Introduction
Gene amplification is a well defined cause of oncogene activation during tumor development, and some genomic regions are recognized to be more frequently amplified than others [
1].
Amplification of chromosome locus 11q13 occurs at high frequencies in certain human cancers, including lung, bladder, breast and ovarian carcinomas, as well as in head and neck squamous cell carcinomas (HNSCCs) [
2‐
6]. Approximately 15% of primary breast cancers are affected by this specific amplification, which is associated with poor prognosis [
7‐
10]. Four distinct core regions or amplicons within the 11q13 locus have been identified, and these can be amplified independently or concurrently in various combinations [
7,
11]. A number of oncogenes or potential cancer-related genes have been mapped to the 11q13 chromosomal region. The
CCND1 and
CTTN oncogenes have been putatively proposed as candidate genes for the emergence and maintenance of this amplification event in breast cancer [
3,
7]. These genes map to two different amplification cores, located within 0.8 megabases of each other at 11q13.3 [
2,
11], and their co-amplification has been reported in breast cancer [
1,
2,
12]. The core region comprising
CCND1 is the most frequently amplified and is involved in two-thirds of the amplifications. The
CCND1 gene is the most extensively studied gene of the 11q13 amplification region, and encodes the cell cycle regulatory protein cyclin D
1, which is important both for development of mammary tissue and in mammary carcinogenesis [
2].
In breast cancer, amplification and over-expression of cyclin D
1 has been associated with worse prognosis [
13,
14], but high expression of cyclin D
1, in contrast, has also been associated with better prognosis [
15,
16]. Breast cancer patients exhibiting estrogen receptor (ER)-α expression, with concurrent over-expression of cyclin D
1, have been reported not to benefit from treatment with the selective estrogen receptor modulator tamoxifen, which is in contrast to the evident response in ER-α-positive breast cancers with moderate and low cyclin D
1 expression [
15]. Jirstrom and coworkers reported that
CCND1 amplification was associated with a potential agonistic effect of tamoxifen in ER-α-positive premenopausal breast cancer patients, even when not accompanied by protein over-expression [
17].
INT2,
FADD,
PAK1, and
EMSY are other candidate genes reported to be included in the 11q13 amplicon [
7] and thier amplification or protein over-expression has been associated with a poor prognosis in various cancers [
1,
2,
8,
9,
18‐
20].
It has been reported that in HNSCC amplification of 11q13 involves a loss of distal chromosome 11q (from 11q14.2 to 11qter) through a breakage-fusion-bridge cycle mechanism [
21]. In this process, genes with important roles in, for instance, the DNA damage response are lost in the deletion step preceding the amplification. Frequent allelic deletions at chromosome 11q24–q25 have been reported in both breast and ovarian cancer and have been associated with a worse clinical outcome [
22,
23].
A potential adverse effect of tamoxifen in CCND1 amplified breast cancers is indeed intriguing. As noted, over-expression of cyclin D1 protein has been linked to lack of tamoxifen response but not to a direct agonist effect. Hypothetically, another gene co-amplified with CCND1 might be responsible for the agonistic effect of tamoxifen. Furthermore, genes deleted in the 11q13 amplification event might also affect breast cancer outcome and treatment response.
In order to elucidate the importance of 11q-associated genes with regard to tamoxifen response and CCND1 amplification, we identified previously described candidate genes at the 11q chromosomal region for further investigation, using array comparative genomic hybridization (CGH) analysis of breast cancer samples. Protein expression levels of the various cancer-related gene products were then analyzed in a tissue microarray from a randomized trial of premenopausal breast cancer patients receiving 2 years of adjuvant tamoxifen treatment or no adjuvant treatment. By comparing treated and untreated patients, we were able to delineate the response to tamoxifen irrespective of prognostic features, and a potential agonistic or diminished effect of tamoxifen could be identified. Furthermore, the tumor material allowed us to study prognostic features, relations with different clinicopathological parameters, and associations with CCND1 amplification, in different subgroups defined by the expression of the 11q13 and distal 11q gene products. The results indicate that many 11q13-associated gene products are over-expressed in conjunction with cyclin D1 but are not linked to an agonistic effect of tamoxifen. Conversely, deletion of the distal end of chromosome 11q, defined as downregulation of the marker Chk1 (checkpoint kinase 1), was associated with an impaired tamoxifen response, and with low proliferative breast cancer of low grade.
Materials and methods
Comparative genomic hybridization
Array CGH was performed essentially as was previously described [
24]. Raw data and normalized data are available through National Center for Biotechnology Information Gene Expression Omnibus [GEO: GSE12759].
Patient materials
Between 1986 and 1991, a total of 564 premenopausal breast cancer patients with invasive stage II disease were enrolled in a Swedish trial (SBII:2a), in which they were randomly assigned to 2 years of adjuvant tamoxifen (
n = 276) or no adjuvant treatment (control;
n = 288). The aim of the original study was to compare 2 years of tamoxifen treatment (20 or 40 mg/day) versus no adjuvant treatment. Patients were included irrespective of hormone receptor status. All patients were followed up for recurrence-free survival (RFS) and overall survival. Recurrence was defined as local, regional, or distant recurrence, and breast cancer-specific death, whereas contralateral breast cancer was excluded. Surgery was modified radical mastectomy or breast conserving surgery, followed by radiotherapy and, in a few cases, adjuvant polychemotherapy (in <2% of cases). The time of surgery defined time point zero in this study. The patient median follow-up time without breast cancer event was 13.9 years. Detailed description of the SBII:2a study design can be further viewed in a previous report [
25]. Informed consent was obtained from the patients for the initial randomized study, and the ethics committees at Lund and Linköping Universities that approved the study did not require additional consent for the present study.
Tissue specimens and immunohistochemistry
Formalin-fixed and paraffin-embedded tumor material was available from 500 of the 564 patients in the trial. Areas representative of invasive cancer were selected and assembled in a tissue microarray. Two 0.6 mm tissue cores from each donor block were placed in recipient paraffin blocks by using an automated tissue arrayer (Beecher Instruments Microarray Technology, Woodland, MD, USA). Sections (4 μm) from this block were mounted onto slides before they were deparaffinized, rehydrated, and microwave treated in target retrieval solution pH 9.9 (Dako, Glostrup, Denmark), before undergoing processing in an automated immunostainer (Techmate 500; Dako, Copenhagen, Denmark), using the Envision software (Dako, Glostrup, Denmark). The antibodies used were mouse monoclonal anti-human cortactin (1:50, clone 30; BD Biosciences, Erembodegem, Belgium), mouse monoclonal anti-human FADD (Fas-associated death domain; 1:50, clone A66-2; BD Biosciences), and mouse monoclonal anti-human Chk1 (1:100, clone 2G1D5; Cell Signaling, Danvers, MA, USA). For Chk1, both nuclear staining intensity and fraction positive nuclei were evaluated. The variable designating fraction Chk1 positive nuclei was best suited for describing the appearance of Chk1. Staining was evaluated by two independent observers (one pathologist), in order to obtain a result as correct and representative as possible. Conflicting observations were low (<5%) for all three evaluations made. All immunohistochemical (IHC) evaluations were performed without knowledge of tumor characteristics. In cases of no evaluation, cores were either nonrepresentative (contained no invasive tumor cells) or missing.
Data for expression of ER-α (a combination of IHC and enzyme immunoassay) were available from a previous study, in which ER-α positivity was assessed according to the Swedish clinically established cutoff of 10% positively stained nuclei [
25].
Data for expression of Pak1 (p21-activated kinase 1) [
26], cyclin D
1, and
CCND1 gene amplification status (done by fluorescence
in situ hybridization [FISH] analysis) [
17] were also available. When the ratio of intensity of the
CCND1 probe to the centromere probe was greater than 1 in at least 20% of the tumor cells, the gene was considered to be amplified. In addition, expression of the proliferation marker Ki67 had also been evaluated in a previous study [
27].
Chromogenic in situhybridization
Chromogenic
in situ hybridization (CISH) was performed in accordance with the Zymed SPoT-Light Cyclin D1 Probe protocol, which is well suited to CISH [
28], using the SPoT-Light Cyclin D1 Amplification Probe (Zymed laboratories, Invitrogen immunodetection; San Francisco, CA, USA). Pretreatment procedures included heat pretreatment and enzyme digestion to optimize the CISH performance.
Cell lines, Western blot, and immunocytochemistry analyses
The human breast cancer cell lines CAMA-1, MCF-7, T-47D, MDA-MB-468, and MDA-MB-231 (ATCC, Manassas, VA, USA) were used to verify the reactivity of the cortactin, FADD, and Chk1 antibodies, by Western blot and immunocytochemistry (ICC). For detailed culturing conditions, and ICC and Western blot analyses, we refer to the methods described by Holm and coworkers [
26]. MCF-7 cells were grown in Improved MEM (Minimum essential media) zinc option (Gibco, Grand Island, NY, USA) supplemented with 5% fetal bovine serum, and all culture media were supplemented with 1% penicillin/streptomycin. For ICC, an array of these cell lines was constructed and stained with the cortactin, FADD, and Chk1 antibodies separately.
For Western blot, 20 μg of each protein sample was resolved on SDS-polyacrylamide gels and transferred to Hybond ECL nitrocellulose membranes (Amersham Pharmacia Biotech, Amersham, Buckinghamshire, UK). Membranes were incubated with cortactin (1:1,000), FADD (1:250), Chk1 (1:1000), and polyclonal goat anti-human β-actin (1:500; Santa Cruz, Biotechnology, Santa Cruz, CA, USA) antibodies for 2 hours, followed by incubation with secondary horseradish peroxidase-conjugated anti-mouse (Amersham Life Science, Aylesbury, UK) and anti-goat antibodies (Sigma, Gothenburg, Sweden) for 1 hour. Membrane-bound antibody was detected by using the ECL+ system (Amersham Life Science).
Statistical methods
Statistical analyses were performed using SPSS software (version 15.0; SPSS, Chicago, IL, USA). Fisher's exact test was employed to determine the statistical significance of associations between cortactin, FADD, cyclin D1, and Pak1 protein expression, and CCND1 amplification. The Spearman's rank-order correlation coefficient (ρ), Kruskal-Wallis, and the Wilcoxon/Mann-Whitney tests were used for associations with clinicopathological parameters. To study RFS, the Kaplan-Meier method was used, and the log-rank test was applied for comparison of RFS survival among different treatment groups. A Cox proportional hazards regression model was used for the estimation of relative risk in univariate analysis. All P values corresponded to two-sided tests, and a P value less than 0.05 was considered statistically significant.
Discussion
The amplification event at chromosome locus 11q13 has in a number of different cancers been associated with unfavorable prognosis [
6,
7]. Several genes included in this region have been identified as driver genes, or most frequently amplified genes [
3,
7,
18]. These driver genes are putative promoters of biologic processes such as oncogenesis and multidrug resistance [
18,
29].
CTTN has been proposed to be a strong candidate gene driving 11q13 amplification [
7], and clear evidence for involvement of the F-actin binding gene product cortactin in tumorigenesis have been reported [
1,
3,
30]. FADD has been implicated in cell survival as well as growth control, and consequently it may play a role in tumor progression [
31]. Furthermore,
FADD has been reported to be a candidate driver gene for the 11q13 amplification in HNSCC [
18].
In this study, the search for candidate genes co-amplified with CCND1 and responsible for impaired tamoxifen response was conducted by assessing the expression of two proteins with corresponding genes at the 11q13 locus, and one gene harbored at the distal chromosome 11q, namely CHK1. Importantly, the protein expression should not be seen as an exact reflection of the amplification event, but a certain conformity of amplification and expression level is expected. Protein over-expression is not exclusively caused by amplification, but it can be the result of several different genetic alterations. The co-variance between the expression of cortactin, FADD, cyclin D1, and Pak1 might be interpreted as a reflection of the amplification event. Over-expression of more than one of the four 11q13 proteins was highest in subgroups expressing high levels of both cortactin and FADD, and might be explained by the chromosomal location, where the CTTN and FADD genes are located in close proximity. The concurrent protein over-expression with the lowest overlay was between cyclin D1 and Pak1, which may be due to the location of these two genes on either side of region harboring CTTN and FADD. The positive correlation between expression of cortactin, FADD, cyclin D1, and Pak1 further suggests a possible link to co-amplification of the different core regions at the 11q13 locus. However, no correlation between cyclin D1 and Pak1 was observed, and this further confirms that these two genes may not be co-amplified to the same extent as genes in closer proximity.
In contrast, CCND1 amplification was positively correlated with expression of all four proteins, indicating a link between amplification of the four genes. Of the CCND1 amplified tumors, between 28.9% and 44.0% over-expressed one of the four proteins, indicating that amplification of CCND1 might be of importance for amplification of any of the other three genes we examined at the 11q13 locus. The fraction of Pak1 over-expressing tumors was (in the CCND1 amplified subgroup) again the lowest of the four 11q13 proteins, indicating a lower frequency of co-amplification of CCND1 and PAK1.
Chk1 is one of the key regulatory components of the DNA damage checkpoint and its corresponding gene has been mapped to 11q24, and was used as our marker for the deletion occurring at distal 11q. The inverse correlation between Chk1 protein expression and
CCND1 amplification might be interpreted such that
CCND1 amplification also involves a deletion of distal chromosome 11q, and in this case a loss of heterozygosity (LOH) of the
CHK1 gene, resulting in a lower level of protein expression. Mapping of
CHK1 to this chromosomal region of frequent LOH in human tumors indicates that this gene is a putative tumor suppressor gene [
32]. The positive correlation between Chk1 protein expression and tumor grade, tumor type, tumor size, and Ki67 expression defines Chk1 as a marker for tumor aggressiveness. The subgroup exhibiting low Chk1 expression could hypothetically have a defective DNA damage response, resulting in a more aggressive tumor. However, when analyzing prognostic features of Chk1 in untreated premenopausal breast cancer patients, we did not observe any link between Chk1 and increased recurrence rate.
The definition of Chk1 expression in relation to proliferation characterized a subgroup of Chk1 low-expressing tumors exhibiting a high proliferation rate, excluding false-negative tumors. Within this subgroup, only the inverse correlation with amplification of CCND1 was observed, suggesting an accurate classification and a better representation of CHK1 deleted tumors.
This tumor material has been used in several different studies thus far, and extensive sectioning has led to a significant number of missing tumor cores. When analyzing the tumors that were missing for the 11q genes (102 out of 514), there was a tendency toward over-representation of lobular cancers and cancers of low grade and low proliferation, but there was no difference in breast cancer recurrences.
Because the overlay between CGH and the IHC analyses consisted of only 56 tumors, the statistical power of these analyses was limited. However, the significant correlation between the gene and protein expression of PAK1/Pak1 indicates that an association between the expression of the other 11q genes and their protein products is also likely. As previously described, the correlation between gene and protein expression was confirmed for CCND1/cyclin D1 with the FISH data available. No correlation between gene and protein expression for CHK1/Chk1 was found. Using the definition of Chk1 as normal or deviant did not reveal any further information about the association between the CGH and IHC data.
Tamoxifen resistance is commonly encountered in breast cancer therapy, with approximately one-third of ER-α-positive breast cancers resistant to the drug. Hence, clarification of the underlying cause of resistance could prove vital in augmenting treatment strategies. Because we previously found amplification of CCND1 to be associated with an adverse effect of tamoxifen in premenopausal patients, it may be expected that over-expression of cyclin D1 protein would show the same trend. During the amplification event at chromosome locus 11q13, more than one region can be amplified, meaning CCND1 may be expressed in addition to several other genes. It is entirely plausible that one such gene may be responsible for the adverse effect of tamoxifen.
In this cohort of randomized premenopausal breast cancer patients, we observed that – with high expression of cortactin and low expression of Chk1 – there was a tendency toward an impaired response to tamoxifen. Interestingly, patients with tumors exhibiting Chk1 expression defined as deviant clearly did not have an improved RFS when treated with tamoxifen.
Because over-expression of either cyclin D1 or Pak1 impaired the tamoxifen response in the patient cohort, it was of interest to investigate the influence of combined over-expression of cyclin D1 and Pak1 on the outcome of tamoxifen response. However, only eight patients with ER-α-positive breast cancer exhibited combined over-expression, limiting the statistical power of this analysis. In clinical material including a larger number of patients, this type of analysis might reveal pertinent information regarding combined over-expression of cyclin D1 and Pak1, and also for cyclin D1 over-expression in combination with deviant Chk1 expression.
The underlying mechanism for tamoxifen resistance in subgroups of tumors exhibiting deviant Chk1 expression is unclear. In the case of cyclin D
1 and Pak1, the resistance mechanism could be explained by both cyclin D
1 and Pak1 potentiating the activity of the ER-α [
33‐
35]. Notably, Chk1 was used in this study as a marker for distal 11q deletion, but we can not state whether loss of
CHK1 alone or deletion of a larger chromosomal region of 11q is the important event in breast cancer that leads to decreased tamoxifen response. Nevertheless, there is a multitude of gene products that are potentially downregulated in conjunction with a loss of the distal end of 11q, and it is unlikely that loss of
CHK1 is the only important event on 11q.
Climent and coworkers [
36] reported that deletion of 11q in node-negative breast cancer was associated with earlier relapse in patients not receiving anthracycline-based chemotherapy, as compared with patients receiving this kind of treatment. The distal part of chromosome 11q harbors a number of genes that are involved in DNA repair, and hence a deletion of this region might contribute to defective repair machinery. Two major genes that are involved in DNA repair and cell cycle control are the ataxia-telangiectasia mutated (
ATM) gene and the
CHK1 gene [
37,
38]. These two genes, together with a number of additional ones involved in DNA repair, have been proposed as candidate targets for the 11q deletion [
39,
40].
Several genetic alterations that differed between the tamoxifen-sensitive breast cancer cell line MCF-7 and the tamoxifen-resistant clone CL-9 were identified in a study based on CGH analyses [
41]. One of the alterations that was seen exclusively in the tamoxifen-resistant cell line was the deletion of 11q24, indicating that genes that are involved in development of tamoxifen resistance are potentially harbored in this chromosomal region. The mechanism of endocrine responsiveness in breast cancer is thought to be controlled by complex interactions between steroid hormones and numerous signaling pathways, such as growth factor signaling, which in turn most likely can be affected by different genetic alterations [
42]. The identification of genes responsible for, or involved in, tamoxifen resistance is one approach to clarify the underlying mechanisms in this intricate series of events. To date, few reports have dealt with genetic alterations and anti-estrogen resistance; it is therefore of great importance to continue the search for possible markers involved in this elusive field of breast cancer biology.
Presence of the progesterone receptor (PR) is considered to be an indicator of a functional ER, and thus determines the extent of the response to hormonal therapy [
43]. The PR gene maps to chromosomal region 11q22–23 [
44] and is consequently likely to be involved in the LOH occurring at distal 11q, where the whole end telomeric to 11q14 commonly is lost. In a previous report by Stendahl and coworkers [
45] it was observed that expression of the PR was a stronger predictor of tamoxifen response than the ER. This indicates that the mechanism responsible for loss of this receptor gene is important and, although CGH analyses for this particular genetic event is beyond the scope of the present study, it suggests a productive area for future research. Interestingly, we observed an association between expression of PR and Chk1 defined as normal or deviant, with borderline significance (
P = 0.054; data not shown).
When considering candidate genes responsible for altered disease outcome in different cancers related to 11q13 amplification, the deletion of distal 11q should clearly be considered an event that might be equally important for disease progression and tamoxifen response. Even though a number of genes involved in tumorigenesis have been proposed so far, further studies investigating LOH of genes at distal 11q would be needed to characterize the main candidates with roles in tumor behavior and treatment response. The biological mechanism that interconnects the protein expression of Chk1 and reduced tamoxifen sensitivity needs to be explored. It is apparent that future studies are necessary to determine the intricate mechanisms underlying tamoxifen resistance, and to elucidate the cause of events rendering CCND1 amplified premenopausal breast cancers not only resistant but possibly stimulated by this selective ER modulator.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
KL carried out the IHC assessments, performed the statistical analyses and drafted the manuscript. KH performed the CGH analyses, interpreted the data from these analyses, and revised the manuscript. BN contributed to the planning and performance of the clinical trial and revised the manuscript critically. ÅB supervised the CGH study design and revised the manuscript. GL participated in the study design and interpretation of the data, took part in the IHC assessments, and helped to draft the manuscript. All authors read and approved the final manuscript.