NEW YORK (GenomeWeb News) – Adding data on 10 genetic variants to non-genetic risk models of breast cancer offers little improvement for predicting breast cancer risk, new research suggests.
Based on a study of nearly 6,000 women with breast cancer and about as many controls, researchers from the National Cancer Institute and elsewhere concluded that adding information on 10 breast cancer risk SNPs to a non-genetic risk model only slightly improve breast cancer risk prediction, reclassifying just over half of women into higher- or lower-risk groups. The work appeared in today's issue of the New England Journal of Medicine.
"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," lead author Sholom Wacholder, a senior investigator with NCI's cancer epidemiology and genetics division, said in a statement.
Wacholder and his co-workers compared genetic risk models with models based on traditional risk factors with or without the addition of genotyping information. The team used 10 SNPs previously associated with breast cancer and included data on 5,590 women with breast cancer and 5,998 unaffected controls. Women assessed in the study were between 50 and 79 years old and had been involved in four American cohort studies and a Polish case-control study.
The researchers found that a model relying on SNP data alone provided slightly better risk information than a non-genetic model called the Gail model (which is based on information on a woman's medical, reproductive, and family history).
But, they noted, "non-genetic clinical variables are available at essentially no cost, whereas the costs of obtaining genetic information are likely to be substantial."
Meanwhile, adding genetic information the Gail model only slightly improved its predictive value, leading to reclassification of 32.5 percent of women to a higher risk group and 20.4 percent to a lower risk group.
"Our results indicate that the recent identification of common genetic variants does not herald the arrival of personalized prevention of breast cancer in most women," the researchers concluded. "Even with the addition of these common variants, breast cancer risk models are not yet able to identify women at reduced or elevated risk in a clinically useful way."
Still, researchers are not giving up on the possibility of finding additional genetic variants that may eventually have better predictive value and clinical utility.
"We can expect to identify more genetic determinants of breast cancer, and to learn more about those we have already found," Wacholder said in a statement. "This information, along with our increasing knowledge of non-genetic factors, should allow us to steadily improve our risk prediction models for breast cancer."
In a NEJM editorial, Leiden University pathology and human genetics researcher Peter Devilee and Netherlands Cancer Institute epidemiology researcher Matti Rookus said "[t]heir disappointment seems a little premature."
"Clearly the 10 SNPs assessed by Wacholder et al. are no more than the tip of the iceberg," the pair wrote. "A more pressing question is why, after the completion of several genome-wide association studies of breast cancer, only a dozen risk alleles have been identified."
While Devilee and Rookus called for studies aimed at finding additional breast cancer risk variants, another pair of researchers this week discussed the state of another type of genetic information — prognostic gene expression signatures — in lung cancer.
In a review in the Journal of the National Cancer Institute online earlier this week, JCI researchers Jyothi Subramanian and Richard Simon assessed 16 studies of gene expression-based prognostic signatures for non-small cell lung cancer published between January 2002 and February of last year and did not find a study in which gene expression signatures improved outcome prediction over other risk factors.
"[W]e found little evidence that any of the reported gene expression signatures are ready for clinical application," they wrote.
Moreover, they noted, several of the studies they evaluated as part of the review had "serious problems" in their design and analysis. As such, they proposed a set of guidelines for future prognostic gene expression signature studies, discussing everything from data collection and statistical validation to the presentation of results.
"These guidelines emphasize the importance of focused study planning to address specific medically important questions and the use of unbiased analysis methods to evaluate whether the resulting signatures provide evidence of medical utility beyond standard of care-based prognostic factors," they explained.