The work suggests that the genetic causes of cancer are even more complex than previously thought and highlights the difficulty of developing new drug treatments. However, it also points toward new avenues of research for treatments and diagnostic tests.
Scientists at the Johns Hopkins Kimmel Cancer Center examined DNA from 22 patients' brain tumors and 24 patients' pancreatic tumors.
The type of brain tumor studied, called glioblastoma multiforme, affects some 20,000 Americans each year -- including, most famously, Massachusetts Sen. Ted Kennedy -- and the median survival time is between 7 and 17 months. Some 38,000 Americans are diagnosed each year with pancreatic cancer, with a survival rate of only about 5 percent.
In two studies published this week in the journal Science, the researchers sequenced the DNA of more than 20,000 genes in each of the tumor cells. They found that the typical glioblastoma contained about 60 mutations, and the average pancreatic tumor contained 63. However, no two patients had the exact same set of mutations.
"If you have 100 patients, you have 100 different diseases," said Johns Hopkins researcher Bert Vogelstein, one of the authors of the papers.
That finding suggests that it will be difficult to develop effective drugs for solid tumors that work like Novartis' celebrated leukemia treatment, Gleevec, which targets only one mutation.
"Our study indicates that it's going to be even more difficult than expected to derive cures from such therapies," Vogelstein said.
Instead, he suggested, more short-term research should focus on using the new knowledge of cancer genetics to develop genetic screening tests that could catch cancers early, before they metastasize and while they can still be cured with simple surgery.
The studies did, however, also suggest a new avenue for drug treatment research. The researchers found hundreds of mutations, but most of those mutations clustered into about a dozen "pathways" -- cellular processes, controlled by genes, that regulate functions such as rates of cell division and the death of abnormal cells.
"Cancer is a disease of pathways, where many different individual genes can be broken at the genetic level, but where these genes when considered together affect a much smaller underlying number of mechanisms," Vogelstein said.
Researchers, then, may be able to concentrate their drug research efforts on pathways rather than individual genes.
"Right now treatments are really only designed against specific molecules that are members of pathways," said Webster Cavanee, director of the Ludwig Institute for Cancer Research at the University of California, San Diego, who was not involved in the research. "But there might be ways you could modulate a pathway instead."
Another study published this week, this one in the journal Nature, also examined the genetics of brain cancer. The research, conducted by a large consortium of scientists under the NIH-funded Cancer Genome Atlas Research Network, examined many more tumors -- from more than 200 patients -- but a smaller number of genes, only ones previously suspected to be associated with brain cancer.
That approach doesn't net scientists any information about genes not previously associated with the cancer. For example, the Johns Hopkins group discovered that the gene IDH1, not previously associated with cancer at all, was mutated in 12 percent of glioblastomas and particularly in younger patients, while the Genome Atlas group didn't find that mutation.
However, the Genome Atlas approach gives scientists more information about the genetic diversity of cancer between patients, said researcher Lynda Chin of Harvard Medical School, and is more likely to find mutations that affect only a very small percentage of patients. For example, the group discovered a mutation that may be the cause of about 7 percent of brain cancer patients' resistance to the drug Temador, one of the few treatments available for glioblastomas.
The two research groups' work is complementary, Chin said, but both are limited by technology -- right now, it is too difficult and expensive to sequence a large number of genes from a large number of patients. However, it may soon be possible to do both.
"We're already piloting the technology that will have the bandwidth to sequence every gene in 550 tumors," Chin said. "That will be the next phase of the research."