GnRH and LHRgene variants predict adverse outcome in premenopausal breast cancer patients

The diagnosis of breast cancer is made one million times each year worldwide. About one-quarter of these women are premenopausal at time of diagnosis, which is associated with poor prognosis compared to postmenopausal women [1, 2]. It is anticipated that, as a result of changing demographic and lifestyle factors, more and more women will be diagnosed at a younger age with breast cancer [3, 4]. In addition to age and family history, several factors relating to increased or prolonged cumulative estrogen exposure have been identified as important risk factors for breast cancer development and progression [5, 6]. Polymorphic variation in genes regulating estrogen production may partly explain differences in susceptibility, clinical presentation and outcome of breast cancer between individuals or populations [6, 7].

In premenopausal women, estrogens predominantly arise from the ovaries, where production is regulated by the neuro-endocrine system consisting of hypothalamus, pituitary and gonads: the HPG axis. Internal and external stimuli are integrated in the brain, resulting in the pulsatile secretion of the hypothalamic neuropeptide gonadotropin-releasing hormone (GnRH). GnRH reaches the gonadotroph cells in the anterior pituitary through the hypophysial portal circulation, where it stimulates de novo synthesis and secretion of the gonadotropins follicle stimulating hormone and luteinizing hormone (LH), which reach the ovaries in women through the circulation. LH, acting through the LH receptor (LHR), stimulates production of androgen precursors in theca cells that surround antral follicles. Follicle stimulating hormone subsequently regulates the granulosa cell enzyme aromatase, which converts these androgens to estrogens. In turn, ovarian sex steroid and peptide hormones (inhibin A and B) provide negative feedback regulation, either in the pituitary or hypothalamus. The menopausal transition, that is, the cessation of menstrual cycling, is characterized by disruption of this tightly balanced HPG axis system and is accompanied by continuously increased serum LH and follicle stimulating hormone in combination with decreased levels of ovarian sex steroid hormones [8, 9].

We have previously reported, in a training set of 266 Australian breast cancer patients, an association between a common polymorphic CTCCAG (Leu-Gln (LQ)) insertion (LHR insLQ) in the signal peptide of the LHR gene and poor survival [10]. No associations between its ligand, LH, genotype and clinical parameters were observed in this study. In our subsequent validation study on a large independent breast cancer cohort of 751 retrospectively collected Dutch patients with long detailed follow-up, we have confirmed the association of the LHR insLQ gene variant with a shorter disease-free survival (DFS) [11]. Furthermore, we have shown the functional importance of the LQ insertion in the signal peptide, that is, an increased activity for the LHR insLQ variant compared with the LHR non-LQ protein. We hypothesized an ovary-dependent increase in cumulative estrogen exposure that may influence breast cancer outcome in patients with the LHR insLQ genotype [10]. Interestingly, the GnRH gene also carries a common signal peptide polymorphism (Trp16Ser) [12], which has been associated with altered bone mineral density, an indirect marker for estrogen exposure [13].

In line with the hypothesis that possible associations of the above mentioned polymorphisms with outcome would depend on HPG-regulation of ovarian function, we have investigated associations of the LHR insLQ and GnRH 16Ser alleles with premenopausal breast cancer outcome in the present study.

Breast cancer is a heterogeneous and complex disease. Many gene variants, acting in concert with each other and with environmental factors, may influence its susceptibility, prognosis and response to treatment [19]. In light of their possible role in the variability of estrogen exposure, variants of genes involved in the HPG axis are likely candidates to contribute to differences in clinical phenotype and outcome. In the present study we tested this hypothesis. In a genetic association approach we used stratification for ovarian activity and explored cooperative action of two HPG gene variants. We show an association between the LHR insLQ allele and shorter DFS in premenopausal patients, especially in the hormone receptor positive subset. A common GnRH gene variant, GnRH 16Ser, showed an association with lymph node involvement. In addition, coincident carriership of the GnRH 16Ser and LHR insLQ variants, present in almost one-quarter of the patients, resulted in a more than doubled risk of recurrence of disease in (hormone receptor positive) premenopausal patients with a long (>10 years follow-up of still living patients). Multivariate analyses showed that these associations with poor DFS were independent of known prognostic factors.

Whether variants of the GnRH gene differ in function remains to be elucidated. Substitution of Trp16 by the less hydrophobic serine may change the efficiency of the GnRH signal peptide. Using the same in vitro assay for the insLQ signal peptide variant, as described by us previously [11], we were unable to detect a difference in signal peptide efficiency between the GnRH 16Ser and GnRH 16Trp signal peptide constructs (data not shown). Furthermore, in silico analysis of the variants using the program SignalP 3.0 [20] did not result in any difference in signal peptide characteristics. On the other hand, Iwasaki and colleagues [13] have described an association between the GnRH 16Trp allele and higher bone mineral density, considered to be an indirect marker for estrogen activity, in 384 Japanese postmenopausal women; this suggests there is higher estrogen activity in women bearing this variant. In contrast, we observed a significant increased lymph node involvement and shorter DFS in Caucasian breast cancer patients carrying the other variant, the GnRH 16Ser allele, which we hypothesize to result from a higher level of cumulative estrogen exposure. Possible reasons for the apparent conflicting results are numerous, including differences in sample size, technical approach, and ethnicity, or so far unknown differences in the interaction of genetic and environmental factors between Japanese and Caucasian subjects. Consequently, further studies are needed to identify the exact mechanisms of the effect of the GnRH 16Ser polymorphism, including linkage to other polymorphisms in the GnRH gene that may affect regulation of expression.

There are several hypotheses as to how GnRH and LHR gene variants may affect tumor features and clinical outcome in breast cancer as demonstrated in this study. In view of the abundant data on the direct effects of GnRH modulation on sex steroid hormone-dependent cancers, a direct effect of locally produced GnRH via GnRH receptors expressed in breast cancer tissue cannot be ruled out [21‐24]. Local co-expression of mRNAs for GnRH and the GnRH receptor in breast cancer tissue has been shown [25, 26], and direct growth inhibition of cultured breast cancer cells have been reported as well [27]. However, to our knowledge, local production of GnRH has not been shown. Hypothalamic GnRH is unlikely to reach the breast via the peripheral circulation given its low concentration and short half-life [28]. Furthermore, the effects of GnRH agonist treatment regimens are most likely explained by down-regulation of pituitary GnRH receptors and subsequent shutdown of the HPG axis [29]. A few studies have shown LHR expression in normal and breast tumor cells. However, it is less likely that direct effects of LH explain the adverse association of LHR insLQ with DFS, since this was not seen in the postmenopausal patients (data not shown) in whom circulating LH levels are high.

In premenopausal women, epithelial proliferation in the non-pregnant, non-lactating breast is maximal approximately one week after ovulation in the luteal phase of the menstrual cycle. During the luteal phase, which can be considered a risky period for carcinogenesis in the breast, the corpus luteum is the main site of estrogen and progesterone hormone production, which is, in large part, dependent on LHR action. Therefore, increased cyclic hormonal stimulation of early breast cancer in women carrying activating HPG gene variants may enhance dedifferentiation and worsen prognosis. In premenopausal breast cancer patients with advanced disease, reduction of estrogen levels by a GnRH agonist in combination with the ER antagonist tamoxifen improves clinical outcome compared with the use of GnRH agonists or tamoxifen alone [29, 30]. Randomized trials assessing whether the combination of a GnRH agonist with an aromatase inhibitor as adjuvant therapy improves outcome compared to treatment consisting of a GnRH agonist and tamoxifen are ongoing [31]. Regimens for endocrine therapy are still largely empirically based [32]. Recently, it has been suggested that, as for adjuvant systemic therapy, the role of genetic factors in breast cancer treatment outcome should be considered [33]. Our current study identified almost 25% of premenopausal patients with a genetic background associated with clearly significant poor outcome and hypothetically associated with altered treatment outcomes. Exploratory studies in our subset of patients that received adjuvant endocrine therapy (n = 4) were too underpowered to detect differences in response. Finally, it can be hypothesized that the GnRH and LHR variants may play a role in chemotherapy-induced amennorhoea [33], since 110 of the patients studied were treated with chemotherapy. Although adjuvant chemotherapy was equally divided over patients with and without polymorphisms, possible differences in the chance of ovarian failure resulting from chemotherapy may bias the outcome. However, since chemotherapy is likely to inhibit ovarian activity, this is not anticipated.

We have shown a strong and independent association with DFS (HR = 2.1) in almost one-quarter of (hormone receptor positive) premenopausal women carrying LHR insLQ and GnRH 16Ser genotypes. The observations strongly suggest that the adverse outcome in patients with these variants occurs through enhanced HPG-mediated ovarian estrogen production. Prospective studies, including serum estrogen analyses, are needed. When validated in independent studies, the observed results raise the possibility that LHR insLQ and GnRH 16Ser genotyping may provide additional prognostic information for premenopausal breast cancer patients in clinical practice and may result in tailored endocrine treatments for these patients.