In May 2006 Dwayne Berg woke up on a gurney in a Seattle emergency room, an IV in his arm and a team of doctors and nurses working him up. The last thing the 42-year-old financial executive could remember was running on a treadmill at his gym, part of his regular fitness regimen. He had suffered a seizure and tumbled off the machine, and although he had not hurt himself in the fall, doctors had asked for an MRI scan of his brain to see if they could find a cause for the seizure.

They did, and the news was not good: the scan showed a large mass in the left frontal lobe that turned out to be a malignant glioma, a brain cancer that is almost invariably fatal. Berg underwent standard treatment: an operation to remove the tumor, followed by chemotherapy and radiation to eradicate any cancer cells that might remain.

Today Berg is back to his fitness regimen, without any sign that the tumor has returned. But almost certainly it will, and when it does, current treatments offer little chance of cure.

More than 25,000 Americans are diagnosed with a malignant glioma every year, according to the Central Brain Tumor Registry of the United States. About 60 to 70 percent of these cancers occur in the deadliest form, glioblastoma, the type that took the life of Massachusetts Senator Edward Kennedy in August 2009, 15 months after he was diagnosed. Like Berg, Kennedy’s first sign of trouble was a seizure, and of course Kennedy also received the best treatments medicine has to offer. Berg had a less aggressive form called oligodendroglioma. Other common types include astrocytomas and oligoastrocytomas, names derived from normal brain cells called oligodendrocytes and astrocytes that resemble cells in these tumors.

Progress in developing new treatments for brain cancers has been agonizingly slow. Over the past 30 years medicine has found cures for some leukemias and lymphomas and markedly reduced the death rates for breast, prostate and colon cancer through early detection and treatment. Yet only three new drugs that treat brain cancer have been approved by the U.S. Food and Drug Administration in the past 35 years, and these prolong the lives of patients by only a few months. Despite decades of research, the life expectancy of a person diagnosed with glioblastoma remains 12 to 14 months, roughly the same as it was a century ago.

But that picture may soon brighten. New research focusing on a tiny population of tumor-regenerating stem cells within brain tumors has fundamentally changed our understanding of how brain tumors develop. The discovery and characterization of these cells brings new hope that these deadly cancers can be successfully treated, possibly with drugs that are already on pharmacy shelves.

Missing Link
The goal of current brain cancer treatments is to rid the body of as many tumor cells as possible. When feasible, the first step is to surgically remove all accessible tumor tissue. Such “debulking” operations often require cutting near critical areas of the brain that, if damaged, could leave the patient severely disabled. In Berg’s case, the tumor was close to his brain’s speech center, Broca’s area. Berg was kept awake during his operation so that I and the rest of his surgical team could talk to him and determine if we were encroaching on this area. Our strategy worked, but swelling caused by the surgery still left Berg unable to talk for two weeks, after which his speech returned slowly over several months.

Surgery is typically followed by chemotherapy and radiation, which target rapidly dividing, cancerous cells. These treatments will slow tumor growth and in some cases will shrink or eliminate the mass by triggering cell death. But most cancers recur, sometimes years later, often having acquired resistance to treatment.

Cancer recurrence after treatment has long been a mystery. Traditionally, scientists thought tumors consisted of a largely homogeneous group of rapidly proliferating cells. As a result, standard cancer therapies were designed to target and kill those cells. But recently researchers have realized that this dogma may be dramatically wrong—and in a way that explains the mystery of recurrence.

The story began in the mid-1990s, when molecular geneticist John Dick of the University of Toronto and his colleagues made an astounding observation: only a tiny fraction of the leukemia cells in a patient’s blood were capable of seeding a new leukemia when they were transplanted into another animal. The rest would proliferate in a test tube but could not regenerate the vast population of leukemic cells. The regenerative minority became known as cancer stem cells.

Cancer stem cells have since been found in a variety of solid tumors, including those of the skin (melanoma), breast, ovaries, pancreas, liver, prostate, colon, and head and neck. Like the normal stem cells that help to maintain healthy tissues, the cancerous stem cells can replicate themselves and never die out. Moreover, like normal stem cells, cancer stem cells can mature into different types of cells—in this case, the cell types found in a tumor rather than in healthy tissue. In one theory, cancer stem cells are considered to be stem cells “gone bad.” According to this idea, genetic damage in normal stem cells (or in progenitor cells, which are slightly more developed than stem cells) leaves them with some of their standard reproductive traits but causes them to generate malignant instead of healthy progeny.

Many researchers now believe that cancer stem cells form the lifeblood of a cancer, sustaining the mass and giving rise to millions of new malignant cells. In addition, these stem cells share traits with normal ones that make them highly resistant to standard cancer therapies. Thus, cancer stem cells explain why standard cancer treatments so often fail: those therapies target the wrong cells.

Growth in the Brain
But for years after the discovery of cancer stem cells, few scientists thought that they played a role in brain cancer. Most cancers, such as those of the breast, lung and colon, involve active tissues in which cell proliferation is common, so biologists easily warmed to the idea that stem cells would be present to replenish those tissues. In contrast, the brain was thought to be different: it was supposedly static—devoid of growing, dividing cells, much less stem cells.

There was evidence, however, that the animal brain was not static. In the early 1980s neuroscientists Steven Goldman (now at the University of Rochester) and Fernando Nottebohm of the Rockefeller University found cells in the adult canary brain that could form new brain cells, migrate through the brain and mature into specific cell types. They speculated that these cells could be involved in learning (which appears to be true, at least in some cases).

Other researchers subsequently identified similar cells in adult rodent brains and in adult monkey brains. Then, in a 1998 publication, neuroscientist Fred Gage of the Salk Institute for Biological Studies in La Jolla, Calif., and his colleagues revealed that they had discovered comparable cells in the brains of adult humans, demonstrating for the first time that the adult human brain can generate new tissue throughout life.

Finally, in 2003, neurosurgeon Peter Dirks of the University of Toronto and his colleagues reported spotting cancer stem cells in human brain tumors. The researchers examined human brain tumor tissue for cells bearing a protein on their surface called CD133, which is found on normal neural stem cells. Not only were cells sporting this marker present in the tumors, but those cells could replicate, proliferate and differentiate (develop into mature tissue types) much as normal stem cells do. In addition, the stem cells that copied themselves the most frequently came from patients with aggressive brain tumors.

Since that discovery, scientists have learned more about the properties of brain tumor stem cells. They make up only about 3 percent of the cells in a brain tumor and behave quite unlike other tumor cells: they do not proliferate rapidly and are relatively quiescent. But when small numbers of these cells are transplanted into a mouse brain, they can construct a tumor that is almost an exact copy of the original. Indeed, Dirks and his colleagues demonstrated in 2004 that creating a brain tumor in a mouse requires just 100 brain cancer stem cells, compared with the one million regular cancer cells ordinarily required.

The discovery of stem cells in brain tumors could explain several mysteries about gliomas, such as where they appear in the brain, how they spread and why they are so difficult to treat. Most gliomas, for example, tend to arise from areas where neural stem cells reside—in particular, an area at the base of the brain called the subventricular zone of the lateral ventricles. Gliomas also often spread by traveling along the white matter tracts (bundles of nerve fibers) that connect different areas of the brain; these are the same pathways that stem cells and their offspring follow. And the ability of these brain tumor stem cells to migrate and spread throughout the brain helps to explain why surgery alone cannot cure malignant gliomas: many cells simply elude the scalpel.

Moreover, other properties of cancer stem cells can clarify why malignant gliomas do not respond to standard chemotherapy and radiation. First of all, tumor stem cells divide much less often than most cancer cells. Thus, they are less vulnerable to many chemotherapy agents, which directly sabotage cell division by targeting its molecular machinery and thereby triggering cell death, and to radiation, which also preferentially kills fast-dividing cells by damaging their DNA.

Second, data suggest that glioma cells bearing the CD133 biomarker actively resist the effects of chemotherapy and radiation. Neurosurgeon John S. Yu and his colleagues at Cedars-Sinai Medical Center in Los Angeles, for example, showed in a 2006 paper that CD133-bearing cells turn on a gene for a protein that pumps chemotherapy drugs out of the cell. These cells also produce proteins that help to repair chemotherapy-induced DNA damage and that prevent cell suicide. Meanwhile a group led by neurologist Jeremy Rich, then at Duke University, reported in 2006  that when exposed to radiation, glioma cells carrying CD133 can activate a specific system that fixes radiation-induced DNA damage and can therefore initiate repair more effectively than the vast bulk of glioma cells. Thus, although standard radiation and chemotherapy treatments kill the proliferating cells that make up most of the tumor, they leave behind, unscathed, a remnant capable of regenerating the deadly mass.

Aiming at the Enemy
Nevertheless, the discovery of brain tumor stem cells offers hope to victims of brain cancer, because it suggests that treatment strategies that specifically target those cells could kill the cancer and prevent it from recurring. One of the first challenges is to find better ways to isolate brain cancer stem cells. The molecular flags on the cells—which include characteristic DNA, RNA and proteins—found as yet are not foolproof identifiers. Not all glioma cells that sport CD133 are brain cancer stem cells, and not all brain cancer stem cells carry this marker. Thus, attempts to isolate these cellular time bombs may miss some of them.

Distinguishing brain cancer stem cells from normal stem cells is important for designing therapies that eradicate the former while sparing the latter, which are crucial for regeneration, for repair and (in the brain) maybe for learning. For example, doctors might employ monoclonal antibodies—Y-shaped proteins that help to destroy invading bacteria and viruses—that target surface biomarkers unique to brain cancer stem cells. Such molecular tags might also reveal whether a brain cancer is more or less aggressive and which drugs are most likely to eradicate it. After treatment, tests that look for the presence of certain biomarkers in the blood or spinal fluid may also make it possible to detect a recurring tumor before it has had time to grow.

Researchers are also trying to understand the molecular changes that transform normal cells into a brain cancer. Their efforts could lead to drugs that prevent the cancers from developing, shrink them or stop them from spreading. To become cancerous, a cell must sustain genetic damage that alters proteins in one or more of the molecular pathways that control a cell’s growth and behavior. Often a disrupted pathway interacts with other processes, leading to additional failures—and those interactions may change as the tumor grows or responds to treatment. A new approach called systems biology can help make sense of such complexity. Systems biology combines technology that can quickly analyze the activity of thousands of genes in tumor tissue with supercomputers that can identify patterns among abnormal genes, proteins and molecular pathways and then link those patterns to clinical information such as treatment and tumor type.

In 2009 neuro-oncologist Markus Bredel, who directs the Brain Tumor Institute Research Program at Northwestern University, and his colleagues used a systems biology approach to unearth a network of genes that appears to play an important role in malignant glioma. In an analysis of gliomas from 501 patients, they identified the most common genes and genetic abnormalities among the cancerous cells, along with their patterns of expression. Many of the most active genes, they discovered, are involved in a complex system of interacting signaling pathways that tells a cell when to grow and when to stop. Certain patterns of gene activity in these interacting networks, they further learned, were associated with better or worse patient survival. They also identified what they called “hub” genes that seemed to be key elements in these networks, providing possible targets for future medications. A larger effort to dissect the molecular anatomy of brain cancer is under way at the Allen Institute for Brain Science in Seattle, where researchers will be creating a 3-D genetic map of these tumors.

Other researchers are finding drugs that temper the toxicity of brain tumor stem cells by coaxing them into a less hazardous form. In a 2006 paper, for example, cell biologist Angelo Vescovi of the University of Milan-Bicocca in Italy and his colleagues studied the effect that a growth factor called bone morphogenetic protein (BMP) had on glioblastoma cells. In the normal brain, BMP directs cells to differentiate, mature and specialize. In their study, Viscovi’s team showed that BMP had a similar effect on human glioblastoma stem cells, causing them to abandon their stem cell–like behavior and become less aggressive. In test tube experiments, BMP shrank the number of stem cells within a tumor. It also prevented the cancer cells from growing into a tumor when they were later implanted in a mouse brain. And administering BMP after a glioblastoma had been transplanted into the brain of a mouse could block the growth of the tumor and save the mouse’s life.

Taking a separate tack, some investigators are targeting the vascular hideouts in which brain tumor stem cells thrive. Like neural stem cells, brain cancer stem cells appear to prefer to occupy areas with a rich blood supply. (The subventricular zone near the bottom of the brain is one such location.) To survive in less vascular regions of the brain, brain cancer stem cells release growth factors that stimulate blood vessel growth.

Some brain tumor patients are already treated with anticancer drugs that block these growth factors to inhibit this growth. One drug, bevacizumab (Avastin), which has been used to treat other cancers for many years, has now been approved for glioblastoma. Unfortunately, although tumors often appear to shrink in response to Avastin, they inevitably grow back.

Off-the-Shelf Treatments?
Intriguing new findings hint that drugs used to treat certain common psychiatric disorders may also be effective against brain tumors—again, by targeting brain tumor stem cells. In a study published in 2009 Dirks and his colleagues created cultures of glioma neural stem cells on which they tested the efficacy of various medications. In a trial-and-error screen of 450 approved drugs, the researchers found that 23 drugs used to treat mental illnesses such as depression, anxiety and schizophrenia killed the glioma stem cells.

These drugs all block or alter the transmission or reception of neurotransmitters (substances that pass information between neurons), and that mechanism probably underlies their toxicity to brain tumors. During brain development, normal neural stem cells need certain chemical signals from their surroundings to transform into mature nervous system cells. Similarly, brain tumor stem cells depend on chemical input to survive and grow. Thus, such neuromodulatory drugs may interfere with the molecular messages that brain tumor stem cells need to multiply and mature.

Testing of these neuromodulatory drugs is still in the very early stages. Currently a major effort is under way to identify which of these compounds appear most promising by screening them against tumor cells in laboratory studies. Once the most promising drugs are identified, however, clinical trials should start reasonably quickly because many of these drugs have already been tested for safety and approved by the FDA for other purposes.

Could an antidepressant treat brain cancer? Dwayne Berg would certainly like to know. Developing a new drug typically takes decades, time that Berg and other brain cancer patients do not have. The promise of combating their disease with available medications is immediately appealing. Other new treatments that target brain cancer stem cells, too, remain unproved. Clinical trials for many of them are just getting under way. But for the first time in a long while, our new understanding of brain cancer is giving patients and doctors some degree of hope.