Oct4, Nanog) against cell proliferation markers — look for an increase in cell cycle regulators like p21, Cyclin D2, and TP53, paired with a drop in proliferative markers like Cyclin B1, cdc20, and Myc (Moore and Lyle, 2011).
  • Label-retention/chase experiments — such as tritiated thymidine (3H-TdR) or 5-bromo-2-deoxy-uridine (BrdU) — are another good option, and offer an in vivo alternative (Moore and Lyle, 2011).
• Genetically engineered mouse models that spontaneously form tumors let you study tumor-initiating cells. This approach avoids most of the artificial bias seen in cross-species cell engraftment or high-passage cancer cells. Artificial plastic culture conditions may also skew in vitro studies, since they lack the mechanical cues found in a real tumor microenvironment. You can help control for this by running clonal analysis on a 3D scaffold system that mimics the tumor’s actual location (Pastrana et al, 2011).
  • Keep in mind that sphere formation assays may select against tumor-initiating cells that simply don’t form spheres (Read and Wechsler-Reya, 2012).

Suggested Reading/References:

The cancer stem cell: premises, promises, and challenges. Clevers H. Nature Medicine 2011 Mar 7:17(3).

Cancer Stem Cells: Impact, Heterogeneity, and Uncertainty. Magee JA, Piskounova E, Morrison SJ. Cancer Cell 2012 Mar 20:(21).

The developing cancer stem-cell model: clinical challenges and opportunities. Vermeulen L, De Sousa e Melo F, Richel DJ, Medema JP. Lancet Oncology Feb: (13):e83-89.

Quiescent, Slow–Cycling Stem Cell Populations in Cancer: A Review of Evidence and Discussion of Significance. Moore N and Lyle S. Journal of Oncology, 2011.

Spheres without Influence: Dissociating In Vitro Self-Renewal from Tumorigenic Potential in Glioma. Read TA, Wechsler-Reya RJ. Cancer Cell, 2012 Jan 17: (21).

Eyes Wide Open: A Critical Review of Sphere-Formation as an Assay for Stem Cells. Pastrana E, Silva-Vargas V, Doetsch F. Cell Stem Cell May6:(8).