Introduction
Improvements in mammography screening now permit the detection of early breast lesions such as ductal carcinoma
in situ (DCIS). These lesions represent the most rapidly growing subgroup of breast cancers and comprise nearly 20% of all newly diagnosed cases of breast cancer. Although DCIS is considered a noninvasive lesion, if it is left untreated, invasive carcinoma will develop in 40% to 50% of DCIS cases [
1]. Thus, effective treatment of DCIS could substantially reduce the incidence of invasive breast cancer. However, developing strategies for DCIS treatment has been difficult, partly because of a lack of an experimentally amenable
in vivo model.
Over the past several years, our group has developed and characterized a mouse model for DCIS that shares biological, morphological, and molecular characteristics with human DCIS [
2,
3]. In addition, this model meets the criteria for mouse premalignant mammary intraepithelial lesions set by the NIH Annapolis Pathology Panel of the Workshop on Mouse Models of Human Breast Cancer [
4]. The definition of mammary intraepithelial neoplasia (MIN) is based on morphological criteria and the biological behavior of the lesions assessed by ectopic and orthotopic transplantation in syngeneic recipient mice. A MIN lesion is incapable of ectopic growth but capable of orthotopic growth (in a gland-cleared mammary fat pad), and this growth has a consistently high rate of transformation to 'malignancy', defined as the ability to grow in ectopic and orthotopic locations.
The model described here was derived by transplanting focal mammary lesions from mouse mammary tumor virus (MMTV) polyomavirus middle T (PyV-mT) transgenic mice to syngeneic wild-type host mammary fat pads. The Tg(
PyV-mT) model is an attractive human breast cancer model not only because of its molecular similarities to the human breast cancer, but also because of the similar morphology and histology to human breast cancer [
5].
PyV-mT is a membrane-bound viral oncogene. The signaling pathways activated by
PyV-mT include those of Ras, Shc, and phosphotidylinositol 3-kinase, which are frequently activated in human breast cancer (reviewed in [
6]) and are also activated by
ErbB2 (
Her2/neu), a receptor tyrosine kinase that is overexpressed in 30% of breast cancer and is associated with poor outcome [
7]. The molecular profile of Tg(
PyV-mT) mouse mammary tumors is more similar to that of
Neu/ErbB2 and
myc transgenic mouse mammary tumors than to other transgenic mouse tumors [
8]. The histology of Tg(
PyV-mT) mammary tumors resembles Tg(
Neu/ErbB2) mammary tumors and human breast cancer much more closely than other transgenic mammary tumors [
9]. Stages of Tg(
PyV-mT) mammary tumor development also recapitulate human breast cancer progression histologically as well as in the expression of biomarkers associated with poor prognosis [
5].
Stable mammary intraepithelial neoplasia outgrowth (MIN-O) lines have been established by microscopically identifying and dissecting premalignant dysplastic foci from Tg(
PyV-mT) mammary fat pads and serially transplanting them into the gland-cleared mammary fat pads of wild-type FVB/N host mice [
2,
10]. The transplanted MIN tissue grows to fill the host mammary fat pad, and after a certain latency period, tumor foci arise within the MIN-Os. The lines have been maintained over several years by serial transplantation of MIN-Os to new host fat pads. Therefore, the MIN-O lines provide the biology of the tumor progression found in the original Tg(
PyV-mT) mammary fat pad in an experimentally reproducible setting.
Estrogen exposure is an important breast cancer risk factor. Seventy percent of breast cancers express the estrogen receptor (ER) [
11]. The absence of ER in breast cancer is associated with poor prognosis. In the Tg(
PyV-mT) model, relatively high numbers of ER-positive cells are found in the early MIN stage, but invasive tumorigenesis is associated with loss of ER [
5]. In a Tg(
PyV-mT) tumor explant model with a low to moderate expression of ER, tumor grew slower in ovariectomized animals than in intact animals, whereas estrogen supplementation stimulated rapid tumor growth, suggesting that this model is sensitive to estrogen level [
12].
In ER-positive human breast cancer, a selective estrogen receptor modulator (SERM), tamoxifen, is typically used in adjuvant therapy. The US Food and Drug Administration (FDA) has approved tamoxifen, and studies have shown that in ER-positive cancer, treatment for 5 years reduces recurrence by 47% and the risk of death by 26% [
13]. Unfortunately, tamoxifen therapy is associated with undesirable side effects, including endometrial cancer, thromboembolic events, and liver cancer (as seen in animal models [
14]) as well as hot flashes, insomnia, vaginal discharge, and vaginal dryness [
15]. Therefore, the search for other SERMs has continued. Recently, third-generation aromatase inhibitors, which block the conversion of androgens to estrogens, have been shown to be potentially more efficacious than tamoxifen [
15], but recently reported side effects, including bone loss, joint pain, and cardiac events, have caused concern [
16].
Therefore, a model that recapitulates human DCIS, with progression to invasive carcinoma with demonstrated ER-positivity and estrogen-dependency would allow for preclinical studies to assess the efficacy of this new generation of SERMs and aromatase inhibitors. Here we have used one transplantable MIN-O line, 8w-B, because it has a relatively short tumor latency, has uniform histopathology, has consistent molecular architecture over multiple transplant generations, and is ER-positive [
3]. To further explore the ovarian-hormone-dependency of the MIN-O line, growth of the MIN-O transplant and tumor incidence were determined after ovariectomy. To demonstrate the utility of this model in antiestrogen therapy, we treated the animals with tamoxifen, the gold standard and FDA-approved adjuvant therapeutic for ER-positive cancer, and a less extensively studied SERM, ospemifene. Ospemifene (FC-1271a; Z-2-[4-(4-chloro-1,2-diphenyl-but-1-enyl)phenoxy]ethanol), is structurally similar to tamoxifen but has a more benign side-effect profile, including proestrogenic in the bone and neutral in the endometrium [
17]. It is currently in phase III clinical trials for the urogenital sequelae, but there are limited data demonstrating its effectiveness in breast cancer. In our study, both tamoxifen and ospemifene treatments resulted in a similar level of suppression of the MIN-O growth and tumor incidence. In addition, Ki-67, a marker for proliferation and a potential efficacy biomarker, showed decreased expression with treatment. Interestingly, we found that tamoxifen treatment decreased the ER status of the MIN-Os while ospemifene treatment did not. This suggests that these two related SERMs act differently and that there may be differential long-term effects. Overall, these studies show that this MIN model can be used for preclinical trials and is a suitable preclinical model for antiestrogen therapy.
Materials and methods
Mice
Standard techniques for mammary gland clearing and transplantation, and the establishment and characterization of the MIN-O line, have been described previously [
2]. Three-week-old FVB/J female mice were purchased from Charles River Laboratories (Wilmington, MA, USA). The surgery and treatments were carried out in the animal housing facility on the University of California Davis campus following the approved procedures.
Ovariectomy study
8w-B premalignant MIN-O tissue (1 mm
3) [
2] was transplanted in the gland-cleared no.4 fat pads of 3-week-old virgin FVB/J female mice. In addition, at the time of transplantation, the experimental group (8 animals) was ovariectomized. The control group (4 animals) received the transplants and sham surgery without ovariectomy. The tumor latency was measured by weekly palpation. The extent of MIN-O growth was assessed at 5 weeks after transplantation by exposing the transplanted mammary fat pad and visually inspecting the transplanted MIN-Os under a dissecting microscope. The animals were palpated weekly until a tumor was detected. The experiment was concluded at day 99.
Chemoprevention study
SERM treatments were prepared by suspending ospemifene (Hormos Medical Corporation, Turku, Finland) or tamoxifen citrate (Orion Corporation, Orion-Pharma, Espoo, Finland) in dimethyl sulfoxide (DMSO) and then diluting with peanut oil to the final concentration of 1% DMSO. Animals received 50 mg/kg of tamoxifen or ospemifene in 100 μl peanut oil daily by oral gavage. These concentrations have been shown to be effective in mice (MW DeGregorio, unpublished results). The control group received 100 μl of 1% DMSO in peanut oil.
Three-week-old gland-cleared FVB/J female mice were treated with tamoxifen, ospemifene, or vehicle control starting 1 week before the MIN-O transplantation, ensuring therapeutic levels of the SERM at the time of transplantation. In the first experiment, animals (n = 4 per group) were sacrificed 3 weeks after transplantation. In the second experiment, animals (n = 9 for tamoxifen, n = 12 for ospemifene, and n = 13 for control) were sacrificed 10 weeks after transplantation. For both experiments, at the time of sacrifice, the MIN-O growth was recorded as gross percentage of area filled in the fat pad (% fat pad filled), and the tumor foci were measured in two dimensions under a stereomicroscope. Each fat pad was fixed in 10% formalin for histological analysis. Tumor incidence was assessed from histological sections of each fat pad. In addition, serum was obtained from each animal for HPLC analysis. At autopsy, each mouse was assessed for any other abnormalities.
Histology
Methods for whole-mount preparation of fixed mammary fat pads and immunohistochemistry on the paraffin-embedded fat pads have been described previously [
2]. The following primary antibodies were used with the VECTASTAIN ABC Elite Kit (Vector Laboratories, Burlingame, CA, USA): rabbit anti-ER (1:600, LabVision, Fremont, CA, USA), anti-Ki-67, anti-cleaved caspase-3 (1:250, Promega, Madison, WI, USA), and anti-PyV-mT (B4Rat7; 1:50; Dr Gernot Walter, UC San Diego).
ER positivity was scored according to the method of Harvey and colleagues [
18]. The frequency of ER-positive cells was scored on a scale of 0 to 5, where 0 = no ER staining, 1 = up to 1% of cells with ER positivity, 2 = 1% to 10%, 3 = 11% to 33% (one-third), 4 = 34% to 66%, and 5 = more than two-thirds of cells with ER positivity. The intensity of staining was scored on a scale of 0 to 3, where 0 = no staining, 1 = weak staining, 2 = intermediate staining, and 3 = strong staining. The final ER score was calculated by adding the frequency score and the intensity score. ER staining in proliferating edge and differentiated center zone was scored separately for each MIN-Os (
n = 17 for control,
n = 16 for ospemifene, and
n = 11 for tamoxifen).
For quantification of Ki-67 and cleaved caspase-3 staining, the positive nuclei were counted and the total epithelial area was measured in Image-Pro PLUS (MediaCybernetics, Silver Spring, MD, USA) from at least four different 40×-objective images from each treatment group. The number of positive nuclei was normalized to the epithelial area of each image.
HPLC analysis
Ospemifene and tamoxifen serum samples were quantified using a previously published assay [
19,
20]. Toremifene citrate (5 μg/ml in methanol (Orion Corporation, Orion Pharma, Espoo, Finland) was used as the internal standard for the ospemifene samples, and nafoxidine hydrochloride (50 μg/ml in methanol; Sigma, St Louis, MO, USA) served as the internal standard for the tamoxifen and control group samples.
Statistical analysis
An unpaired t-test was used for the MIN-O size analysis in the ovariectomy and SERM treatment experiments and immunohistochemistry analysis. For the analysis of tumor latency in the ovariectomy study, a log-rank test was used. For the tumor incidence analysis, Fisher's exact test was used. All statistical analysis was performed using GraphPad Prism version 4.00 for Windows (GraphPad Software, San Diego, CA, USA).
Discussion
The study reported here is to our knowledge the first chemoprevention study to compare the efficacy of two SERMs, using a transgenic mouse transplant model that fulfills the operational definition of premalignancy. We demonstrate that the transplanted premalignant lesions (MIN-Os) are ER-positive and their growth is ovarian-hormone-sensitive. Two related SERMs, ospemifene and tamoxifen, had an inhibitory effect on this mouse model for DCIS on the growth of premalignant mammary lesions and on the tumor incidence. The primary effect of these two agents was on proliferation, while apoptosis was less affected. In comparing the two SERMs, no significant differences were found in the growth of the premalignant lesion, tumor incidence, and markers for proliferation or apoptosis. In contrast, there was a significant difference in the ER status between the two SERM-treated groups. The ER score in the MIN-O lesions was reduced after treatment with tamoxifen but increased by treatment with ospemifene. Overall, these studies provide a foundation for studying and comparing the effects of hormonal manipulation in a chemoprevention setting in an immunocompetent transplantable mouse model for DCIS.
The MIN-O line, 8w-B, originally isolated from dysplastic lesions in the Tg(
PyV-mT) mammary glands, was found to be ER-positive and ovarian-hormone-sensitive in this study. Recent studies with Tg(
PyV-mT) mice have demonstrated that the distribution and number of ER-positive cells change with progression to malignancy [
5]. In our studies, we observed a specific distribution of ER-positive cells in the MIN-Os, showing a higher ER score near the leading edge of the growing outgrowth. Tumors, on the contrary, typically had cells with weak ER positivity randomly scattered throughout the tumor. To functionally assess this observation that the MIN-O is estrogen-positive, the host animals were ovariectomized to study the effect on MIN-O growth and tumor incidence. Ovariectomy significantly reduced the growth of the MIN-O and it significantly increased the tumor latency. This data, coupled to the receptor levels, suggests that MIN-O growth and progression to invasive tumor is ovarian-hormone-dependent, albeit not exclusively. This is reminiscent of human breast cancer, in which oophorectomy decreases recurrence and the incidence of contralateral invasive breast cancer [
22,
23].
Both SERMs exerted an inhibitory effect on the premalignant tissue growth. As each initial transplant contains 1 mm3 of tissue, the results at 3 weeks of treatment show that the transplants are growing, albeit at a much slower rate. In addition, this decreased growth effect is not due to decreased PyV-mT expression in the treatment groups, as the PyV-mT expression was not different between the SERM groups and the control group, based on immunohistochemistry.
The SERMs also inhibited progression to invasive carcinoma. Typically, a high proportion of the MIN-Os will have tumors by 10 weeks after transplantation. The SERM-treated animals showed diminished MIN-O growth with significantly fewer tumors than controls at this time point. The major effect of the SERMs was on proliferation as seen by Ki-67 staining, rather than on apoptosis. In human breast cancer, reduced Ki-67 staining is seen after treatment with SERMs and aromatase inhibitors and has been used as efficacy marker in antiestrogen therapy [
24]. No statistically significant differences in MIN-O growth, tumor incidence, and rates of proliferation and apoptosis were observed between the two SERM treatments. One exception was the effect on ER expression. The ospemifene treatment increased the ER score, whereas tamoxifen decreased it. In ER-positive mammary tumors from Wnt-1 transgenic mice, tamoxifen treatment resulted in significant reduction of ER expression [
21]. In a portion of human tumors, decreased ER levels in tumors after tamoxifen treatment has been shown to predict tamoxifen resistance [
25]. Therefore, these differences in ER modulation between the two related SERMs may suggest different mechanisms of action and may portend different long-term outcomes. This may be reflected in the slight differences in cell proliferation and apoptosis rate between the two treatment groups.
The study detailed here is the first to compare these two related SERMs in a mouse mammary premalignant transplant model and shows that ospemifene has equal effects to tamoxifen in mammary lesions. This is important, because ospemifene is currently in phase III clinical trials for urogenital sequelae but limited data demonstrate its effectiveness in breast cancer. As phase I and II studies have shown it was well tolerated in healthy postmenopausal women, it may offer alternative hormonal options to women with breast cancer or at high risk for breast cancer. In particular, ospemifene is not known to cause menopausal symptoms, such as hot flashes, insomnia, [
26,
27], melancholy, nervousness, dizziness, while it has some proestrogenic effects on the bone [
17] and vaginal tissue [
26] and, unlike other SERMs, does not cause vaginal dryness. Recently, aromatase inhibitors have been shown to provide better chemoprevention to the breast than tamoxifen, but similar to tamoxifen, they have significant side effects including bone loss, muscle and joint sequelae, and cardiac events [
15].
These studies provide evidence that both tamoxifen and ospemifene have effects on decreasing growth and progression in our model of DCIS. Previously, the chemopreventive effects of ospemifene have been studied in a dimethylbenz [a]anthracene (DMBA)-induced rat and mouse tumor models, where it reduced the incidence of mammary tumors [
28]. Although the DMBA-induced models have been utilized by many investigators in chemoprevention studies, they have significant drawbacks in that they can only test the ability of an agent to affect the progression to invasive carcinoma, rather than examining the effects on the early preneoplastic disease. In addition, the tumors that do arise from DMBA treatment are commonly heterogeneous and involve many organs. Moreover, a large portion of breast carcinomas derived in DMBA-treated animals are adenocanthomas, which do not represent or model typical human invasive carcinoma [
29]. The chemopreventive effects of tamoxifen have been studied in various mouse models, including transgenic mice with activated
neu expression. In Tg(
MMTV-neuN) mice, which exhibit estrogen-sensitive tumor development [
30,
31], tamoxifen treatment reduced the mammary tumor incidence and size when the treatment was initiated before subclinical tumors had developed [
32,
33]. More recently, tamoxifen was shown to delay tumorigenesis in an ER-positive Tg(
P53-/-) mammary premalignant transplant model [
34].
The MIN-O model illustrated in these studies offers many advantages over other mouse mammary carcinoma models for chemoprevention studies. In typical transgenic mouse mammary models, tumorigenesis occurs in a multifocal manner, that is, multiple tumor foci develop in a mammary fat pad arising independently and at different starting times. Thus, in a given mammary fat pad, multiple lesions at different stages of tumorigenesis can be seen. Since no two fat pads are the same with respect to the development of the lesions, interpreting the results of chemopreventative interventions can be very complicated. Moreover it may require a significant amount of animals to distinguish the effect of an intervention. In contrast, in the MIN-O model, the proliferation of the 'premalignant' growth begins upon transplantation, and therefore the time to malignant transformation is easily measurable. Chemopreventative interventions can be applied before transplantation, at transplantation, or at a defined time after the time of transplantation. In particular, line 8w-B, a MIN-O line used in this study, has a defined tumor latency period [
2]. This relatively short latency affords the opportunity to perform chemoprevention experiments rapidly. Secondly, the premalignant MIN-O and the invasive tumor mimic the histopathology of, respectively, human DCIS and invasive tumor [
2,
10]. Third, since each experimental subject receives tissue from the same MIN-O, the comparison of the experimental and control groups is less prone to error due to differences in the biological potential of the tissue. Fourth, the outgrowths continue to maintain the same biological characteristics, such as tumor latency, histopathological characteristics, and molecular profiles, over multiple serial transplant generations [
2,
3]. This phenotypic stability affords the opportunity to compare experiments over time, regardless of the transplant generation.
Acknowledgements
The ospemifene and tamoxifen citrate used in this study were kindly provided by, respectively, Hormos Medical Corporation, Turku, Finland, and the Orion Corporation, Orion-Pharma, Espoo, Finland. This work was supported by the following grants: National Cancer Institute grants CA89140-01 (RD Cardiff and JP Gregg) and R01CA81376 (CL MacLeod), Department of Defense grant DAMD 17-03-1-0666 (CL MacLeod), California Breast Cancer Research Program grant 9FB-0212 (R Namba), National Centers for Research Resources grant U42RR14905, and the IF Smith Foundation. We would like to thank the mutant mouse pathology laboratory at UCDavis for histology sample processing and Dr. Robert Munn for assistance with image acquisition and processing.
Competing interests
MD has patents on ospemifene and tamoxifen on subject matter not related to this manuscript and has received research funding within the past 5 years for projects unrelated to this manuscript. The authors declare that they have no other competing interests.
Authors' contributions
RN carried out the SERM treatment studies, assisted with the ovarian ablation study, analyzed the data, and drafted the manuscript. LJTY carried out the transplantation and ovarian ablation surgeries and assisted with all the animal studies. JEM carried out the ovarian ablation study. LJTY and JEM also developed the MIN-O model. ETM carried out the transplantation surgeries and assisted with the SERM treatment studies. SL assisted with the animal studies and the drafting of the manuscript. GTW and MWD participated in the design of the SERM study and carried out the serum analysis. ADB and RDC assisted in the histology analysis and made contributions to the interpretation of data. CLM and RDC conceived of the MIN-O model and applications of the model and contributed to the drafting of the manuscript. JPG conceived and coordinated the study and helped to draft the manuscript. All authors read and approved the final manuscript.