Association of hormone receptor-based pathological features with prognosis in high-grade serous ovarian carcinoma: a retrospective cohort study and nomogram construction

Abstract

Background

High-grade serous ovarian cancer (HGSC) causes high mortality rates worldwide due to its insidious onset and poor prognosis. The statuses of hormone receptors (ER/PR) have been shown to have significant prognostic value in hormone-related tumors such as breast cancer, but their roles in HGSC remain unclear.

Methods

This study retrospectively analyzed the relationship between estrogen receptor (ER) and progesterone receptor (PR) expression status, clinical features, and survival outcomes in 176 patients with HGSC. Kaplan-Meier analysis and univariate and multivariate Cox regression were used to identify prognostic factors. Propensity score matching (PSM) was applied to control for confounding, and a nomogram incorporating ER/PR status and other independent prognostic factors was developed. Internal validation was performed using repeated stratified 5-fold cross-validation and 1,000-bootstrap resampling.

Results

ER/PR expression status was significantly associated with survival. The ER(+)/PR(+) group had the best prognosis, whereas the double-negative group had the worst. Combined ER/PR expression remained an independent prognostic factor in multivariate analysis (HR = 11.610, 95% CI: 6.225–21.653). The nomogram showed good predictive performance for 1-, 3-, and 5-year OS, with favorable discrimination and calibration in internal validation.

Conclusion

Combined ER/PR expression status is an independent prognostic factor in HGSC. A nomogram integrating ER/PR status with clinicopathological variables may provide a practical tool for individualized risk assessment.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13048-026-02172-0.

Introduction

High-grade serous ovarian cancer (HGSOC) is the most aggressive form of ovarian cancer. It accounts for approximately 70% of all ovarian cancer cases. This type of cancer is often diagnosed at an advanced stage, resulting in a poor prognosis with a five-year survival rate below 45% [1]. Notably, the development of HGSOC is complex and associated with multiple factors, including genetic variations, alterations in the tumor microenvironment, and hormone receptor status. Currently, the main treatments for HGSOC include surgical resection and chemotherapy. However, these methods often have limited effectiveness due to patient heterogeneity and the tumor’s biological characteristics [2]. Therefore, it is imperative to develop more effective prognostic assessment tools and personalized treatment strategies to enhance patient survival rates and quality of life.

Ovarian cancer is considered a hormone-driven malignant tumor, especially because the expression of estrogen (ER) and progesterone (PR) receptors is closely linked to patient prognosis. A systematic assessment showed that ER positivity is significantly higher in low-grade serous ovarian cancer (LGSOC) compared to HGSOC [3]. This suggested that differences in hormone receptor expression among ovarian cancer subtypes may influence clinical presentation and prognosis.Among the various subtypes, endometrioid carcinoma and low-grade serous ovarian cancer have been shown to be associated with ER and PR expression, and the presence of these receptors may influence patient prognosis [4, 5]. However, the relationship between HGSOC and ER/PR expression remains unclear, complicating the analysis of prognostic factors for this subtype.

PR expression is recognized as a favorable prognostic biomarker in HGSOC [6]. Although ER positivity in HGSOC can reach 86%-92%, its prognostic value varies widely, and not all such tumors respond to anti-estrogen therapy [7]. Notably, patients with recurrent HGSOC and peritoneal metastasis who exhibit better clinical outcomes showed an ER positivity rate as high as 86%, suggesting ER may influence disease progression, though the mechanisms remain unclear [8]. In hormone-related tumors, breast cancer studies showed that patients with the ER+/PR+ phenotype had the greatest survival benefits [9], providing a theoretical basis for studying combined biomarkers in HGSOC. Several exploratory studies on hormones have indicated that AR (androgen receptor), ER, and PR, both individually and in combination, have predictive and prognostic value [10]. Nonetheless, no systematic evaluation has been conducted on the link between ER/PR combined expression subtypes and clinical features.

This study addresses the clinical significance of ER and PR expression profiles in HGSC patients based on identified research gaps. It also explores their relationships with prognosis. We include combined ER and PR expression status in a multifactor analysis to assess its association with clinical, pathological factors, and survival outcomes. Predictive tools such as nomograms will be introduced to integrate molecular profiles with clinical variables. These tools aim to improve risk assessment and clinical usability, providing both a theoretical basis and practical support for individualized prognosis assessment and management of HGSC patients.

Methods

Study population and data collection

This study included patients with epithelial ovarian cancer treated at the Obstetrics and Gynecology Department of the First Affiliated Hospital of Wannan Medical College from August 2012 to July 2020. All patients underwent surgery and received systemic chemotherapy. The first-line chemotherapy regimens were mainly paclitaxel plus carboplatin; a small number of patients who could not tolerate this regimen received docetaxel plus carboplatin. All patients completed 6–8 cycles of chemotherapy. Exclusion criteria included low-grade serous, mucinous, endometrioid, and clear cell ovarian cancers, as well as insufficient follow-up data. A total of 176 cases of HGSC were selected, as shown in the flowchart in Fig. 1. Clinical and pathological data were extracted from Electronic Medical Records (EMRs) and pathology reports, including age, p53FIGO stage, tumor size, ascites, vascular cancer thrombus, and the status of ER, PR,, and Ki67.

Fig. 1.

Flowchart of the patient selection

All patients were followed up via regular outpatient visits combined with telephone contact, starting from the date of diagnosis and completion of initial treatment, with the last follow-up in July 2025. The follow-up schedule was as follows: every 3 months within the first 2 years after treatment, every 4–6 months from the 3rd to 5th year, and annually thereafter. Follow-up assessments included history taking, physical examination, tumor marker testing, and imaging examinations.

Immunohistochemistry (IHC) testing

Standardized IHC protocols stained formalin-fixed paraffin-embedded tissue sections. ER, PR, p53, and Ki67 status was classified as positive or negative. Positive staining for ER and PR was defined as ≥ 1% of tumor cell nuclei being immunoreactive, according to the Ovarian Tumor Tissue Analysis (OTTA) consortium’s research protocol [11]. To be consistent, nuclear staining for all p53 and Ki67 antibodies exceeding 10% was considered positive [12, 13]. All sections were independently evaluated by two pathologists who were blinded to the clinical outcomes.

Statistical analysis

Overall survival (OS) is calculated as the time from diagnosis until death or last follow-up. Survival differences between ER/PR subgroups were assessed using Kaplan-Meier curves and log-rank tests. Independent prognostic factors were identified by univariate and multivariate Cox proportional hazards regression models. Variables with a univariate P-value less than 0.05 were included in the multivariate model. Propensity score matching (PSM) was performed in a 1:1 ratio, based on covariates using logistic regression. The balance between matched groups was assessed by standardized mean differences below 0.1. A predictive nomogram integrating ER/PR status and clinicopathological variables was constructed using the rms package in R software.

Internal validation was performed using repeated stratified 5-fold cross-validation based on the prespecified final nomogram variables, including pathologic stage, tumor size, cancer emboli, p53, Ki67, and ER/PR group. The cohort was stratified by survival status and randomly split into 5 folds; in each iteration, 4 folds were used for model refitting and the remaining fold was used for validation. This procedure was repeated 50 times using different random splits. Model discrimination was assessed using the C-index and time-dependent AUCs at 1, 3, and 5 years, and overall prediction error was evaluated using the Brier score. Calibration curves were generated from averaged out-of-fold predicted survival probabilities. In addition, bootstrap internal validation with 1000 resamples was performed to estimate optimism-corrected model performance. Because a large proportion of survivors were administratively censored at day 1825, the 5-year performance was evaluated at day 1824 to avoid boundary instability.

Results

Clinical and pathological features

This study included 176 patients with HGSC. They were divided into four groups based on estrogen receptor (ER) and progesterone receptor (PR) expression status: ER(+)/PR(+) with 61 cases (34.7%), ER(+)/PR(−) with 71 cases (40.3%), ER(−)/PR(+) with 13 cases (7.4%), and ER(−)/PR(−) with 31 cases (17.6%). No significant differences were observed between the groups regarding clinical features including age, ascites, tumor size, and pathological staging. Similarly, there were no significant differences in pathological features such as p53 distribution. However, the groups showed statistically significant differences in vascular cancer emboli and Ki67 (P = 0.028 and P = 0.024) (Table 1). Immunohistochemical analysis showed positive ER expression in 132 patients (75.0%) and positive PR expression in 74 patients (42.0%). ER was predominantly located in the nucleus, exhibiting a brownish granular distribution with moderate to high intensity (Fig. 2A); PR was also located in the nucleus but showed weaker or patchy distribution in some cases, indicating greater expression heterogeneity (Fig. 2B).

Table 1.

Clinicopathological characteristics of the study cohort

Characteristics	ER(+)/PR(+)	ER(+)/PR(-)	ER(-)/PR(+)	ER(-)/PR(-)	P value
n	61	71	13	31
Age(years), n (%)					0.124
<60	44 (25%)	43 (24.4%)	5 (2.8%)	19 (10.8%)
≥ 60	17 (9.7%)	28 (15.9%)	8 (4.5%)	12 (6.8%)
Pathologic.stage, n (%)					0.232
Stage I	11 (6.2%)	8 (4.5%)	4 (2.3%)	2 (1.1%)
Stage II	9 (5.1%)	5 (2.8%)	3 (1.7%)	4 (2.3%)
Stage III	39 (22.2%)	55 (31.2%)	5 (2.8%)	23 (13.1%)
Stage IV	2 (1.1%)	3 (1.7%)	1 (0.6%)	2 (1.1%)
tumor size(cm), n (%)					0.784
≤ 5	13 (7.4%)	20 (11.4%)	3 (1.7%)	9 (5.1%)
>5	48 (27.3%)	51 (29%)	10 (5.7%)	22 (12.5%)
cancer emboli, n (%)					0.028
Negative	37 (21%)	39 (22.2%)	8 (4.5%)	9 (5.1%)
Positive	24 (13.6%)	32 (18.2%)	5 (2.8%)	22 (12.5%)
Ascites, n (%)					0.993
Negative	33 (18.8%)	37 (21%)	7 (4%)	17 (9.7%)
Positive	28 (15.9%)	34 (19.3%)	6 (3.4%)	14 (8%)
p53, n (%)					0.806
Negative	21 (11.9%)	23 (13.1%)	6 (3.4%)	10 (5.7%)
Positive	40 (22.7%)	48 (27.3%)	7 (4%)	21 (11.9%)
Ki67, n (%)					0.024
Negative	7 (4%)	18 (10.2%)	6 (3.4%)	9 (5.1%)
Positive	54 (30.7%)	53 (30.1%)	7 (4%)	22 (12.5%)

Fig. 2.

The expression status of ER and PR in immunohistochemistry. A The expression of ER in each group. B The expression of PR in each group

Prognosis and survival analysis of hormone receptor expression

After 5 years of follow-up, we analyzed the impact of ER and PR expression status on patient prognosis. Kaplan-Meier survival curves were used for this analysis. The median survival time was 55 months for ER-positive patients, significantly longer than the 47 months for ER-negative patients (Fig. 3A). For PR status, the median survival time was 58 months in positive patients, significantly longer than 48 months in negative patients (Fig. 3B). The results show that positive expression of both ER and PR is significantly associated with better prognosis.

Fig. 3.

Kaplan-Meier analysis of OS in HGSC patients with different hormone receptors. (A) ER-positive and ER-negative. (B) PR-positive and PR-negative. (C) ER(+)/PR(+), ER(+)/PR(−) or ER(−)/PR(+), and ER(−)/PR(−). (D) ER(+)/PR(+), ER(+)/PR(−), ER(−)/PR(+), and ER(−)/PR(−). The I–II stage (E) and III–IV stage (F) in the pathological staging subgroup

To investigate how combined receptor expression affects prognosis in HGSC, patients were divided into three groups based on ER and PR status: ER(+)/PR(+), ER(+)/PR(−) or ER(−)/PR(+), and ER(−)/PR(−). The ER(+)/PR(+) group had the longest median survival time of 58 months, while the ER(−)/PR(−) group had the shortest at 38 months, with a statistically significant difference between these groups (P < 0.001)(Fig. 3C). These results indicate that patients with both ER and PR positive expression have the best prognosis, while those with double negative expression have the worst. Further analysis of the intermediate group showed no statistically significant difference in prognosis between ER(+)/PR(−) and ER(−)/PR(+), although the latter had a longer median survival time (55 months vs. 50 months)(Fig. 3D).

Subgroup analysis by pathological stage showed that survival differences among the three groups were statistically significant in early-stage patients (stage I–II) (P = 0.036)(Fig. 3E). In advanced-stage patients (stage III–IV), the median survival time for the ER(+)/PR(+) group was 54 months, significantly better than the 36 months for the ER(−)/PR(−) group (P < 0.001)(Fig. 3F), which indicated that hormone receptors were more clinically valuable for predicting outcomes in advanced tumors.

Independent prognostic analysis and PSM for risk models

To identify independent prognostic factors in patients with high-grade serous ovarian cancer, this study first performed univariate Cox regression on clinical pathological variables. The univariate analysis showed that advanced pathological stage (stage III–IV, HR = 6.001, P < 0.001), tumor size > 5 cm (HR = 2.011, P = 0.004), positive cancer emboli (HR = 2.193, P < 0.001), positive p53 (HR = 4.298, P < 0.001), positive Ki67 (HR = 3.111, P < 0.001), and ER/PR status (compared to the ER+/PR+ group, mixed group HR = 1.571, P = 0.043; double-negative status HR = 3.532, P < 0.001) were significant adverse prognostic factors.

Next, we included the statistically significant variables (P < 0.05) from the univariate analysis in a multivariate Cox proportional hazards mode (Table 2). All variables met the proportional hazards assumption (Supplementary Fig. 1). The results of the multivariate analysis indicated that, after adjusting for other confounding factors, advanced pathological stage (HR = 9.146, P < 0.001), tumor size > 5 cm (HR = 3.307, P < 0.001), cancer emboli positivity (HR = 1.919, P = 0.002), positive p53 (HR = 7.353, P < 0.001), positive Ki67 (HR = 3.217, P < 0.001), and ER/PR expression status remained independent prognostic factors affecting overall survival. Of these, double-negative ER/PR expression was one of the strongest independent risk factors, with a mortality risk 11.610 times higher than that of patients with double-positive ER/PR. The mortality risk for patients with mixed ER/PR expression was also significantly higher than that for the double-positive group.

Table 2.

Univariate and multivariate Cox regression analysis of clinicopathological characteristics and OS

Characteristics	Total(N)	Univariate analysis		Multivariate analysis
		Hazard ratio (95% CI)	P value	Hazard ratio (95% CI)	P value
Age(years)	176
<60	111	Reference
≥ 60	65	1.027 (0.703–1.500)	0.892
Pathologic.stage	176
Stage I or Stage II	46	Reference		Reference
Stage III or Stage IV	130	6.001 (3.209–11.221)	< 0.001	9.146 (4.756–17.585)	< 0.001
tumor size(cm)	176
≤ 5	45	Reference		Reference
>5	131	2.011 (1.252–3.231)	0.004	3.307 (1.922–5.690)	< 0.001
cancer emboli	176
Negative	93	Reference		Reference
Positive	83	2.193 (1.511–3.183)	< 0.001	1.919 (1.282–2.873)	0.002
Ascites	176
Negative	94	Reference
Positive	82	1.133 (0.784–1.637)	0.506
p53	176
Negative	60	Reference		Reference
Positive	116	4.298 (2.663–6.938)	< 0.001	7.353 (4.374–12.363)	< 0.001
Ki67	176
Negative	40	Reference		Reference
Positive	136	3.111 (1.776–5.449)	< 0.001	3.217 (1.790–5.784)	< 0.001
group	176
ER(+)/PR(+)	61	Reference		Reference
ER(+)/PR(-) or ER(-)/PR(+)	84	1.571 (1.015–2.430)	0.043	3.674 (2.252–5.992)	< 0.001
ER(-)/PR(-)	31	3.532 (2.106–5.924)	< 0.001	11.610 (6.225–21.653)	< 0.001

HR hazard ratio, CI confidence interval

To control for the impact of confounding factors in baseline data on prognostic analysis and reduce selection bias, this study employed propensity score matching (PSM). A 1:1 match was performed between the ER/PR double-positive group and the double-negative group (Supplementary Table 1). This approach aimed to make the comparison more valid and verify the reliability of the results. The matching variables included pathological stage, tumor size, cancer emboli, p53, and Ki67 status. Before matching, there were significant differences between the double-positive group (n = 61) and the double-negative group (n = 31), notably in cancer emboli (P = 0.004) and Ki67 expression (P = 0.036). After PSM matching, each group included 31 patients. All clinical pathological characteristics—such as pathological stage, tumor size, vascular cancer emboli, p53, and Ki67—were well balanced. No statistical differences were found between the groups, indicating successful matching and comparability of baseline characteristics. Subsequently, we conducted multivariate Cox regression and prognostic analyses on patients after PSM (Supplementary Table 2). The results showed that ER/PR double-negative expression remained the strongest independent risk factor associated with overall survival (HR = 11.122, P < 0.001), and the survival rate was significantly lower than that of the double-positive group (Supplementary Fig. 2).

Development of a predictive nomogram incorporating clinical risk factors

We constructed a nomogram to predict the 1-, 3-, and 5-year survival probabilities of HGSC patients (Fig. 4). The nomogram includes clinicopathological factors related to hormone receptors (ER/PR) identified through multifactor analysis. The prognostic model generated risk scores that classified patients into low-risk and high-risk groups. Kaplan-Meier analysis revealed a significant difference in survival time between high-risk and low-risk patients (Fig. 5B). Additionally, a risk factor heatmap showed that high-risk patients were more likely to be at advanced stages (III/IV), have larger tumors, positive vascular cancer emboli, positive p53 and Ki67, and negative ER/PR status (Fig. 5A). These factors correspond to the higher scores in the nomogram. Calibration curves demonstrated good agreement between the nomogram’s predicted survival probabilities and the actual observed outcomes (Fig. 5C).

Fig. 4.

Nomogram of the prognostic model based on hormone receptors for individual survival prediction

Fig. 5.

Evaluation of the performance and clinical benefits of prognostic models. (A) Heatmap of risk factors in prognostic models. (B) Survival differences between the high-risk group and low-risk group in prognostic scoring. (C) Calibration curve of the predictive model. (D) Time-dependent ROC curve of high-risk group and low-risk group. (E) Time-dependent AUC. (F) C-index of the nomogram from 1 to 5 years. DCA of the prognostic model at 1 (G), 3 (H), and 5 (I) years. ROC, Receiver Operating Characteristic; AUC, area under curve; DCA, Decision Curve Analysis

We used time-dependent ROC curves to evaluate the prognostic model’s ability to distinguish between 1-year and 3-year survival (Fig. 5D). The prognostic model maintained high AUC and C-index values throughout the 1- to 5-year period, reflecting its robust performance in predicting HGSC prognosis over 5 years (Fig. 5E-F). Decision Curve Analysis (DCA) showed that the model provided better net benefits for clinical decision-making at various thresholds compared to individual predictive factors. In patients with advanced HGSC and poor prognosis, the model demonstrated consistent net benefits across a broad range of decision thresholds, regardless of clinical decision preferences (Fig. 5G-I).

Internal validation of nomogram variables

Internal validation demonstrated stable discrimination and acceptable calibration of the nomogram. In repeated stratified 5-fold cross-validation (50 repeats), the mean C-index was 0.852 (95% empirical interval, 0.842–0.862). The 1-, 3-, and 5-year time-dependent AUCs were 0.851 (0.776–0.912), 0.921 (0.892–0.936), and 0.909 (0.892–0.923), respectively. The corresponding Brier scores were 0.028, 0.081, and 0.121. Bootstrap validation with 1000 resamples further showed an optimism-corrected C-index of 0.855, with optimism-corrected 1-, 3-, and 5-year AUCs of 0.836, 0.922, and 0.915, respectively (Supplementary Table 1). Calibration curves based on out-of-fold predictions showed acceptable agreement between predicted and observed survival probabilities (Fig. 6A-C).

Fig. 6.

Calibration plots at 1 year (A), 3 years (B), and 5 years (C) for the nomogram

Discussion

HGSC has a low overall survival rate because it is often diagnosed at an advanced stage and exhibits strong resistance to conventional chemotherapy [14]. Unfortunately, endocrine therapy is generally ineffective for gynecological tumors other than breast cancer. This ineffectiveness stems from a lack of prognostic studies on combined hormone receptor markers in the histological subtypes of gynecological tumors. As a result, precise targeted therapies are scarce. Surprisingly, breast tumors and high-grade serous ovarian tumors share significant molecular similarities, including BRCA gene mutations, which indicate common causes and potential treatment targets [15]. Studies show a negative correlation between ER/PR-positive breast cancer and the risk of ovarian cancer. This means that ER/PR-positive patients have a lower likelihood of developing ovarian cancer [16]. Therefore, analysis of ER and PR expression patterns and their associations with patients’ clinical characteristics and survival outcomes, along with the construction of a survival prediction model integrating molecular markers and clinical parameters, could help develop personalized endocrine therapies [17].

ER/PR function as nuclear receptors that regulate the transcription of target genes at the molecular level. This regulation modulates cell proliferation, differentiation, and apoptosis. Ki67 is a cell proliferation marker that typically indicates a more active cell cycle [18]. The ER/PR signaling pathway, when activated, regulates downstream genes related to proliferation. This regulation influences the growth rate of tumor cells. Transcriptomic studies of breast cancer have shown that ER/PR-positive status is often associated with higher differentiation and lower proliferation. Conversely, absence of ER/PR increases tumor invasiveness by strongly suppressing antigen presentation via MHC-II [19]. Notably, heterogeneity in PR expression is linked to multiple mechanisms. These include gene methylation, splice variants, and microenvironmental signals. The variation in expression among tumor subtypes may reflect their complex adaptations to hormone dependence and resistance [20]. Additionally, the formation of vascular cancer thrombi indicates that tumors have an enhanced ability to invade blood vessels. ER/PR signaling may promote local tumor infiltration and metastasis by regulating molecules such as matrix metalloproteinases (MMPs) [21]. ER/PR expression is closely linked to the mechanisms underlying proliferation, differentiation, and vascular invasion, which highlights the molecular basis of tumor heterogeneity. Interestingly, no significant difference exists in the age at diagnosis between early- and late-stage patients, suggesting that biological behavior rather than screening delays is the primary cause of late diagnosis [22].

This study shows that patients positive for ER or PR have a longer median survival time. Among them, those positive for both receptors have the best prognosis, while patients negative for both have the worst. Mechanistically, ER/PR-mediated pathways not only directly regulate the tumor cell cycle, apoptosis, and DNA repair but also modulate key pathways such as PI3K/AKT and MAPK to influence sensitivity to chemotherapy and endocrine therapy [20]. The absence of hormone receptors often causes insensitivity to the apoptosis signals, resulting in continuous tumor proliferation and drug resistance. ER/PR-negative tumors exhibit greater genomic instability and aggressiveness, an effect that is especially pronounced in tumors with p53 mutations [23]. Consistent with previous large multicenter studies, ER/PR positivity generally predicts better survival, although the protective effect varies across histological subtypes [24]. We analyzed patients with HGSC in detail and found that ER/PR expression status more accurately predicts prognosis in advanced-stage tumors.

Multivariate Cox regression analysis identified several independent prognostic factors. These include advanced stage, tumor volume, tumor thrombus, p53, Ki67, and ER/PR status. Among these, ER/PR double-negative status was the strongest risk factor. Clinical staging reflects tumor burden and metastatic potential. Tumor volume is linked to local stress and hypoxia, factors that contribute to tumor progression and chemotherapy resistance [25]. P53 mutations lead to impaired apoptosis and genomic instability, while high Ki67 expression indicates active cell cycle activity. Both factors are closely related to tumor progression and chemotherapy resistance [26]. The absence of ER/PR may interact with the high-risk factors mentioned above, jointly enhancing tumor aggressiveness. This interaction leads to increased proliferation, higher metastatic potential, and primary resistance to endocrine therapy. Advanced stage, tumor size, and molecular markers are consistently included in risk scoring across different cohorts and international prediction models. However, the weighting of these factors varies across ethnic groups and geographic regions [27, 28]. In retrospective cohort studies, PSM is widely used to address confounding factors such as clinical characteristics and treatment methods. When the sample size is sufficient, PSM can reduce bias more effectively than other methods such as inverse probability weighting and stratified analysis [29, 30]. However, PSM has drawbacks, including sample size loss and sensitivity to variable selection. Some literature suggests combining multiple statistical methods in multi-center, large-sample studies to improve robustness [31, 32]. Therefore, we applied PSM solely to eliminate confounding from baseline characteristic imbalance, while the prognostic model was developed using the original sample to verify the adverse effect of ER/PR double negativity.

Endocrine therapy has gradually been introduced for ovarian cancer in recent years. However, its effectiveness is significantly reduced due to complex resistance mechanisms [33, 34]. Research shows that there are significant differences in drug sensitivity among individuals. Prognostic models that combine hormone receptors with clinical and pathological features help identify patients who may benefit from endocrine therapy and those at high risk of primary or acquired resistance. In developing prognostic assessment tools, we integrated ER/PR and several clinical and pathological parameters to establish a visual nomogram model, which shows great predictive accuracy and is clinically useful. Multi-factor risk integration, based on the concept that a single biomarker cannot fully capture tumor diversity, improves the performance of predictive models by combining multiple parameters [35]. Compared to previous nomogram models for ovarian cancer and other solid tumors, multi-factor models incorporating molecular and clinical factors generally outperform traditional staging systems in metrics such as AUC (area under the curve), C-index, and DCA (decision curve analysis) clinical net benefit [28, 36, 37]. This model is innovative not only in integrating biomarkers for outcome prediction, but also in providing a framework for personalized prognostic assessment.

Although this study controlled for confounding factors to a significant extent through rigorous propensity score matching and multivariate regression, it was still a single-center retrospective study. Therefore, selection and information biases during patient screening and follow-up might still exist, potentially affecting the generalizability of the results. Additionally, the small sample sizes in some ER/PR-defined subgroups, particularly the ER(-)/PR(+) subgroup, may have reduced statistical power and limited the stability of subgroup-specific estimates. The subjectivity of immunohistochemical scoring might also affect the reliability of conclusions regarding PR expression variability. Therefore, some findings require further validation in larger, multicenter, prospective studies.

Conclusion

In this cohort of HGSC patients, combined ER/PR expression status served as an independent prognostic indicator. A nomogram integrating this biomarker with clinicopathological characteristics showed potential value for individualized risk assessment. Future multicenter prospective studies are needed to further validate these findings and clarify their broader clinical applicability.

Supplementary Information

Supplementary Material 1. ^{(314.6KB, docx)}

Abbreviations

AR

androgen receptor

AUC

area under the curve

DCA

Decision Curve Analysis

ER

estrogen receptor

HGSC

High-grade serous ovarian cancer IHC Immunohistochemistry

LGSOC

low-grade serous ovarian cancer

MMPs

matrix metalloproteinases

OS

overall survival

OTTA

Ovarian Tumor Tissue Analysis

PR

progesterone receptor

PSM

Propensity score matching

ROC

Receiver Operating Characteristic

Authors’ contributions

Chengcheng Zhu and Hengliang Sun developed the project, designed the methodology, curated and analyzed data, and drafted the original manuscript; Yonghong Luo, Dandan Ge, Shun Yao, Yi Wang and Meng Yan contributed to data curation, validation, investigation and manuscript revision; Yuanyuan Lyu and Jiangli Liu supervised the project, revised and edited the manuscript, and provided conceptual guidance. All the authors have read and approved the final manuscript.

Funding

This study was supported by the Clinical and Translational Research Project of Anhui Province(No: 202427b10020084, 202427b10020096)and the Anhui Province University Middle-aged and Young Teachers’ Training Action Project (No: YQYB2025017).

Data availability

All data supporting the findings of this study are available within the paper and its Supplementary Information.

Declarations

Ethics approval and consent to participate

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. The study was approved by the ethical committee of Wannan Medical College (approval number: 2024207), in which informed consent was waived for retrospective observational study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.