Like a crossword-puzzle solver who uses the letters in some answers to figure out others, researchers at Dana-Farber Cancer Institute and an international group of collaborators have used data on genes involved in inherited forms of breast cancer to identify a gene linked to non-hereditary cases of the disease.
Their approach, described in a study posted online by the journal Nature Genetics, didn't require any new experiments or lab procedures, but used standard analytic methods and "data mining" techniques to discover new cancer-related genes in information gathered from previous studies. The technique may help researchers find genes associated with other forms of cancer as well, plus non-cancerous conditions that have a genetic basis, the authors stated.
"More than 85 percent of breast cancer cases don't have a hereditary link. That is, they arise in women who haven't inherited an abnormal gene that increases their risk for the disease," said study co-author David E. Hill, PhD, of Dana-Farber's Center for Cancer Systems Biology, where the research was based. "Science knows very little about which genes are involved in such cancers. Even among the 15 percent of breast cancers that do have a hereditary component, the responsible genes are known in only about three-quarters of cases."
In this study, Hill said that the researchers wondered if they could use the developing tools of network modeling (which probes the connections between genes and between proteins) and expression profiling (which focuses on patterns of gene activity) to guide them in finding candidate breast cancer genes.
Beginning with a list of four genes known to cause breast cancer when inherited in an abnormal form == BRCA1, BRCA2, ATM, and CHEK2 == the investigators gathered information from existing studies on connections between those "hub" genes and their gene partners. Those studies were of five kinds: expression profiles of genes whose activity mirrors that of the four hubs; surveys of gene activity when BRCA1 is shut down; published studies of biochemical interactions among proteins specified by the hub genes; expression profiles using yeast and worm versions of the hub genes; and the Dana-Farber team's own work in mapping protein-protein interactions.
"Although inherited and non-inherited (or 'somatic') breast cancers develop by markedly different mechanisms, it's likely that some of the same genes are involved," said study co-author Michael Cusick, PhD, of Dana-Farber. "We hope that by beginning with genes known to cause hereditary cases, we can derive clues to genes associated with hereditary disease as well as non-hereditary."
The evidence the researchers collected provided a "global perspective" on the functions of breast cancer genes. "This technique enabled us to integrate existing data to find connections or patterns that others might have missed," added Cusick.
Using this approach, the researchers narrowed the list of candidate breast cancer-susceptibility genes from several thousand to about 150. Within that smaller group, they ranked the genes according to how closely their activity resembled that of the four hubs, and therefore how likely they are to play a role in the disease. The highest-ranked gene that hadn't been previously investigated was one called HMMR.
Confirmation of the validity of the list came serendipitously, when researchers in the lab of Dana-Farber's David Livingston, MD, identified two genes as potentially related to breast cancer. Both were high on the team's list.
Surprisingly, one of these genes turned out to be the frog version of HMMR. Biochemical studies of both the human and frog proteins, carried out by the Livingston lab, confirmed the association of BRCA1 and HMMR.