Clinicopathological Features and Survival Trends of Non-Epithelial Ovarian Cancer: Analysis of the Surveillance, Epidemiology, and End Results (SEER) Database Free

Ovarian cancer is a common gynecological disease among women and is divided into epithelial and non-epithelial types, of which epithelial cancers constitute by far the majority. Ovarian non-epithelial tumors arising from the ovary are rare and account for 3–8% of ovarian cancers [1‒3]. Non-epithelial ovarian cancers are broadly classified into 2 major groups: malignant ovarian germ cell tumors (MOGCTs) and sex cord-stromal tumors (SCSTs), and the yearly-adjusted incidence rate is 3.7/1,000,000 and 2.1/1,000,000, respectively [4]. Although MOGCTs and SCSTs are rare, they are usually detected at an early stage and have a more favorable prognosis than ovarian epithelial tumors. The first choice of treatment is surgery. Most MOGCTs and SCSTs can be treated surgically at an early stage of tumor progression and recurrence [5]. Concerning the prognosis of these two types of non-epithelial ovarian cancer, to date, there have been a few small population-based studies. In a Netherlands study, the 5-year OS of MOGCTs was 82% and of SCSTs was 78% [3]. Unfortunately, to the present date there still has been the absence of established practical guidelines for clinicians to deal with such cases as the current lack of data from large samples of a population-based study. In the current study, we obtained information on 4,062 patients from the Surveillance, Epidemiology, and End Results (SEER) database to compare clinical features and the results between patients with MOGCTs and those with SCSTs to provide a benchmark for clinical therapy and therapeutic activities.

These data were taken from the National Cancer Institute program SEER, which gathered data from 17 population-based cancer registries that encompassed approximately 30% of the entire US population. Cases of MOGCTs and SCSTs were screened in the SEER database from 2000 to 2019. As the data used were extracted from SEER dataset (public), ethics approval and consent to participate were retrieved from SEER. The cases were coded using ICD‐O‐3. Inclusion criteria were as follows: histology code: dysgerminoma (9060/3), embryonal carcinoma (9070/3), yolk sac tumor (9071/3), malignant teratoma (9080/3), mixed germ cell tumor (9085/3), granulosa cell tumor (8620/3), Sertoli-Leydig cell tumor (8631/3, 8634/3, 8640/3, 8650/3). The exclusion criteria included (1) unknown survival time, (2) not the first malignant primary tumor, (3) unknown cause of death, and (4) unknown surgical treatment. A patient selection criteria flowchart is shown in Figure 1.

We analyzed patient demographic data, clinical characteristics, and treatment patterns, including the age at the time of diagnosis (grouped into <40, 40–59, and >59 years), marital status (married, widowed/separated/divorced, single, and unknown), race (black, white, other, and unknown), laterality (bilateral, only one side, and unknown), tumor size (grouped into <93, 93–219, >219 mm, and unknown), SEER stage (distant, regional, localized, and unknown), grade (grade I-II, grade III-IV, and unknown), International Federation of Gynecology and Obstetrics (FIGO) stage (I, II, III, IV, and unknown), cancer antigen 125 (CA125) status (elevated, normal, and unknown), lymph nodes status (positive, negative, no examined, and unknown),surgery (yes and no), and chemotherapy (yes and no). CA125, although not as specific as inhibin (A or B) and AFP for non-epithelial tumors, was not able to obtain additional tumor markers from the SEER database. In the SEER database, the overall survival (OS) is defined as the length of time between the diagnosis of cancer and the date of death of the patient. Cancer-specific survival (CSS) will be calculated from the last day of treatment until death from MOGCTs and SCSTs. To assess OS and CSS, we extracted follow-up data on time since diagnosis, the vital status of the patient, and the cause of death.

X-tile software was used to determine the optimal cut-off values for age of diagnosis and tumor size [6]. For age, the best minimum and maximum cut-off values were assigned at 40 and 59 years, respectively. For tumor size, the best minimum and maximum cut-off values were assigned at 93 mm and 219 mm, respectively (shown in Fig. 2). Comparisons of demographic and clinical features between patients with MOGCTs and patients with SCSTs were performed using Fisher’s exact test or χ² test. To balance confounding factors, we conducted propensity-score matching to create a 1:1 matched set. For the MOGCTs and SCSTs, the Kaplan-Meier survival curve was used for comparing the rates of survival before and after propensity-score matching [7]. Cox proportional hazard regression analysis was used to identify independent predictors of survival. This study employed R version 4.1.1 to conduct all statistical analyses. Hazard ratios and 95% confidence intervals were calculated. Statistical significance was set at p < 0.05.

The study collected data for 4,062 patients with a diagnosis of MOGCTs or SCSTs from 2000 to 2019 from the SEER database, which consisted of 2,506 patients with MOGCTs and 1,556 patients with SCSTs. According to the histological type documented patient demographics, which are shown in Table 1. Compared with patients with MOGCTs, those with SCSTs were more probable between the ages of forty and fifty-nine (48.3% vs. 5.3%; p < 0.001). Additionally, individuals with SCSTs were more likely to be black (23.3% vs. 11.6%; p < 0.001) as well as married (46.9% vs. 23.1%) or separated/widowed/divorced (18.4% vs. 3.7%; p < 0.0001) compared with those with MOGCTs. Furthermore, MOGCTs and SCSTs occurred more in the single-sided (95.1% vs. 91.1%; p < 0.0001), and the tumor size of MOGCTs was bigger at the time of diagnosis than the tumor size of SCSTs (93–219 mm, 44.9% vs. 25.4% and >219 mm, 12.1% vs. 8.7%; p < 0.001), respectively. Most MOGCTs and SCSTs patients had a diagnosis of FIGO stage I disease (57.5% vs. 53.9%; p < 0.001) and SEER localization stage (51.6% vs. 48.4%; p < 0.001), respectively. Surgery was performed in the vast majority of patients with both MOGCTs and SCSTs (98.1% vs. 94.5%; p < 0.001), meanwhile patients in the MOGCTs group received more chemotherapy than those in the SCSTs group (56.1% vs. 32.2%; p < 0.001), respectively. Higher levels of CA125 (20.5% vs. 12.0%; p < 0.001) and more common lymph node-positive (8.4% vs. 1.9%; p < 0.001) were seen in patients with MOGCTs compared with SCSTs.

Kaplan-Meier survival curves were used to describe the survival difference between patients with MOGCTs and those with SCSTs. The 1-, 3-, and 5-year OS rates of MOGCTs were 96.9%, 95.2%, and 94.3%, respectively, while those of SCSTs were 95.3%, 89.8% and 86.2%, respectively. In contrast, as far as CSS is concerned, SCSTs at 1, 3, and 5 years had worse rates of survival than MOGCTs (all p < 0.0001; shown in Fig. 3). Age at the time of diagnosis (years), laterality, SEER stage, grade, FIGO stage, CA125, surgery, chemotherapy, and lymph node status are prognostic factors independent of OS and CSS in univariate Cox regression models that were limited to women who underwent MOGCTs (Table 2). Cox multivariable regression analysis revealed that age at diagnosis, grade of differentiation (III-IV), FIGO III-IV, and surgery were independent risk factors for OS and CSS with MOGCTs (shown in Fig. 4, 5). The results of the univariate and multivariate Cox regression analyses for SCSTs to identify independent risk factors are shown in Table 3. In multivariable Cox regression analysis for SCSTs, laterality, tumor size, SEER stage, grade of differentiation (III-IV), and CA125 were independent risk factors for CSS and OS, in contrast to MOGCTs. Surgery was identified as an independent predictor of OS, while chemotherapy was identified as an independent predictor of CSS (shown in Fig. 6, 7). Furthermore, the results of the SEER database indicate that positive lymph node status was a worse prognosis factor for SCSTs. There was no difference in OS and CSS between patients with FIGO stage I-IV SCSTs who did and did not receive any type of chemotherapy (shown in Fig. 8). In patients with MOGCTs, chemotherapy led to improvements in OS and CSS in both II and III stages of FIGO (shown in Fig. 9).

The study matched 563 MOGCT patients with 563 SCST patients, as shown in Table 4. p values were all greater than 0.05, as expected, demonstrating that the data were balanced. Continuing this line of investigation, the K-M method was used for survival analysis. The rate of 1‐year and 3‐year OS in patients with SCSTs was 96.1% and 90.2%, respectively, which was higher than that of MOGCTs (with 1-year OS rate of 92.9% and 3-year OS rate of 89.9%), which is in disagreement with the previous study that at 5 years, 87.6% of patients had OS rates that were poorer than those of MOGCTs (with 5-year OS rate of 89.2%) (p = 0.0038; shown in Fig. 10a). Patients with SCSTs had higher 1-year CSS rates (with 1-year CSS rate of 96.3%) but poorer 3-year and 5-year CSS rates (with 3-year CSS rate of 90.9% and 5-year CSS rate of 88.8%) compared to patients with MOGCTs (with 1-year CSS rate of 94.0%, 3-year CSS rate of 91.9%, and 5-year CSS rate of 91.5%) (p = 0.0022; shown in Fig. 10b).

Because of the scarcity of non-epithelial ovarian cancer, the collection of a vast amount of data to advance medical knowledge regarding non-epithelial ovarian cancer is challenging. As far as we know, this is the largest population-based study investigating the epidemiology and clinical outcomes of female patients with non-epithelial ovarian cancer. We enrolled 4,062 patients from the SEER database and assessed survival outcome trends in patients with MOGCTs and SCSTs from the SEER database between 2000 and 2019 using both Kaplan-Meier and Cox regression analyses.

There is variation in the percentage of ovarian tumors that are non-epithelial in origin in different research and there seems to be some geographic variation. A study based on the basis of a global population of ovarian tumors found that non-epithelial tumors account for approximately 5–6% of all North American cases in Oceania and Europe, and the ratio has tended to be more elevated in Central and South America as well as in Asia (approximately 9%) [2]. MOGCTs and SCSTs accounted for 2.3% and 1.4% of all ovarian cancers in the SEER database, respectively, in our study. Non-epithelial ovarian cancers tend to have a much better prognosis than the bulk of their counterparts in the epithelium [8]. However, the probability of survival of MOGCTs and SCSTs varies noticeably between studies, and there are only a small number of previous investigations that have looked at the time course of survival of these cancers, as well as the influence of stage at diagnosis [3, 8, 9]. Consistent with our findings, several researchers have suggested that patients with SCSTs have a poor survival rate when compared with MOGCTs [1, 3, 10]. In our research, PSM was used to adjust the confounders, increasing the robustness of the results [11]. Torre et al. [12] found that both MOGCTs and SCSTs have high 5-year survival rates of 99% for MOGCTs and 98% for SCSTs. Survival rates remain relatively high even for FIGO stage IV disease. A large multicenter study of global ovarian cancer survival comparisons reported wide worldwide variations in the 5-year survival of non-epithelial ovarian cancers, ranging from 59% in Japan to 100% in Korea for SCSTs and 42% in China to 76% in Australia for MOGCTs [8].

MOGCTs are derived from the primitive germ cell of the embryonic gonad, approximately 1–2% of all ovarian cancers [13‒15]. 70% of patients develop FIGO stage I disease at the time of diagnosis [16]. In terms of clinical characteristics, MOGCTs mainly take place during the first 30 years of life [17]. Siegel et al. [18] found that 3% of cases occurred from birth to the age of 14 years and that 11% of cases in those occurred between the ages of 15 and 19. In our study, we found that MOGCTs had a younger age distribution (<40 years old) and more often had FIGO stage I disease (57%). Guidelines from the Gynecologic Cancer Inter Group (GCIG) recommend that the gold standard for early MOGCT patient management is fertility preservation [16]. In our study, we received surgical intervention that reduced the risk of death for patients with MOGCTs. A 5-day regimen of bleomycin/etoposide/platinum (BEP) is recommended by the guidelines of the European Society for Medical Oncology (ESMO) as the adjuvant chemotherapy of choice for MOGCTs [19]. In our study, we found that there was no statistical difference in the survival rate between patients with or without chemotherapy for FIGO stage I and IV patients with MOGCTs.

SCSTs originate from either stromal cells or the sex cord, or both, and comprise 1.3% of all ovarian cancers [1]. Over 70% of SCST patients are diagnosed at an early stage, which is usually indolent [20]. SCSTs occur across age groups, with granulosa cell tumors occurring mainly in peri- and postmenopausal women and Sertoli-Leydig cell tumors largely occurring in young females [19, 21]. SCSTs can be treated with surgery, chemotherapy, radiation, and targeted therapy. The standard management is reductive debulking surgery, which may be followed by platinum-based adjuvant chemotherapy. The prognosis for patients with stage I SCST is excellent; adjuvant therapy is not required in patients with FIGO stage I clinical disease who have received surgical treatment [22]. Although the use of chemotherapy remains a subject of debate in FIGO stage IC, a study demonstrated that adjuvant chemotherapy was not able to prolong disease-free survival in FIGO stage IC [23]. Adjuvant chemotherapy is recommended for patients with stage II or greater [24]. However, another study found that adjuvant chemotherapy did not lead to improved survival in patients with FIGO stage II-IV SCSTs [25]. In our study, there was no statistically significant difference in FIGO stage I-IV SCSTs response with or without chemotherapy, which is in agreement with the results of Oseledchyk’s study [26].

In clinical decision-making, based on findings on tumor biomarkers, ultrasonography, and physical examination, women undergo surgery to diagnose and treat adnexal mass. For example, more specific tumor biomarkers such as fetoprotein α, human chorionic gonadotropin β-subunit (β-hCG), and lactic dehydrogenase (LDH) were useful in diagnosing non-epithelial ovarian cancer [27]. Given the rarity of non-epithelial ovarian cancer, data on the role of CA125 in the preoperative diagnosis of these rare tumors are scarce. In earlier research, Pitta et al. [28] found that women with non-epithelial ovarian cancer did not express increased levels of CA125 as did women with epithelial ovarian cancer. As part of our research, we found elevated CA125 levels accounted for 20.5% of MOGCTs and 12% of SCSTs.

This study has several limitations that need to be acknowledged. The largest strength of the current study was the large sample size provided for studying non-epithelial ovarian cancer by the SEER database. The SEER database gives information about whether or not subjects received any form of chemotherapy. Nevertheless, there is no indication of the type of chemotherapy or the number of chemotherapy cycles received. In addition, details of the timing and location of tumor recurrence are not available in the database. Finally, our study focused solely on OS and CSS of patients, unaccounted for disease-free survival or cancer recurrence, which limits the applicability of our findings in clinical research. Further research with a larger sample population is needed to deal with these limitations.

We had signed the SEER research data agreement. The data in this research were obtained from the SEER database following approved guidelines. The information on patients had been studied by the US Department of Health and Human Services. The data is publicly available and deidentified after permission. We confirm that the research was performed in accordance with the principles stated in the Declaration of Helsinki. No personally identifying information was used in the study, which eliminated the requirement for Institutional Review Board approval or informed patient consent.

The authors have no conflicts of interest to declare.

There are no funding sources to declare.

C.X.: conception, design of the study, acquisition of data, analysis and interpretation of data, and drafting the article; X.X.: critical revision for important intellectual content. All authors contributed to the article and approved the submitted version.

Publicly available datasets were analyzed in this study. These data can be found here: https://seer.cancer.gov/data/. Further inquiries can be directed to the corresponding author.