A clinically applicable molecular classification for high-grade serous ovarian cancer based on hormone receptor expression

Zheng Feng, Hao Wen, Rui Bi, Xingzhu Ju, Xiaojun Chen, Wentao Yang, Xiaohua WuScientific 2016 · DOI

Abstract

To establish an effective hormone receptor-based molecular classification of high-grade serous ovarian cancer (HGSC), we retrospectively examined 875 consecutive HGSC patients who underwent primary surgery at our hospital and constructed tissue microarrays from these specimens. The expression levels of the hormone receptors were as follows: ER 64.4%, PR 12.6%, AR 35.6%, FSHR 54.5%, LHR 34.8%, and GnRHR 88.3%. Based on clustering of their expression patterns, we classified patients into five subgroups with distinctive clinical features (PR+, PR - ER + AR+, PR - ER + AR-, PR - ER - AR+, and PR - ER - AR-). Patients in the PR + group were younger compared to those in the other groups (p < 0.001). More patients were of advanced stage in the PR - ER + AR- group than the other groups (p = 0.020). A greater proportion of patients were sensitive to platinum-based chemotherapy in the PR - ER - AR + group compared with the other groups (p = 0.034). A trend of increasing risk of death was observed among these subgroups (p < 0.001). In the multivariate analysis, patients also had orderly increased hazard ratios for death in the PR + (HR = 2.256, 95% CI, 0.983-5.175), PR - ER + AR + (HR = 2.188, 95% CI, 1.004-4.796), PR - ER - AR- (HR = 2.316, 95% CI, 1.097-5.082) and PR - ER + AR- (HR = 2.928, 95% CI, 1.366-6.276) subgroups compared to the PR - ER - AR+ subgroup. Our classification could help predict patient clinical outcomes, guide individual treatments and stratify patients in future clinical trials.

Introduction

High-grade serous ovarian cancer (HGSC) represents the most prevalent and lethal subtype of ovarian cancer. Despite standard treatment involving debulking surgery and platinum-based chemotherapy, clinical outcomes remain highly heterogeneous, with approximately 50% of patients experiencing relapse within 16 months. While hormone receptors such as estrogen (ER) and progesterone (PR) are established biomarkers in breast cancer, their role in HGSC has been obscured by small study sizes and inconsistent methodologies. This study aims to establish a standardized, clinically applicable molecular classification system for HGSC based on hormone receptor expression to better predict prognosis and guide personalized therapy.

Methods

The study analyzed a retrospective cohort of 875 consecutive HGSC patients from the Fudan University Shanghai Cancer Center (2005–2013). Researchers utilized tissue microarrays (TMA) and immunohistochemistry (IHC) to evaluate eight biomarkers: ER, PR, Androgen Receptor (AR), FSHR, LHR, GnRHR, HER2, and Ki67.

Unsupervised hierarchical clustering was employed to identify expression patterns, leading to the development of a classification tree. This model prioritized PR, ER, and AR as the primary markers for defining five distinct patient subgroups.

Classification tree for the five HGSC subgroups based on PR, ER, and AR expression.

Results

The analysis identified five subgroups with significantly different clinical profiles:

  1. PR+: Associated with younger patients and a moderate prognosis.
  2. PR− ER+ AR+
  3. PR− ER+ AR−: Exhibited the worst prognosis and was associated with more advanced stages.
  4. PR− ER− AR+: Demonstrated the best prognosis and the highest sensitivity to platinum-based chemotherapy (86.1%).
  5. PR− ER− AR−

Survival analysis revealed a clear trend in the risk of death across these groups ($p < 0.001$). Notably, AR emerged as a positive prognostic factor, while ER expression was unexpectedly correlated with poorer survival outcomes in this HGSC cohort. Compared to the favorable PR− ER− AR+ group, the PR− ER+ AR− group faced nearly triple the risk of death (Hazard Ratio = 2.928).

Kaplan-Meier survival analysis showing distinct overall survival outcomes across the five hormone receptor-based subgroups.

Discussion and Limitations

This classification system provides a robust framework for stratifying HGSC patients. The identification of the AR+ subgroup as highly chemo-sensitive suggests that androgen signaling may play a protective or sensitizing role in HGSC. Conversely, the poor outcomes in ER+ groups (in the absence of PR) highlight the complexity of endocrine signaling in ovarian tissue compared to breast tissue.

While promising, these findings are limited by the retrospective, single-center nature of the study and the specific ethnic composition of the cohort. Future prospective trials are necessary to validate these subgroups and explore the efficacy of targeted endocrine therapies—such as AR agonists or anti-estrogens—within these specific molecular contexts.

Summary updated

Abstract

To establish an effective hormone receptor-based molecular classification of high-grade serous ovarian cancer (HGSC), we retrospectively examined 875 consecutive HGSC patients who underwent primary surgery at our hospital and constructed tissue microarrays from these specimens. The expression levels of the hormone receptors were as follows: ER 64.4%, PR 12.6%, AR 35.6%, FSHR 54.5%, LHR 34.8%, and GnRHR 88.3%. Based on clustering of their expression patterns, we classified patients into five subgroups with distinctive clinical features (PR+, PR - ER + AR+, PR - ER + AR-, PR - ER - AR+, and PR - ER - AR-). Patients in the PR + group were younger compared to those in the other groups (p < 0.001). More patients were of advanced stage in the PR - ER + AR- group than the other groups (p = 0.020). A greater proportion of patients were sensitive to platinum-based chemotherapy in the PR - ER - AR + group compared with the other groups (p = 0.034). A trend of increasing risk of death was observed among these subgroups (p < 0.001). In the multivariate analysis, patients also had orderly increased hazard ratios for death in the PR + (HR = 2.256, 95% CI, 0.983-5.175), PR - ER + AR + (HR = 2.188, 95% CI, 1.004-4.796), PR - ER - AR- (HR = 2.316, 95% CI, 1.097-5.082) and PR - ER + AR- (HR = 2.928, 95% CI, 1.366-6.276) subgroups compared to the PR - ER - AR+ subgroup. Our classification could help predict patient clinical outcomes, guide individual treatments and stratify patients in future clinical trials.
BibTeX
@inproceedings{feng2016clinically,
  author = {Zheng Feng and Hao Wen and Rui Bi and Xingzhu Ju and Xiaojun Chen and Wentao Yang and Xiaohua Wu},
  title = {{A clinically applicable molecular classification for high-grade serous ovarian cancer based on hormone receptor expression}},
  booktitle = {Scientific},
  year = {2016},
  doi = {10.1038/srep25408},
}
Figure