Breast Cancer Mutation Probability Models Accurate

High false-positive, false-negative rates outside high-risk populations

THURSDAY, Oct. 18 (HealthDay News) -- Seven models can adequately predict the probability of a harmful mutation in two breast cancer genes, but with high false-positive and false-negative rates outside high-risk populations, according to study findings published in the Oct. 2 issue of the Annals of Internal Medicine.

Giovanni Parmigiani, Ph.D., from Johns Hopkins University in Baltimore, and colleagues examined the validity of seven models to accurately predict the probability of a mutation in the BRCA1 or BRCA2 gene in 3,342 families.

The researchers found that all models could identify women who probably carry a harmful mutation to differing degrees. The better models had a c-statistic of about 80 percent. However, they note that all of the models had high false-negative and false-positive rates outside of high-risk populations.

"At this time, we should use model-based predictions in conjunction with clinical judgment," Susan M. Domcheck, M.D., from the University of Pennsylvania in Philadelphia, and Antonis Antoniou, Ph.D., from the University of Cambridge in the United Kingdom, write in an accompanying editorial. "However, we can't use these models responsibly unless families keep track of their medical history and physicians ask about it."

Full Text (subscription or payment may be required)

Physician's Briefing