Experts seek genetic key to cancer's door
The Oregonian, Portland, Ore.
Jun. 23--Top U.S. cancer scientists want the government to start a massive project to map the genetic code of cancer, which could revolutionize treatment of the dreaded disease.
Knowing the DNA mutations behind common cancers would let doctors individualize care in unprecedented ways, advocates say. Patients could get medicines tailored to their genetic profile, and medicines would be targeted to harm only cancerous cells and avoid healthy ones, limiting side effects. A few existing cancer drugs have blazed this trial, including Gleevec, the leukemia medicine tested at Oregon Health & Science University.
"In my opinion, and in the opinion of many others right now, what we really need is a cancer genome project," said Dr. Brian Druker, the OHSU Cancer Institute expert who pioneered Gleevec.
National Institutes of Health leaders are debating the venture, which would be as ambitious as the Human Genome Project, and perhaps as costly. Labs would need many years and perhaps billions of dollars to decode the DNA of roughly 30,000 genes in each of many hundred tumors. Set mutations in certain of those genes can spur the out-of-control cell growth of cancer. While scientists have found several such mutations, they suspect many more.
"A cancer genome project would be an effort to find all those cancer-causing changes," said Dr. Matthew Meyerson, a Harvard researcher who supports the effort.
In April, Meyerson and other cancer experts met with NIH officials to discuss a cancer genome project. He said the federal officials "definitely think the idea is good. They're thinking about how they exactly want to do it." Geoff Spencer, a spokesman for the National Human Genome Research Institute, said the idea is being discussed "in its early conceptual stage."
Druker, who serves on a committee that advises the National Cancer Institute on technology projects, said mapping cancer's DNA "is going to be our No. 1 recommendation." Matching patient, drug
For decades, doctors have diagnosed tumors based on where they grow, such as the lining of the lungs or colon. Cancers arising from similar tissues were treated with similar medicine -- even though they might be fueled by different genetic mutations. That's like assuming all broken-down Oldsmobiles have electrical problems, and making the same repairs to every one.
"What we're going to get to in the coming decade is the ability to match the right patient with the right drug," Druker said. "And we won't talk about these any more as breast, colon, lung, prostate. We'll talk about them by their genetic defects . . . understand what's broken and develop strategies to treat it."
Gleevec showed that model works. Genes help cells make proteins, molecules that do work in the body. When specific genes mutate in set ways, they can make rogue proteins that let cells grow too much, like a brick on a gas pedal.
One such mutation drives 95 percent of cases of an uncommon blood cancer, called chronic myelogenous leukemia. Druker's tests of Gleevec showed that the drug focused on that mutation, killing cancer cells and sparing healthy ones. That was an advance over older cancer drugs that hurt both healthy and sick cells.
Some doctors feared that approach would fail with more complex solid tumors, or those caused by multiple, sometimes unknown mutations. But Gleevec's use against a stomach cancer caused by a different mutant protein quieted some fears.
A new lung cancer drug called Iressa should answer remaining skeptics, Druker said. That drug helped few patients in early trials, but a small number of users improved dramatically. In the past two months, Meyerson and other researchers found a specific gene that is mutated in patients who responded well.
Such success has led a few researchers to start their own, limited cancer genome projects.
Meyerson is part of a "Kinase Project" at Dana-Farber Cancer Institute and the Broad Institute in Boston. Kinases are kinds of proteins -- including the one targeted by Gleevec -- that help cells live and grow.
Since 2001, Meyerson and colleagues have scanned samples of lung, blood, prostate and brain tumors to find mutations in the genes that make a subset of kinase proteins called tyrosine kinases. There are about 90 such genes. And the group must decode each of those in tumor samples from about 100 different patients to uncover mutations that happen at least 2 percent of the time.
Eventually, the team wants to scan all the genes that make kinase proteins, roughly 518 genes, Meyerson said. That expansion would take significant new funds, he said.
Johns Hopkins University's Dr. Victor Velculescu is on a similar hunt through colorectal tumors. Last year, he and colleagues screened the tyrosine kinase genes in 182 tumors and found seven gene mutations that might influence colon cancer. They moved on to other kinds of kinase genes and recently found one that is mutated in nearly a third of the colon cancers, and a few other tumors.
"This was a shock, because it would make it one of the most highly mutated genes in cancer," Velculescu said. He added that existing drugs might interfere with the mutation, which "looks like an excellent therapeutic target."
A group linked to England's Sanger Institute is making a similar effort, focusing on gene families such as kinases in cancers of the breast, lung, colon, stomach and testicle, said Dr. Andy Futreal, one of the researchers.
Scientists behind those efforts are the first to say that a large-scale cancer genome project faces major financial and technical obstacles. It would mean screening not just hundreds of genes but all 30,000 human genes, since no one knows which might cause cancer. That is complex and pricey work.
"DNA sequencing is still too expensive to support a vigorous attack on the cancer-mutation problem," said Maynard Olson, a University of Washington genetics expert who attended the April NIH meeting. "Costs need to be brought down by at least a factor of 10, preferably more" for the project to realistically succeed.
Faster technology would help, too. The Human Genome Project invented efficient new ways to scan DNA, which scientists are still improving. Velculescu said recent software breakthroughs cut the predicted time to sort his tyrosine kinase data from 9 years to "a couple months." Still, it would take five to 10 years to screen for mutations in all 30,000 genes in colon cancer alone, he predicted.
Olson said it could also be tough to find enough large, well-preserved tumor samples. Scientists would need samples from 200 volunteers for each type of cancer under study. They also must screen healthy tissue from each person, since mutations emerge only when tumor DNA is compared with the same person's healthy genes.
Finding mutations is only the start of the work. Researchers would have to prove which mutations were really influencing the cancer and not just random events. Then comes the work of creating tests for the key mutations and hunting down drugs to treat them.
Iressa is a good example. Dr. Helen Ross, a lung cancer expert at Providence Portland Medical Center, said patients are asking about the drug. She explains that more work is needed to prove which mutations Iressa affects and to figure out how to screen patients for those DNA changes.
"A lot more testing needs to be done," she said. "And you can't just go to your doctor and get this screening done tomorrow."
Such challenges mean small, existing cancer genome efforts won't succeed alone, Velculescu said.
"Given the effort and the cost, we certainly couldn't do this without some federal funding," Velculescu said. But "in terms of money well spent, or money invested, I think this is the type of project that has a high chance of payoff."
Olson said it is "an open question as to whether or not this proposal can compete successfully" for NIH funds now. But he hopes it will, for patients and researchers alike.
"It is a bold idea at a time of some retrenchment away from bold ideas," he said. "Science needs big goals."
To see more of The Oregonian, or to subscribe the newspaper, go to http://www.oregonian.com.