Common Genetic Variants Have Little Effect on Breast Cancer Prediction

Incorporating common genetic risk factors into the models typically used to calculate a women’s breast cancer risk has little impact on clinical decision-making, such as whether an individual should consider earlier or more frequent mammography or prophylactic drugs, according to a paper published today in the New England Journal of Medicine. The results follow similar studies for diabetes and cardiovascular disease, echoing what has now become a common criticism of genome-wide association studies; that this approach is unlikely to identify genetic risk factors of diagnostic value.

Over the last few years, researchers have used DNA-studded microarrays to quickly search tens of thousands of human genomes for common genetic variations linked to various diseases, dubbed genome-wide association studies. Despite the enormous size of these studies, they have only identified a fraction of the source of the genetic risk of disease. And the vast majority account for a very small change in risk in a given individual. (This is in contrast to rare genetic variants, like BRCA1, which substantially increase a women’s risk of developing breast cancer.)

To analyze the potential impact of common, breast cancer-linked variations, researchers from the National Cancer Institute combined data from five studies of breast cancer. Taken together, these studies compared 5,590 breast cancer patients to 5,998 women without cancer, mostly white and age 50 and 79. The team employed a commonly breast cancer risk model, which uses medical, reproductive and family history, to calculate an individual’s risk of developing cancer over the next five years. They found the risk score was similar to that calculated using 10 breast cancer variations recently identified in genome wide association studies. But combining the two risk models had little impact. “When we included these newly discovered genetic factors, we found some improvement in the performance of risk models for breast cancer, but it was not enough improvement to matter for the great majority of women.” said Sholom Wacholder, Ph.D., senior investigator in NCI’s Division of Cancer Epidemiology and Genetics (DCEG), in a statement.

For most women in the study, the inclusive model did not substantially change their personal estimated risk of developing breast cancer beyond the Gail model calculations. Overall, using the inclusive model reclassified 26 percent of women to a higher risk category; 28 percent to a lower risk category; and left 46 percent in the same category of risk score. The shifts from one category to another were generally too small to influence clinical decision-making.

That’s probably not welcome news to direct-to-consumer genetic testing companies, such as Navigenics and DecodeMe, which screen for these types of variants. Kari Stefansson, founder of Decode, argues that the results do show that common variants are useful, but that we still have a way to go.

My hope is that this entire argument is soon moot. Whole genome sequencing, which can identify both rare and common variants, seems finally poised to fulfill its role in illuminating the genomics of disease. A handful of papers published over the last few months have demonstrated that sequencing can identify genetic variants linked to some rare diseases, and scientists hope the same approach can be applied to more common ones. While a $48,000 genome sequence–the cost of Illumina’s personal sequencing service–is still a lot compared to Decode’s $500 cancer screen, the price is dropping rapidly. (No one knows yet how much more bang you’d get for your buck.) Complete Genomics, a startup in California, will soon offer bulk sequencing services for about $20,000 a genome, with a $5,000 price tag not far behind.