News

Women with breast cancer who carry a mutation in either the BRCA1 or the BRCA2 gene have a substantial risk of a developing a second primary breast tumour or ovarian cancer in the absence of prophylactic intervention such as mastectomy, oophorectomy or tamoxifen treatment. Risk prediction models to estimate the probability that a woman carries a BRCA mutation have mostly been developed using large families and include family history as part of the analysis. Such models might not be well suited for women with few or no female relatives.

A study of 306 women with breast cancer diagnosed under the age of 50, and no first- or second-degree relatives with breast or ovarian cancer, has found that mutation-prediction models do indeed perform poorly in this situation [Weitzel JN et al. (2007) JAMA 297, 2587-95]. Although none of the women had close affected relatives, half had a family structure considered ‘adequate’ for risk analysis (that is, an adequate number of maternal and paternal female relatives), and half were classified as having ‘limited’ family structure (fewer than 2 females who lived to age 45 in each lineage). The probability that a woman with early-onset breast cancer and limited family structure had a BRCA mutation was found to be 13.7%, compared to a 5.2% probability for those with adequate family structure. None of the three risk prediction models tested (the Couch, Myriad and BRCAPRO models) performed well in predicting mutation risk for the women with limited family structure.

Comment: The results of this study suggest that criteria for considering BRCA mutation testing in women with early-onset breast cancer may need to be modified in cases where family structure is limited, if knowledge of mutation status is likely to affect medical management. Women in this situation may also wish to know their mutation status if they have children or are thinking of starting a family. Current UK guidelines recommend that genetic testing should only be offered in cases where there is at least a 20% probability of finding a mutation; the 13.7% probability identified in this study falls short of that figure but is higher than the 10% threshold used in some US guidelines.