This retrospective cohort study used data from the Surveillance, Epidemiology, and End Results (SEER) program. The SEER database is a population-based cancer registry covering nearly 30% of the US population and collecting cancer demographics, incidence, survival, and treatment data. The SEER*Stat software version 8.3.8 was used for the analysis (9). This study was performed according to the STROCSS guidelines (Strengthening the reporting of cohort, cross-sectional and case-control studies in surgery) (10).

Female patients diagnosed with the first primary malignant ovarian cancer (site codes: C56.9) between 2000 and 2017 were extracted from the SEER 18 database (2020 submission) (11). Only patients with unilateral-originated ovarian cancer were included because patients with bilateral origin might have lost their opportunity to undergo unilateral oophorectomy. Patients diagnosed only through autopsy or death certificates were excluded. We further excluded patients without complete follow-up information, including follow-up duration and age at diagnosis. To accurately evaluate the effects of surgical operations, we further excluded patients with advanced-stage cancer or those with a cancer of unknown stage ( Figure S1 ).

Since it is a publicly available database, access to the SEER data required a signed research data agreement form. The Institutional Review Board of Zhongnan Hospital of Wuhan University waived the institutional review board approval for the data obtained from the SEER database, as the study did not directly involve human subjects, and all data were anonymized. The requirement for informed consent was waived.

All patients were followed between the time of the first primary diagnosis of ovarian cancer and the time of their death, exiting the study alive, or the end of the study (December 31, 2017). Among the patients included in this study, we evaluated the following variables: age at diagnosis, race, year of diagnosis, cancer stage, American Joint Committee on Cancer Staging (AJCC) N stage, AJCC T stage, surgical therapy, cause of death, histological types, urban/rural residency at diagnosis, median household income, follow-up time, and vital status at the end of follow-up.

As the SEER database records the survival duration in months, and a month was the shortest time interval available for analysis, survival durations shorter than 1 month were recorded as 0 months in the SEER program. Age at cancer diagnosis was divided into 4 groups for comparison: “15-39 years,” “40-59 years,” “60-79 years,” and “80+ years”. Patients aged 15-49 years were selected for specific analyses, as this proportion of women had a greater desire for fertility preservation.

For ovarian cancer, the SEER program derived TNM values of the stage from the International Federation of Gynecology and Obstetrics (FIGO) stage. Thus, FIGO information of this study was inferred from TNM-stage values. TNM-stage values were extracted from AJCC 3^rd stage codes for patients diagnosed between 2000 and 2003, AJCC 6^th stage codes for patients diagnosed between 2004 and 2009, AJCC 7^th stage codes for patients diagnosed between 2010 and 2015, and SEER combined stage for patients diagnosed in 2016 and 2017 (12). We excluded the patients with stage IV ovarian cancer.

The SEER program provided detailed site-specific surgical information for the included patients (13–15). Surgical operations for ovarian cancer were divided into two major groups: unilateral and bilateral oophorectomy. To avoid confusion, local excision/destruction and unknown surgical operations were excluded (surgery codes: 17 and 90-99). Unilateral oophorectomy was defined as total removal of the tumor or (single) ovary, and unilateral (salpingo-) oophorectomy with or without hysterectomy (surgery codes: 25-28 and 35-37). Bilateral oophorectomy includes bilateral (salpingo-) oophorectomy with or without hysterectomy, cytoreductive surgery, and pelvic exenteration (surgery codes: 60-74). Surgical operations with unknown laterality were excluded for accuracy (surgery codes: 55-57 and 80) (13).

Causes of death of patients with ovarian cancer were classified into two major groups: death from cancer (i.e., a second primary cancer) and death from non-cancer comorbidities (i.e., deaths from any medical cause other than cancer). Causes of death were defined by the SEER cause-specific death classification variable from death certificates (15–17). Non-cancer causes were categorized into 26 major groups. These groups were further divided into seven broad categories: infectious diseases, cardiovascular diseases (CVD), respiratory diseases, gastrointestinal and liver diseases, renal diseases, external injuries, and other non-cancer causes.

We estimated the characteristics of the patients with ovarian cancer. Trends in surgical operations were characterized by age at diagnosis and year of diagnosis. Moreover, we analyzed the overall survival (OS) and disease-specific survival (DSS) of patients using the Kaplan-Meier method. The OS rate was defined as the percentage of survivors (all causes of death) after follow-up. The DSS rate was defined as the percentage of patients who have not died from ovarian cancer (rather than from other causes) in a defined period of time (18). Cox regression models were used to assess the significance of differences in the OS and DSS analyses. The cumulative mortality rate (CMR) was estimated for non-cancer comorbidities (15).

All analyses were performed using SEER*Stat software version 8.3.8 (9) and R 3.6.3 (19). Tests were two-tailed, with a p-value of less than 0.05 considered statistically significant.

In this population-based study involving 28,480 women with stage I-III ovarian cancer, 11,517 (40.4%) deaths were recorded, with a median follow-up time of 4.1 years (range: 0–17.9 years) ( Figure S1 and Table 1 ). Most of the patients were aged 40–79 years (83.3%) and were white (82.6%). Of the cancers, 46.8% were stage-I tumors. Serous ovarian cancer accounted for the majority of tumors (51.9%), followed by endometrioid carcinoma (20.7%) ( Table 1 ).

Of the patients, 95.5% (N = 27,197) had undergone surgical operations, among whom 7.9%, 50.3%, and 41.8% underwent unilateral oophorectomy, bilateral oophorectomy, and other surgical procedures, respectively ( Table S1 ). Patients who underwent unilateral oophorectomy were younger. Most patients who underwent unilateral oophorectomy (70.9%) were younger than 60. Of the patients aged 15–39 years, 33.7% and 22.7% underwent unilateral oophorectomy and bilateral oophorectomy, respectively ( Table 1 ). Moreover, there was a decreasing trend in the unilateral oophorectomy rate by age at cancer diagnosis ( Figure 1A ). Similarly, the unilateral oophorectomy rate decreased by FIGO stage ( Figure 1B ), especially for younger patients ( Figure 1C ). Most tumors treated by unilateral oophorectomy were in the Ia stage (53.1%), followed by Ic stage (19.4%). The Hispanic population had a higher unilateral oophorectomy rate (11.3%) than the non-Hispanic population (7.0%) ( Table 1 ).

The OS and DSS of patients who had undergone surgery were significantly better than those of patients who did not (all p < 0.001) ( Figure 2 and Figure S2 ). The prognostic superiority of surgical operation could be observed in ovarian cancer at all stages ( Figures 2B, C and Figure S2 ).

To examine the therapeutic effects of unilateral oophorectomy, we performed survival analyses according to the type of surgical intervention ( Figure 3 and Figure S3 ). In stage-Ia tumor, unilateral oophorectomy was associated with remarkably better OS and DSS compared with bilateral oophorectomy, with a 5-year OS rate of 89.9% for unilateral oophorectomy and 87.9% for bilateral oophorectomy (OS: p < 0.001; DSS: p = 0.01) ( Figure 3A and Figure S3A ). For stage-Ib and stage-Ic ovarian tumor, there was no significant difference between the OS and DSS of patients treated by unilateral oophorectomy and those of patients treated by bilateral oophorectomy (stage Ib: OS: p = 0.6; DSS: p = 0.8; stage Ic: OS: p = 0.06; DSS: p = 0.2) ( Figures 3B, C and Figures S3B, C ). For stage-IIa tumors, the OS and DSS after unilateral oophorectomy were significantly worse than those after bilateral oophorectomy (5-year OS: 50.3% vs. 72.0%, p < 0.001; 5-year DSS: 61.6% vs. 72.0%, p < 0.001) ( Figure 3D and Figure S3D ). For stage-IIb/IIc tumors, there was no significant difference between the OS and DSS of patients treated by unilateral oophorectomy and those of patients treated by bilateral oophorectomy (OS: p = 0.6; DSS: p = 0.6) ( Figure 3E and Figure S3E ). For stage-III tumors, the OS and DSS after unilateral oophorectomy were significantly worse than those after bilateral oophorectomy (OS: p < 0.001; DSS: p < 0.001) ( Figure 3F and Figure S3F ).

For low-grade and high-grade stage-I ovarian tumors, there was no significant difference between the OS and DSS of patients treated by unilateral oophorectomy and those of patients treated by bilateral oophorectomy ( Figures 4A–D and Figures S4B, C ). The OS of patients with low-grade and high-grade stage-IIa ovarian tumors undergoing unilateral oophorectomy were worse than those of patients undergoing bilateral oophorectomy (low-grade: p = 0.03; high-grade: p < 0.001) ( Figures 4E, F ). The DSS of patients with high-grade stage-IIa ovarian tumor undergoing unilateral oophorectomy were worse than those of patients undergoing bilateral oophorectomy (high-grade: p < 0.001) ( Figure S4F ).

We further analyzed the survival after unilateral oophorectomy or bilateral oophorectomy in patients with ovarian cancers of different histology ( Figures S5–S8 ). We found that, except for high-grade stage-Ic serous ovarian cancer (Due to the inadequate cases with stage-Ib tumors, stage-Ib tumors were not included for further analyses hereafter), unilateral oophorectomy was comparable to bilateral oophorectomy in low-grade and high-grade stage-I ovarian cancer of any histology, including serous, mucinous, endometrioid, and clear cell carcinoma ( Figures S5–S8 ). For high-grade stage-Ic serous ovarian cancer, the OS after unilateral oophorectomy was significantly worse than that after bilateral oophorectomy (p = 0.03) ( Figure S5D ).

Female patients of reproductive age had a greater desire to preserve fertility; thus, we investigated the prevalence and therapeutic effects of unilateral oophorectomy in this peculiar population ( Figure 5 and Figure S9 ). Among the reproductive-age women younger than 50 years, 22.8% received unilateral oophorectomy, and 27.9% underwent bilateral oophorectomy. We found that OS and DSS of patients with low-grade and high-grade stage-I receiving unilateral oophorectomy were comparable to those of patients receiving bilateral oophorectomy (all p > 0.05) ( Figure 6 and Figure S9 ). For patients aged 15–59 years with high-grade stage-Ic ovarian tumor, the OS and DSS of patients receiving unilateral oophorectomy were similar to those of patients receiving bilateral oophorectomy (OS: p = 1; DSS: p = 0.7) ( Figure 5F and Figure S9F ).

For those aged 50 years and older, the OS of patients with low-grade and high-grade stage-I receiving unilateral oophorectomy was significantly worse than that of patients receiving bilateral oophorectomy (low-grade: p < 0.001; high-grade: p < 0.001) ( Figures 6A, B ). The DSS of patients aged 50 years and older with low-grade stage-I receiving unilateral oophorectomy were similar to those of patients receiving bilateral oophorectomy (p = 0.2) ( Figure S10A ). The DSS of patients aged 50 years and older with high-grade stage-I receiving unilateral oophorectomy were worse than that of patients receiving bilateral oophorectomy (p < 0.001) ( Figure S10B ). For patients aged 50 years and older with low-grade stage-Ia ovarian tumor, the OS and DSS of patients receiving unilateral oophorectomy were significantly worse than those of patients receiving bilateral oophorectomy (OS: p < 0.001; DSS: p = 0.01) ( Figure 6C and Figure S10C ).

A comorbidity analysis was carried out on the causes of death for the patients with ovarian cancer ( Figure 7 , Figure S11 , and Figure S12 ). In stage-I ovarian cancer, the CMR of cancer-related deaths was significantly lower in patients who underwent unilateral oophorectomy than in those who underwent bilateral oophorectomy (p < 0.001) ( Figure 7A ). CVDs were also remarkably decreased in patients who underwent unilateral oophorectomy (5-year CMR: unilateral oophorectomy, 1.5%; bilateral oophorectomy, 1.7%; p = 0.04) ( Figure 7C ). For stage-II and stage-III tumors, there were no significant differences between the CMR of unilateral and bilateral oophorectomy for both cancer-related deaths and non-cancer comorbidities ( Figure S11 and Figure S12 ).