BRCA1 mutation carriers have a lower number of mature oocytes after ovarian stimulation for IVF/PGD

Contradicting results have been published on a potential influence of mutations in the BRCA1 and BRCA2 genes on ovarian reserve. Mutations in the BRCA genes are primarily known for their predisposition to breast and ovarian cancer [1]. The BRCA genes act as tumor suppressor genes and are involved in DNA double-strand break repair [2]. An impaired function leads to an accumulation of intracellular DNA damage. This may affect cellular growth mechanisms, leading to carcinogenic transformation [3]. Alternatively, accumulating DNA damage may induce growth arrest, leading to apoptosis [4]. Hypothetically, this may be illustrated in non-dividing cell populations, e.g., the ovarian follicle pool.

Oktay et al. [5] were the first to observe a reduced ovarian response to ovarian stimulation for in vitro fertilization (IVF) in BRCA1 mutation-positive cancer patients undergoing fertility preservation. This was not confirmed by another report on the ovarian response to IVF stimulation in a combined group of BRCA1/2 mutation carriers undergoing fertility preservation because of breast cancer and asymptomatic BRCA1/2 mutation carriers undergoing IVF with preimplantation genetic diagnosis (PGD) [6]. Contradicting results have also been published when assessing ovarian reserve in BRCA1/2 mutation carriers using other endpoints. Several studies on age of natural menopause reported an earlier menopause in both BRCA1 and BRCA2 mutation carriers [7‐9]. The majority of studies using anti-Müllerian hormone (AMH) as an indicator for the number of (pre-)antral follicles in the ovaries detected lower levels of AMH in BRCA1 mutation carriers, not in BRCA2 mutation carriers [10‐13]. Studies using several other reproductive outcome parameters (e.g., parity) did not point to a reduced fecundity in BRCA1/2 mutation carriers [14‐18].

Ovarian response to stimulation for IVF is a strong indicator for ovarian reserve status [19]. Sufficient ovarian response is particularly important in PGD, where transfer criteria primarily involve genetic results. After a second selection on embryo quality, only a minority of the obtained embryos will be available for transfer. If a mutation in the BRCA1 and/or BRCA2 gene is associated with a lower ovarian reserve, this may have a negative effect on success chances of mutation carriers undergoing IVF for infertility reasons, for fertility preservation, as well as for PGD. PGD for BRCA1/2 mutations has been performed for a decade now and the number of couples treated each year has been growing steadily [20, 21].

A retrospective, observational cohort study was carried out in five centers: Maastricht University Medical Center (center 1) and affiliated IVF centers University Medical Center Utrecht (center 2), University Medical Center Groningen (center 3), and Academic Medical Center Amsterdam (center 4), united in the Dutch consortium for PGD, and Universitair Ziekenhuis Brussel, Brussels, Belgium (center 5). The study period lasted from the introduction of PGD for hereditary cancer syndromes (i.e., 2006 for Brussels and 2008 for The Netherlands) until September 2015.

The exposed group consisted of couples who underwent IVF/PGD because of a pathogenic mutation in the BRCA1 or BRCA2 gene in the female (the “BRCA group”). All mutations were proven pathogenic by means that they had a verified significant disturbing effect on protein translation. The control group consisted of couples who underwent PGD because of an autosomal dominant or recessive disorder not known to be associated with a reduced ovarian reserve. For the selection of controls, frequency matching was used: control couples were selected blinded for outcome, based on treatment center and treatment period in order to obtain an equal distribution in both groups [22]. For this purpose, a chronological overview of PGD treatments performed per PGD center for autosomal dominant and recessive disorders (excluding conditions known for a (potential) effect on ovarian reserve (e.g., fragile X syndrome, myotonic dystrophy type 1) and male BRCA1/2 mutation carriership) was created. Matching was done per PGD center: PGD treatments for female BRCA1/2 mutations were identified, and (if available) four PGD treatments for autosomal dominant or recessive disorders chronologically performed closely before or after the PGD treatment for BRCA1/2 were included as controls. In order to rule out bias from repetitive cycles, only first treatment cycles were included. First cycles with and without oocyte pick-up were included in order to assess the cancelation rate because of poor ovarian response in both groups.

Only treatments in a long gonadotropin-releasing hormone (GnRH) agonist-suppressive protocol, with stimulation with at least 150 IU follicle stimulating hormone (FSH) or human menopausal gonadotropin (hMG) per day, were included in order to obtain a homogenous study population with optimal ovarian stimulation [23]. Other inclusion criteria for both groups were: female age < 43 years, female body mass index (BMI) < 35 kg/m², and female endogenous FSH < 15 IU/l. Exclusion criteria were a history of invasive (breast) cancer up to 2 years prior to IVF/PGD treatment, ovarian surgery, chemotherapy, pelvic radiation, polycystic ovary syndrome that conforms the Rotterdam criteria [24], and known endocrine, autoimmune, or genetic abnormalities (potentially) associated with a reduced ovarian reserve (e.g., fragile X premutation carriers, myotonic dystrophy type 1).

Final oocyte maturation was induced when sufficient dominant follicles were seen at ultrasound (i.e., at least four follicles > 14 mm in the Netherlands and at least three follicles > 17 mm in Brussels). The number of mature oocytes was assessed at the moment of intracytoplasmic sperm injection (ICSI). ICSI was used for fertilization in order to avoid contamination of the zona pellicuda with residual spermatozoa. Embryo biopsy was performed on day 3 after fertilization. Single-cell analysis of the removed blastomeres was performed using multiplex polymerase chain reaction (PCR), as described elsewhere [20, 25, 26]. Data were extracted from medical files.

Diverse studies have been reported on a possible diminished ovarian reserve in BRCA mutation carriers, using different primary outcomes and study designs. Oktay et al. [5] were the first to report a lower yield of oocytes in eight BRCA1, but not in four BRCA2-mutated breast cancer patients. A case-control study by Shapira et al. [6] found no difference in oocyte yield according to BRCA mutation status in 62 BRCA mutation-positive women. However, the inclusion of cancer patients and patients stimulated in different IVF protocols and the lack of clarity regarding minimal stimulation doses applied may have obscured an existing difference.

Previous studies on ovarian reserve in BRCA1/2 mutation carriers using non-IVF-related parameters did not show consistent results. It is challenging, however, to study ovarian reserve in BRCA1/2 mutation carriers because of the presence of several confounding factors in this particular population. Firstly, breast cancer [27, 28] as well as its potential gonadotoxic treatment [29] has a negative effect on ovarian reserve. Secondly, many BRCA1/2 mutation carriers opt for a risk-reducing salpingo-oophorectomy. The timing of this event may be influenced by personal cancer history and related to the menopausal transition. As a consequence, studies on age at natural menopause in BRCA1/2 mutation carriers have important limitations, as set out by van Tilborg et al. [30]. Two studies reported a younger age of natural menopause in both BRCA1 and BRCA2 mutation carriers [8, 9]. A third study found a younger age of menopause in BRCA1 mutation carriers with and without breast cancer [7]. Two other studies did not detect a difference in the age of natural menopause between carriers and non-carriers [30, 31]. Studies were troubled by both the inclusion [7] and exclusion of breast cancer patients [8], the exclusion of women who experienced menopause due to other reasons than natural menopause [8], bias resulting from informative censoring due to risk-reducing salpingo-oophorectomy uptake [9], the inclusion of only few women who had actually reach natural menopause [31], and/or other forms of bias [30]. Three studies have found a lower AMH in BRCA1 mutation carriers and not in BRCA2 mutation carriers [10, 12, 13], while two other studies did not detect a difference between BRCA1 and BRCA2 mutation carriers and controls [11, 32]. Differences in outcome may be the result of variances in study design, in particular the inclusion of breast cancer patients [10] and women with irregular menstrual cycles and/or polycystic ovarian syndrome [11‐13], the lack of appropriate adjustment for potential confounding factors in the analysis [10, 11], and/or power issues [32]. Pregnancy rate and parity in BRCA1/2 mutation carriers were not different from controls [14‐16]. Some studies even report more pregnancies and children born per mother among BRCA1/2 mutation carriers [17, 18].

Our study provides additional evidence for a reduced ovarian reserve in BRCA1 mutation carriers, although the effect size was rather small and the oocyte yield was in the range of a normal response for all subgroups. Consequently, our finding may be more interesting from a biological point of view than relevant for clinical practice. The strengths of our study are (a) the large homogeneous cohort of BRCA1/2 mutation carriers without recent malignant disease [27, 28], (b) the use of the same IVF protocol including only first cycles, and (c) the application of frequency matching, resulting in a representative control group. Our study also has limitations, mainly associated with the retrospective study design although the most important outcome data were complete for all inclusions (supplemental Table 2). Firstly, during the study period, different IVF protocols were used in the participating centers. In order to obtain a homogenous stimulated cohort, we only included couples treated in a long GnRH agonist-suppressive protocol with at least 150 IU gonadotropins per day. This selection led to a smaller cohort than initially powered. Nevertheless, the effect size in the BRCA1 subgroup was large enough to be detectable. Additionally, this strategy may have introduced bias due to the exclusion of expected hyperresponders (treated with lower doses of FSH per day) and the inclusion of an excess of suspected poor responders, since in center 5, this IVF protocol was only the first choice in this subgroup of patients. However, this may have had an effect on both the BRCA and control groups and a sensitivity analysis excluding center 5 did not change the primary outcome. Secondly, since the poor response rate was (non-significantly) higher in the control group, this could have biased our primary outcome. Thirdly, we did not have the opportunity to correct for lifestyle factors (e.g., smoking). Finally, the BRCA1 and BRCA2 subgroups were still relatively small. Consequentially, the absence of an effect of BRCA2 dysfunction on ovarian response may have been the result of insufficient power.

Despite these limitations, our finding of an impaired response to ovarian stimulation in BRCA1 mutation carriers and not in BRCA2 mutation carriers is interesting and confirms several previous studies. The absence of a(n) (detectable) effect of BRCA2 dysfunction on ovarian reserve in most studies may be the result of a true lack of a difference, of insufficient power, and/or of either a later-in-life-occurring or more subtle decline in ovarian reserve, corresponding to the lower risk and higher age at diagnosis of breast and ovarian cancer associated with BRCA2 mutations [33]. Both BRCA genes are involved in DNA double-strand break repair, but their biological functions differ. The association between BRCA1 and BRCA2 and (reproductive) aging is demonstrated by the involvement of the BRCA genes in telomere maintenance: telomeres shorten with age [34, 35]. In human oocytes, DNA double-strand breaks are more prevalent with increasing age, while BRCA1 expression is reduced by then [10]. BRCA1 plays an important role in meiotic spindle formation in mice, and BRCA1 mutant mice had fewer primordial follicles, produced fewer oocytes in response to ovarian stimulation, had a smaller litter size, and showed more DNA double-strand breaks in their oocytes with increasing age than wild-type mice [10, 36]. BRCA2 dysfunction in mice has been associated with insufficient spermatogenesis, a depletion of germ cells in female mice, and a higher frequency of nuclear aberrations in mutant oocytes [37]. However, the involvement of BRCA2 in DNA double-strand break repair is probably less comprehensive than BRCA1 involvement [38]. Consequentially, it can be hypothesized that the effect of BRCA2 dysfunction on ovarian reserve is less powerful than the effect of a BRCA1 mutation and potentially only becomes visible at increasing age.

If BRCA1/2 mutation carriers are affected with a reduced ovarian reserve, this might have several clinical consequences, such as a higher need for fertility treatment, a worse treatment outcome, an urge for more treatment attempts, and/or higher doses of fertility drugs. However, the size of the effect found in our study is probably too small to be of clinical relevance. Future clinical and molecular studies are needed to provide more insight into the role of the BRCA(1) gene(s) in the maintenance of the ovarian pool.

A reduced yield of mature oocytes was found in BRCA1 mutation carriers undergoing IVF/PGD, suggesting a role of the BRCA1 gene in the maintenance of ovarian reserve.