As of this Wednesday, roughly 350 genes had been implicated in the development of cancer. "These genes have been discovered over the course of 25 years, using many different strategies" cancer researcher Michael Stratton told reporters at a press conference.
On Thursday, in the March 8, 2007, issue of Nature, Stratton and colleagues at a dozen European, American and Australian research institutions added roughly a hundred genes to the cancer gene pool, through a new strategy that he termed "in principle very simple:" they systematically sequenced all 518 kinase genes of cancer cells and compared them to those of normal cells.
The researchers focused on kinase genes because despite the advances of genomic technologies over the past decades, sequencing the entire genome of a sufficient number of cells remains technically infeasible at this point - though Stratton said that new technologies might make such a project possible in the future.
But for now, "to tackle the genome, one has to bite off a chunk," Andrew Futreal said. Stratton and Futreal are co-leaders of the human cancer genome project at the Wellcome Trust Sanger Institute in Cambridge, UK, which had a major role in the research now published in Nature.
Kinase genes were an obvious choice because, as the authors wrote in their paper, "the protein kinase is the domain most commonly found among known cancer genes1 and because inhibitors of mutated protein kinases," such as Gleevec (imatinib mesylate, Novartis AG), "have recently shown remarkable efficacy in cancer treatment."
The researchers focused on kinases partly for their therapeutic potential, but the pattern of their findings may be somewhat sobering, at least for those hoping for blockbusters. The research argued against there being a whole lot of blockbusters around.
"We tend to see infrequently mutated genes distributed pretty broadly across all the tumor types we looked at," Futreal said, describing the overall pattern of the data. Though some kinases were mutated in up to 6 percent of the samples, such a high number was "more the exception than the rule."
Even though each individual mutation existed in relatively few samples, their cumulative number was nevertheless impressive. In roughly 200 tissue samples from major cancer types, the scientists found no fewer than a thousand mutations. Of these, roughly 100 - which the researchers termed "driver" mutations - seemed to have a causative role in cancer, while the other 900 "passenger" mutations seemed to result from either defective DNA repair mechanisms the cancer or treatments.
The researchers used several methods to distinguish driver from passenger mutations. "Driver mutations are likely to change the sequence of the proteins they encode," Stratton said, so those mutations that changed the amino acid sequence of a protein were more likely to be drivers.
The fact that the kinases are extremely well studied protein family further helped sort drivers from passengers. "We know an awful lot about the bits of the protein that are functionally involved in modulating and generating the signals. So we can find those mutations, which are present in those critical domains of the kinase, and using that strategy, we can identify driver mutations," Stratton explained.
Even though driver mutations appear to account for only about 10 percent of the mutations the scientists found, even that number was surprising to their discoverers. Futreal said that based on previous data, his expectation of the number of new point mutations they would find was closer to 10 then to 100.
Overall, the results suggested that there may be many more cancer genes than previously suspected - a finding that both validates the approach of the cancer genome atlas and suggests that it should look for strength in numbers. "More is definitely going to be better here - we need to expand the number of tumors, the number of gene sequences, the number of types of tumors," Futreal said.
Stratton added that "on the basis of these data . . . We really ought to be aiming for the complete sequences of the genomes of maybe 10 to 20,000 individual human cancers, covering all the common and many of the less common types of cancer."
The sheer volume of their data allowed the researchers to go beyond looking at the roles of individual kinases and look at whole kinase networks. "The way these things function is that one kinase turns on another one, which turns on another one, which turns on another one and so forth . . ." and to see whether there were any mutations that popped up frequently. One pathway they identified, which "wasn't particularly well implicated before," was the JNK/Map kinase pathway, which appears to have an intriguing connection to colorectal cancer, Futreal said. That is a connection "that we probably would not have found, had we not gone and sequenced all the kinases through a large number of tumors."