| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© Oxford University Press 2006.
NEWS |
Tumor Complexity Prompts Caution About Sequencing
In mid-September, the National Cancer Institute and the National Human Genome Research Institute announced a $100 million effort to sequence the genomes of lung, brain, and ovarian cancer, the three cancers with the best tissue repositories. It is just the first step in the larger endeavor called the Cancer Genome Atlas, a $1.5 billion attempt to find all the mutations in cancer.
The study was controversial from the outset, with some researchers questioning spending a billion dollars in an uncertain research area. Now new studies have shown that individual tumors are more unique than previously anticipated, which has fueled the controversy again. Supporters of the genome atlas say this unexpected diversity emphasizes the need for mapping cancer genomes to pick out what's important. Critics say that the genome atlas has methodological challenges and are concerned that a big sequencing effort is not the most efficient use of resources in this tight-budget climate.
"I like this kind of science. I've done a lot myself. The thing that people sometimes forget is there's a fundamental distinction between sequencing the human genome and sequencing many cancers," said geneticist Richard Kolodner, Ph.D. of the University of California–San Diego Ludwig Institute for Cancer Research.
|
The same month the pilot project was announced, Science published a paper looking at genomic changes in breast and colon cancers. The researchers found 189 genes that contribute in various degrees to cancer, less than a quarter of which had already been implicated in the disease.
"When you look at every nook and cranny, you find all sorts of new genes," says study coauthor Victor Velculescu, M.D., Ph.D. of Johns Hopkins University Kimmel Cancer Center in Baltimore.
|
The level of complexity surprised the researchers. All 189 genes were not mutated in every sample—each tumor contained a subset of 11 mutated genes on average. And the subsets of mutated genes did not overlap as much as expected among tumors. For example, any two tumors of the same cancer type shared on average only six mutated genes of those 11 mutated genes per tumor. Overall, breast and colon cancer shared only two genes in common—p53, one of the most famous of all cancer genes, and obscurin, a gene of unknown function, as the name implies.
To come up with the 189 genes, first the researchers sequenced two-thirds of the DNA from 22 individual cancers (half breast and half colon). The team found 800,000-plus mutations in 13,000 genes. They pared that down to about 1,300 mutations in 1,149 genes by eliminating things such as silent mutations that do not cause a change in the protein. They then sequenced those 1,149 genes in another 24 tumor samples, found more DNA changes, and sorted again. This approach resulted in an additional 365 mutations.
Armed with the 1,149 genes and 1,600-plus mutations scattered among them, the team narrowed down how many of the genes may contribute to cancer specifically. They scored each gene on the basis of characteristics such as what percentage of tumors harbored the mutant gene (for example, one gene might show up mutant in 80% of tumors, whereas another gene might be mutated in only 3% of tumors) and how many different mutations each gene sported. Generally, cancer genes were considered relevant if they were mutated in 5% or 10% of the tumors. Each tumor harbored on average 100 gene mutations, only some of which could be classified as cancer genes. The number of mutated genes suggests that the number of steps it takes for cells to become cancerous is a lot more than researchers originally thought, Velculescu says. The researchers are continuing to analyze the data to determine whether the mutations are linked within particular biochemical pathways.
Not Ready for Prime Time?
But not everyone thinks the complexity found by Velculescu's group bodes well for the cancer genome project. "From my perspective, it's hopeless," says cancer biologist Lawrence Loeb, M.D., Ph.D., of the University of Washington School of Medicine in Seattle. "It's a wonderful study, but there's no core of genes associated with a particular type of cancer. It started with a wonderful optimistic view, but now it presents enormous complexity." This broad sequencing of tumors overlooks random mutations that arise as tumors continue to develop past their clonal origins. "The random mutations present in only some cells may or may not drive malignancy and are probably responsible for drug resistance," Loeb says.
Also, some worry that the project is ahead of the game. The details of the paper itself hint that the technology is "not ready for prime time," said Kolodner. To eliminate false positives from the vast number of mutations, the research team visually inspected hundreds of thousands of sequence traces. This step removed more than 350,000 changes. "It is stultifying to look at those graphs," Kolodner says. "It suggests they could not reliably automate it. And if you're going to expand the scale of [a study this size], can you expand the scale of that?"
Francis Collins, M.D., Ph.D., remains optimistic that the project will allow "very interesting themes to emerge" across cancers that will lead to individualized medicine. He says developing sequencing technologies should drop costs significantly and that the 3-year length of the study will allow researchers to take advantage of technological advances. "There's nothing like a really important biomedical problem to drive technology," says Collins, director of the National Human Genome Research Institute.
|
"As far as therapy goes, it's not as dire as it sounds," says Velculescu. He pictures a future where a cancer patient comes into a clinic and has her tumor analyzed. "Then she is treated based on a spectrum of her mutations with a cocktail of drugs. It doesn't mean a new drug for each person, just a different combination of drugs." Current therapies merely do damage to rapidly dividing cells, says Collins, but customizing treatment based on individualized cancers "will go after what's going on in the tumors."
Noisy Silence
Determining how many of the mutated genes are cancer genes doesn't provide information on which particular ones are important, said Bernard Strauss, Ph.D. emeritus professor of genetics at the University of Chicago. By eliminating silent mutations, the authors of the Science paper lost important information, he says. Strauss points out that of the more than 21,000 mutations that have been found in cancer for the protein p53, only 4% are silent. Rather than eliminating the silent mutations, Strauss would like to see the percentage of silent mutations mapped for each gene. Those genes containing few silent mutations could very well be the important ones.
|
The number of mutations the authors found relevant to cancer represents 0.2% of the mutations they started with, Strauss adds. "It seems like a very inefficient way of finding out what's going on in tumors," Strauss says. "They are sifting through the haystack for the needle. Wouldn't it be better to find some sort of magnet to pass through the haystack?"
Some researchers say a good alternative to mass sequencing is the effort by molecular geneticists P. Andrew Futreal, Ph.D., and Michael Stratton, Ph.D., of the Sanger Institute in Hinxton, United Kingdom. They are looking at a well-defined set of genes, which several researchers said makes more sense than global sequencing. They sequenced 518 protein kinase genes in 200 cancers. Interestingly, their results were similar to those of the Velculescu group—cancers exhibit a complex set of mutations in a wide range of kinases rather than a unified front. "We didn't get a giant spike [in one or a few of the genes]. It was sort of a rumble," Futreal says. "All of us have been somewhat amazed at the level of complexity we're dealing with."
The Money Issue
Calling the Cancer Genome Atlas "a dream come true," Collins says objections to the huge sequencing effort have for him "a little flavor of déjà vu" hearkening back to the late 1980s when the Human Genome Project first got off the ground. "They all ask is it going to end up costing more than planned and will it take away from more meritorious research?" he says. He points out that the Human Genome Project came in 2.5 years early and ended up costing less than expected. "Ahead of schedule and under budget is our method," he says.
"What worries me is not that it isn't interesting but whether it's ahead of itself," says Kolodner, who is skeptical that the technology is up to the "astronomical technical problem" presented by the effort. "My bet is there's lots of really interesting smaller-scale projects that could be done by giving people access to, say, a hundred sequencers."
Researchers such as Stephen Elledge, Ph.D., of Harvard Medical School in Boston and others critical of putting so much money into one basket hope that Collins is right about the future. "Since we are going down this sequencing road, I hope they can make the sequencing much less expensive so as to free up valuable National Cancer Institute grant resources," Elledge says.
Although cancer biologists are glad to see new money coming from the National Human Genome Research Institute, they point out that NIH's budget does not resemble the heyday of the Human Genome Project. "Funding is down to the 10%–12% level, so that means 90% of grants are not being funded," Loeb says. Kolodner agrees, saying, "There is a severe slow bleed on the funding front for most labs."
And although established scientists might be able to keep their labs limping along, the younger set might be adversely affected by diminishing resources. "NCI funds come out of R01s in a very tight time where the next generation of cancer researchers is trying to gain a foothold in a very difficult funding environment," Elledge says. "They are the future."
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||