Human development is the work of stem cells. Embryonic stem cells give rise to the progenitors of the body's fundamental tissue types (endoderm, mesoderm, and ectoderm), while more restricted stem cell populations, such as neural and hematopoietic stem cells, provide the seeds for discrete tissue and cell types.
As it turns out, stem cells – or at least stem-like cells -- may also have much to do with the disorderly growth associated with tissue development's polar opposite: cancer.
The cancer stem cell hypothesis, as it's called, is changing the way pharmaceutical companies and researchers think about anti-cancer therapies. Some are pursing strategies targeting the molecular clothes that cancer stem cells may wear; others are focusing on the internal molecular pathways that give them their stem-like properties. Whichever approach one favors, says cancer stem cell expert Peter Dirks of the Hospital for Sick Children in Toronto, the general concept is here to stay. "The idea of thinking of cancer as an aberrant organ, [of] considering the stemness behavior of cancer, is still very, very important," he says.
THE CANCER STEM CELL HYPOTHESIS
The cancer stem cell hypothesis posits that tumors, for all their apparent dysfunctionality, are actually hierarchically organized, with a relatively small number of cells at the top of the cellular pyramid that can both self-renew and differentiate into the other cells (the "bulk cells") of the tumor that are destined to stop dividing within a few generations.
"A tumor is like an organ, but it doesn't follow the rules completely," says Jeremy Rich, Chair of the Department of Stem Cell Biology and Regenerative Medicine at the Cleveland Clinic.
The nature of cancer stem cells themselves is still somewhat mysterious. It isn't clear if they cells are "normal" stem cells gone wild, or somatic cells that have reverted to a stem-like state in a manner akin to Shinya Yamanaka's induced pluripotent stem cells. "There's good data to suggest both [pathways] can be true, and indeed the latter is more commonly documented," says Connie Eaves, Distinguished Scientist in the Terry Fox Laboratory at the British Columbia Cancer Agency, who studies stem cells in chronic myeloid leukemia.
Whatever their origin, though, these cells are currently identified functionally, says Rich – that is, as a tumor cell subpopulation with the capacity to recapitulate the cancer in mice. Yet they are characterized molecularly, using surface antigen phenotypes. Unfortunately, when it comes to cancer stem cells, those phenotypes aren't exclusive: They may be present on other cells, too, meaning cancer stem cell selection strategies capture both stem cells and non-stem cells. For instance, as few as 100 CD133+ cancer stem cells can induce a brain tumor in immunodeficient mice, but it takes about 100,000 ABCB5+ cells, which include melanoma stem cells, to grow a melanoma. [1,2] Rich likens the process to population profiling: "It enriches for cancer stem cells," he says.
Cancer stem cell populations have been identified in tumors from blood to brain to pancreas, and researchers have ascribed to them all manner of properties – that they drive or support metastasis, promote survival in new niches, and mediate radiation resistance, for instance. One recent study by Jeffrey Rosen, the C.C. Bell Professor of Molecular and Cellular Biology & Medicine at Baylor College of Medicine, illustrates the point.
Using both genetically engineered mouse models and a human breast cancer xenograft model in mice, Rosen demonstrated that upon treatment with ionizing radiation, the percentage of cancer stem cells in the remaining tumors actually increased, possibly due to enhanced DNA repair mechanisms in these cells. But, heating the tumors to 42oC in vivo by laser irradiation of injected gold nanoparticles (in addition to radiation treatment) reversed that trend, making the cells more sensitive to radiation. 
Melanoma stem cells seem particularly nasty. Markus Frank, Assistant Professor of Pediatrics and Dermatology at Harvard Medical School, has demonstrated over the past few years that these cells facilitate immune evasion, drug resistance, and vasculogenesis – three traits that contribute to the disease's pathology. "That renders the cancer stem cell concept relevant to the clinic," Frank observes.
Of course, cancer isn't a single disease, and what seems well established in one tumor type is perhaps less clear in others. Even within a given tumor there is variation in the cells present due to how advanced the disease is and the different mutations that have accumulated along the way, says Eaves. "As with many phenomena we don't understand, we start by using very crude descriptions of what we think is going on to get a rough feeling about what the significant mechanisms are. But as you get closer up, the old descriptions may gloss over important differences and hence become less useful for solving a particular problem," Eaves says. Because the cancer stem cells are continuously evolving and may eventually come to dominate the tumor, the utility of the cancer stem cell concept is not yet clear.
Yet in broad strokes, the cancer stem cell hypothesis’ implications are profound. For one, the hypothesis potentially explains why so many tumors recur. Most conventional therapies aim to destroy as much of the tumor as quickly as possible, for instance by targeting highly proliferative cells. But if those therapies miss the cancer stem cells (also called tumor-initiating cells), for instance because those cells are slower growing or more drug-resistant than the more "differentiated" mass of the tumor, then killing off the bulk tumor will have little long-term benefit, as the tumor can simply grow back -- the oncologic equivalent of pruning back a fast-growing weed without removing its roots, as well.
At the same time, the cancer stem cell hypothesis also suggests a new therapeutic approach: Kill off the cancer stem cells, as well as the bulk tumor mass, and the cancer should be gone for good. "The cancer stem cell hypothesis provides a rationale for several therapeutic strategies beyond traditional anti-proliferative agents," Dirks wrote in a 2009 review in Nature Reviews Drug Discovery. 
TARGETING SURFACE MARKERS
There are fundamentally two ways researchers can exploit cancer stem cells therapeutically. One is to target the cells based on their surface marker phenotypes.
Several such proteins have been identified, from CD34 in acute myeloid leukemia, to CD133 in pancreatic, colon, and central nervous system cancers. (See a recent review by Frank for a complete list of known cancer stem cell markers )
Surface markers, says Rich, can be like the racing stripe on a car. They may serve no clear biological function to the stem cell per se – that is, they don't make the cells stem-like -- but they do make the cells easier to find. "Racing stripes don't make a car go faster, but they can help you identify cars that go faster in general," he says. Of course, that's not always true; Cancer stem cell markers are not exclusively reserved to stem cells. "You can also paint a stripe on a Pinto," Rich says. In that sense, targeting therapies through cellular markers carries the risk of side effects, in that not every cell expressing a particular marker is a CSC.
Complicating matters, even well established markers aren't always present, says Craig Jordan, Professor of Medicine at the University of Rochester Medical School. Jordan studies stem cells in acute myeloid leukemia, one of the cancers in which the stem cell hypothesis is best worked out. AML stem cells are generally CD34+, says Jordan, but not exclusively; sometimes, they express other surface antigens instead. "You can get this squirrelly behavior where they are not consistent," he says. As a result, he says, "If your therapy is based on targeting only a single surface antigen, your likelihood of success is minimal."
Nevertheless, some researchers are pursuing the targeted approach.
Ravi Majeti, an Assistant Professor at Stanford University's Institute for Stem Cell Biology and Regenerative Medicine, for instance, has used antibodies to target stem cells in acute myeloid leukemia.
AML stem cells express the antigen, CD47. According to Majeti, this protein is not "just a spot on a dog" – that is, biologically meaningless except as a molecular target; CD47 essentially tells the immune system not to attack the tumor, what Majeti calls a "don't-eat-me signal." He speculates that as the tumor develops, it incurs cellular damage that prompts it to express a cell surface marker that promotes phagocytosis by immune cells. The tumor compensates by overexpressing CD47, which effectively cancels out that signal. By targeting the CD47 protein with an antibody, Majeti and his team essentially restored immune targeting of the tumor.
"We saw a really profound ability for the antibody to eliminate the leukemia in the mouse experiments," Majeti says. In the article, he and his colleagues wrote, "These results establish the rationale for considering the use of an anti-CD47 monoclonal antibody as a novel therapy for human AML." 
Majeti's experience notwithstanding, most researchers have adopted the strategy of targeting cells via the cell signaling pathways that give cancer stem cells their "stem-ness." Several such pathways have already been identified, including Notch, hedgehog, and Wnt, and several have been tested in patients. (Some of these pathways, of course, are also active in normal stem cell populations, says Rich; in that case, the goal is to find a "therapeutic index" of the drug that can hit the cancer cells while causing minimal damage to normal cells.)
Jenny Chang, Director of the Methodist Cancer Center in Houston, and colleagues, for instance, recently completed a Phase I/II trial of a small molecule inhibitor of the Notch pathway, called MK-0752, in 30 patients with breast cancer. MK-0752 is a gamma-secretase inhibitor, which blocks Notch-driven gene expression. "We definitely saw responses, much higher than we expected," Chang says. Side effects, she adds, were mild; though they anticipated gastric dysplasia – that is, gastrointestinal problems -- the only significant side effect they observed was lethargy. "It was a well-tolerated drug," she says.
A second study, using multiple pathway inhibitors, was less successful. Patricia LoRusso of the Barbara Ann Karmanos Cancer Institute in Detroit, Mich., and colleagues (including Chang) tested a combination regimen of a gamma-secretase inhibitor for the Notch pathway, plus GDC-0449, a Hedgehog antagonist, in a phase 1 study of advanced breast cancer. According to ClinicalTrials.gov, the study, which was to include up to 36 subjects, has been suspended for "potential risk of severe or life-threatening arrhythmia."
Other studies have highlighted the potential power of still other targets, including TGF-beta, poly (ADP-ribose) polymerase-1 (PARP-1), and others.
Naturally, drug developers are pursuing these targets, too. OncoMed Pharmaceuticals, for instance, in April received approval from the US Food and Drug Administration to initiate a Phase I clinical trial of OMP-18R5, a monoclonal antibody targeting the Wnt pathway in tumor-initiating cells. And Verastem, whose founders developed a method to screen compounds targeting cancer stem cells, announced recently that it had raised $32 million to drive its drug development program forward and said its first candidate “is projected to enter the clinic in 2012.”
Given the dynamic and protean nature of cancer stem cells, effective treatment strategies will likely require a multipronged approach, says Rosen. "The question really is how to combine [cancer stem cells targeting] with conventional therapies," he says. "Rather than using just a single pathway inhibitor, you'll probably have multiple pathways you need to target to prevent [tumor] resistance and recurrence."
 S.K. Singh et al., "Identification of human brain tumour initiating cells," Nature, 432:396–401, 2004.
 T. Schatton et al., "Identification of cells initiating human melanomas," Nature, 451:345–9, 2008.
 R.L. Atkinson et al., "Thermal enhancement with optically activated gold nanoshells sensitizes breast cancer stem cells to radiation therapy," Sci Transl Med, 2:55ra79, 2010.
 B.B.S. Zhou et al., "Tumour-initiating cells: Challenges and opportunities for anticancer drug discovery," Nat Rev Drug Disc, 8:806–23, 2009.
 N.Y. Frank, T. Schatton, M.H. Frank, "The therapeutic promise of the cancer stem cell concept," J Clin Invest, 120:41–50, 2010.
 R. Majeti et al., "CD47 is an adverse prognostic factor and therapeutic antibody target on human acute myeloid leukemia stem cells," Cell, 138:286–99, 2009.