Establishment of a predictive survival prognosis model for epithelial ovarian cancer: a study based on the SEER database

1. Introduction

Epithelial ovarian cancer (EOC) is one of the most aggressive malignancies among gynecological tumors globally, accounting for more than 90% of all ovarian cancer cases [1]. EOC includes various subtypes such as serous, mucinous, endometrioid and clear cell. This heterogeneity in pathological and molecular characteristics poses significant challenges to effective clinical management [2, 3]. Due to its insidious onset and nonspecific early symptoms such as mild abdominal distension, digestive discomfort and general malaise, approximately 75% of patients are diagnosed at an advanced stage (Stage III or IV) [4]. Consequently, the overall prognosis remains poor, especially in advanced-stage cases, with a 5-year survival rate of less than 30% [5]. Despite advancements in surgical techniques, chemotherapy, targeted therapy and immunotherapy, improvements in survival outcomes have been modest, underscoring the urgent need for more precise and individualized treatment approaches [6, 7]. The Surveillance, Epidemiology and End Results (SEER) database, maintained by the United States National Cancer Institute (NCI), is one of the most comprehensive and authoritative cancer registries worldwide [8, 9]. It compiles extensive cancer-related data from multiple geographic regions across the U.S., including patient demographics, tumor pathology, treatment modalities and survival outcomes. This makes it an invaluable resource for conducting large-scale epidemiological and prognostic studies in oncology [10].

In this study, we utilize the SEER database to explore the key prognostic factors affecting survival in patients with EOC. By establishing a COX proportional hazards regression model and combining it with a nomogram for survival prediction, we aim to identify variables most closely associated with prognosis. This may serve as a reference for personalized treatment strategies and contribute to the advancement of precision medicine in the management of EOC.

2. Materials and methods

3. Results

3.1 Comparison of general characteristics

After screening the database, the study included 915 patients diagnosed with EOC in the final analysis. The patient selection process is illustrated in Fig. 1. Among these patients, 95 (10.38%) were deceased at the time of the final follow-up, while 820 (89.62%) remained alive. Comparative analysis between the deceased and surviving groups revealed statistically significant differences in several variables including radiation status, chemotherapy, diagnosis to treatment Time, Regional lymph nodes positivity, operation and Grade pathological (p < 0.05), Table 1.

Fig. 1.Flowchart of case inclusion. SEER: Surveillance, Epidemiology and End Results.

Table 1.Comparison of general characteristics between the deceased group and the survived group.

Type	Deceased group (n = 95)	Survived group (n = 820)	χ2	p
Age
	<60 yr	62 (65.3)	558 (68.0)	0.302	0.582
	≥60 yr	33 (34.7)	262 (32.0)
Marital
	Unmarried	27 (28.4)	177 (21.6)	2.296	0.130
	Married	68 (71.6)	643 (78.4)
Income
	<100,000$	50 (52.6)	454 (55.4)	0.257	0.612
	≥100,000$	45 (47.4)	366 (44.6)
Radiation
	No	88 (92.6)	805 (98.2)	11.132	0.001
	Yes	7 (7.4)	15 (1.8)
Chemotherapy
	No	18 (18.9)	245 (29.9)	4.966	0.026
	Yes	77 (81.1)	575 (70.1)
Diagnosis To Treatment Time
	<7 d	54 (56.8)	641 (78.2)	21.207	<0.001
	≥7 d	41 (43.2)	179 (21.8)
Tumor Size
	<80 mm	25 (26.3)	257 (31.3)	1.009	0.315
	≥80 mm	70 (73.7)	563 (68.7)
Tumor Laterality
	Unilateral	72 (75.8)	678 (82.7)	3.439	0.179
	Bilateral	21 (22.1)	135 (16.5)
	Paired	2 (2.1)	7 (0.9)
Regional Lymph Nodes Positive
	No	65 (68.4)	686 (83.7)	13.437	<0.001
	Yes	30 (31.6)	134 (16.3)
Pathological Grade
	Well differentiated	18 (18.9)	439 (53.5)	53.284	<0.001
	Moderately differentiated	21 (22.1)	118 (14.4)
	Poorly differentiated	35 (36.8)	111 (13.5)
	Unknown	21 (22.1)	152 (18.5)
FIGO staging
	1	0 (0.0)	264 (32.2)	160.734	<0.001
	2	26 (27.4)	275 (33.5)
	3	41 (43.2)	263 (32.1)
	4	28 (29.5)	18 (2.2)
Operation
	No	19 (20.0)	64 (7.8)	15.351	<0.001
	Yes	76 (80.0)	756 (92.2)
FIGO: International Federation of Gynecology and Obstetrics.

3.2 Factors influencing OS prognosis in EOC patients

A univariate COX proportional hazards regression analysis was performed to evaluate the impact of individual clinical variables on the overall survival (OS) of EOC patients. Survival time was treated as the time variable, and OS status (alive or deceased) was used as the event indicator. Variables that demonstrated statistically significant differences between the deceased and surviving groups in the baseline comparison were included in the analysis Stratification was conducted based on Grade pathological. The results indicated that FIGO stage, operation, radiotherapy and the time interval from diagnosis to treatment were significantly associated with the OS prognosis of EOC patients (p < 0.05), as shown in Table 2. Further survival curves revealed that patients with Grade 3 tumors exhibited a notably more rapid decline in survival probability within the first 36 months compared to patients with lower-grade tumors as. Additionally, patients with Grade 2 tumors had worse survival outcomes than those with Grade 1 tumors, highlighting the impact of tumor differentiation on prognosis, as illustrated in Fig. 2.

Table 2.Univariate COX regression analysis of overall survival in EOC patients.

Type	β	SE.	Wald χ2	p	HR	95% CI
FIGO staging	1.049	0.135	60.765	<0.001	2.856	2.194–3.718
Operation	−0.662	0.267	6.154	0.013	0.516	0.306–0.870
Radiation	0.954	0.398	5.735	0.017	2.596	1.189–5.667
Chemotherapy	0.166	0.271	0.375	0.540	1.180	0.694–2.007
Diagnosis To Treatment Time	0.728	0.216	11.367	0.001	2.072	1.357–3.164
Regional Positive lymph nodes	0.353	0.232	2.308	0.129	1.423	0.903–2.244
SE.: Standard Error; HR: Hazard Ratio; CI: Confidence Interval; FIGO: International Federation of Gynecology and Obstetrics.

Fig. 2.Survival curves.

3.3 Construction of the nomogram for prediction model

Based on the results of the multivariate COX regression analysis, a nomogram for predicting the 1-year, 2-year and 3-year OS of EOC patients was established using R software, incorporating the three prognostic factors of FIGO staging, operation, time interval from diagnosis to treatment, and tumor differentiation grade, See Fig. 3.

Fig. 3.Nomogram for OS in EOC patients. FIGO: International Federation of Gynecology and Obstetrics.

3.4 Model validation

The performance of the nomogram prediction model was assessed using both calibration curves and ROC curves. The calibration curve demonstrated a high degree of agreement between the predicted and actual survival outcomes, indicating that the model was well-calibrated for 1-year, 2-year and 3-year overall survival in EOC patients (Fig. 4). The ROC curve was employed to assess the discriminatory ability of the model. The area under the curve (AUC) values for 1-year, 2-year and 3-year OS predictions were 0.853, 0.840 and 0.847, respectively. These results suggest that the model has good predictive accuracy. In a secondary analysis or subgroup setting (if applicable), AUC values of 0.624, 0.615 and 0.667 were observed, indicating a moderate predictive performance in that context (Fig. 5).

Fig. 4.Calibration curve of the COX regression model.

Fig. 5.ROC curves for predicting 1-year, 2-year and 3-year OS using the model. OS: overall survival; AUC: area under the curve; ROC: Receiver Operating Characteristic.

4. Discussion

Ovarian cancer remains one of the most common and lethal malignancies affecting the female reproductive system. Its global incidence continues to rise, posing a substantial threat to women’s health. A major challenge in the management of ovarian cancer is the lack of specific symptoms in the early stages, leading to delayed diagnoses, with most patients presenting at advanced or late stages, where treatment options are limited and the prognosis is generally poor [12, 13]. Based on histological subtypes, ovarian cancer can be classified into various types, such as epithelial, sex cord-stromal and germ cell tumors, with EOC being the most prevalent EOC accounts for approximately 90% of all ovarian cancers cases [14]. Due to the complexity of its biological characteristics and pathological subtypes, EOC exhibits significant heterogeneity in terms of etiology, disease progression and treatment response. Although cytoreductive surgery combined with platinum-based chemotherapy remains the standard treatment regimen for EOC, the survival benefits differ markedly between individuals [15, 16]. This variability underscores the need for reliable prognostic tools that can help tailor treatment approaches and improve patient outcomes [17]. Accurate prognostic assessment in the early stages of EOC, not only aids in identifying high-risk patients for more aggressive treatment strategies but also avoids unnecessary treatments for low-risk patients, thereby reducing treatment burden and adverse reactions.

In this study, an in-depth analysis of the survival prognosis of EOC patients was conducted based on data from the SEER database, comparing multiple variables between the survival and deceased groups. There was no significant difference in age, marital status, income, tumor size, left and right distribution of tumor and lymph node status between the two groups. Age is widely recognized as an important factor in the prognosis of ovarian cancer [18], but the results of this study showed that mortality rates were comparable between patients aged ≥60 years and those under 60 years. This contrasts with previous studies that have suggested advancing age is associated with reduced immune function, diminished treatment tolerance, and subsequently worse survival outcomes in ovarian cancer patients After adjusting for potential confounding factors, patients aged 70 and older had a 1.4-fold increased risk of cancer-specific mortality [19]. The reason for the insignificant difference between the survival and deceased groups in this study may be related to the age threshold. Additionally, Zhuxuan Fu’s research [20] suggested that the specific mechanism by which age affects the prognosis of EOC patients may involve lifetime ovulation duration and menopause age. Therefore, assessing patient prognosis solely based on age may yield different results among studies. Additionally, broader evidence indicates that the association between age and prognosis in EOC may not be uniformly strong across all populations or study designs [21]. Pathological grade is an important indicator reflecting the malignancy of a tumor. Previous studies have shown that married patients often have better social support systems and better psychological states, which may help improve prognosis [22]. However, this study did not directly indicate differences in marital status between the survival and deceased groups, suggesting that the specific impact of marital status on ovarian cancer prognosis may vary between studies and requires comprehensive consideration of other factors for analysis. In this study, income did not have a significant impact on survival, which may be related to regional differences and income distribution characteristics in the SEER data. Although previous studies have shown that individuals with better economic conditions may have access to better medical resources, the prognosis of EOC patients may depend more on the effectiveness of treatment itself [23]. Tumor size can reflect the severity of the disease to some extent, as larger tumors are often linked to higher malignancy and aggressiveness, making them more prone to metastasis and recurrence. However, in this study showed, no statistically significant difference in tumor size was observed between the survival and deceased groups. This suggests that factors such as tumor differentiation grade and degree of invasion may play a more critical role in determining prognosis, and that tumor size alone may not be a reliable predictor of patient outcomes. Prognostic differences have also been noted between unilateral and bilateral tumors. Studies have shown that patients with bilateral ovarian cancer generally have poorer prognoses, which may be related to the fact that bilateral cancer usually represents more severe disease progression [24]. Although this study did not find a significant association between laterality and survival, the presence of bilateral tumors may still warrant closer clinical attention when assessing prognosis. In contrast, regional lymph node positivity was confirmed as a critical prognostic indicator in this study.

The results of this study revealed statistically significant differences between the survival and deceased groups in several key variables, including radiotherapy, chemotherapy, time from diagnosis to treatment, regional lymph node positivity, pathological grade, FIGO staging and surgical intervention. Notably, the proportion of patients with poorly differentiated tumors was significantly higher in the deceased group than in the survival group. Consistent with previous research, tumor cells with lower pathological grades (i.e., poorly differentiated) exhibit higher proliferative capacity and greater aggressiveness, contributing to a worse prognosis [25]. Chemotherapy, as one of the primary treatment methods for ovarian cancer, was associated with significantly improved survival outcomes in this study. Patients who received chemotherapy had notably higher survival rates than those who did not, aligning with earlier findings that support the role of chemotherapy in eliminating residual cancer cells and preventing recurrence [26]. These results underscore the importance of initiating chemotherapy as early as possible following diagnosis to enhance treatment effectiveness. Conversely, patients who received radiotherapy showed significantly lower survival rates. This is likely due to the clinical context in which radiotherapy is typically employed: it is not a first-line treatment for ovarian cancer and is generally reserved for locally advanced or symptomatic cases to reduce tumor burden and alleviate discomfort [27]. As such, patients selected for radiation therapy are often in the advanced or terminal stages, which may account for their poorer survival outcomes.

The study found that patients who began treatment within 7 days of diagnosis had better survival outcomes. Early treatment initiation likely contributes to slowing disease progression, thereby improving prognosis [28]. In contrast, treatment delays may lead to tumor progression and metastasis, thereby reducing treatment effectiveness and prognosis. These findings emphasize the critical importance of minimizing the diagnostic-to-treatment interval as a key strategy in improving survival in ovarian cancer patients. Furthermore, the presence of positive regional lymph nodes was associated with lower survival rates, reinforcing its role as a marker of poor prognosis. This aligns with the previous studies that have shown that lymph node metastasis is indicative of increased tumor aggressiveness and advanced disease stage [29]. Therefore, lymph node status should be carefully assessed and integrated into prognostic evaluations to guide clinical decision-making.

Further development of the COX proportional hazards model identified four independent prognostic factors, including grade pathological, FIGO staging, operation and time from diagnosis to treatment. Based on these variables, a nomogram was constructed to predict overall survival in epithelial ovarian cancer (EOC) patients. The model calibration curve had high consistency and good prediction accuracy. Compared with the existing prognostic models (such as FIGO staging, C-Reactive Protein (CRP), albumin lymphocyte index), our nomogram introduces a new clinical tool that incorporates both the treatment timing and surgical intervention status. Zhang Y [30] reported that advanced tumor stage is associated with lower surgical compliance, which itself is a crucial and independent predictor of overall survival in patients undergoing ovarian cancer (OC) treatment. Wang W [31] suggested that tumor staging is an independent prognostic factor for epithelial ovarian cancer (EOC) patients, and the combination of tumor staging with the CRP-albumin-lymphocyte index can effectively predict patients’ overall survival. High-grade EOC patients generally have poorer prognoses because these tumors are more prone to recurrence and metastasis. Therefore, Grade Pathological emerged as a significant prognostic factor in the COX regression model, with a measurable impact on both the survival duration and overall survival rate in EOC patients. In the traditional treatment, surgery remains the first-line treatment option for ovarian cancer Patients who do not undergo surgical intervention often present with more advanced disease, have poor general health, or possess tumor characteristics that preclude complete resection, making them ineligible for surgery. In such cases, radiotherapy is typically administered as a palliative measure, aiming to relieve symptoms and achieve local disease control, rather than to provide curative intent. As a result, the overall prognosis for patients who do not undergo surgery is generally poor, reflecting both the severity of their disease and the limitations of non-surgical treatment modalities [32]. Although radiotherapy may offer short-term symptom relief and local disease control, its impact on prolonging overall survival in patients with advanced EOC appears to be limited [33]. Timely initiation of treatment following diagnosis is widely recognized as a critical factor in cancer management. This study showed that patients who received treatment within 7 days after diagnosis had significantly better survival prognoses than those who experienced treatment delays. Given that epithelial ovarian cancer (EOC) is frequently diagnosed at an intermediate or advanced stage, and is characterized by rapid disease progression, early intervention is vital for improving overall survival [34]. Studies have shown that the extent of residual disease following initial cytoreductive surgery is a key prognostic indicator in advanced ovarian cancer [35]. Additionally, opportunistic salpingectomy has been shown to positively impact survival outcomes, particularly in serous ovarian cancer [36]. Delays in initiating treatment may reduce the likelihood of optimal surgical resection and limit the effectiveness of therapeutic interventions. The results of this study further confirm the importance of timely treatment in EOC management and highlight the need for clinicians to minimize diagnostic-to-treatment intervals wherever possible. This model facilitates the identification of high-risk patients (e.g., those experiencing delayed treatment, and presenting with high-grade tumors) who may benefit from intensified monitoring. However, challenges remain in accounting for real-world variability in treatment practice and validating the model across diverse patient populations.

5. Conclusions

This study used large-scale, multicenter data from the Surveillance, Epidemiology and End Results (SEER) database to develop and validate a predictive model for the survival prognosis of patients with EOC. The results demonstrated that treatment delay following diagnosis, tumor stage and operation, radiotherapy are significant factors influencing patient outcomes. Using COX regression models and ROC curve analysis, the predictive efficacy of the model was assessed at multiple time points, confirming its robustness and validity in predicting 1-year, 3-year and 5-year survival rates. Compared with existing literature, this study offers updated data support for enhancing precision medicine and prognosis assessment in EOC. By identifying high-risk patient groups, the model enables clinicians to formulate targeted and individualized treatment plans, with the potential to improve both survival rates and quality of life for EOC patients. However, this study has certain limitations. Firstly, the data from the SEER database are from specific regions in the United States, potentially limiting their applicability to other populations. Secondly, the database lacks detailed molecular biological indicators, preventing in-depth exploration of the impact of molecular factors such as gene mutations on prognosis. Therefore, future studies should integrate multicenter datasets and include molecular-level variables to further improve the predictive accuracy and applicability of survival models for EOC.

Availability of data and materials

The authors declare that all data supporting the findings of this study are available within the paper and any raw data can be obtained from the corresponding author upon request.

Author contributions

XLY, YF—designed the study and carried them out; prepared the manuscript for publication and reviewed the draft of the manuscript. XLY, YF, YL—supervised the data collection; analyzed the data; interpreted the data. All authors have read and approved the manuscript.

Ethics approval and consent to participate

This article does not contain any studies with human participants or animals performed by any of the authors.

Acknowledgment

Not applicable.

Funding

This research received no external funding.

Conflict of interest

The authors declare no conflict of interest.