The largest study yet of genetic changes across a broad range of cancers has turned up some unexpected results. Mutations thought to cause one kind of cancer also seem to be important in others, suggesting that the same drugs might be useful for patients with very different kinds of cancer. But all the mutations in the study–even those shared among cancers–occurred at low frequencies, suggesting that the search for cancer genes may be more difficult than researchers had hoped.

Researchers led by Levi Garraway, assistant professor of biological chemistry at Harvard Medical School, surveyed 1,000 samples from 17 different types of tumors. They searched for mutations known to be important to cancer, whether by allowing cells to divide unchecked or by helping tumors recruit blood supplies or spread throughout the body. Using a technique called genotyping, the researchers sought 238 mutations in 17 genes. Genotyping is a relatively inexpensive way to use what is known about one kind of cancer to learn about others, says Garraway. But previous genotyping studies have not been as comprehensive, focusing on one kind of cancer or one particular gene.

The most exciting results, says Stephen Chanock, head of the genomic variation section in the pediatric branch at the National Cancer Institute’s Center for Cancer Research, include the discovery that several mutations are not specific to the cancers in which they were discovered. Mutations in a gene called RAS, for example, have been implicated in lung cancer. But in their genotyping study, Garraway and his group looked for mutations in the gene in all 1,000 tumors–and found that 10 percent of mesotheliomas (a cancer involving the lining of the chest or abdomen) have RAS mutations.

“All of the mutations we studied are known to occur somewhere, not necessarily in the cancers we studied,” says Garraway. “The fact that these mutations exist at all in other cancers could change how we think about treatment.” Garraway’s group deliberately studied the best-characterized cancer genes and emphasized those for which there are already targeted drugs. The results suggest that drugs designed and approved for particular cancers might help other cancer patients as well.

The new results could also change how researchers think about cancer. “Are there similarities between breast cancer and melanoma?” asks Chanock. “There may be more than we have considered [possible] in the past. We may need to look at cancer differently.”

But Chanock says the study also provides “a good reality check.” Thirty percent of the tumors in the study had none of the mutations Garraway’s group looked for. Chanock says this study forces researchers to address the question, “What’s it going to take to make sense of [the genetic changes behind cancer]?” The answer, he says, is even larger genotyping studies, on the order of 10,000 tumors. And, as those tumors with no known mutations in this study suggest, many of the genetic changes that drive cancer remain to be discovered.

The genotyping study underscores the value of a project at the National Institutes of Health called the Cancer Genome Atlas (see “Genomic War on Cancer”). Project scientists are currently working to sequence ovarian and lung cancers and glioma, a cancer of the brain. Todd Golub, head of the cancer program at the Broad Institute, in Cambridge, MA, one of the centers charged with the sequencing, says the Harvard work is encouraging because it implies that mutations found in the Cancer Genome Atlas will likely be important in other tumor types.

The rarity of particular cancer-causing mutations also means that it will be critical to determine what mutations are present in individual patients’ tumors to guide their treatment, says Golub. “There is no conceptual reason why genotyping couldn’t be done on patients,” says Garraway.