Introduction
Hormone dependence is a fundamental hallmark of the majority of breast cancers, and tumour growth can be inhibited either by deprivation of circulating oestrogens or by antagonising the effect of these hormones on their receptors [
1]. The selective oestrogen receptor (ER) modulator tamoxifen has long been the most commonly used adjuvant therapy for patients with advanced hormone-sensitive breast cancer [
2]. In recent years, however, aromatase inhibitors have become an alternative treatment option for postmenopausal women with breast cancer. An aromatase inhibitor acts by interfering with the enzyme that converts androgens to oestrogen, and reduces tumour and systemic oestrogen concentration [
3]. The third-generation selective aromatase inhibitor anastrozole (Arimidex) reduces serum oestradiol to nanomolar concentrations [
4]. The Arimidex, Tamoxifen, Alone or in Combination (ATAC) trial was designed to compare the efficacy of anastrozole alone or in combination with the established adjuvant treatment, tamoxifen for 5 years, as adjuvant treatment for postmenopausal women with operable breast cancer [
5]. The study demonstrated that the efficacy of anastrozole was higher compared with tamoxifen alone, and also superior to the combination of both agents [
5,
6]. After a median follow-up of 10 years, 5 years after completion of treatment, the significant advantage for anastrozole over tamoxifen as initial adjuvant therapy for postmenopausal, ER-positive breast cancer patients was confirmed [
7].
In breast cancer, genetic alterations such as amplifications and deletions occur within the tumour at high frequencies, and a number of these alterations are closely related to poor clinical outcome. One such region of amplification is 11q13, harbouring the cyclin D
1 gene
CCND1 [
8‐
10]. Cyclin D
1 plays a crucial role as a cell cycle regulator, promoting progression through the G
1-S phase, following complex formation with CDK4/6 and phosphorylation of the retinoblastoma protein [
11]. Various studies have described the oncogenic capacity of cyclin D
1 in vitro, and overexpression
in vivo results in tumour formation [
12‐
14]. Overexpression of cyclin D
1 is observed in approximately 50% of breast cancers [
15,
16], and cyclin D
1 is one of the most commonly overexpressed proteins in this form of cancer. A number of studies report cyclin D
1 overexpression to be a predictor of worse prognosis [
17,
18], while others have found an association with an ER-positive phenotype and a better clinical outcome [
19‐
23]. In about 15% of all primary breast cancers, overexpression is due to amplification of the corresponding gene
CCND1 [
15,
24,
25], and this specific amplification has been linked to poor prognosis [
23,
26].
Despite the presence of ERα, approximately 50% of breast cancers develop resistance to hormonal treatment, a major clinical limitation of breast cancer therapy [
27,
28]. The mechanisms behind this phenomenon have been extensively studied, and imply a complex signalling network governing ER function and interaction with various co-regulators [
29‐
32]. Cyclin D
1 is one such co-factor, known to interact with ERα and, independently of oestrogen, activate the receptor and potentially modify oestrogen/anti-oestrogen responses [
33,
34]. Overexpression of cyclin D
1 has been reported to result in a conformational change in ERα that induces receptor activation in the presence of the novel selective ER modulator arzoxifene, which in turn promotes growth of MCF-7 cells - indicating a change from antagonist to agonist [
28]. This study also suggests that different mechanisms are required to confer resistance depending on the specific anti-oestrogen administered, and that changes in the conformation of ERα play a crucial role in anti-hormonal insensitivity. A similar study demonstrated that overexpression of cyclin D
1 reversed the growth inhibitory effect of tamoxifen in two ER-positive breast cancer cell lines [
35]. In line with these experimental findings we have previously observed that cyclin D
1 overexpression was associated with tamoxifen resistance in premenopausal and postmenopausal breast cancer [
21,
36]. Worryingly, amplification of
CCND1 was further linked to a potentially detrimental effect of tamoxifen in premenopausal breast cancer patients, when compared with randomised control patients not receiving any adjuvant therapy [
36].
The aim of our study was to characterise the association between CCND1 amplification and cyclin D1 protein expression and breast cancer recurrence in a large randomised cohort of postmenopausal patients with ER-positive breast cancer treated with endocrine therapy. In addition, we aimed to assess whether there was a significant difference in response to anastrozole versus tamoxifen according to cyclin D1 gene and protein status, and thereby to address any potentially unfavourable effects of tamoxifen in subgroups of breast cancer defined by CCND1 amplification.
Discussion
Amplification of the
CCND1 gene has been associated with a poor patient outcome in previous studies [
19,
26], whilst controversy regarding overexpression of cyclin D
1 protein in relation to patient survival still exists. Cyclin D
1 has been reported to be a prognostic marker in invasive breast cancer and has been associated with both a less aggressive ER-positive phenotype [
20,
22] and also with an adverse clinical outcome [
18]. These conflicting findings can potentially be explained by the low patient numbers analysed and/or methodological discrepancies. To clarify the importance of cyclin D
1 in breast cancer we therefore analysed the expression of cyclin D
1 in different subcellular localisations, using a previously validated antibody [
36], as well as the gene amplification status by the well-established CISH technique in a large, well-characterised randomised patient cohort including more than 1,000 patients with ER-positive breast cancers. Our data support studies indicating that low cyclin D
1 protein expression as well as
CCND1 amplification are linked to tumour aggressiveness and increased risk of disease recurrence in ER-positive postmenopausal breast cancer [
23,
26]. Similar findings have been observed for HER2, where both high expression linked to amplification and low expression are linked to poor outcome [
43].
Amplification of
CCND1 was observed in 8.7% of the tumours, which is slightly lower than the frequency of 10 to 15% generally reported, even though some groups have demonstrated a lower percentage of
CCND1-amplified tumours [
15,
44,
45]. The slightly lower fraction of
CCND1-amplified cases may be due to all patients being ER-positive or due to methodological differences and different cutoff points for defining amplification between studies. In addition, the use of TMAs has certain limitations; however, this technique is indispensable when analysing large patient materials and is today a well-accepted approach for large-scale tumour sample analysis. In agreement with previous studies, gene amplification of
CCND1 was associated with an overall adverse clinical outcome. The observed positive correlation between nuclear cyclin D
1 expression and tumour grade and proliferation suggests a link between cyclin D
1 and aggressive disease. In contrast, both higher nuclear expression and high cytoplasmic expression of cyclin D
1 was instead associated with a decreased recurrence risk. Despite a positive association between cyclin D
1 protein and
CCND1 amplification status, both low nuclear fraction of cyclin D
1 and
CCND1 amplification were linked to earlier disease recurrence independently of other clinicopathological parameters - hence both factors serve as prognostic markers in endocrine-treated, ER-positive postmenopausal breast cancer. In patients not displaying
CCND1 amplification, cytoplasmic expression of cyclin D
1 was also an independent marker for longer TTR, indicating that the true prognostic value of cyclin D
1 protein expression may be obscured by the
CCND1-amplified cases: the clinicopathological significance of cyclin D
1 expression might thus be best considered separately for amplified and nonamplified cases.
Apart from the role as a prognostic marker, cyclin D
1 has been proposed as a predictive factor for tamoxifen response, as illustrated by poor clinical outcome in patients with ER-positive tumours with high cyclin D
1 expression treated with tamoxifen [
46]. These findings together with numerous experimental reports [
33‐
35] support that cyclin D
1 overexpression might abrogate the response to tamoxifen, as previously reported by our group and others [
21,
46]. Our earlier discoveries have nevertheless been made in cohorts where patients were randomly assigned to receive either no adjuvant treatment or to receive tamoxifen [
21,
36]. In the present study we compared the two endocrine therapies anastrozole and tamoxifen in relation to disease recurrence and cyclin D
1 status, but there were no untreated control patients. There was no significant difference in treatment response between these two adjuvant therapies by stratification for cyclin D
1 status, indicating that cyclin D
1 is not a predictive marker for differences in response to anastrozole versus tamoxifen. No conclusions can be drawn, however, regarding cyclin D
1 as a marker for general endocrine treatment resistance, since no untreated patients were available for analysis within the TransATAC study. Moreover, differences in tamoxifen response in relation to cyclin D
1 in postmenopausal versus premenopausal breast cancer might exist. Our previous study reporting a potential unfavourable effect of tamoxifen included premenopausal patients exclusively, whereas this study focused exclusively on postmenopausal breast cancer cases and has shown no detrimental effect since the results for the two endocrine therapies were similar.
The relationship between cyclin D
1, proliferation and prognosis is quite complex, with a positive correlation between cyclin D
1 and Ki67 - Ki67 is associated with shorter TTR, while cyclin D
1 is associated with longer TTR. Patients showing low expression of Ki67 had a longer TTR with the highest levels of cyclin D
1 expression, whereas patients exhibiting higher levels of Ki67 had the shortest TTR with the lowest levels of cyclin D
1 expression. Multivariate analysis identified the fraction cyclin D
1-positive nuclei as a predictor of outcome independently of other clinicopathological parameters such as Ki67. These results suggest that, irrespective of proliferation status, intermediate to high expression of cyclin D
1 results in a prolonged TTR in ER-positive, postmenopausal breast cancer patients. Similar results were observed in our previous study of randomised material from premenopausal breast cancer patients [
47]. The relationship between cyclin D
1, proliferation and prognosis hence seems to be complex, and this could in part be explained by potential additional functions for cyclin D
1 unrelated to proliferation control, as well as co-amplification and co-deletion of specific genes on chromosome 11q13, the locus harbouring
CCND1.
Competing interests
JC has acted as a consultant and/or on advisory boards for AstraZeneca and received commercial research grants from AstraZeneca. AH has acted as a consultant and/or on advisory boards for AstraZeneca, GSK, Pfizer, Roche and Amgen, and has received honoraria from AstraZeneca. MD receives paid advisory boards and research funding from each of Novartis, AstraZeneca and Roche. The remaining authors declare that they have no competing interests.
Authors' contributions
KL carried out the immunohistochemical and CISH assessments, participated in the study design and drafted the manuscript. MB assisted with evaluation of the immunohistochemistry. SP performed the statistical analyses. JC designed the clinical trial, assembled trial data, and planned and performed the statistical analyses. JS assembled trial data. LZ assembled trial data. AH designed the clinical trial and collected clinical data. MD designed the clinical trial and collected clinical data, and initiated and conceived the current study. GL initiated and conceived the study, supervised interpretation of the data, took part in the immunohistochemical assessments and evaluation of CISH, and helped to draft the manuscript. All authors took part in data interpretation, writing the report and approval of the final version of the manuscript.