Statistics from Altmetric.com
The functions of the breast cancer susceptibility genes BRCA1 and BRCA2 are not fully elucidated, but appear to include the regulation of X chromosome activity. Xist is a non-coding RNA that accumulates on the inactive X chromosome and is required for X chromosome inactivation during the silencing step.1 The RING domain of BRCA1 and Xist interact in mammalian cells and it has been suggested that BRCA1 contributes to the initiation of X chromosome inactivation.2 Women with ovarian cancer possessing germline mutations in BRCA1 have been found to frequently demonstrate non-random X chromosome inactivation.3 In the light of these findings, a recent report by de la Hoya et al is of interest.4 In this study of 68 Spanish breast/ovarian pedigrees they reported that 67% of the children of women who carried a BRCA1 mutation were female. By contrast, only 54% of the offspring of BRCA2 carriers and 52% of the offspring of non-carriers were female. This highly skewed sex ratio in the offspring of BRCA1 carriers from Spain prompted us to ask whether this is true in other populations as well.
To address this question, we examined the sex ratios of the offspring of a total of 1040 women (229 BRCA1 carriers, 104 BRCA2 carriers, and 707 non-carriers) from five different studies which were conducted in Montreal, Toronto, and Oslo between 1993 and 2003. In four studies the carriers were identified through hospital-based and population-based ascertainment and were not selected for family history,5–8 whereas one of the studies (the single study from McGill University, Montreal, Quebec, Table 1) was composed of Ashkenazi Jewish women ascertained via a high risk clinic. Three of the studies were also restricted to Ashkenazi Jews. In these studies the participants were tested for the three Ashkenazi Jewish founder germline mutations in BRCA1 and BRCA2. In the fourth study of ovarian cancer patients in Ontario, BRCA1 was screened in its entirety using PTT and DGGE and BRCA2 was screened in part using PTT.5 The fifth study, restricted to women with ovarian cancer diagnosed in southern Norway, was conducted in Oslo, and the results presented here are restricted to mutation analysis of known Norwegian founder BRCA1 mutations as previously reported for this series of cases,8 and the results of an extended mutation search in patients with a positive family history of breast and ovarian cancer. We counted all the female and male births reported by the proband in each pedigree.
We did not find any evidence for sex ratio distortion in the overall sample (Table 1). Among 229 BRCA1 carriers, there was no statistically significant excess of female births (221 males vs 234 females, p = 0.54). The p value was derived from the binomial distribution, using an exact method, comparing the observed proportion with the expected proportion of 0.50. The proportion of female offspring was quite similar for children of BRCA1 carriers (51.4%), of BRCA2 carriers (50.2%), and of non-carriers (51.6%). This is in contrast to the striking results of the previous report (65 males vs 133 females offspring among confirmed BRCA1 carriers, p⩽0.001).4 Three subgroup results were of borderline significance—there were slight excesses of females among children of Ashkenazi Jewish women with ovarian cancer and a BRCA1 mutation7 and among Ashkenazi Jewish women with either no founder mutation or a BRCA2 mutation in the McGill University, Montreal, Quebec study (Table 1). The differences could be related to ascertainment bias, or may be chance findings as a result of the large number of comparisons made. This latter possibility is strengthened by a further analysis of a group of 23 non-Ashkenazi women who were diagnosed with ovarian cancer at McGill University and were seen in high risk clinics, and were found to be BRCA1 carriers: the number of male and female children from these 23 women was identical (44 in total).
BRCA1 may have a role in X-inactivation.
Some previous evidence presented supporting non-random X-inactivation in BRCA1 carriers.
A recent study demonstrated sex ratio bias, favouring female births.
We studied 1040 women tested for BRCA1/2 in three centres in Canada and Norway.
Overall, there was no consistent evidence for a bias in the sex ratio of offspring of BRCA1 or BRCA2 carriers.
Women with BRCA1 mutations and a previous diagnosis of ovarian cancer had significantly fewer children than did women with ovarian cancer and a BRCA2 mutation, implying that parity affects ovarian cancer risk differently in the two groups.
Studies of offspring number, sex ratio, and transmission ratio among BRCA1/2 carriers are all subject to several possibly hidden biases. It is difficult to exclude them all, particularly when using clinic-based ascertainment strategies.
It is noteworthy that BRCA1 carriers affected by ovarian cancer had significantly fewer children (whether male or female) than all other groups combined (1.89±1.07 and 2.16±1.30, respectively, p = 0.006), possibly as a result of the diagnosis of ovarian cancer being at sufficiently young age to adversely affect fertility, but more likely because nulliparity is a risk factor for ovarian cancer. Interestingly, BRCA2 carriers with ovarian cancer had significantly more children (2.29±1.44) than BRCA1 carriers with ovarian cancer (1.89±1.07, p = 0.04). This difference could be due to parity being less protective against ovarian cancer in BRCA2 carriers than in BRCA1 carriers. Another possibility is that fertility might be more likely to be prematurely terminated by a diagnosis of a BRCA1-related ovarian cancer than it would be by a diagnosis of a BRCA2-related ovarian cancer. In support of this idea, the average age at diagnosis of ovarian cancer was 59.2 years (±9.64 years) for BRCA2 carriers and 51.2 years (±9.41 years) for BRCA1 carriers (p<0.0001) in the three available series.5,7,8 Finally, it may be merely a chance finding.
Overall, these results indicate that subtle ascertainment biases are likely to be present in any study that involves a tested population whose disease status and family history may influence the decision to attend a high risk clinic, or be enrolled in a study. Therefore we consider that the parsimonious explanation for the different results obtained in the studies in Spain, Canada, and Norway is that neither BRCA1 nor BRCA2 influence the male/female sex ratio of offspring, and that the observed effects observed are more likely to be due to ascertainment bias. Several different ascertainment biases might be important; for example, women with female offspring may be more likely to seek breast/ovarian cancer susceptibility testing than women with male children only or with no children.9,10
Conflict of interest: none declared