In preparation for minor surgery, John Leventhal needed a routine chest x-ray.When the New Haven, CT, doctor joined the radiologist who was examining the film, he was shocked by what he saw: an opaque blotch deep in his lung. “As a physician,” says Leventhal, “you’re taught in medical school that when you see a mass like that, it means lung cancer.” Leventhal’s medical training also taught him that to confirm the diagnosis, his doctors would need to crack open his rib cage to get a piece of the suspect tissue that would be closely examined by a pathologist-an extremely painful and hazardous operation. The weekend before that surgery, Leventhal went off on a family ski vacation. He recalls thinking, “This is the last time I will go skiing for a long, long time.”
That was five years ago. Today the medical profession’s way of dealing with cancer could be about to change. Around the same time that Leventhal underwent
surgery, researchers at Stanford University and Santa Clara, CA-based startup Affymetrix were beginning to build the first “DNA microarrays.”More commonly known as DNA chips, these are DNAcovered silicon, glass or plastic wafers capable of analyzing thousands of genes at a time to, for example, identify the ones that are active in a sample of cells. Now these microarrays appear poised to join the war on cancer. DNA chips, predicts National Cancer Institute director Richard Klausner, are “going to have a huge effect” on the diagnosis and treatment of the disease.
One reason for the excitement is that DNA chips offer a whole new-and potentially much earlier, easier and more precise-way of detecting cancerous cells. Most forms of cancers go unnoticed until lumps, coughs or pains develop, at which point it is often too late. And even then, once a pathologist gets a biopsy from a tumor, distinguishing one form of cancer from another can be difficult or even impossible with existing techniques, which involve noting distortions in the cells’ architecture under a microscope. Better diagnostic information could be used to make better treatment decisions, perhaps making the difference between life and death.
ithin the next two years, pathologists expect to begin using DNA-chip-based tools to spot genetic differences among cells; these telltale differences could be used to help detect cancerous cells long before symptoms develop and to distinguish one type of cancer from another. In short, the chips will provide a genetic profile of a cancerous cell that can be read like a criminal’s rap sheet. The physician will know where the cancerous cell originated, how far it has progressed, and which therapies will work best to halt its further growth and spread.
Leventhal was lucky. His lung biopsy was negative, and he was back on the slopes the next winter. But it took him a month to recover from the biopsy surgery, and today he has an angry scar down the middle of his chest to remind him of the ordeal. By the end of the decade, it is likely that a patient like Leventhal will be able to skip invasive diagnostic procedures altogether. A DNA-chip-based device might be able to read a sputum sample right in the doctor’s office, checking for the genetic changes in the lung cells that are naturally sloughed off into the viscous fluid. If the news is bad, the patient might well have a host of new treatment options. That’s because DNA chips are also speeding the discovery of new and better cancer drugs. “We’re on the threshold of a new era,” says Klausner. “Technologies like DNA chips will tell us not only that something may be amiss, but what it is and what we can do about it.”
With one out of every two men and one out of every three women in the United States likely to get cancer at some point in their lives-and about 560,000 Americans expected to die of the disease this year alone, according to the American Cancer Society- advances can’t come fast enough. As many as 500 research laboratories in academia and industry are already employing DNA chips to develop sweeping new genetic pictures of different cancers. In 1999, the National Cancer Institute alone provided $4.1 million to 24 U.S. academic cancer institutions to set up or upgrade microarray centers.Meanwhile, the pharmaceutical and biotech industries are drawing on information gleaned from DNA chips to develop new and better diagnostic tests and more effective anticancer drugs with fewer side effects. Indeed, all the major drug companies and at least a dozen biotech firms are already using DNA chips to tackle cancer.
At the same time, large manufacturing companies such as Agilent Technologies, Corning and Motorola are seeing the potential of DNA chips. All three have allied with academic research centers to come up with DNA chips that will analyze genes related to specific cancers. And while at the moment DNA chips are far too expensive to compete with existing diagnostic technologies, the involvement of these manufacturers and their production facilities could drop prices as low as $10 for a chip, once large-volume production gears up.
Of course, for DNA chips to help win the war on cancer, it will take considerable effort-and years of further development. For one thing, DNA chips generate tons of data, and researchers will need to beef up their computing capabilities and nail down data standards in order to make sense of it all (see “Gene Babel,” TR April 2001). And any new drugs or diagnostic devices will have to prove themselves in clinical trials. But the initial fruits of the efforts to apply DNA chips to cancer-new diagnostic tools-could begin saving lives as early as the end of next year. The first anticancer drugs developed using DNA chips will enter human trials around the same time, with dozens more to follow. With all those new tools available, currently untreatable forms of cancer may, one day, no longer mean death sentences.
Profiles in Cancer
The first step toward that grand vision is generating a profile of the genes that are activated or shut down when a normal cell becomes cancerous.While most genes are quiet in any given cell at any given time, the ones that are active, or “expressed,” tell a lot about that cell’s health.And though many of us tend to think of diseases as being caused by particular genes-say the gene for Huntington’s disease or cystic fibrosis-most diseases actually involve complicated interactions among a large set of different genes. However, just as a person’s fingerprints can be distinguished from virtually all others by just a small number of differences, a sort of genetic fingerprint, perhaps involving a hundred active genes or even fewer, could distinguish cells showing even the very earliest signs of cancer.
The beauty of using a technology like DNA chips to find those fingerprints, says Klausner, is that “we’re not limited by preconceived knowledge or notions.” In other words, cancer investigators no longer have to bias their experiments by looking individually at the genes they suspect might be involved with a particular cancer. “Instead of focusing on one gene,” explains National Cancer Institute researcher Louis Staudt,”with microarrays we get to look at the entire genome and let the cancer cell tell us what the important genes are.”
The flagship in the National Cancer Institute’s efforts to demonstrate the validity of the DNA-chip approach is the so-called Lymphoma/Leukemia Molecular Profiling Project, which is directed by Staudt. The study is looking at diffuse large B-cell lymphoma, a relatively common cancer of the white blood cells that affects more than 15,000 people in the United States each year. When oncologists give those patients standard chemotherapy treatments, about 40 percent respond rapidly. Their cancer melts away, and the majority are still alive five years after diagnosis. But of that other 60 percent, most are not so lucky. The cancer may go into remission briefly, but when it returns, it comes back with a vengeance. A few patients benefit at that point from radiation treatments and bone marrow transplants, but for most it is already too late to halt the spread of the disease. Clearly there is something different about the two groups, but under a pathologist’s microscope their cancer cells look identical.
The surprising answer is that these patients respond differently to treatment because, in fact, they are suffering from completely different types of lymphoma. Using what they dubbed a “Lymphochip,” a customized Affymetrix DNA chip, Staudt and a group at Stanford led by geneticist David Botstein discovered distinctive genetic differences between the cancers in the patients with large B-cell lymphoma who died and those who survived. “I was blown away by what we found,” says Botstein. Effectively, they were looking at two different illnesses. “It’s remarkable,” says Staudt. “We found something in this disease that was missed for all the years pathologists were looking at it.”
Similar projects are now under way to profile various forms of cancer, from different types of melanoma to colon cancer. Most other cancers present pictures similar to that of lymphoma: some patients get better and some do not, but predicting who will respond to therapies is impossible. If there were some way to identify the patients who won’t respond to standard chemotherapy, doctors could turn immediately to alternative treatments-and save lives. Indeed, says Pat Brown, a Stanford University School of Medicine geneticist who helped invent one of the two main types of DNA chips, “The same story is coming out for a bunch of cancers we look at-cancers with different clinical outcomes have different molecular subtypes.” And knowing the precise subtype of cancer afflicting a patient could help doctors pick the right treatments, right from the start.
Once researchers know the fingerprints of different cancers, they’ll be able to craft customized DNA chips that doctors can use to diagnose patients with previously unheard-of accuracy. Says Staudt, “The textbooks on cancer diagnostics are going to be rewritten over the next three to four years.[DNA-chipbased diagnostics] will very soon become routine technology.”
But the ability to read subtle genetic changes could allow doctors to do more than pinpoint the exact identity of a cancer; it could also help them read early warning signs that normal cells are about to turn cancerous-long before such changes are evident to a pathologist. That’s what University of South Florida cancer geneticist Melvyn Tockman is hoping, anyway.He and his colleagues are working on an early-detection method for lung cancer-a method that could make John Leventhal’s scar a relic of the medical dark ages.
The researchers take sputum samples from ex-smokers and use DNA chips to analyze which genes are active in the lung cells. By comparing the genetic profile of these damaged cells to profiles from both healthy and cancerous lung cells, Tockman hopes to find the fingerprint that indicates a cancer is just about to form. In the future a patient at risk for lung cancer might take a simple DNA-chip-based test for this genetic fingerprint each time he went for his regular checkup.
That’s a few years in the future, but the initial payoff of DNA chips in detecting cancer may come even sooner. Researchers are already using the chips to identify telltale proteins that can be detected by conventional cancer-screening tools. “If a cancer has one hundred uniquely expressed genes,” explains Mohan Iyer, the vice president of business development at Santa Clara, CAbased diaDexus, “the home run hit is to find one [of the proteins those genes code for] that can be used in a simple blood test to screen individuals for cancer.” If a protein were found to be unique to a certain cancer, says Iyer, standard hospital equipment could easily detect it in a blood sample.
With DNA-chip tools now helping to identify the proteins associated with breast, lung, colon and ovarian cancer, to name a few, Incyte Genomics, Corning and a handful of other companies are developing new protein-based screening methods for diagnosis of the diseases. These new tests should begin to reach diagnostic laboratories in the next two years or so.
But better diagnostics will begin to make a real difference only when they’re coupled with more effective treatments, treatments that are fine-tuned to combat particular types of cancer. Even “if you can distinguish 50 different lymphomas,” says Yale University School of Medicine pathologist Michael Kashgarian, “what does it matter whether it’s type A or type Z if the therapy is the same?”
In this area of cancer drug discovery,DNA chips are also playing a key role. Just as the rapid analysis of a large number of genes is helping to profile cancer for better diagnostics, it is also providing valuable clues to how to attack cancer cells.
Researchers have long believed that developing new therapies would begin with finding cancer-associated genes, but the past two decades have been filled with disappointment. Stephen Friend, once an oncology researcher at the Whitehead Institute for Biomedical Research in Cambridge, MA, and now chief executive officer of Rosetta Inpharmatics in Seattle, blames what he calls the “my-favorite-gene approach.” Biomedical researchers would spend years tracking down one gene associated with a particular cancer, then proceed on the assumption that that gene, or the protein it coded for, would make a great target for new drugs. But, Friend says, “the chances were in 999 out of 1,000 cases that you’d be wrong.” Very few genes work alone and in such simple and direct relationships with the body to cause disease. “God, were we stupid!” he says.
Friend is now convinced that technologies such as DNA chips that allow researchers to find all the genes involved in a disease are the way to go. (Rosetta plays a role in such research by selling software and other services for reading microarrays.) Not only can DNA chips help identify all the potential drug targets for a given type of tumor, they can also help rule out the genes that are active in healthy tissues. That way, drugmakers can develop precisely targeted treatments that kill cancer cells without damaging other parts of the body.”Drugs,” says Friend, “are going to be developed at a tenth the cost and in a third of the time by improving targeting and making sure compounds don’t pick up unwanted side effects.”
Eos Biotechnology, a South San Francisco company that is developing new cancer therapies using DNA chips from its partner Affymetrix, is betting he’s right. In the company’s labs, vice president of genomics research David Mack holds up one of those chips, which contains virtually the entire set of human genes. “The ability to generate the human genome on a chip today is incredible,” he says. Eos uses the chips as a platform on which to compare genetic activity in normal human cells and, say, a breast-cancer cell. Computers can then sort out the genes that are active only in the diseased cell.Moreover, they can select just those genes that present the best targets for drugs.
Under the traditional drug-development paradigm, once researchers identify a set of potential targets, they begin to stumble ahead into animal and human trials, with educated guesses about which potential drugs might be effective against a given target, and which of those drug candidates might have toxic side effects.Very often, it’s only much later in the process that a candidate’s problems become apparent-at a huge cost in time and money.
By contrast, Eos continues to use microarrays and other high-volume genomic techniques to test the drugs, better predicting which will be the most effective and the least toxic before more costly testing even begins.According to Mack,”We’re seeing data-driven science now, which hasn’t been the previous paradigm.”Thanks in part to the use of DNA chips, the company plans in the coming year to begin clinical testing of its first drug-which attacks a tumor’s ability to generate its own lifesustaining blood supply-with more than a dozen other anticancer drugs expected to follow rapidly. “The promise of these technologies to impact patients is here-finally,” he says.
While DNA chips have only been around for five years or so, they have already helped to get a number of new drugs into pharmaceutical-company pipelines, and to identify many potential new drug targets and sources of earlier diagnosis.With these advances, it is likely that cancer therapy will become both more complex and more effective over the next decade. Eventually, each patient’s cancer fingerprint will be met with just the right drug “cocktail,”or combination of therapies. Doctors will have new tools to diagnose and treat cancers much earlier-when the chances of cure are far better-and to monitor a patient’s progress, ensuring that tumors don’t develop resistance to the treatment.
It may take more than a decade before such practices become the norm, but if and when they do, they will change everything for people like John Leventhal.His (mis)diagnosis of lung cancer came when he was the age at which his own father got news of the cancer that eventually killed him-a fact that ratcheted up Leventhal’s terror when he learned he might have cancer. But should his children ever find themselves in the same shoes, perhaps they won’t have nearly as much to fear.
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
The Biggest Questions: What is death?
New neuroscience is challenging our understanding of the dying process—bringing opportunities for the living.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
How to fix the internet
If we want online discourse to improve, we need to move beyond the big platforms.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.