Mapping the Risk of Breast Cancer in Thousands of Women

Some women inherit relatively rare mutations in the breast cancer susceptibility genes BRCA1 and BRCA2 that raise their risk of developing the disease significantly. But for most women, the predisposition to breast cancer appears to be determined by a large collection of common and uncommon genetic variations, each with a small individual effect on cancer risk.

This view has emerged from large-scale genomic studies, including genome-wide association studies, that have begun to pinpoint chromosome regions associated with breast cancer risk. In recent years, a first generation of studies, which included thousands of women, conclusively linked six chromosome regions to breast cancer risk, and now four more regions have been identified.

"These studies have shown us that there are dozens, if not hundreds, of regions that each contribute a small amount to a woman's overall breast cancer risk," said Dr. David Hunter of the Harvard School of Public Health and a leader of NCI's Cancer Genetic Markers of Susceptibility (CGEMS) project, which has conducted genome-wide association studies for breast cancer and prostate cancer. "This idea was hypothesized 3 or 4 years ago, and now we have the beginnings of the proof."

In a planned follow-up to earlier breast cancer genome-wide association studies, the CGEMS team identified and validated regions of interest on chromosomes 1 and 14. A second study, led by Dr. Douglas Easton of Cancer Research UK, similarly identified regions on chromosomes 3 and 17 in close collaboration with the CGEMS initiative. Findings from both studies appeared online in Nature Genetics on March 29.

"These two studies represent the second wave of genome-wide association studies for breast cancer involving very large numbers of women," said Dr. Stephen Chanock of NCI's Division of Cancer Epidemiology and Genetics and a leader of CGEMS. "We now know that we are going to continue to reap novel insights from this strategy, and also that we will need to do further follow-up work to understand the basic mechanisms underlying risk."

The CGEMS team also confirmed previous reports that risk factors for breast cancer may reside on six other regions, on chromosomes 2, 5, 8, 10, and 16.

Identifying the Landmarks

The first step in all of these studies was to compare DNA from women with breast cancer and healthy women using single nucleotide polymorphisms (SNPs) as markers. These are places in the genome where a single unit of DNA varies from person to person with high enough frequency to be tested effectively. Tracking the distribution of SNPs in women with and without cancer can reveal inherited variants that are significantly more common in those with an increased or decreased risk of disease.

Because of their unbiased approach, genome-wide association studies have frequently pointed to parts of the genome where no one had thought to look previously for cancer risk factors. While this has generated new leads, it has also underscored a challenge facing the entire field: How do researchers make sense of risk signals in poorly characterized regions of the genome?

"Even when the genomic architecture of a region has been explored, mapping and then understanding the source of the risk can be difficult," noted Dr. Chanock, who last month hosted a scientific meeting on this subject for experts in genome-wide association studies.

"We know that each of these regions is complex," he continued. "A risk signal may be a marker for a particular set of variations or structural changes in the genome. The challenge is to dissect each region and understand how changes in that region contribute to breast cancer risk."

These types of studies are not designed to measure all of the variation in the human genome, so when a signal is found, much more work is needed to discover its source, added Dr. John Witte of the University of California, San Francisco, who uses the approach to investigate prostate cancer genetics.

In just the last few years, new technologies have revealed a surprising amount of both normal and abnormal variations across the genome, including gains and losses of DNA and variation in the number of gene copies. "The human genome is extremely complicated, and there are many sources of variation that need to be considered as underlying causes of cancer risk," Dr. Witte said.

These association studies have provided important new clues to genetic mechanisms underlying cancer risk. In the case of prostate cancer, researchers have suspected that genetic factors contribute to risk because of familial patterns, but until genome-wide association studies there had been essentially no biomarkers of inherited risk.

Scanning the Horizon

Even with the recent progress, these are early days. In breast cancer, for example, the chromosome regions identified to date account for only a small percentage of the overall inherited risk for the disease. But as the new results suggest, collaborations involving tens of thousands of DNA samples and the careful validation of results should reveal additional sources of risk.

"The lesson we've learned from genome-wide association studies is that because the strength of the signal at each marker is relatively weak, no single study has the power to pick up more than a small number of all the regions that are out there, so larger sample sizes and pooling of data are needed," said Dr. Hunter.

"But by bringing together resources from within NCI and its external collaborators," he continued, "we were able to do in several years what it would take an individual group many more years to achieve."

—Edward R. Winstead