Background
Breast cancer is one of the leading causes of cancer morbidity and mortality among women worldwide [
1]. The majority of breast cancer patients have tumors that express hormone receptors [
2,
3] and can thus be offered endocrine therapy such as tamoxifen and aromatase inhibitors (AIs). However, endocrine resistance is a major obstacle to optimal treatment effect [
4]. Several genetic markers for tamoxifen response have been proposed, although no consensus has yet been reached [
5‐
11]. For AIs, data on genetic markers are sparse [
10‐
13]. The response rates to AIs vary between 35 and 70 % in the neoadjuvant setting [
4,
14,
15] and may be lower in advanced disease [
16]. By identifying mechanisms of resistance as well as treatment predictive factors, patients may be offered more effective personalized medicine and be spared side-effects of ineffective treatment [
17].
Only a few studies have investigated the association between polymorphisms in Cytochrome P450 (CYP)
CYP19A1 (aromatase) and disease-free survival in breast cancer [
10,
18,
19]. There are currently only a few studies published with a proposed polymorphism for predicting AI response in the adjuvant setting, and these have contradictory results [
11,
13]. Some studies have investigated the impact of
CYP19A1 polymorphisms on treatment response in the metastatic- [
20] and in the neoadjuvant settings [
21,
22]. However, the results have been inconsistent. Therefore, it is currently unknown whether single nucleotide polymorphisms (SNPs) in
CYP19A1 are associated with a risk of early events in patients treated with AI as first line treatment.
The formation and metabolism of estrogens in the steroidal sex hormone metabolism is complex and involves several enzymes. In addition to CYP19A1, some examples include CYP1A1, CYP1A2, COMT, and CYP3A4 [
23]. Several of these enzymes are also involved in the metabolism of AIs [
24,
25]. Furthermore, AIs interfere with some of these enzymes; letrozole has been shown to inhibit CYP2A6 and CYP2C19 in vitro [
26], anastrozole has been shown to inhibit CYP1A2, CYP2C9, and CYP3A in vitro [
27], and exemestane has been shown to be metabolized by CYP4A11 and CYP1A1/2 in vitro [
28]. Polymorphisms in the corresponding genes may be a mechanism behind primary (
de novo) resistance of AI as estrogens are known risk factors for recurrence of breast cancer and the enzymes that metabolize estrogens are tightly linked to AI metabolism. Two of these genes involved in estrogen metabolism,
CYP1A1 and
CYP1A2, share a common promoter [
29] and are under regulatory control of the aryl hydrocarbon receptor (AhR) [
30]. These genes may therefore be of interest to study in relation to AI response.
To find new markers beyond the candidate genes for AI resistance, it might be useful to expand the search to other known genes involved in Absorption, Distribution, Metabolism, and Elimination (ADME-related genes). High-throughput, drug metabolism enzymes and transporters (DMET™) chips genotype several SNPs at the same time [
31]. The Affymetrix DMET Plus Premier Pack includes 1931 SNPs in 227 genes in ADME-related genes on a single array. We hypothesized that SNPs in the aromatase gene
CYP19A1 and SNPs in other genes for drug and estrogen metabolism may be used as treatment predictive markers for adjuvant treatment with AI in primary breast cancer patients. The aim of the study was: 1) to perform an exploratory analysis using the DMET™ chip to find new treatment predictive markers in a subset of the cohort and 2) to examine these potential markers with a special focus on
CYP19A1 in relation to a risk for early events in the extended cohort of AI-treated breast cancer patients.
Discussion
The present study investigated the association between SNPs in ADME-related genes and the risk of early breast cancer events in AI-treated patients with primary breast cancer. The main finding was that CYP1A2 rs762551 was significantly associated with risk of early breast cancer events among AI-treated patients with ER+ tumors, both in the exploratory analysis and in the extended cohort. This suggests that CYP1A2 rs762551 may be a predictive marker for early AI-response. To the best of our knowledge, this has not been reported before.
The DMET™ chip was selected because the included SNPs are involved in genes of importance for drug metabolism and transportation. This approach increases the chance that a finding is of biological relevance for AI response. The cut-off for the
P-value in the DMET™ analysis was chosen to allow for identification of potentially new candidate genes while keeping the number of false positive findings low. As this was an exploratory analysis of nearly 2000 SNPs, a Bonferroni correction would have been too stringent and the risk for false negative findings substantial. The
CYP1A2 rs762551 was the only SNP that met the predetermined cut-off and the enzyme is involved in the metabolic pathways of AIs or is inhibited by AIs [
25,
27,
28], which increases the chance that the finding may be of biological relevance.
CYP1A2 is a phase I pathway for drug metabolism and elimination [
40]. An in vitro study reported a significant role of CYP1A2 in exemestane metabolism [
28]. Moreover, CYP1A2 catalyzes the conversion of estradiol to hydroxylated metabolites, primarily 2-hydroxylated estradiol [
23], which has been shown to act as a weak or even as an anti-estrogenic substance [
41]. In a subset of 59 patients in the current cohort, the
CYP1A2 rs762551 C-allele was associated with a low 2OHE-to-16alphaOHE1 plasma ratio both pre- and post-operatively [
42]. However, none of these patients were treated with AIs at the time of blood draw. Since AIs block estrogen formation, it is unlikely that there are measurable estrogen metabolite plasma levels in the 201 AI-treated patients.
The
CYP1A2 rs762551 is located in intron 1 of the
CYP1A2 gene and carries a -163C > A substitution. CYP1A1/2 expression is regulated by the AhR and a number of transcription factors and might be influenced by transcriptional coactivators and corepressors [
43]. The A/A genotype of
CYP1A2 rs762551 is highly inducible especially by smoking [
44] and coffee consumption [
45]. Neither smoking nor coffee consumption accounted for the association between
CYP1A2,
AhR, and risk for early events (data not shown). Furthermore, all multivariable models were adjusted for smoking. While the
CYP1A2 rs762551 has been shown to influence inducibility, it has not been shown to significantly alter the gene expression [
46]. The results are conflicting as to whether the SNP influences CYP1A2 enzyme activity [
43,
46]. Aklillu et al. have performed extensive characterization of
CYP1A2 genotype phenotype correlations [
47]. Cell transfection experiments showed that there was no significant difference in the constitutive transcriptional activity depending on the
CYP1A2 rs762551 SNP. Further, electrophoretic mobility shift assay analysis could not identify any specific transcription factor whose binding could be affected by rs762551. However, a xenobiotic response element (XRE) containing an invariant CACGC core sequence, recognized by AhR, is present in
CYP1A2 intron 1 further downstream of the rs762551 site [
47]. In the current study, a multiplicative association between having at least one
CYP1A2 rs762551 C-allele and at least one
AhR Arg554Lys A-allele on the risk for early events in AI-treated patients was observed. Helmig et al. reported that the
AhR A-allele (Lys) confers lower expression of AhR compared to the G-allele (Arg) [
48]. Further, there is cross-talk between AhR and ERα. An animal rat model showed that ligand-activated AhR confers anti-estrogenic effects partly due to lower ERα levels in ductal epithelial cells [
49]. CYP1A2 is mainly expressed in liver cells but has also been detected in the ER+ breast cancer MCF-7 cell-line after induction [
30]. In the current study, neither
AhR nor
CYP1A2 were associated with prognosis among the patients with ER+ tumors who had not been treated with AIs but either received tamoxifen or no endocrine treatment (data not shown). This suggests that the association of
AhR Arg554Lys and
CYP1A2 rs762551 on prognosis may be exclusive for the AI-treated patients where the ER is still open as opposed to tamoxifen-treated patients where the ER is blocked. The cross-talk between AhR and ER signaling may be one mechanism behind these findings. Taken together, this suggests that
AhR G/G carries may have both lower ERα levels and more effective
CYP1A2 transcription and expression. This may be especially pronounced in patients with the highly inducible
CYP1A2 A/A genotype, since AhR regulates
CYP1A2 expression, thus leading to a lower risk for early events during AI-treatment.
In addition to AhR, other c
is- or
trans acting loci may also regulate
CYP1A2 gene expression levels [
50]. The CYP1A2 enzyme activity level and gene expression is clustered with CYP2C8, CYP2C9, and CYP3A4 [
51]. Further,
CYP1A2 share a common promoter with
CYP1A1 [
29]. However,
CYP1A1,
CYP2C8,
CYP2C9, and
CYP3A4 were not significant in the analysis based on the DMET™ chip data, therefore no further analyses was performed here.
While the role of
CYP1A2 rs762551 with respect to breast cancer risk seems weak or non-significant [
52], unless coffee consumption was taken into account [
34,
53]. A meta-analysis showed that the association between the
AhR Arg554Lys and breast cancer risk differs between studies of women with different ethnicities, although the overall result was no association [
54]. The
CYP1A2 rs762551 is associated with the metabolism of several drugs and also with efficacy and toxicity [
43]. Although the mechanism behind the finding of the present study is not fully understood, the current study provides new insight into how the
CYP1A2 rs762551 combined with the functional
AhR Arg554Lys variant is linked to prognosis in AI-treated breast cancer patients. Moreover, AhR appears to be involved in the regulation of
CYP19A1 both in ovarian and adrenocortical cells via different mechanisms [
55]. Further mechanistic and translational studies of the AhR and CYP1A1/2 signaling pathway with respect to
CYP19A1 regulation and AI-response are therefore warranted.
In the present study, a special focus was placed on
CYP19A1 since aromatase is the target of AIs. However, genotypes, haplotypes, and diplotypes of
CYP19A1 SNPs rs4646, rs10046, Aro1, and Aro2 were not significantly associated with a risk of early breast cancer events in the AI-treated patients. These results are in line with the recent study by of Leyland-Jones et al. [
11] that investigated
CYP19A1 tumor genotype data in relation to endocrine treatment response in the BIG 1–98 trial, but in contrast to a previous study on genomic
CYP19A1 genotypes regarding the risk of recurrence in AI-treated patients [
13]. In the recent abstract by Umamasheran et al., an increased risk for breast cancer recurrence was observed among 191 Indian letrozole-treated T/T carriers of rs4646 [
13]. As mentioned in the background, several SNPs in
CYP19A1 have previously been associated with AI response in neoadjuvant and metastatic settings. The rs4646 is among the most frequently studied SNPs. The A-allele has previously been associated with longer time to progression in the metastatic setting [
20] but with poor response in the neoadjuvant setting [
22]. The differences may be due to ethnicity, different study types, and small study populations. Furthermore, different types of AIs may yield different results. The majority of the patients included in the present study had received anastrozole. A recent study by Lunardi et al. reported no association between plasma estrone concentrations during treatment with letrozole and rs4646 or rs10046 [
56]. Since circulating estrone levels may influence the risk for early events in breast cancer, this is in line with the findings of no significant association between these SNPs and early events in the present study.
In the present study, there was a significant interaction between CYP1A2 rs762551 and CYP19A1 rs4646 with a high risk for breast cancer events, especially within five years, among patients with any C-allele of CYP1A2 rs762551 and C/C-carriers of CYP19A1 rs4646. This subgroup was quite small, and the results need to be interpreted with caution. There was no linkage between CYP19A1 rs4646 and AhR Arg554Lys that could explain the results (data not shown). All findings in the present study warrant validation, preferably within a randomized clinical trial. In such a trial, it would be possible to elucidate whether these genotypes are associated with AI-response. If so, this could guide selection of endocrine treatment for more personalized medicine in the clinical setting.
The Skåne University Hospital in Lund has a catchment area that includes almost 300,000 inhabitants. This study is population-based since patients with breast cancer diagnoses are not referred to other hospitals for surgery. The vast majority of the patients who are diagnosed in Lund are Swedes, but no data on ethnicity was collected. Thus, studies with different study populations are warranted. Since the subset of the cohort that was analyzed with the DMET™ chip also was part of the extended cohort from which they originated, all findings warrant validation in an independent cohort. Further, the follow-up period was relatively short—patients with ER+ tumors tend to relapse later [
57]. Therefore, the long-term effects of
CYP1A2 rs762551 or
CYP19A1 rs4646 could not be evaluated. However, the main impact of these SNPs was observed during the 5-year period when endocrine treatment is administered. This suggests that these SNPs may be involved in primary rather than acquired resistance.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
MS has been involved in acquisition of data, analysis and interpretation of data, in drafting the manuscript, and has revised the manuscript critically for important intellectual content. SV performed statistical analysis of DMET™ chip data and has revised the manuscript critically for important intellectual content. AM has been involved in acquisition of data, analysis and interpretation of data, and in revising the manuscript critically for important intellectual content. CI and CR have been involved in conception and design of the study, analysis and interpretation of data, and have revised the manuscript critically for important intellectual content. HJ has been involved in acquisition of data, in conception and design of the study, analysis and interpretation of data, in drafting the manuscript, and has revised the manuscript critically for important intellectual content. All authors have read and approved the final version of the manuscript.