DURING CANCER—NOT JUST BUMPING UP SURVIVAL RATES or making incremental progress on a few kinds of malignancies, but wrestling the thing itself to the ground and driving a stake through its heart—has proved maddeningly quixotic. At one time or another, surgery, radiation and chemotherapy each seemed to do the trick. Yet even in combination, those often can’t eliminate stray cancer cells that appear endlessly able to seed new tumors, so researchers have come up with ever more ingenious approaches. With immunotherapy the goal is to stimulate the body’s immune response to recognize cancer cells as dangerously foreign. Antiangiogenesis uses drugs to try to shut off a tumor’s blood supply. In targeted therapies, so-called smart drugs are designed to block specific pathways of cancer growth.

All of these approaches have had their successes. Yet patients continue to relapse and die, and now another potentially game-changing—but controversial—cancer theory could explain why. This latest idea runs counter to the prevailing view that any cancer cell in a tumor is as likely as any other to propagate a new tumor. In this alternative model, only a select few cells—cancer stem cells—are able to sustain the disease, and they’re the ones anticancer strategies should target.

But cancer stem cells are a tough lot, says Max Wicha, a cancer biologist at the University of Michigan. They can expel therapeutic agents sent to kill them, and they can repair damage to their DNA caused by drugs, radiation or surgery. Normal adult stem cells rest quietly until they receive signals from wounded, infected or dying cells that rouse them to differentiate into whatever type of tissue cell is needed, and stem cells often migrate to remote locations that need repair. So it’s conceivable that anticancer therapies could inadvertently awaken similar survival and repair mechanisms in cancer stem cells, stimulating them to metastasize.

“If the cancer-stem-cell hypothesis is true, treating the majority of dividing cancer cells will shrink a tumor but won’t cure the cancer unless we can target the cancer stem cells themselves,” says Daniel Haber, who directs the Massachusetts General Hospital Cancer Center. That could explain why tumor shrinkage—the gold standard for measuring a drug’s effectiveness—doesn’t always translate into longer survival for patients.

Scientists spotted the first cancer stem cell in 1994, in a form of leukemia, and during the past decade, reports of cancer stem cells in diverse solid tumors have proliferated, although much of the latest research has yet to be replicated and confirmed. What’s more, the cancer-stem-cell hypothesis—even the idea that such cells exist in most cancers—is far from universally accepted. “There are lots of ideas about which cancers follow the model and which don’t,” says Sean Morrison, a colleague of Wicha’s at the University of Michigan. “Dogmas are being created at a furious rate, with ideas that people find intuitively attractive getting repeated to the point that nobody remembers that they’ve never actually been tested.”

But discovering the truth about cancer stem cells is crucial, Haber says, “because either our current strategy of developing therapies aimed at rapidly shrinking a tumor is wrong or insufficient, or we’ve been on the right path and shouldn’t be distracted. So we’d better know, one, whether cancer stem cells exist, two, how to find them, and three, what their defenses are.”

THE LEAST CONTROVERSIAL PART OF THIS WHOLE DISCUSSION is the original group of experiments in leukemia, performed in John Dick’s laboratory at the Hospital for Sick Children in Toronto, that defined the cancer-stem-cell model. “That we could begin to talk about cancer stem cells at all was because we’d had decades of rigorous research on the blood system,” says Dick, director of the Cancer Stem Cells Program at the Ontario Institute for Cancer Research.

That earlier work had shown that while most blood cells can give rise to only their own kind, a rare few can turn into any type of blood cell. These anomalies, hematopoietic stem cells (HSCs), were the first kind of adult tissue stem cell to be discovered, and they have two crucial hallmarks of “stemness”—they’re self-renewing, and they can create any of the many types of differentiated cells in blood or in a particular body tissue. When called upon to divide, an HSC produces one daughter cell that’s a copy of itself (self-renewal) and another that sets off on a path toward specialized functions (differentiation).

But it was difficult to coax these cells to differentiate in a petri dish, and in any case Dick wanted to know how they behaved in a living animal. He turned to a strain of mice, developed during the 1980s, that was known by the acronym SCID (for severe combined immunodeficiency). Dick reasoned that SCID mice probably wouldn’t reject human cells, so he transplanted human HSCs into the mice after irradiating their bone marrow to destroy their innate blood-making system. He found that the transplanted stem cells transformed themselves into different types of human blood cells and reconstituted the blood system.

Next, Dick tried a similar experiment with acute myeloid leukemia. First he established that a random sampling of human leukemia cells always generated cancer when transplanted into immunocompromised mice. Then he transplanted smaller and smaller numbers of cells, trying to determine whether most AML cells could cause cancer. Dick ultimately discovered that only one in a million AML cells could initiate leukemia, and he wondered whether a defining feature of that very rare cell was that it had stemlike properties.

Previously, researchers had learned to recognize HSCs by looking at the cells’ surface proteins. All HSCs had a CD34 protein, which normal blood cells lacked, and they lacked CD38, which normal blood cells had. Dick used this CD34 positive/CD38 negative marker to separate the leukemia cells into two classes, and in a 1994 experiment he demonstrated that only those with stem cells’ signature characteristics could cause leukemia in immunocompromised mice. He also discovered that transplanting just the CD34+/CD38- leukemia cells from a first set of mice into a second group would create cancer as well. That showed that the transplanted cells had self-renewal properties and could propagate cancer indefinitely. Moreover, the leukemia in the second group of mice re-created the entire population of leukemia cells in the original cancer.

That meant that the presumed leukemia stem cells—those with the CD34+/CD38- marker—that had initiated the cancer had differentiated into the many other cell types that occur in leukemia as the disease progressed in the new mice. “From that we could infer a hierarchy, one with a rare stem cell at the top that created those downstream, differentiated cells,” Dick says.

OTHER SCIENTISTS SOON REPLICATED AND EXPANDED Dick’s research, applying his methods to other blood-borne leukemias and lymphomas. Thus emerged a stem cell model in which a few self-renewing “super cells” that seemed able to live indefinitely sat at the top of a pyramid made up of those stem cells’ differentiated, mortal progeny. Importantly, those progeny at the bottom of the pyramid couldn’t reconstitute a tumor when they were transplanted into mice, so it was proposed that they couldn’t cause metastasis or relapse either.

A spate of recent research has seemed to find the same mechanism at work in a variety of solid tumors. In 2003, Michael Clarke and his colleagues (including Morrison and Wicha) at the University of Michigan identified cancer stem cells in breast cancer, and in 2004, Peter Dirks, at the Hospital for Sick Children in Toronto, detected them in a kind of brain cancer. Dick then found stem cells in colon cancer in 2007, and the following year Tobias Schatton and Markus Frank, at Brigham and Women’s Hospital in Boston, discovered them in melanoma. Cancer stem cells have also been reported in cancers of the head and neck, prostate, testicles and ovaries.

The markers that identify these cells vary from cancer to cancer, and so does the ratio of cancer stem cells to other tumor cells. But with the same basic principles appearing to hold sway, some scientists began to think it added up to a new unifying theory for cancer. If most kinds of cancer have stem cells that can continue to seed new tumors indefinitely, that could explain why patients so often suffer relapses—and it could point the way to new treatment strategies.


BUT THERE HAVE BEEN SKEPTICS ALL ALONG. Kornelia Polyak at Dana-Farber Cancer Institute in Boston found that although many cancer cells show stemlike properties, sometimes there are no identifiable stem cells in a tumor. Also, she suggests, the presence of cancer stem cells may not solely determine a cancer’s aggressiveness or resistance to treatment. In some estrogen-sensitive breast cancers, for example, it’s the more differentiated cancer cells, rather than stem cells, that are toughest for chemotherapy to kill. And while such breast cancers have a high percentage of stem cells, those tumors respond better to chemotherapy than do tumors with proportionately fewer of the special cells. “Either we don’t have good markers for cancer stem cells and so we don’t always recognize them, or many kinds of cancer cells have potential to develop into a tumor,” she says.

Polyak also questions the validity of conclusions based on transplanting human cancer cells into mice. “Looking at how a single cancer cell behaves if it’s transplanted into a mouse won’t necessarily tell you what it does in its natural environment,” she says. “We don’t know whether the cells that cause cancer in a mouse are the ones that cause patients to relapse.”

Other researchers share Polyak’s concern about drawing conclusions from mouse studies. Even extremely immunocompromised mice have remnants of an immune system, and it’s possible that what differentiates cells that generate tumors from those that don’t is their ability to survive the attack of mouse immunity, regardless of whether they’re cancer stem cells.

To investigate that possibility, the University of Michigan’s Morrison has used mice with an additional genetic change that further cripples their immunity. His results, reported in 2008 and 2010, cast further doubt on the idea that cancer stem cells are a key to the spread of most tumors.

Morrison found that at least one-quarter of the melanoma cells he took from patients with different stages of the disease initiated tumors in his special mice. Moreover, there was no correlation between cells carrying surface markers identified as badges of cancer stem cells and cells that could create new tumors, or between how many apparent stem cells there were in a patient sample and how fast tumors grew in the mice. “As far as we could tell, all kinds of cells can generate a tumor,” says Morrison. (Two other labs performed similar experiments involving melanoma and obtained similar results.)

Morrison thinks that using mice with a greater level of immunodeficiency might change the results of some of the experiments that appeared to confirm the importance of cancer stem cells. He also thinks researchers should re-examine their assumptions about the cell surface markers they use to identify presumed cancer stem cells. “In many cases, claims about solid tumors come from one paper about work in one lab with compelling data on cells from only a few patients,” he says. “Rarely have additional studies tested the same markers in larger numbers of patients. Markers that work in one patient may not work in another patient.”

Dick agrees that “people have universalized from a very small number of studies. They have lost sight of where the original work came from and how we were able to make these inferences based on rigorous quantitative analysis of single cells.”

Caveats about inadequate experimental methods have worried several other researchers, including Haber, who is using some of the markers Morrison has called into question in his own work to develop methods of detecting circulating tumor cells in the bloodstream. “The entire field depends on key observations that have now been called into doubt,” says Haber. “We have to ask, how open are we to reinvestigating? It’s rare to see an entire field teetering like this.”

AROUND THE TIME MORRISON WAS LOSING FAITH in the cancer-stem-cell model, Robert Weinberg at the Whitehead Institute in Cambridge, Mass., who had always considered himself a cancer-stem-cell agnostic, made a discovery that turned him into a believer—but in an unorthodox interpretation of the model that also might explain some of the contradictory evidence Polyak has noted.

To understand what drives metastasis, Weinberg’s lab had been studying a phenomenon called the epithelial-to-mesenchymal transition (EMT) in mammary tissue. In this process, epithelial cells, which line body tissues and cavities and are normally stationary, become more like the mesenchymal cells in connective tissues. Mesenchymal cells are intrinsically motile and invasive because one of their jobs is to burrow into tissues and remodel them. They are also more resistant to apoptosis, or cell death, than epithelial cells. “These are the same properties that enable cancer cells from primary tumors to migrate to distant sites and disseminate cancer,” Weinberg says.

Still, he was surprised in 2008 when his laboratory demonstrated that cells undergoing EMT gained the attributes of adult stem cells. “Such cells can renew themselves and give rise to more differentiated progeny,” he says. “We then discovered that the EMT works the same in malignant breast cancer cells as it does in normal epithelial cells. That suggests that cancer cells don’t invent new stem cell programs but rather take over those that already exist.” Weinberg thinks the same process probably happens in other carcinomas that arise from malignancies involving various kinds of epithelial cells.

Weinberg’s EMT explanation for the origin of at least some cancer stem cells makes sense to many researchers on both sides of the cancer-stem-cell debate, but it also changes the understanding of the model. It means that a run-of-the-mill tumor cell could undergo an EMT and become a cancer stem cell. Normal mammary stem cells and breast cancer stem cells can also undergo a reverse transition, from mesenchymal to epithelial, and lose their stemness, which may explain observations, by Polyak and others, that any hierarchy that does exist in a tumor is potentially unstable.

In this version of the cancer-stem-cell model, a cell’s state is fluid and reversible, at least to a degree. But if that’s true, does it make sense to target cancer stem cells in cancer treatments? Suppose there were a therapy that could kill cancer stem cells. What would prevent a nonstem cancer cell from undergoing EMT and becoming a cancer stem cell?

The feasibility of such a two-way transition may explain findings in Polyak’s laboratory and elsewhere that undifferentiated cells can acquire stemlike properties as cancer progresses. Similarly, Polyak says, stemlike cancer cells themselves may undergo genetic changes. So, whatever hierarchy there is, it tends to evolve, as diverse cells within a tumor reproduce and some either acquire or lose stemlike characteristics.

New research suggests that even stem cells in leukemia—the cancer for which the existence of the hierarchy is most firmly established—may evolve. Dick and members of another laboratory independently reported in Nature, in January 2011, that in acute lymphoblastic leukemia, patients’ “cancer-initiating cells”—Dick has not yet proved to his rigorous standards that these are actually cancer stem cells, though they do cause cancer when transplanted into mice—continue to undergo mutations and send off progeny with varied genetic profiles. Moreover, the genetic makeup of these branches of stemlike cells varies from patient to patient, and even within the same patient, it varies in different stages of the cancer’s progression.

That raises the essential question of whether it’s possible to identify and target stem cells in treating cancer. If stem cells aren’t predictable enough for that, studying them may reveal much about cancer biology without providing a path to new therapies. “If cancer stem cells turn out not to be rare, and if they’re reversible, do you gain anything from being able to identify them?” asks Morrison. “Then you’re no longer looking for a needle in a haystack—it’s a haystack in a haystack. If there’s no hierarchy, we’re back to where we started. We have to kill all the cancer cells after all.”

BUT THAT’S NOT THE ONLY POSSIBILITY. It could be that current therapies may be missing cancer stem cells, and that although we can’t focus just on that subpopulation—because to do so would miss other cancer cells that may be able to become stem cells—it may be necessary to find therapies that can also eliminate cancer stem cells. “In leukemia, for example, we know our current treatment kills most of the leukemia cells, but not the leukemia stem cells,” Dick says. “We want to include the leukemia stem cells in therapeutics.”

The same may hold true for at least some other cancers. Wicha, in separate collaborations with Dana-Farber and Baylor College of Medicine and with OncoMed Pharmaceu OncoMed Pharmaceuticals (a company he co-founded), is testing molecules he believes can inhibit cancer stem cells in patients with breast cancer. These clinical trials are comparing the effectiveness of chemotherapy alone to chemotherapy that includes the inhibitors.

The relevance of the cancer-stem-cell model could turn out to depend very much on the type of cancer and the genetic profile of an individual patient’s tumor. And although the reality is more complex than those embracing the original theory might have hoped, the model remains a very hot area of research. “While understanding and targeting cancer stem cells won’t be sufficient by itself, it may prove to be essential to achieve long-term remission for some currently intractable cancers,” Weinberg says. Patients could need to mix cancer-stem-cell therapies into cocktails that also might include targeted therapies, antiangiogenesis drugs, immunotherapy and chemotherapy.

Whether it’s still too early or too naive to hope for a unified theory for cancer—which comprises, after all, a multifaceted universe of diseases—remains to be seen. The key is to remain grounded in solid evidence. “The history of cancer biology is littered with dashed hopes following breathless predictions about how this discovery or that paradigm shift will lead to widespread cures,” says Morrison. “Cancer is endlessly resourceful. That’s why none of these latest things—including the cancer- stem-cell model—turns out to be the whole answer.”