Genetic Testing for Cancer Risk Not Clinically Useful

Genetic Testing for Cancer Risk Not Clinically Useful
by Tim Wogan on March 17, 2010 5:01 PM |

For women, genetic screening offers the hope of better understanding the likelihood that they'll develop breast cancer. But reality doesn't match that dream, at least not yet. Scientists at the U.S. National Cancer Institute (NCI) today report that DNA doesn't predict breast cancer risk much better than a questionnaire. The small improvement does not yet justify the cost of introducing the technique into the clinic, they say.

In recent years, several gene mutations have been discovered that increase a woman's risk of breast cancer. Best known are mutations in two tumor-suppressor genes called BRCA1 (breast cancer susceptibility gene 1) and BRCA2, which are thought to be present in 0.3% of the U.K. population. A harmful mutation in either gene increases a woman's lifetime risk from 12% to about 60%. Eighteen other genes have been discovered that more subtly influence a woman's risk of breast cancer.

In theory, testing for these genes could allow women to make more informed choices about how often to undergo routine mammograms, for example, or, more radically, whether to take anticancer drugs like tamoxifen prophylactically. These decisions are currently made by patients, in consultation with clinicians, based on a predicted risk of cancer provided by the so-called Gail model. This model calculates a risk based on the answers to seven questions, including the age at which a woman began menstruating, the age at which she had her first child, and the number of relatives with breast cancer.

To find out how well genetic screening measured up to the question-based Gail model, a team of cancer epidemiologists at NCI pooled data from five of the studies originally used to isolate the breast cancer genetic risk factors. Four were cohort studies in which a healthy population was genetically screened at the outset and followed for 15 years to see who developed breast cancer and who did not.

In the new work, published today in The New England Journal of Medicine (NEJM), the researchers identified from those studies 5590 women who went on to develop breast cancer and 5998 who did not. Then they retrospectively calculated a prediction of cancer risk based on each woman's data for the 10 genetic risk factors known at the outset of the study. They next asked a simple question: What is the probability that a woman selected at random from the group that did go on to develop cancer would have a higher risk prediction than a randomly selected woman who did not? For a completely useless model, the answer would be 50%; for a perfect model, the answer would be 100%.

The answer for the genetic screening was 59.7%, whereas the answer for the question-based Gail model was 58%. By combining the two, the researchers were able to produce a model with a predictive power of 61.8%. But that combination didn't impact the prediction of risk, also called the score, very much for most individual patients. “There were very, very few cases in which the new score was hugely different from the old score,” says cancer epidemiologist Patricia Hartge of the National Cancer Institute in Bethesda, Maryland, a study co-author. She and her colleagues conclude that, given the cost involved, genetic screening is not worthwhile in a clinical context.

Nevertheless, Hartge remains optimistic about the future. She points out that the common genetic variants they tested were discovered less than 3 years ago. "Isn't it fascinating that we get the same ability to predict from these that we got from 40 years of painstaking research on the other risk factors?" Discoveries of more mutations, including the eight found since this study began, should improve the reliability of genetic tests, she says.

Cancer epidemiologist Paul Pharoah of the University of Cambridge in the United Kingdom, who published a similar analysis 2 years ago in NEJM based on just seven genetic risk factors, agrees that genetic tests don't add a whole lot to the Gail model. But he questions the new paper's assessment that screening has to be expensive: "The cost of one of these genetic tests in reality is trivial," Pharoah says. So genetic tests could be a cost-effective way to decide whom to screen further, he says.

This article has been corrected to reflect that Patricia Hartge is currently affiliated with the National Cancer Institute of the National Institutes of Health and not the George Washington University.
http://news.sciencemag.org/sciencenow/2010/03/genetic-testing-for-cancer-risk-.html