Hormone receptor expression profiles differ between primary and recurrent high-grade serous ovarian cancers

Hormone receptor status assessment is necessary for selecting cancer patients who might potentially benefit from endocrine therapy. To determine whether hormone receptor status changes during tumor progression, we retrospectively examined 107 high-grade serous ovarian cancer (HGSC) patients with paired primary and recurrent tumor specimens. Hormone receptor expression discordance rates between primary and recurrent tumors were as follows: estrogen receptor (ER) 34.9%, progesterone receptor (PR) 12.4%, androgen receptor (AR) 41.7%, follicle stimulating hormone receptor 46.6%, luteinizing hormone receptor 50.5%, and gonadotropin releasing hormone receptor 20.0%. Hormone receptor discordance was not associated with patient survival. The proportion of the PR-ER+AR- subgroup, which exhibited the worst prognosis, was higher in recurrent than primary tumor specimens. Our study demonstrated that paired primary and recurrent HGSC specimens exhibit differing hormone receptor profiles. Thus, to most effectively identify patient-specific therapies, biomarker status re-assessment is required for recurrent patients.


INTRODUCTION
Ovarian cancer is one of the most common and lethal cancers affecting women globally [1]. After primary debulking or staging surgery, and subsequent platinumbased chemotherapy, most patients relapse within a median of 16 months [2]. Patients with recurrent ovarian cancer are usually offered several consecutive therapies with variable response rates and prognoses [2]. Effective clinicopathological biomarkers to identify the most appropriate alternative therapy for a given patient are urgently needed.
Hormone therapy is considered a salvage therapy for recurrent disease, and endocrine therapy tends to be more efficacious in hormone receptor positive subgroups [3][4][5][6]. Hormone receptor status assessment is necessary for selecting patients who would potentially benefit from endocrine therapy. However, hormone receptors or HER2 status can change during breast cancer progression [7][8][9], although whether hormone receptor expression differs between primary and recurrent ovarian cancers is currently unknown.
High-grade serous ovarian cancer (HGSC) is the main histologic subtype of epithelial ovarian cancers with high hormone receptor levels [10,11]. Our previous study classified patients into five subgroups with distinctive clinicopathological features via immunohistochemistry [12]. In the present work, we analyzed hormone receptor expression in 107 HGSC patients with paired primary and recurrent tumor specimens, and investigated the clinical significance of hormone receptor status discordance between primary and recurrent HGSC.

Hormone receptor status changes between primary and recurrent specimens
Representative images for hormone receptors are shown in Supplementary Figure 1. Most primary and recurrent ovarian cancers expressed estrogen receptor (ER; 67.0% and 72.9%, respectively) or gonadotropin releasing hormone receptor (GnRHR; 87.9% and 85%, respectively), while progesterone receptor (PR) remained at low levels in both primary and recurrent specimens (9.3% and 6.7%, respectively) ( Table 2). Androgen receptor (AR) was downregulated from 33.6% to 17.5%, respectively. Approximately half of the patients expressed follicle stimulating hormone receptor (FSHR) or luteinizing hormone receptor (LHR) in both specimens.
Hormone receptor discordance rates were as follows: ER 34.9%, PR 12.4%, AR 41.7%, FSHR 46.6%, LHR 50.5%, and GnRHR 20.0% (Table 3). Quantitative hormone receptor change was calculated as quantitative expression in the recurrent specimen minus that in the primary specimen in each case ( Figure 1). PR expression remained relatively unchanged, while other hormone receptor levels fluctuated throughout tumor progression.

Prognostic impact of hormone receptor changes
In survival analyses, patients whose recurrent specimens showed hormone receptor upregulation were compared with those whose recurrent specimens showed downregulation. Similarly, patients with hormone receptor expression loss in recurrent specimens were compared with patients who showed hormone receptor upregulation. Hormone receptor discordance was not associated with PFS or OS (Table 4).
We then analyzed the distribution of hormone receptor-based molecular subgroups according to our previous study (Table 5) [12]. As we previously reported, a trend of increasing risk of death was observed for the following four subgroups: PR-ER-AR+, PR+, PR-ER+AR+, PR-ER-AR-and PR-ER+AR- [12]. In our current study, the proportion of PR-ER+AR-cases increased in recurrent vs. primary specimens (54.9% vs. 37.7%, P=0.020).

DISCUSSION
Our study demonstrated hormone receptor status instability in a portion of HGSC patients during ovarian cancer progression. We also associated patient risk of death with tumor hormone receptor status. Receptor status discordance between primary and recurrent breast cancers has been addressed in recent work [19]. Approximately 10-40% of breast cancer patients exhibited unstable receptor statuses during tumor progression [7][8][9]. Different mechanisms for receptor discordance have been proposed, such as intratumoral heterogeneity and selection through    previous treatments [20,21]. Receptor status instability has also been correlated with patient survival, and could influence therapeutic decision making in the management of recurrent patients [7][8][9]. The present study identified hormone receptor status discordance between paired primary and recurrent ovarian cancers for the first time. Furthermore, we observed that the proportion of the PR-ER+AR-subgroup, which had the worst prognosis, increased in recurrent vs. primary HGSC cases.
Primary and recurrent ovarian cancer tissue samples collected from the same patient may exhibit distinctive morphological, molecular, and/or genetic features [22,23]. Marques, et al. [23] found that chemotherapy reduced PARP1 expression in ovarian cancer, while Despierre, et al. [22] reported that folate receptor alpha expression remained unchanged in epithelial ovarian cancer after chemotherapy. Our results indicate that recurrent ovarian cancers present a more aggressive phenotype compared with primary tumors. Further investigations should focus on mechanisms that drive genetic disparity between primary and recurrent ovarian cancers.
Hormone therapy (tamoxifen, aromatase inhibitors, and others) is considered a salvage therapy for recurrent ovarian cancer patients, although results have been unsatisfactory [3][4][5]. However, a phase II study of letrozole in ER+ relapsed ovarian cancer patients had more promising response rates [6]. As hormone receptor status may change during ovarian cancer progression, endocrine therapy administration should be based on hormone receptor status in recurrent patients. Additionally, FSHR or GnRHR targeted agents have been developed using corresponding ligands as targeting moieties [24,25]. Targeted therapy relies on corresponding receptor expression, and biopsies are necessary to confirm this expression.
In conclusion, our study demonstrated that paired primary and recurrent ovarian cancer specimens exhibit discordant hormone receptor statuses. Thus, biomarker status re-assessment in recurrent patients is required to most effectively identify patient-specific therapies.

Clinical data
This study was conducted according to the Declaration of Helsinki and was approved by the Committee at Fudan University Shanghai Cancer Center. Written informed consent was obtained from all study participants. Our retrospectively study included 107 women who underwent both primary staging or debulking surgery and secondary cytoreduction to treat HGSC at Fudan University Shanghai Cancer Center between April 2005 and June 2013. Patients were excluded if they had received neoadjuvant therapy prior to primary surgery, were found to have other histological diagnoses on pathological review, or if paraffin-embedded tissue samples were not available.
Clinical and pathological data were obtained from medical records, cancer registries, and pathology reports. Patient characteristics, including age, FIGO stage, surgical outcomes, date of primary and secondary surgeries, date of progression or recurrence, date of last follow-up, and disease status at last contact, were collected. Patient follow-up for this study ended on December 31, 2014.
R0 was defined as the absence of macroscopic residual disease (RD) after surgery. Chemosensitivity was defined as a time interval of six months or longer between completion of platinum-based chemotherapy and detection of relapse. Chemoresistance was defined as disease progression during adjuvant chemotherapy or within the six-month interval between completion of platinum-based chemotherapy and detection of relapse. Progression-free survival (PFS) was defined as the time interval between primary surgery and disease progression or recurrence. Overall survival (OS) was defined as the time interval between primary surgery and death or last follow-up.

Tissue microarray and immunohistochemistry
Histological diagnoses were based on the WHO criteria [13], and all paired primary and recurrent specimen slides were reviewed by two experienced gynecologic pathologists. A microarray (1 mm) with triplicate tissue samples from each tumor was prepared [11,14]_ENREF_20. We collected ovarian masses as the primary specimens. We also collected recurrence specimens: 92/107 (86.0%) from pelvic masses, 13/107 (12.1%) from metastatic lymph nodes, and 2 (1.9%) from isolated thoracic masses. Intra-class correlation coefficient (ICC) was calculated to evaluate the internal consistency of immunoscores of three cores from each individual tumor sample. Cronbach's α indexes were approximately 0.9, which meant that there were no differences in parameter expression among the different cores. Sections (3μm) of the completed tissue microarray were analyzed by standard immunohistochemistry methods.
Results were blindly and independently judged, evaluated, and scored by two experienced gynecologic pathologists. Results were reported as the numerical means of the triplicate scores. Hormone receptor expression was determined using the following criteria: ER, PR, and AR levels: >10% of cells showing positive nuclear staining of any intensity was considered positive [15,16]. FSHR and LHR levels: evaluation of the cytoplasmic staining reaction was performed in accordance with the immunoreactive score (IRS). IRS was defined as staining intensity (SI) multiplied by the percentage of positive cells (PP). SI was defined as 0 (negative), 1 (weak), 2 (moderate), and 3 (strong). PP was defined as 0 (negative), 1 (≥10% positive cells), 2 (11-50% positive cells), 3 (51-80% positive cells) and 4 (>80% positive cells). IRS≥3 was considered positive [17]. GnRHR level: Cytoplasmic GnRHR staining was recorded as negative, weak, moderate or strong. Staining of any intensity was considered positive [18].
Discordance rate was defined as the proportion of patients who displayed differential hormone receptor status between primary and recurrent specimens. Discordance included cases in which a primary sample presented as positive for a particular hormone receptor, but recurrent specimens were negative, and cases that initially presented as negative, but turned positive in recurrent specimens.

Statistical analyses
SPSS software (version 21.0) and GraphPad Prism (version 6.0) were used for statistical analyses. Descriptive statistics were used for demographic data and were summarized as means ± standard deviations (SD), medians with interquartile ranges (IQRs) or ranges, or frequencies with percentages. Categorical data were compared via chisquare or Fisher's exact tests as appropriate. A two-tailed McNemar test was used to evaluate biomarker discordance in the same patient before and after chemotherapeutic treatment. OS was analyzed via the Cox regression method, which is expressed as hazard ratios (HRs). P<0.05 was considered statistically significant, and all reported P-values were 2-sided.