Abstract
Background: Cell culture studies have disclosed that the mitotic Aurora kinase A is causally involved in both tamoxifen and aromatase inhibitor resistant cell growth and thus may be a potential new marker for endocrine resistance in the clinical setting.
Material and methods: Archival tumor tissue was available from 1323 Danish patients with estrogen receptor (ER) positive primary breast cancer, who participated in the Breast International Group (BIG) 1-98 trial, comparing treatment with tamoxifen and letrozole and both in a sequence. The expression of Aurora A was determined by immunohistochemistry in 980 tumors and semi quantitively scored into three groups; negative/weak, moderate and high. The Aurora A expression levels were compared to other clinico-pathological parameters and outcome, defined as disease-free survival (DFS) and overall survival (OS).
Results: High expression of Aurora A was found in 26.9% of patients and moderate in 57.0%. High expression was significantly associated with high malignancy grade and HER2 amplification. High Aurora A expression was significantly more frequent in ductal compared to lobular carcinomas. We found no significant association between Aurora A expression and DFS or OS and no evidence of interaction between Aurora A expression and benefits from tamoxifen versus letrozole.
Conclusions: Aurora A expression in breast tumors was associated with high malignancy grade III and with HER2 amplification. A trend as a prognostic factor for OS was found in patients with high Aurora A expression. No predictive property was observed in this study with early breast cancer.
Background
Adjuvant endocrine therapy has improved survival of estrogen receptor α (ER) positive breast cancer patients, but recurrent disease is found in 19.1 and 22.7% of the patients at 10 years follow-up after five years of treatment with aromatase inhibitor or tamoxifen, respectively [Citation1]. Thus, new markers are required to identify the ER positive breast cancer patients who may need supplementary treatment, e.g., tailored biological treatment.
Studies of cell culture models mimicking ER positive breast cancer resistant to the antiestrogens tamoxifen and fulvestrant and also to aromatase inhibitors have disclosed that Aurora kinases are important for growth of both antiestrogen and aromatase inhibitor resistant breast cancer cells [Citation2–4]. Aurora kinases (A, B and C) are key regulators of mitosis and multiple signaling pathways [Citation5,Citation6]. Gene amplification and protein overexpression of Aurora kinases have been found in both hematologic malignancies and solid tumors and deregulation of Aurora kinases has been linked to tumorigenesis [Citation7]. Aurora A is consistently associated with cancers and Aurora B may also contribute to tumorigenesis, whereas the role of Aurora C is not yet clarified [Citation8].
Aurora A has been found to be a prognostic marker indicating poor prognosis in ER positive node-negative breast cancer [Citation9] and Aurora A outperforms other proliferation markers including the surrogate tissue-based proliferation marker Ki-67 [Citation10,Citation11].
Screening analyses with a library of 195 kinase inhibitors identified the dual Aurora A and B inhibitor JNJ-7706621 as a preferential and efficient inhibitor of a panel of tamoxifen and aromatase inhibitor resistant breast cancer cell lines derived from the ER positive breast cancer cell lines MCF-7 and T47D by long term treatment with the endocrine agent [Citation2,Citation3]. siRNA (small interfering RNA) mediated specific knock-down of Aurora A and B revealed that Aurora A, but not B, was important for growth of the tamoxifen resistant breast cancer cells. Furthermore, sensitivity to tamoxifen treatment was restored in tamoxifen resistant cell lines with siRNA mediated knock-down of Aurora A or by treatment with the Aurora A/B inhibitor JNJ-7706621 [Citation2], indicating a causal role for Aurora A in tamoxifen resistance. The finding that tamoxifen resistant breast cancer cell lines express increased level of Aurora A compared to their parental cell lines [Citation2,Citation12] and that ectopic overexpression of Aurora A renders ERα positive breast cancer cell lines less sensitive to tamoxifen treatment via phosphorylation of ERα, suggest that high expression level of Aurora A may be a marker for tamoxifen resistance [Citation12]. In agreement, our pilot study with 244 patients, who received adjuvant tamoxifen therapy, disclosed that high tumor expression level of Aurora A was associated with reduced disease-free survival (DFS) [Citation2]. Today, treatment with aromatase inhibitor is the recommended adjuvant endocrine therapy for postmenopausal breast cancer patients and high expression level of Aurora A may also be a marker for reduced benefit from adjuvant aromatase inhibitor treatment, as our cell culture model studies have shown that Aurora A and also Aurora B are important for growth of aromatase inhibitor resistant breast cancer cells [Citation3]. Therefore, in this study, we have evaluated whether high expression level of Aurora A is associated with shorter DFS and overall survival (OS) in 980 Danish postmenopausal breast cancer patients who have participated in BIG 1–98, a randomized phase III clinical trial comparing adjuvant endocrine therapy with tamoxifen, the aromatase inhibitor letrozole or sequential tamoxifen and letrozole. We have also investigated whether Aurora A is a predictive marker for response to treatment with tamoxifen and with letrozole.
Material and methods
Patients
Archival formalin-fixed and paraffin-embedded primary tumor tissues from Danish breast cancer patients participating in the international randomized double-blinded clinical phase III trial, Breast International Group trial 1-98 (BIG 1-98), were included in the study. All patients gave written, informed consent before inclusion in the study.
In the BIG 1-98, ER-positive, early breast cancer patients were randomized to five years of tamoxifen or letrozole monotherapy, or sequential treatment, two years tamoxifen or letrozole followed by three years of letrozole or tamoxifen [Citation13]. In total, 1396 Danish patients were randomized in BIG 1-98 between 1998 and 2003 and primary tumors from 1323 patients were available for tissue microarray (TMA) preparation as previously described [Citation14]. The study was conducted according to the Helsinki declaration and approved by The Danish National Committee on Biomedical Ethics in Denmark in 1997 (KF02 1 178/97) and an addendum approved in 2004 (KF 12 – 142/04). The Danish Data Protection Agency also provided an approval (RH-2015-166).
Immunohistochemical (IHC) analysis and evaluation
TMAs were constructed from formalin-fixed paraffin-embedded tumor blocks, with two cores of 2 mm tissue from each tumor. Immunohistochemical staining of Aurora A was performed with the procedure described in our pilot study [Citation2]. The Aurora A primary antibody (Cell Signaling Technologies, Danvers, MA,USA, 4718) was selected for its usefulness in both western blot analysis and immunohistochemistry. The specificity of the antibody was documented by knocking-down of the expression of the 48 kDa Aurora A protein in human breast cancer MCF-7 cells treated with Aurora A specific silencing RNA constructs, as shown in Thrane et al. [Citation2]. In the present validation study, the enhancement system was improved. Antigen retrieval was performed in a microwave oven for 15 min in Envision FLEX Target Retrieval Solution High pH (50x) (DAKO Denmark A/S, Glostrup, Denmark Cat. no K8004), Aurora A primary antibody was diluted 1:100 and applied over night at 4 °C. High Definition Detection HRP Polymer System (AH Diagnostics A/S, Tilst, Denmark, Cat. no 954D-30) was used for enhancement and staining was performed with DAB substrate (DAKO, Cat. no K3468). Nuclei were counterstained with hematoxylin before mounting in pertex.
Aurora A expression in tumor cells was evaluated in both cores from each tumor with a modified Allred procedure [Citation15], including scoring of the fraction of Aurora A positive tumor cells and of Aurora A expression intensity. Regarding the fraction of Aurora A positive tumor cells, three levels were defined (1 point: 0–10%; 2 points: 11–50% and 3 points: 51–100% of tumor cells), as well as three levels of Aurora A expression intensity (1 point: low; 2 points: medium and 3 points: high). These two parameters combined resulted in a three-tiered score (0–2 points: negative or weak; 3–4 points: moderate and 5–6 points: high; ). Cores with less than 100 tumor cells were excluded from the analysis.
Figure 1. Representative pictures showing immunohistochemical staining for Aurora A: negative/weak, moderate and high staining.
Statistical analysis
All clinical data were collected and monitored by the International Breast Cancer Study Group (IBSCG) Data Management Center (Buffalo, NY, USA) and the Danish Breast Cancer Cooperative Group (DBCG). Baseline data including data on follow-up until October 2010 [Citation16] were provided by the IBCSG datacenter. The statistical analyses were conducted at DBCG by the author MJ.
Follow-up time was quantified in terms of a Kaplan-Meier estimate of potential follow-up. Kaplan-Meier plots were used to illustrate the primary end-point DFS, defined as the time from randomization to the earliest of any of the following events: recurrence of the disease at a local, regional or distant site; a new invasive cancer in the contralateral breast; a new secondary non-breast cancer or death without a previous cancer event. Secondary end-point was OS, defined as the time from randomization to death, irrespective of cause of death. Time to event outcomes DFS and OS were analyzed according to the intention to treat principle (ITT). Follow-up in the sequential treatment arms was censored at two years, i.e., at the time of scheduled treatment change.
Baseline characteristics were compared using the two-sided Fisher’s exact test (excluding unknown) or Wilcoxon rank sum test. The associations of Aurora A expression and time to event endpoints, DFS and OS, were analyzed by Kaplan-Meier estimates, and statistical significance was estimated by log-rank test stratified by two- or four-arm random assignment option and treatment arm (tamoxifen vs. letrozole). Hazard ratio (HR) estimates were obtained from analysis by the Cox proportional hazards ratio model. Multivariate analysis included age at randomization, tumor size, histological type and grade, nodal status, ER and HER2 status and random assignment. The HER2 normal population was used to estimate treatment effect. Level of statistical significance was set to 5%.
Results
Aurora A expression and correlation to other tumor markers
In this study, archival primary tumor tissues from 1323 of the 1396 Danish breast cancer patients enrolled in the BIG 1-98 study were investigated. By central assessment, 1244 patients were ER-positive (>1%). Aurora A data were available from 980 of the ER-positive patients. The exclusion of 416 patients was mainly because of lack of tumor samples or too few tumor cells in the TMA-cores.
Clinico-pathological parameters for included and excluded patients are presented in . The excluded patients had a higher frequency of small tumors (0–20 mm) than the included patients (54.3 vs. 46.7%). The estimated median potential follow-up time was nine years. 353 patients experienced a recurrence and 252 had died. In the letrozole arm, slightly but significantly more patients were included (p = .04) as compared to the other treatment arms (tamoxifen alone, tamoxifen followed by letrozole and letrozole followed by tamoxifen, data not shown).
Table 1. Patient characteristics in study population and patients excluded from the total cohort of Danish patients in BIG 1-98.
High Aurora A expression was found in 26.9% of the study population, moderate expression in 57.0% (). High expression of Aurora A was more frequent in invasive ductal carcinomas than in lobular carcinomas (p < .0001). The majority of tumors with high malignancy grade III and with HER2 amplification expressed high Aurora A level (p < .0001). Tumors with intermediate ER scores (Allred 3–6) were less frequently expressing high Aurora A compared to tumors with high ER expression (p < .0001). With regard to expression of Aurora A and treatment arms, the patients were well balanced (data not shown).
Table 2. Aurora A’s association with other prognostic variables.
Aurora A as a prognostic factor
The association of Aurora A expression and DFS and OS is shown by Kaplan-Meier estimates . DFS at five year follow-up was 77.6% (95% CI 70.3–83.4) for negative or weak expression, 81.0% (95% CI 77.5–84.0) for moderate expression and 76.3% (95% CI 70.7–81.0) for high expression; the corresponding numbers for OS were 85.9% (95% CI 79.4–90.5) versus 89.8% (95% CI 87.0–92.0) versus 85.8% (95% CI 81.0–89.5). No significant association was found between Aurora A expression and DFS (HR 1.22 (CI 0.96–1.55), p = .28) whereas OS was borderline significant (HR 1.41 (1.07–1.87), p = .06; ) for patients with high Aurora A expression compared to the moderate group. Restricting the patient population by excluding patients with amplified HER2 revealed no statistical significant association between Aurora A expression and DFS and OS (p = .51 and .13; ). Adjusted analysis (multivariate analysis) included ER status (Allred scores 3–6 vs. 7–8), tumor grade, lymph node status, histological type, age and tumor diameter as covariates and neither the whole group of patients nor the HER2 normal subpopulation disclosed significant difference ().
Figure 2. Kaplan-Meier survival curves demonstrating DFS (A) and OS (B) for patients with high, moderate and negative/weak Aurora A expression. p values are procured from log-rank testing.
Table 3. Association between Aurora A expression and disease-free survival (DFS) and overall survival (OS).
Aurora A as a predictive marker for treatment with letrozole versus tamoxifen
In the group of patients with HER2 normal expression, no statistical significant difference in DFS (HR 0.95 (CI 0.72–1.25), p = .70) or OS (HR 1.03 (CI 0.73–1.44), p = .89) was found in relation to treatment with letrozole versus tamoxifen (). Treatment effect heterogeneity (letrozole versus tamoxifen) according to Aurora A expression levels negative/weak, moderate or high, did not show significant effects, p = .20 and .89 for DFS and OS, respectively.
Table 4. Treatment effect of letrozole versus tamoxifenTable Footnote a.
Discussion
In this large subset of BIG 1-98, we were unable to demonstrate a prognostic or predictive effect of Aurora A expression. Our results are in contrast to previous results including a prior pilot study from our group [Citation2,Citation9–12,Citation17]. Our pilot study included both HER2 positive and negative patients [Citation2] and in the BIG 1-98 patients, we found a trend for significance of Aurora A expression as a prognostic marker for OS in the univariate analysis including all patients, p = .06. However, when we omitted patients with amplified HER2, no significant association between Aurora A expression and DFS or OS was observed. High Aurora A expression was associated with high tumor grade (grade III compared to I) and high Aurora A level was more frequent in ductal compared to lobular tumors (29.0 vs. 13.9%). The higher frequency of high Aurora A expression in ductal carcinomas is in agreement with a study with 1359 ER positive breast tumors, in which high Aurora A expression was of found in 27% of the ductal and in 12.8% of the lobular tumors [Citation10]. Our finding of more frequently high Aurora A expression in HER2 positive tumors compared to HER2 negative tumors was seen also in other studies [Citation10,Citation18]. The association between Aurora A expression and the independent variables; histological type, tumor grade and HER2 status may explain the lack of significance in the multivariate analysis.
The BIG 1-98 study included 8010 patients with early breast cancer, and reduction in recurrence and mortality was obtained by letrozole monotherapy compared to tamoxifen monotherapy [Citation16]. We have analyzed tumors from a subgroup of only the Danish patients who had participated in the BIG 1-98 study. Overall there was no significant treatment effect comparing tamoxifen and letrozole. This may be due to more extensive cross-over from tamoxifen to letrozole in Denmark when it became evident that letrozole was superior to tamoxifen. Our analysis for potential treatment effect heterogeneity according to Aurora A expression level did not reveal significant difference. Of note, centrally reviewed Ki-67 labeling index of 2685 primary tumors from the BIG 1-98 study receiving monotherapy disclosed that treatment benefit from letrozole versus tamoxifen was greater among patients with high Ki-67 than among patients with low Ki-67 [Citation19].
Several studies have shown that the mitotic Aurora kinase A is a poor prognostic marker in breast cancer [Citation9–11,Citation17,Citation18,Citation20], and we found borderline significance for reduced OS in this group of breast cancer patients with early breast cancer treated with adjuvant endocrine therapy. Aurora A is not only a cell cycle regulator but is also involved in many different signaling cross-talks within the cells [Citation5–7], including activation of the ER via phosphorylation [Citation12]. Our finding in this study which included 980 early breast cancer patients showed that Aurora A did not show significance as prognostic or predictive marker for response to tamoxifen or to letrozole. This indicates that Aurora A overexpression alone may not be sufficient to render breast cancer cells resistant to adjuvant endocrine therapy. It should be mentioned that our cell culture models have been established from the human breast cancer cell lines MCF-7 and T47D, both of which are established from breast cancer metastases. MCF-7 and T47D cells respond to treatment with tamoxifen and with aromatase inhibitor, but few cells survive treatment. Resistant cell lines have been established from colonies of surviving cell, which have acquired the ability to grow in presence of either tamoxifen or an aromatase inhibitor [Citation2,Citation21,Citation22]. Thus, the cell culture models in which Aurora A has a causal role for resistant cell growth mimic patients with advanced disease who after an initial response to the endocrine therapy regress during treatment. This study indicates that the mechanisms, which render early breast cancer resistant to treatment may differ from the mechanisms involved in acquired resistance. Determination of Aurora A expression in primary breast tumors from patients with advanced disease and in the corresponding metastases developed after an initial response to treatment with either tamoxifen or an aromatase inhibitor may clarify whether high Aurora A expression is associated with acquired resistance.
In summary, we have found high Aurora A expression in 26.9% of the 980 breast tumors included in this study. Aurora A expression was associated with ductal carcinomas, high malignancy grade and HER2 amplification. A trend was found for decreased OS in patients with high Aurora A expression. In this study with early breast cancer, Aurora A was not a predictive marker neither for response to tamoxifen nor to letrozole. We suggest that high expression of Aurora A may be specific for acquired resistance to treatment with tamoxifen and with aromatase inhibitor.
Acknowledgments
We kindly acknowledge the participating patients, investigators, study nurses and thank the Danish Pathology Departments for supplying paraffin embedded tumor material: Aalborg Sygehus, Aarhus Hospital, Bispebjerg Hospital, Esbjerg Sygehus, Gentofte Hospital, Herlev Hospital, Hillerød Hospital, Hjørring Sygehus, Holstebro Sygehus, Hvidovre Hospital, Nykøbing Falster sygehus, Naestved Sygehus, Odense Hospital, Randers Sygehus, Rigshospitalet, Roskilde Sygehus, Skive Sygehus, Slagelse Sygehus, Svendborg Sygehus, Sønderborg Sygehus, Vejle Sygehus and Viborg Sygehus. Thanks to Annette Bartels and Nils Brünner from Section for Molecular Disease Biology, University of Copenhagen for the help rendered regarding immunohistochemical analysis.
Disclosure statement
No potential conflicts of interest were disclosed.
Additional information
Funding
References
- Early Breast Cancer Trialists’ Collaborative Group: Aromatase inhibitors versus tamoxifen in early breast cancer: patient-level meta-analysis of randomized trials. Lancet. 2015;386:1341–1352.
- Thrane S, Pedersen AM, Thomsen MBH, et al. A kinase inhibitor screen identifies Mcl-1 and Aurora kinase A as novel treatment targets in antiestrogen-resistant breast cancer cells. Oncogene. 2015;34:4199–4210.
- Hole S, Pedersen AM, Lykkesfeldt AE, et al. Aurora kinase A and B as new treatment targets in aromatase inhibitor-resistant breast cancer cells. Breast Cancer Res Treat. 2015;149:715–726.
- Larsen SL, Yde CW, Laenkholm AV, et al. Aurora B is important for antiestrogen resistant cell growth and a potential biomarker for tamoxifen resistant breast cancer. BMC Cancer. 2015;15:239.
- Otto T, Sicinski P. Cell cycle proteins as promising targets in cancer therapy. Nat Rev Cancer. 2017;17:93–115.
- Sasai K, Treekitkammonkol T, Kai K, et al. Functional significance of Aurora kinase-p53 protein interactions. Front Oncol. 2016;6:247.
- Saeki T, Ouchi M, Ouchi T. Physiological and oncogenic Aurora-A pathway. Int J Biol Sci. 2009;5:758–762.
- Dar AA, Goff LW, Majid S, et al. Aurora kinase inhibitors – rising stars in cancer therapeutics? Mol Cancer Therap. 2010;9:267–278.
- Siggelkow W, Boehm D, Gebhard S, et al. Expression of Aurora kinase A is associated with metastasis-free survival in node-negative breast cancer patients. BMC Cancer. 2012;12:562.
- Ali HR, Dawson SJ, Blows FM, et al. Aurora kinase A outperforms Ki67 as a prognostic marker in ER-positive breast cancer. Br J Cancer. 2012;106:1798–1806.
- Yamamoto S, Yamamoto-Ibusuki M, Yamamoto Y, et al. A comprehensive analysis of Aurora A; transcript levels are the most reliable in association with proliferation and prognosis in breast cancer. BMC Cancer. 2013;13:217.
- Zheng XQ, Guo JP, Yang H, et al. Aurora-A is a determinant of tamoxifen sensitivity through phosphorylation of ERα in breast cancer. Oncogene. 2014;33:4985–4996.
- Thurlimann B, Keshaviah A, Coates AS, et al. A comparison of letrozole and tamoxifen in postmenopausal women with early breast cancer. N Engl J Med. 2005;353:2747–2757.
- Henriksen KL, Rasmussen BB, Lykkesfeldt AE, et al. Semi-quantitative scoring of potentially predictive markers for endocrine treatment of breast cancer: a comparison between whole sections and tissue microarrays. J Clin Pathol. 2007;60:397–404.
- Allred DC, Harvey JM, Berardo M, et al. Prognostic and predictive factors in breast cancer by immunohistochemical analysis. Mod Pathol. 1998;11:155–168.
- Regan MM, Neven P, Giobbie-Hurder A, et al. Assessment of letrozole and tamoxifen alone and in sequence for postmenopausal women with steroid hormone receptor-positive breast cancer: the BIG 1-98 randomised clinical trial at 8.1 years median follow-up. Lancet Oncol. 2011;12:1101–1108.
- Nadler Y, Camp RL, Schwartz C, et al. Expression of Aurora A (but not Aurora B) is predictive of survival in breast cancer. Clin Cancer Res. 2008;14:4455–4462.
- Cirak Y, Furuncuoglu Y, Yapicier O, et al. Aurora A overexpression in breast cancer patients induces taxane resistance and results in worse prognosis. J BUON. 2015;20:1414–1419.
- Viale G, Giobbie-Hurder A, Regan MM, et al. Prognostic and predictive value of centrally reviewed Ki-67 labeling index in postmenopausal women with endocrine-responsive breast cancer: results from Breast International Group trial 1-98 comparing adjuvant tamoxifen with letrozole. J Clin Oncol. 2008;26:5569–5575.
- Xu J, Xing W, Zhou W, et al. Aurora-A identifies early recurrence and poor prognosis and promises a potential therapeutic target in triple negative breast cancer. PLoS One. 2013;8:e56919.
- Thrane S, Lykkesfeldt AE, Larsen M, et al. Estrogen receptor α is the major driving factor for growth in tamoxifen-resistant breast cancer and supported by HER/ERK signaling. Breast Cancer Res Treat. 2013;139:71–80.
- Hole S, Pedersen AM, Hansen S, et al. New cell culture model for aromatase inhibitor-resistant breast cancer shows sensitivity to fulvestrant treatment and cross-resistance between letrozole and exemestane. Int J Oncol. 2015;46:1481–1490.