These data support the position that early-onset triple-negative breast cancer is an indicator that can be used to help to identify candidates for BRCA1 mutation testing. Our study is relatively small (54 patients) but we confirm similar observations made by others. To date, no single study has been large or definitive, and therefore it is important to consider the results of all studies in aggregate. Lidereau et al found that 6 of 70 women (9%) with breast cancer diagnosed at age 35 or below, unselected for family history, carried a BRCA1 mutation . However, the proportion with mutations was 29% (4 of 14) for those with ER-negative, high-grade tumours, compared to only 4% (2 of 56) among women with other tumour types (odds ratio = 11; p = 0.007). Among women in our study who were diagnosed before the age of 35, the mutation prevalence was 8.3%. In a study of 254 white women from the UK, diagnosed with breast cancer before age 36, a germline BRCA1 or BRCA2 mutation was identified in only 6% (ref 16). Of the 15 women with germline mutations, only one had a family history of breast or ovarian cancer. In this early study, the study sample was not subdivided by ER-status or other factors.
A number of studies suggest that breast cancers associated with BRCA1 mutations are likely to be triple-negative and the majority of these are also the basal phenotype [7, 8, 17–19]. Basal-like tumours express certain cytokeratins characteristic of the 'basal' layer myoepithelial cells lining the terminal duct lobular unit (namely cytokeratins 5, 6, 14 and 17) . Basal-like tumours are usually high-grade, exhibit comedeonecrosis, pushing borders and an inflammatory lymphocytic infiltrate.
Other studies included women who were selected for family history or ethnic group. Chang et al. studied women with familial breast cancer (at least one first-degree relative with either breast or ovarian cancer) diagnosed before age 45 . They found that 6 of 24 (25%) patients with an ER-negative, high-grade tumour had a germline BRCA1 mutation. Haffty et al. that reported that eight of 34 women with triple-negative breast cancer had a BRCA1 mutation (24%), but these included women with a strong family history of breast cancer . Foulkes et al. found that of 24% of 72 Ashkenazi Jewish patients diagnosed before age 65 with high grade, ER-negative, HER2-negative breast cancers had a germline BRCA1 mutation . However, the frequency of mutations is known to be high in this group. Given that all Ashkenazi Jewish women with breast cancer are candidates for BRCA mutation testing, we excluded this group from the present study.
Our study has several limitations. Our subjects were derived from various sources, including a population-based series of cases (Ohio), and from a US hospital (North Carolina) and from women who presented for genetic risk assessment to various clinics across the country. The determination of grade, ER, PR and HER2 status was based on review of pathology report and there was no central pathology review (however, this situation accurately reflects actual clinical practise). The purpose of this study was to estimate the prevalence of BRCA1 mutations and the entire BRCA2 gene was not screened. However, the triple-negative phenotype is not characteristic of BRCA2-associated breast cancers. We expect that the majority of BRCA mutations diagnosed in the triple-negative, family history negative subgroup cancers will be BRCA1. In most situations, genetic counsellors will offer to test for both genes, and if a woman is insured, testing for both genes will be covered. However, for women who do not have full coverage, or in countries where resources are limited, some may wish to test triple-negative patients for BRCA1 only.
All five BRCA1 mutations were seen in white women between the ages of 31 and 40; in this subgroup the mutation prevalence was 17%. It is interesting that no mutation was identified in a woman diagnosed before age 30; however this subgroup was small and it is in not clear if this is a chance finding. Further studies will be useful in determining if there are other genetic markers of very early-onset breast cancer.
The criteria used to offer genetic testing to a patient varies from center to center, but in general, testing is usually offered to a woman if the probability of finding a mutation exceeds ten percent . Most guidelines for selecting patients rely on family history, but in many centers, genetic testing is routinely offered to all women diagnosed with breast cancer under the age of 40. In the absence of a family history, fewer than ten percent of patients in this group are expected to carry a BRCA mutation.