ONE of the worst things facing cancer patients is the uncertainty. Often, their doctors cannot give them an accurate picture of how rapidly their disease is likely to progress, how it will respond to treatment or even what treatment would be most appropriate. In some cases doctors may be unable to say exactly what type of cancer a patient has.
But that is about to change. A wave of optimism is spreading among cancer researchers, thanks to new technology that lets them watch how genes operate inside cancer cells and predict how that particular tumour will behave. What’s more, these new insights will transform our view of the disease. In the future, we will not talk of breast cancer or lung cancer. Instead we will classify tumours according to their particular pattern of abnormal gene activity. In effect, we will be seeing cancer for what it really is – a disease of the genes.
If the predictions come true, how we treat cancer now will seem like a crude footnote in history. In its place oncologists will have a system that lets them precisely quantify what a tumour is, know exactly how it is likely to progress, and be able to predict exactly which treatment is likely to have the best outcome.
Advertisement
Until now oncologists knew very little about the workings of a tumour cell. Although decades of hard slog have uncovered many key genes associated with cancer, how these and many other genes interact to drive the basic biology of the tumour is still largely a closed book. That means doctors base diagnoses, treatment and predictions of outcome on educated guesswork combined with subjective interpretation of a patient’s age and health and what the tumour looks like under a microscope.
Todd Golub plans to change all this. A cancer biologist at the Dana-Farber Cancer Institute in Boston and director of the cancer genomics programme at the Whitehead Institute Center for Genome Research in Cambridge, Massachusetts, Golub has a vision – to build a map that shows how all the genes within a cell interact and how this changes when it becomes cancerous. He and his team are using a tool called a DNA microarray to eavesdrop on the activity of thousands of genes simultaneously and so get much closer to what drives cancer at its most basic level.
Although still in its infancy, this approach might make it possible to find a unique gene activity “signature” for each kind of cancer, and to use this signature to predict whether a tumour is likely to spread to another part of the body, or whether it will respond to a particular therapy. It will also name that kind of cancer based primarily on which particular genes are misbehaving rather than its tissue of origin. “It opens up a whole new way of looking at cancer cells,” says cancer biologist Jonathan Pollack at Stanford University in California.
Ultimately, the aim is to understand every interaction of every relevant gene in every tumour type, and then follow the genetic changes over the course of the disease. With this intricate map of interactions in hand, they hope to open up a new era in the fight against cancer.
So far the results have come mainly from small studies, and the findings will have to be replicated on a larger scale. But experts are quietly confident because they have got consistently encouraging results from a variety of cancers, including brain, lung, breast, ovary, prostate, colon and kidney cancer, as well leukaemia and lymphoma.
At the centre of this optimism is the DNA microarray, a tool pioneered in the mid-1990s at Stanford University. A microarray is a glass or nylon slide dotted with thousands of tiny samples of DNA, each representing a different gene. Microarrays rely on the ability of one strand of DNA to stick to another strand with a complementary sequence.
Before they can use microarrays, researchers extract a cell’s messenger RNA – a read-out of all its active genes – and make a DNA copy of it. They “label” this DNA with a chemical that fluoresces under laser light. Any sample that meets its match on the chip will stick to that spot, and the pattern of glowing DNA dots therefore indicates which genes were turned on when the sample was taken (see Diagram).
Cancer biologists were quick to spot the potential of microarrays. Cancer cells, after all, are simply normal cells whose genes are behaving abnormally. Most aspects of a normal cell’s behaviour, such as cell division, are controlled by particular sets of genes. In cancer, these sets are often turned on or off at inappropriate times. What better way to understand cancer than by capturing the faulty wiring in a single snapshot? But questions remained. Would researchers be able to identify relevant genetic activity within those snapshots? Or would they see nothing but a garbled rat’s nest of noise?
In 1998, Golub and his team proved that microarrays could indeed pinpoint faulty genetic activity, when they used microarrays to analyse bone marrow samples from 38 patients with acute leukaemia. For decades researchers had known that this disease came in two major types, acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML), each with its own course of disease. Using a microarray containing 6817 genes, Golub’s team found that each type had its own genetic signature. From these signatures they selected 50 genes as markers for identifying each disease, and used them to predict the subtype of 34 additional samples. The markers made an accurate prediction in 29 of the samples.
Although doctors already knew how to distinguish between ALL and AML based on other tests, the results were exciting. First, they proved that microarrays could work. Second, even though the existing diagnostic approach was efficient, several of the tests didn’t always work. With genetic signatures on the other hand, you have a one-step process that is potentially more accurate.
This was demonstrated dramatically when Golub came across a boy who had been diagnosed with AML. The boy had some of the symptoms of the disease, but his tumour cells looked different from those normally found in AML patients. When researchers tested cells from his bone marrow on a microarray, they failed to find the signature of either AML or ALL. What they did see were high levels of activity in genes relating to muscle cells. More tests confirmed that the patient didn’t have leukaemia at all, but a rare muscle cancer that would need a completely different treatment.
The next startling discovery came in 2000, when another group discovered that a form of cancer called diffuse large B-cell lymphoma (DLBCL) was in fact two very different diseases. DLBCL affects the lymphatic system and blood, so Louis Staudt and his team at the National Cancer Institute in Bethesda, Maryland, looked at the activity of some 18,000 genes associated with normal and abnormal white blood cells. They examined DLBCL tumours and found two distinct patterns of gene activity.
Their findings not only have dramatic implications for our understanding of DLBCL, they also help explain a long-standing mystery. For years doctors treated this disease with a combination of drugs that worked extremely well, but only for 4 out of 10 patients. They wondered if drug resistance could explain the large failure rate, but no one really knew for sure. Nor did they have any way to predict which patients would respond to the drugs, or which ones would need a more aggressive approach. Now doctors know they are dealing with two different diseases, they hope to tailor the treatments accordingly. “The degree of difference took us by surprise,” says Staudt. “This category of large B-cell lymphoma was actually harbouring two diseases that are as unalike as two different types of leukaemia. It was completely unexpected.”
And the surprises are not restricted to leukaemias and lymphomas. Many solid tumours, such as breast or prostate cancer, seem to come in many different types. One recent study uncovered four distinct genetic patterns in breast cancer samples from different patients. Another identified a similar number of different signatures in tumour cells from lung cancer patients. And a team led by Charles Perou, a cancer biologist at the University of North Carolina in Chapel Hill, have discovered that breast cancer can arise from several different kinds of breast cell types – not one as previously suspected. “They are multiple cell types giving rise to similar-looking tumours that are actually biologically very different,” says Perou. “We’re finding diseases within diseases.”
“For the first time we’re beginning to tell the difference between tumours that microscopically look the same,” says Garret Hampton, a cancer biologist at the Novartis Research Foundation in San Diego. This is the key that will one day allow us to tell whether a patient with a particular sort of tumour has a good prognosis or a poor one, and what sort of chemotherapy is appropriate, he says.
Another big question being tackled by the gene eavesdroppers is what happens to patterns of gene expression when you treat cancer with chemotherapy. This knowledge will help scientists design more effective drugs and help them understand how some tumours become resistant to drugs. One recent study found that cancer cells treated with drugs in culture activate different genes than they do inside the body of a person on chemotherapy. Another suggests that gene expression within tumours undergoes much less change after chemotherapy than previously suspected. “That’s a surprise but it’s actually a good thing,” says Pollack. “If a tumour maintains its identity over time it makes it an easier target for specific therapies. It makes it easier to diagnose it as a distinct identity.”
Still, any real forward leaps in cancer care will probably have to wait until researchers have a more complete understanding of basic tumour cell biology. And gene signatures have revealed some big surprises here too. They are toppling assumptions about one of the most fundamentally important questions in tumour biology – why some cancers spread.
This question is particularly important because many patients do not die from the original or primary tumour. Rather, it is the cells that somehow break off and spread to other parts of the body, a process known as metastasis.
Until recently, evidence from tests done on lab animals suggested that metastasis is a process involving a challenging series of steps that cells can perform only after undergoing a specific and complicated sequence of genetic changes. Such cells are extremely rare – they’re estimated to represent only 1 out of every 10 million primary tumour cells. Indeed biologists have come to view them as the decathletes of cancer – rare specimens that can escape the primary tumour, survive in circulation, out-fox the immune system, invade a new site, trigger the growth of new blood vessels and so on. This would explain why only 0.01 per cent of the cells that tumours constantly shed form metastases. It also suggests that the bigger tumours grow, the more likely they are to metastasise.
But results from microarray studies published in January challenge this wisdom. Working with Golub, Harvard Medical School’s Sridhar Ramaswamy originally wanted to identify the genetic differences between normal primary tumour cells and metastatic cells. Surprisingly, he found that the gene signature of metastatic tumours was also present in some primary tumours. This raises the possibility that cells destined to metastasise already bear their distinctive signature at the stage when they are first detected. To test this, Ramaswamy looked for the metastatic gene signature in a separate collection of early primary lung tumour samples. He found that some of the samples carried the metastatic signatures while others didn’t. When he analysed charts from the patients donating the samples, he found that the primary tumours carrying a metastatic signature were more likely to spread and kill.
Ramaswamy has proved his point by successfully predicting the likelihood of metastasis in different forms of cancer from different tissues, including breast and prostate. This suggests that the genetic mechanism driving metastasis is the same in all cancers, and not different as once thought. Together these results suggest that right from the start, cancers that are destined to metastasise are fundamentally different from those that never spread. It also indicates that the potential for cancers to spread may be less dependent on tumour size than previously thought.
Ramaswamy’s findings also tie in with results from last year’s study by Dutch and American scientists who used microarrays to identify genes that can act as a warning flag for metastasis when active in early breast cancer cells. Currently, all breast cancer patients undergo chemotherapy after surgery because doctors can’t identify the 20 to 30 per cent who might be harbouring microscopic clumps of cancer cells that have already spread. Being able to predict which tumours are destined to metastasise would spare many patients the toxic side effects of drugs they didn’t need. With that in mind, the Dutch-American team has begun a clinical trial to see if the metastatic signature can be used to single out those patients in need of more aggressive treatment.
Together these studies have created a buzz among cancer researchers. “It would mean that a lot of information inherent in the biology of the tumour is fixed, or is readable at the time of diagnosis,” comments Staudt. “That is an important, different way of looking at it.”
While the genetic signatures are revealing in themselves, researchers ultimately want to know what they mean. Hence the plan to map all the molecular circuitry of a cell. “We want to be able to take any gene in the genome,” says Golub, “and look it up on a map and say here are all the functional connections that are made under normal conditions or disease.” You could then predict what would happen if you turned a particular gene on or off, he adds.
While such a map would no doubt greatly improve understanding of tumour biology, it might also bring researchers closer to the main aim of cancer biology – better treatments. By knowing the molecular wiring of a tumour intimately, researchers think they’ll be able to develop treatments that strike at the heart of the cancer while sparing the rest of the body.
Still, there are many obstacles to overcome. The technology will have to be cheaper and more user-friendly. Cancer physicians will have to get to grips with tools that didn’t exist when they went to school. But the biggest obstacle may be time. No matter how fast the pace of discovery, testing any new diagnostic aids or treatments will take years of clinical trials.
But given the advances so far, researchers like Golub are more certain than ever that cancer gene signatures will have a huge impact – something they never suspected when they first stated their experiments. “We certainly didn’t have confidence that it would have any real fundamental impact on how we think about cancer,” admits Golub. “But that’s what’s happening now.” And the current crop of discoveries is just the beginning. “There are going to be many more surprises in the future,” says Pollack.