Introduction

Two recent papers have observed surprising findings of distorted sex and transmission ratios associated with BRCA1 mutation carriers.1, 2 Specifically, a significant deficit of sons was observed among 68 Spanish breast–ovarian cancer families segregating a BRCA1 mutation, but not among BRCA2 families nor in BRCA1/2 mutation-negative families.1 Further, in a report from Poland, 91 mothers with breast or ovarian cancer and a founder BRCA1 mutation had 122 daughters. Since 50% would be expected to be mutation carriers at birth, their observation that 61% of daughters who remained free of breast or ovarian cancer into young adulthood were mutation positive suggested a distorted transmission probability, favoring the mutant allele.2 If real, these observations would provide a novel insight into the normal biological function of BRCA1 and have implications for genetic epidemiologic methods used to estimate penetrance, such as kin-cohort,3 because they often rely on the assumption of equal transmission of mutant and wild-type alleles. We addressed these provocative observations in two study settings: a sample of BRCA1/2 mutation-positive families with multiple cases of breast and ovarian cancer,4 and a community-based sample of Ashkenazi Jewish subjects tested for founder BRCA1/2 mutations.5

Materials and methods

We analyzed previously collected data from two study populations. The first consisted of all delineated members of families with multiple cases of breast and ovarian cancer (mean 6.1 cases per family) who were positive for BRCA1 or BRCA2 mutations.4 Among the 28 families segregating BRCA1 mutations and eight families segregating BRCA2 mutations, there were 283 mutation carriers (196 tested directly and 87 inferred positive owing to detection of the mutation in offspring) and 471 subjects who tested negative and who were age 20 or older. We determined the sex of the 730 offspring of mutation carriers and of the 754 offspring of mutation-negative subjects. To calculate the transmission ratio, we determined the mutation status and cancer status of all offspring of BRCA1 and BRCA2 mutation carriers (tested and inferred, females and males).

The second study population included all unrelated Ashkenazi Jewish subjects who participated in a community survey of more than 5000 Washington area Jewish residents who gave permission for future use of their information.5 All 5082 subjects were tested for the three common founder mutations in BRCA1 (185delAG and 5382insC) and BRCA2 (6174delT) and provided detailed family histories, including delineation of the age and cancer status of all first-degree relatives. We determined the sex of the siblings and children of the 58 BRCA1 mutation carriers (194 relatives), the 57 BRCA2 mutation carriers (175 relatives), and the 4864 mutation-negative subjects (16 997 relatives).

We calculated sex ratios by comparing the number of male and female offspring of subjects with known mutation status such that values less than one represent fewer sons than daughters. To determine whether the observed sex ratios deviated significantly from the expected value of 1.0, we calculated approximate confidence intervals using Woolf's method.6 To compare the sex ratios between subjects with differing mutation status within the two study groups, we used the χ2 statistic to contrast all three groups (BRCA1 carriers vs BRCA2 carriers vs noncarriers; χ2 with 2 degrees of freedom) and for each two-way contrast (1 degree of freedom; BRCA1 vs BRCA2 mutation-positive subjects, BRCA1 vs noncarriers, etc.). To assess the transmission ratio of BRCA1 and BRCA2, we calculated the proportion who were mutation carriers among sons, and daughters without breast or ovarian cancer, by attained age (in 10-year intervals) and calculated a χ2 for trend to test whether the proportion positive decreased with increasing age.

Results

Within the NCI BRCA1/2 mutation-positive multiple-case families, the sex ratio among births to BRCA1 mutation carriers was below one (0.85, 95% CI 0.68–1.06) but it was even lower among BRCA2 mutation carriers (0.77, 95% CI 0.47–1.27) (Table 1). We found that the reduced sex ratio among children of mutation carriers was due largely to a distortion among the offspring of male mutation carriers (0.70, 95% CI 0.47–1.05 for offspring of male BRCA1 mutation carriers and 0.42, 95% CI 0.16–1.10 for BRCA2); the ratios were only slightly below 1 for offspring of female BRCA1 and BRCA2 mutation carriers (0.92 and 0.98, respectively). The results for BRCA1 differed when we included only offspring of individuals who were directly tested and found to be positive (ie, excluding those inferred positive); the sex ratio was 1.03. In contrast, the sex ratio was reduced even further, to 0.70, when we included only directly tested individuals from BRCA2 families. Among siblings and children of unrelated subjects in the Washington Ashkenazi Study (WAS), the sex ratio was below one (0.90, 95% CI 0.61–1.34) for BRCA1 mutation carriers, but was even lower for the BRCA2 carriers (0.80, 95% CI 0.53–1.22) and near the expected (1.02) among BRCA1/2-negative subjects. The sex ratio was reduced only slightly among offspring of mutation-negative subjects from the NCI breast–ovarian cancer families (0.94). None of the sex ratios deviated significantly from the expected value of 1.0 (Table 1).

Table 1 Observed sex ratios in NIH BRCA1/2 families and subjects compared to previous studies
Full size table

The sex ratios in BRCA2 families were lower than for BRCA1 families, but they were not statistically different between mutation groups within studies: comparing BRCA1 vs BRCA2, χ2(1 df) was 0.21 (P=0.7) and 0.32 (P=0.6) for the multiple-case families and WAS, respectively. None of the comparisons of sex ratios between BRCA1 vs noncarriers, BRCA2 vs noncarriers, or three-level comparisons (BRCA1 vs BRCA2 vs noncarriers, 2 degrees of freedom) had P-values below 0.1.

To address the question of a potential distorted transmission ratio of mutant BRCA1 alleles, we studied adult daughters without breast or ovarian cancer born to mutation carriers in the NCI multiple-case families; 39% were mutation carriers in BRCA1 families and 44% were mutation carriers in BRCA2 families (Table 2). The proportions of all tested sons who were mutation positive were 54 and 55% for BRCA1 and BRCA2, respectively. When only the offspring of individuals directly tested were considered, the carrier proportions among daughters were 46 and 39% for BRCA1 and BRCA2. Although based on small numbers, there was a lower proportion of carriers among older unaffected daughters, with a statistically significant trend (χ2(1 df)=6.26, P=0.01) (Table 2).

Table 2 Observed transmission ratios in NIH BRCA1/2 families compared to a previous study
Full size table

Discussion

Our data suggest that the sex and transmission ratios associated with BRCA1 mutations are not distorted. We have failed to replicate previous observations1, 2 despite similar to or larger sample sizes and believe they represent chance findings. The sex ratio was only modestly reduced in our multiple-case BRCA1 families (0.85) and in a community-based sample of Ashkenazi Jewish BRCA1 mutation carriers (0.9). These reductions were less pronounced than in BRCA2 families from the same studies, and in none of the subgroups analyzed did the sex ratio deviate significantly from the expected value of 1.0. Most importantly, there were no statistically significant differences between BRCA1-positive vs BRCA2-positive vs mutation-negative families as was observed by de la Hoya et al.1

Assuming equal transmission probabilities for a mutant and wild-type allele and equal prenatal viability, one-half of the offspring of BRCA1/2 mutation carriers will carry the mutation at birth. When one considers only daughters who remain free of breast and ovarian cancer into adulthood, the proportion should be below 50% because women who develop cancer, who will disproportionately be mutation carriers, are removed from the calculation. These considerations make the data from Poland,2 suggesting an increased transmission rate of mutant BRCA1 alleles to female offspring, very striking. If true, it would provide an important insight into the normal biology of this gene. In addition, it would have implications for family-based genetic epidemiologic methods used to estimate penetrance because they rely on the assumption of equal transmission probabilities for the mutant and wild-type alleles.3 We found no evidence to support a distorted transmission ratio. On the contrary, fewer than 50% of unaffected daughters of BRCA1 mutation carriers were themselves positive, and our observations were consistent with the expected trend of fewer carriers among older, unaffected daughters.

Studying sex ratios in families ascertained because of a sex-limited phenotype, such as familial breast and ovarian cancer, is challenging. One might expect the sex ratio to be below one owing to ascertainment criteria alone (ie, families that by chance contain predominantly male offspring will be less likely to demonstrate the clustering of female cancers required to meet study entry criteria). Moreover, since females are disproportionately affected in these syndromes, female relatives may be more likely to be reported by probands and recorded by investigators when family histories are taken. The absolute magnitude of this sex ratio bias would be difficult to predict a priori, even if the family sizes and ascertainment criteria are known, without precise knowledge of penetrance and differential reporting of relatives. Assuming similar penetrance, however, one would not expect the sex ratio to be different between families subsequently found to segregate mutations in BRCA1 vs BRCA2. Hence, while the findings of sex ratios below one were not surprising in the Spanish data,1 that it was significantly lower in BRCA1-positive families compared to BRCA2-positive and BRCA1/2-negative families was noteworthy. We did not observe a significant difference between these mutation subgroups in our study populations.

In calculating sex ratios, we considered all offspring (both mutation known and mutation unknown subjects) only of subjects in whom we had directly tested or inferred their mutation status (ie, the children of subjects whose mutation status was unknown were not included in the calculations). As a result, differences in which subjects decide to get tested in a family may partially explain the reduced sex ratios we observed. Men have less compelling personal health reasons for being tested and instead are more likely to be tested if they have children.7 If men (and/or women) with daughters are more likely to participate in family genetic studies and are more likely to seek testing than men with sons, this would result in a reduced sex ratio. Similarly, if men are even more likely to be tested if they have daughters with breast or ovarian cancer, this would further reduce the sex ratio associated with mutation carriers because such men will often test positive for the family mutation. Our observation of even lower sex ratios among offspring of male mutation carriers supports this idea.

The observed sex ratio among the offspring of BRCA1 mutation carriers differed depending on whether subjects were tested directly (1.03) or inferred to be positive (0.71). The lower sex ratio for offspring of inferred mutation carriers may be an artifact. Many of these subjects were deceased women with breast or ovarian cancer. To be inferred positive these women must have had children, and at least one descendant had to have tested positive. This scenario is weighted toward individuals with daughters, because daughters develop the syndromic cancers and more often seek testing. We did not see the same phenomenon in the BRCA2 families, but these calculations were based on only 15 subjects who were inferred to be positive for BRCA2.

In summary, data from two study settings do not support the hypothesis that the sex and transmission ratios are particularly distorted in BRCA1 families: we observed no significant differences in the sex ratio comparing BRCA1-positive vs BRCA2-positive vs BRCA1/2-negative subjects, and the proportion of daughters without breast and ovarian cancer born to BRCA1 mutation carriers decreased with the age of the daughter and was below 50% overall. The reduced sex ratios we observed were more consistent, and of greater magnitude, in our BRCA2 families, but in no case did the ratios vary significantly from 1.0. We expect that these observations on sex/transmission ratios primarily reflect the biases inherent in family-based studies of a sex-limited phenotype and that the previous observations of significant distortions were due to chance.