Sex Steroid Hormone Receptor Expression Affects Ovarian Cancer Survival12

Abstract

Background and Aims: Although most ovarian cancers express estrogen (ER), progesterone (PR), and androgen (AR) receptors, they are currently not applied in clinical decision making. We explored the prognostic impact of sex steroid hormone receptor protein and mRNA expression on survival in epithelial ovarian cancer. Methods: Immunohistochemical stainings for ERα, ERβ, PR, and AR were assessed in relation to survival in 118 serous and endometrioid ovarian cancers. Expression of the genes encoding the four receptors was studied in relation to prognosis in the molecular subtypes of ovarian cancer in an independent data set, hypothesizing that the expression levels and prognostic impact may differ between the subtypes. Results: Expression of PR or AR protein was associated with improved 5-year progression-free (P = .001 for both) and overall survival (P b .001 for both, log-rank test). ERα and ERβ did not provide prognostic information. Patients whose tumors coexpressed PR and AR had the most favorable prognosis, and this effect was retained in multivariable analyses. Analyses of the corresponding genes using an independent data set revealed differences among the molecular subtypes, but no clear relationship between high coexpression of PGR and AR and prognosis. Conclusions: A favorable outcome was seen for patients whose tumors coexpressed PR and AR. Gene expression data suggested variable effects in the different molecular subtypes. These findings demonstrate a prognostic role for PR and AR in ovarian cancer and support that tumors should be stratified based on molecular as well as histological subtypes in future studies investigating the role of endocrine treatment in ovarian cancer. Translational Oncology (2015) 8, 424–433 Lund University Hospital Research Foundation and an unrestricted educational grant from the Swedish Society for Gynecologic Oncology sponsored by Roche. Funding agencies have no influence on the study design, data collection or analysis, or manuscript writing. Received 30 May 2015; Revised 10 September 2015; Accepted 15 September 2015 © 2015 The Authors. Published by Elsevier Inc. on behalf of Neoplasia Press, Inc. This is an open access article under theCCBY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 1936-5233/15 http://dx.doi.org/10.1016/j.tranon.2015.09.002 Translational Oncology Vol. 8, No. 5, 2015 Sex Steroid Hormone Receptor and Ovarian Cancer Jönsson et al. 425 Introduction Epithelial ovarian cancer accounts for about 3% of female cancers and is the leading cause of death from gynecologic malignancy. Although a somewhat decreased incidence and slightly improved survival have been noted during the last decades, the majority of tumors are diagnosed at advanced stages and the relative 5-year survival is less than 50% [1]. New treatment concepts have shown promising results in clinical trials, but predictive markers are needed for refined therapeutic strategies and will likely need to be stratified in relation to histopathological and molecular subtypes of ovarian cancer [2,3]. Endocrine factors play key roles in ovarian cancer development, with risk reduction related tomultiparity and use of oral contraceptives [4,5]. Likewise, estrogen (ER), progesterone (PR), and androgen (AR) receptors represent prognostic markers and therapeutic targets in, e.g., breast cancer and prostate cancer [6–8]. Estrogen regulates growth and differentiation in the normal ovaries and has been demonstrated to have mutagenic effects. Progesterone, on the other hand, induces apoptosis and decreases cell membrane permeability, leading to decreased invasive potential. Progesterone may however stimulate growth at low concentrations, whereas higher concentrations seem to have growth inhibitory effects [9,10]. The majority of ovarian cancer cases are diagnosed in perimenopausal and postmenopausal women [1]. After menopause, when the estradiol level decreases, androgens are still produced and also seem to influence ovarian cancer development. Androgens promote cell proliferation, and androgen levels are decreased by the use of oral contraceptives [11]. Although the majority of ovarian cancers express ER, antiestrogen treatment has not been successful in ovarian cancer. Several studies have assessed the prognostic value of endocrine receptor expression in ovarian cancer, concluding that expression of PR is prognostically favorable, whereas the results on ERα and ERβ are contradictory. Likewise, the association with other clinical risk factors is variable [12–17]. A review andmeta-analysis by Zhao et al. including in total 35 studies, of which 23 considered the prognostic value of ER, did not find any evidence of an effect of ER on prognosis [18]. Recently though, a multinational study including almost 3000 women with invasive epithelial ovarian cancer showed that ER expression was associated with improved disease-specific survival in endometrioid tumors, whereas PR expression was prognostic in serous tumors [19]. Furthermore, AR expression has been suggested to be associated with a favorable prognosis in serous ovarian tumors and is hypothesized to predict response to antiandrogen treatment [20,21]. Overall, the frequency of ER, PR, and AR expression seems to decrease with increasing malignant potential in ovarian tumors, but reports regarding covariation of the receptors are contradictory [12,16,20,22,23]. In general, however, studies of endocrine responsiveness in ovarian cancer are limited by the relatively small number of cases in each study and have not yet led to clinical application. Ovarian cancer is a highly heterogeneous disease, and increasing evidence suggests that the different subtypes respond differently to targeted treatments and also that prognostic and predictive biomarkers may be subtype specific. In addition to the histopathological classification of ovarian cancers, gene expression profiling has revealed intrinsic molecular subtypes with additional prognostic information [24–26]. Apart from outlining the prognostic value of ER (both the α and β isoforms), PR, and AR in serous and endometrioid ovarian cancer, we aimed to explore the potential additional effect of coexpression of two or more of these receptors. We also sought to compare the immunohistochemical findings with the mRNA levels of the genes encoding each receptor in relation to the previously published molecular subtypes of ovarian cancer, using an independent data set and hypothesizing that the expression and prognostic impact of the genes encoding the sex hormone receptors may vary between the molecular subtypes. To the best of our knowledge, no reports so far have stratified for molecular subtype in relation to endocrine receptor expression in ovarian cancer, either on the mRNA or on the protein level, and this approach has the potential to further increase our knowledge of endocrine signaling in ovarian cancer. Materials and Methods Tumor Material One hundred eighteen patients with epithelial serous (n = 87) and endometrioid (n = 31) ovarian cancer were included in the present study. The patients were recruited from a consecutive cohort study in the southern Swedish health care region between June 1998 and June 2000 (n = 128 patients with ovarian cancer, outlined in Malander et al. [27]) and at the oncogenetic counseling at Lund University Hospital from 1981 to 1997 (n = 18 patients). The small number of patients recruited via oncogenetic counseling reflects the limited extent of counseling service at that time. Of the 146 eligible patients, 4 patients with clear cell tumors and 10 patients with mucinous tumors were excluded because these tumors are generally not expected to express sex steroid hormone receptors [10,13]. Another 14 patients with tumors of unknown primary, mixed histologies, or undifferentiated carcinomas were excluded to reduce external factors, which may potentially bias the analyses. Of the 118 included patients, 30 (25%) had a verified BRCA1 (n = 26) or BRCA2 (n = 4) mutation. Detailed clinical features of included tumors are outlined in Supplementary Table S1. Tumor samples were collected at primary surgery, and the patients had not received chemotherapy before this. Information on amount of residual disease after surgery was not available. Fifty-nine of 118 (50%) patients received postoperative carboplatin (AUC5) and paclitaxel (175 mg/m) treatment, 18/118 (15%) received carboplatin (AUC5) and cyclophosphamide (500 mg/m), and 17/118 (14%) with varying disease stages were reported not to have received any postoperative chemotherapy. Information on chemotherapy treatment was missing for 24/118 (21%) patients, whereas no information on hormonal treatment was available. Eleven of 30 (37%) of the patients with BRCA mutations were also diagnosed with breast cancer either before or after the ovarian cancer diagnosis. All deaths within the follow-up time, however, were related to ovarian cancer. Histopathological subtypes were reviewed by a gynecological pathologist (A. M.). The histologic subtype and grade were determined according to Silverberg and WHO 2003, and hematoxylin and eosin–stained slides were used to assess tumor grade [28,29]. All tumors were staged according to the International Federation of Gynecology and Obstetrics criteria [30]. Ethical approval for the study was granted by the Lund University ethics committee (Sweden), and all patients had given their written informed consent to participate in the study. Immunohistochemical Stainings Existing tissue micro array (TMA) blocks were used for evaluation of protein expression by immunohistochemical staining of ERα, ERβ, PR, and AR. The construction of the TMA blocks is outlined in Figure 1. Examples of immunohistochemical stainings for ERα, ERβ, PR, and AR. Top row from left to right: positive immunohistochemical stainings (≥10% stained cells) for ERα, ERβ, PR, and AR. Bottom row from left to right: negative immunohistochemical stainings (b10% stained cells) for ERα, ERβ, PR, and AR. Magnification, 40×. 426 Sex Steroid Hormone Receptor and Ovarian Cancer Jönsson et al. Translational Oncology Vol. 8, No. 5, 2015 Malander et al. [27]. Three mouse monoclonal antibodies (ERα, DAKO A/S, Glostrup, Denmark, clone 1D5, cat. #M7047, dilution 1:100; ERβ, DAKO, M7292, dilution 1:10; AR, DAKO, M3562, dilution 1:100) and one rabbit polyclonal antibody (PR, DAKO, cat. #A0098, dilution 1:50) were used. For ERα and PR stainings, antigen retrieval was achieved by microwave treatment in 10 mM citrate buffer (pH 6.0, 15 minutes). Antigen retrieval for ERβ and AR stainings was achieved using a pressure cooker and DAKO’s solutions (ERβ pH 6, cat. #S1699, AR pH 9, cat. #S2367), following the manufacturer’s instructions (DAKO). All immunohistochemical stainings were performed using an automated immunostainer (Techmate 500, DAKO), following the manufacturer’s instructions, with application of the Envision systems for visualization (DAKO). Breast cancer tissue known to be positive for the respective receptors was used as positive controls, and ovarian cancer tissue with removal of the primary antibodies was used as negative controls. The ERα and PR nuclear stainings were evaluated independently by authors S. M. and M. N. and ERβ and AR by J. M. J. and N. A. At least three cores per tumor (diameter 0.6 mm) were used to determine the staining pattern. In general, antibodies showed homogenous staining across cores within individual tumors (N80%), but in cases where heterogeneous staining was observed, the staining pattern representing the majority of the tumor cells was used. Stainings were evaluated regarding percent tumor cells with stained nuclei and staining intensity, where 1 = weak, 2 = moderate, and 3 = strong staining intensity. In line with the Swedish national guidelines for breast cancer, 10% stained tumor cells was used as cutoff for positive versus negative stainings [31]. Thus, for statistical comparisons, only the prevalence of stained tumor cells was taken into account, and b10% stained tumor cells were dichotomized as 0 and ≥10% stained tumor cells as 1. Examples of positive and negative nuclear stainings for the respective receptors are shown in Figure 1. Positive versus negative stainings were studied in relation to 5-year progression-free survival (PFS) and overall survival (OS). Sex Steroid Hormone Receptor Gene Expression in Relation to Molecular Subtype To expand the hypothesis, we aimed to outline if the association between a favorable prognosis and receptor expression at the protein level could be detected on the mRNA level. For this purpose, an independent, publicly available data set consisting of 285 high-grade serous and endometrioid, borderline as well as low-grade serous and endometrioid ovarian tumors, fallopian tube, and primary peritoneal cancers was used for studying the expression of the ESR1, ESR2, PGR, and AR genes, coding for ERα, ERβ, PR, and AR, respectively [24]. Gene probes specific for ESR1, ESR2, PGR, and ARwere identified and aligned with the reference genome assembly human build 19 (GRCh37/hg19, released February 2009) using the online BLAST-Like Alignment Tool, identifying DNA sequences with ≥95% similarity of ≥25 bases. When multiple probe sets identified the same gene, the probe set with the highest number of probes identifying the specific gene was chosen and used for further analyses. The data set was originally used for transcriptional subtyping of ovarian cancers, revealing six different subtypes (C1-C6) of which the four subtypes representing high-grade serous ovarian cancer are now commonly termed “mesenchymal” (C1), “immunoreactive” (C2), “differentiated” (C4), and “proliferative” (C5) [25,26]. C3 represents borderline and a few serous tumors of varying differentiation grade, and C6 represents low-grade endometrioid tumors [24]. High versus low expression of the respective genes was studied in relation to PFS and OS in the different molecular subtypes using the median mRNA expression level across all the samples as cutoff. Because of limited follow-up time in this data set, 5-year PFS and OS were not evaluable in all subtypes, and therefore 3-year PFS and OS were used as end points. Statistical Analyses The prognostic value of immunohistochemical expression of ERα, ERβ, PR, and AR was investigated using PFS time and OS time, both Translational Oncology Vol. 8, No. 5, 2015 Sex Steroid Hormone Receptor and Ovarian Cancer Jönsson et al. 427 censored at 5 years, as end points. PFS time was defined as the time interval between date of diagnosis and the first sign of disease recurrence (clinical and/or radiological) or death of any cause, whichever came first. OS time was defined as the time interval between date of diagnosis and death of any cause, obtained frommedical records and the Swedish civil registration register. No patients were lost to follow-up before 5 years, and all patients who died before 5 years of diagnosis died of ovarian cancer, i.e., had persistent or recurrent disease. In the publicly available gene expression data set, high versus low expression of ESR1, ESR2, PGR, and AR in the respective molecular ovarian cancer subtypes, using the median as cutoff, was investigated using PFS time and OS time as end points, here censored at 3 years because of limited follow-up time [24]. PFS time andOS time were defined asmentioned above. For PFS, disease progression or death of any cause was considered an event, and patients lost to follow-up were censored at the time of last notification. For OS, deaths within 3 years were considered as events, and patients lost to follow-up were censored at the time of last notification. Of the patients who died before 3 years of diagnosis, all but two died of ovarian cancer. OS and PFS were for both data sets estimated using the Kaplan-Meier method and compared between groups using the log-rank test. For protein expression, hazard ratios (HRs) with 95% confidence intervals (95% CIs) were calculated in univariable and multivariable analyses using Cox proportional hazards regression, adjusted for clinical factors known to influence ovarian cancer survival (stage, age at diagnosis, histological grade, histology, and BRCA mutation status) [32–34]. A previous Gynecologic Oncology Group study has shown an association between increasing age at diagnosis and tumor progression or death, with the greatest risk among patients ≥70 years at diagnosis [32]. This cutoff was therefore used in the present study. Stage (III-IV vs I-II), age at diagnosis (b70 vs ≥70 years), histology (serous versus endometrioid), and BRCA mutation status (BRCA mutation versus wild type) were treated as binary factors and histological grade as a categorical factor on three levels with grade 3 as reference. Chemotherapy was not adjusted for in the multivariable analyses because treatment data were missing for 21% of the patients, and all patients reported to have received chemotherapy were given platinum-containing combinations. Instead, a stability analysis, i.e., a separate multivariable analysis including the variable postoperative chemotherapy (platinum-containing chemotherapy versus no treatment), was performed for 5-year PFS and OS. To account for histology-dependent differences in receptor expression, which may not be captured by adjusting for histology in the multivariable analysis due to the uneven histological distribution, the independent effect of the receptors on survival was also assessed in serous and endometrioid tumors separately. In the external data set, comparisons between mRNA levels in the different molecular subtypes were performed using Kruskal-Wallis test. Associations between receptor protein expression and clinical parameters were assessed using Fisher exact test, except for the ordinal variables stage and grade where the Mann-Whitney test was used. Statistical analyses of protein expression data were performed in SPSS statistics version 22, and analyses on gene expression data were performed in R version 3.1.0. All P values were two-sided.

Extracted Key Phrases

8 Figures and Tables

Cite this paper

@inproceedings{Jnsson2015SexSH, title={Sex Steroid Hormone Receptor Expression Affects Ovarian Cancer Survival12}, author={Jenny-Maria J{\"{o}nsson and Nicolai Skovbjerg Arildsen and Susanne Malander and Anna M{\aa}sb{\"a}ck and Linda Hartman and Mef Nilbert and Ingrid Hedenfalk}, year={2015} }