The findings, published online on the Nature Genetics Web site, may help relieve a bottleneck between scientists' expanding knowledge of the genetic mutations associated with cancer and the still nascent ability of doctors to use that knowledge to benefit patients. The results constitute an important step toward the era of "personalized medicine," in which cancer therapy will be guided by the particular set of genetic mutations within each patient's tumor, the authors suggest.

"It's universally recognized that cancer is a disease of the genome, of mutations within genes responsible for cell growth and survival, and a great deal of effort has gone into finding those mutations, to the point where several hundred to a thousand are now known," said the study's senior author, Levi Garraway, MD, PhD, of Dana-Farber and the Broad Institute. "The challenge has been how to determine which of them are involved in each of the hundreds of kinds of cancer that occur in humans -- and to develop accurate, affordable methods of detecting key mutations in tumor samples. This study suggests that such a method is feasible on a large scale."

The authors took advantage of a scientific serendipity to devise a simple test to detect important cancer mutations. Mutations in oncogenes (genes linked to cancer) do not occur randomly; rather, they seem to arise most frequently in certain regions of the oncogenes. As a result, researchers didn't necessarily have to scan the entire length of each gene, but could focus instead on the sections most likely to harbor mutations.

They performed these screenings with a technology known as high-throughput genotyping, a fast, relatively inexpensive way of profiling gene mutations within cells. It involves extracting DNA from a tumor sample, copying this material thousands of times, depositing segments of it in tiny "wells" on a small plate, and mixing in reagents that reveal whether each segment carries a specific mutation. Automated equipment then reads the plates to determine which mutations are present in each sample.

In the study, the researchers scanned 1,000 human tumor samples for 238 known mutations in 17 specific oncogenes. (Those 17 were chosen because they are mostly "classic, well-known" contributors to cancer, Garraway stated.) They found at least one mutation in 298 of the samples, or 30 percent of the entire group, which was in keeping with the rates reported in scientific literature for the types of cancer examined.

"Mutations were identified in the percentages we expected," Garraway said, "which indicates this technique is on-target for the mutations we were interested in. Overall, the technique worked very well: we were able to obtain mutation profiles that were accurate, sensitive, and cost-effective." The cost of processing each sample was between $50 and $100, although the figure would probably be somewhat higher if the technology were used in cancer clinics to test for large numbers of oncogene mutations.

The scans produced some surprises as well. Mutations were found in several types of tumors where they had not been previously recognized. Researchers also discovered an unexpectedly large number of instances where the same set of mutations co-occurred within tumor cells, suggesting that oncogenes often work in partnership.

As promising as high-throughput genotyping is for cancer treatment and research, its capacity will need to be expanded so it can handle larger numbers of mutations. The next step, Garraway explained, would be to work with clinical investigators to explore whether use of the technology is feasible in a clinical setting and whether it actually improves doctors' ability to classify and treat individual tumors.

"We've shown the practical potential of this technique," observed Garraway, who is also an instructor in medicine at Harvard Medical School. "It is a step toward the day when cancer patients will routinely have their tumors scanned for specific mutations, and treatment will be based on the cancer's particular genetic profile."