SLCO1B1*5 polymorphism (rs4149056) is associated with chemotherapy-induced amenorrhea in premenopausal women with breast cancer: a prospective cohort study

In Germany, 10 % of all newly diagnosed breast cancers are detected in the age group under 45 years old [1]. Patients younger than 50 years of age, especially those with hormone-insensitive breast cancer, achieve significant benefit from adjuvant systemic chemotherapy in terms of prolonged disease-free and overall survival [2]. However, a considerable number of these young patients will suffer from chemotherapy-induced amenorrhea (CIA) thereafter, depending on age at diagnosis and type of chemotherapy used [3].

Premenopausal women in whom amenorrhea developed as a consequence of adjuvant breast cancer therapy had significantly better disease-free and overall survival than did women without amenorrhea, particularly when the tumor was estrogen receptor (ER)-positive [4]. Interestingly, the dose of the chemotherapy drug delivered was not a key factor in accounting for these differences [5]. Apart from its impact on survival and the loss of fertility, CIA due to premature ovarian failure leads to subjective and objective menopausal symptoms, which can negatively affect short- and long-term quality of life [6, 7].

Apart from its relationship with patient age at the time of therapy, there are no proven patient characteristics that are predictive of CIA [3, 8]. Pretreatment anti-Müllerian hormone (AMH), a biomarker of ovarian reserve, appears to be a potential predictor of CIA in women with early breast cancer [9]. However, further studies are needed in order to define the regimen-specific action of chemotherapy on AMH levels, the percentage of post-treatment recovery of AMH levels, and the relationship between menopausal status and AMH [10]. Because inheritance is recognized as playing a role in age at menarche and natural menopause [11, 12], the development of CIA might also depend on inherited genetic factors; however, studies that explore such a correlation are few and have received scant attention [13].

Previous pharmacogenetic investigations on CIA have focused on alkylating-based regimens. Cyclophosphamide is a pro-drug that requires activation by multiple cytochrome P450 (CYP450) enzymes [14]. The effect of cyclophosphamide on ovarian toxicity in patients with systemic lupus erythematosus has been found to be associated with the CYP2C19 genotype [15–17]. In another study women with CYP3A*1B variant genotype demonstrated a higher risk of ovarian failure after cyclophosphamide-based chemotherapy [18]. In conflict with these data, a pilot study investigating six CYP450 genotypes (n = 20) did not reveal a genetic effect on CIA in women with breast cancer receiving anthracycline- and cyclophosphamide-based therapy [19].

Solute carrier (SLC) transporters mediate the influx of cytotoxic drugs into cells. Organic anion-transporting polypeptides (OATPs) encoded by the solute carrier organic anion (SLCO) genes constitute an important transporter superfamily that mediates transmembrane transport of various clinical drugs and endogenous nutrients. Eleven human OATPs with different transport functions are expressed in various tissues. Many important clinical drugs have been identified as substrates of liver-specific OATP1B1 and OATP1B3 [20, 21].

SLCO1B1 encodes the organic anion-transporting polypeptide OATP1B1, which mediates the hepatic uptake of various drugs [22]. Previous studies have shown associations between genetic polymorphisms in SLCO genes and the uptake pharmacokinetics of their substrates. A common single-nucleotide variation (coding DNA c.521 T > C, protein p.V174A, rs4149056) in the SLCO1B1 gene decreases the transporting activity of OATP1B1, resulting in markedly increased plasma concentrations of many statins, for example [23].

Given the importance of this topic we planned a more comprehensive genotype study in a larger cohort of young women with early breast cancer. Our approach tested the effect of variant polymorphisms in drug metabolism enzymes (DMEs) using a predesigned pharmacogenomics panel in premenopausal women receiving contemporary chemotherapy.

On analyzing the 164 assays of the PGx panel with respect to CIA, no statistics were computed for 89 assays (54.3 %) because results of these targets were constant, and 23 further assays (14 %) were excluded from the analysis because their frequencies were not in Hardy-Weinberg equilibrium. The remained 52 assays of interest were assessed for data quality, resulting in 34 finally evaluated assays (20.7 %) with complete and reproducible data sets (Additional file 2: Table S2, Additional file 3: Table S3).

Among this premenopausal study cohort, a recurrence of breast cancer was detected in 10 cases (20 %) during a median follow-up of 61 months (range 18–120 months). Disease-free survival was not related to occurrence of CIA (log-rank test, P = 0.526), SLCO1B1*5 genotype (P = 0.793), or age group (P = 0.985) using Kaplan-Meier method.

CIA is relevant because 25 % of women diagnosed with breast cancer are premenopausal, and previous studies suggest that CIA may impact breast cancer survival and patient’s quality of life due to premature ovarian failure with associated infertility and menopausal symptoms in women younger than 50 years.

Currently, little is known about the genetics of chemotherapy-induced ovarian failure and CIA. It is possible that the same genetic changes that are associated with premature menopause might also predict treatment-induced menopause [13]. A second category of candidate genes might be those that are implicated in drug metabolism. These include transporters involved in absorption and distribution, and in renal or hepatic elimination, and enzymes of drug metabolism that are usually present in the liver or gastrointestinal tract. SNPs in drug transporters, metabolizing enzymes and drug targets have the potential to affect therapeutic efficacy and toxicity [26, 27].

Our study results are a further example of the importance of variant polymorphisms predicting toxicities from chemotherapy in breast cancer patients. Using a pharmacogenomic approach permitting rapid scan of multiple markers, a significant association with CIA in Causasian women was found only for the SLCO1B1*5 allele distribution (rs4149056) (Table 6). In particular, patients who have the homozygous (T/T) genotype of the rs4149056 polymorphism showed a higher risk for developing CIA compared with patients who have the homozygous (C/C) or heterozygous (C/T) genotype.

In a genome-wide study the SEARCH Collaborative Group has identified the rs4149056 C allele in SLCO1B1 as being strongly associated with an increased risk of statin-induced myopathy [28]. No SNPs in any other region were clearly associated with myopathy. Several clinical studies in the past have investigated associations between rs4149056 SLCO1B1 genotypes and statin pharmacokinetics [22]. Although not all yielded significant results, the collective evidence indicates that statin blood concentrations are higher in people with the C allele [28].

Substrates for OATP1B1 include anticancer agents like methotrexate [29], gimatecan, SN-38 (the active metabolite of irinotecan), and pazopanib [23]. Certain SLCO1B1 variants associated with enhanced clearance of methotrexate increase the risk of gastrointestinal toxicity when this compound is used to treat children with acute lymphoblastic leukemia [30].

Endogenous substrates of SLCO1B1 include steroid hormone conjugates such as estradiol-17β-glucuronide and estrone-3-sulfate [31]. In a nested case-control study in the California Teachers Study cohort, only genetic variation in SLCO1B1 was associated with breast cancer risk. There was also an indication that estrogen-progestin therapy (EPT) may interact with SNPs in SLCO1B1; one variant (rs4149013) was significantly associated with breast cancer risk in EPT users [32].

Furthermore, given the frequent occurrence of overlapping substrate specificities – and the fact that functionally important coding-region SNPs in genes encoding members of some transporters families may not be common – it might be most fruitful to focus on combinations of coding-region SNPs in genes encoding transporters that work in parallel or in series [33]. In our data (Additional file 4: Table S4), alongside the described significant SLCO1B1*5 polymorphism in relation to CIA, there was a trend for a second SLC transporter genotype (SLC15A2: rs2257212, rs2293616) to be associated with menopausal status after completion of chemotherapy (P = 0.093).

In addition to the novel described potential genetic risk factor (SNP rs4149056) for CIA, increased age is unequivocally associated with higher rates of CIA. Older women, often defined as those at least 40 years old, are more sensitive to chemotherapy-related ovarian ablation [3]. This can be attributed to the decreasing number of active ovarian follicles with increasing age [34]. Taking the cut-off of 40 years, we found significantly different incidences of CIA in the age group <40 years (30.4 %) compared with the subgroup ≥40 years (70.4 %) when using contemporary chemotherapeutic regimens. Recently, the pharmacogenetic pilot study by Wessels et al. [19] including common variant alleles for CYP450 enzymes found no differences in genotype frequencies but, as expected, women who experienced CIA were significantly older than those who did not (P = 0.009). The risk of CIA in young patients with breast cancer (≤47 years) does not appear to be related to BRCA1 or BRCA2 mutation status [35]. Again, age at treatment and use of tamoxifen were important predictors of CIA in women who carry a BRCA1 or BRCA2 mutation.

The major limitations of the study are the limited case number (n = 50) leading to a reduced statistical power of 70 %, and the inclusion of patients with goserelin treatment. Previous studies have suggested that temporary ovarian suppression with a GnRH analogue may preserve ovarian function both in humans and animal models [36]. However, the clinical data are conflicting. For example, the ZORO study randomised 60 ER-negative patients with breast cancer to anthracycline- and taxane-based chemotherapy alone or in combination with goserelin and found that there was no statistical difference in the resumption of ovarian function between the two groups [24]. This finding is confirmed in recent publications of other prospective trials [37, 38]. Currently, following publication of the Prevention of Early Menopause Study (POEMS) results [39, 40], the use of GnRH agonists during adjuvant chemotherapy outside clinical trials is being re-evaluated. Our data showed a trend for preservation of menstrual bleeding in patients receiving goserelin treatment concurrently with chemotherapy, but the difference was not statistically significant (Table 2). Administration of goserelin sequentially to chemotherapy as endocrine therapy in ER-positive disease had no impact on the occurrence of CIA.

Other methodologic concerns include inconsistencies in defining CIA and variability in the use of chemotherapeutic agents and tamoxifen in heterogeneous study population, depending on hormone-receptor status. However, our data do not support the hypothesis or previous findings that taxane-based chemotherapy or tamoxifen may have an impact on CIA (Table 2). For example, Najafi et al. [41] have reported that the type of chemotherapy is an important risk factor for CIA and that taxane-based regimens induced more CIA than other regimens. Non-cyclophosphamide treated controls or breast cancer patients without chemotherapy were not included in our study.

The lack of a consistent definition across breast cancer literature complicates incidence comparisons between different regimens and studies. The definition of CIA ranges from one missed period within 1 year to continuous cessation of menses for more than 1 year [3]. In summary, irreversible amenorrhea lasting for several (>12) months following chemotherapy and an FSH level of ≥30 IU/L in the presence of a negative pregnancy test seems to be an appropriate characterization of chemotherapy-induced ovarian failure [42]. We used strict criteria to define CIA as no resumption of menstrual bleeding within 12 months after completion of chemotherapy or goserelin. However, cessation of menses is not synonymous with true ovarian failure because estrogen levels can remain in the premenopausal range even where CIA persists for 1 year or longer (as in 10 % of our cases with CIA).

This pharmacogenomics study revealed a genetic effect on CIA in Caucasian women with early breast cancer receiving standard contemporary chemotherapy. The impact on CIA is due to genotype distribution of the SLC drug transporter gene SLCO1B1*5. A prospective validation of this finding is required in a larger cohort of premenopausal women using clearly predefined inclusion criteria (e.g., no treatment with goserelin). These data illustrate the potential power for predicting CIA based on a more profound understanding of inherited genetic factors.