Age-dependent difference in impact of fertility preserving surgery on disease-specific survival in women with stage I borderline ovarian tumors

Background This study was to determine age-specific impact of fertility preserving surgery on disease-specific survival in women with stage I borderline ovarian tumors (BOTs). Patients diagnosed during 1988–2000 were selected from The Surveillance, Epidemiology, and End Results (SEER) database. The age-specific impact of fertility preserving surgery and other risk factors were analyzed in patients with stage I BOTs using Cox proportion hazard regression models. Data from our hospital were collected during 1996–2017 to determine the prevalence of patients who had undergone fertility preserving surgery. Results Of a total 6295 patients in the SEER database, this study selected 2946 patients with stage T1 BOTs who underwent fertility preserving or radical surgery. Their median age at diagnosis was 45.0 years and the median follow-up time was 200 months. Fertility preserving surgery was performed in 1000/1751 (57.1%) patients < 50 years and in 1,81/1195 (15.1%) patients ≥50 years. Fertility preserving surgery was significantly associated with worse disease-specific survival only in patients ≥50 years. Increased age, stage T1c and mucinous histology were risk factors for overall patients or patients ≥50 years, but not for < 50 years. Data from our hospital showed that fertility preserving surgery was performed in 53.9 and 12.3%patients < 50 and ≥ 50 years with stage I disease, respectively. Conclusion Fertility preserving surgery is safe for women < 50 years with early staged BOTs, but it may decrease disease-specific survival in patients ≥50 years. Conservative surgery is performed at a relatively high rate in patients ≥50 years. Electronic supplementary material The online version of this article (10.1186/s13048-018-0423-y) contains supplementary material, which is available to authorized users.


Background
Borderline ovarian tumors (BOTs) are histologically characterized as atypical epithelial proliferation without the presence of stromal invasion [1]. Serous and mucinous BOTs are the two major histological types [2]. These tumors have a low malignant potential to spread beyond the ovary with peritoneal involvement [1] and have an excellent prognosis [3][4][5]. This disease accounts for 10-15% of all epithelial ovarian cancers [6]. Compared to invasive epithelial ovarian cancers, BOTs occur more commonly, at an early stage, in women of childbearing ages [7].
The majority of BOTs are managed with surgery alone. Fertility preserving surgery is widely adopted for patients who have early-stage tumor development and a desire for fertility. Current consensus states that fertility preserving surgery is associated with an increased risk of recurrence [8][9][10][11][12]. Data from ours and other groups showed that certain styles of fertility preserving surgery may have a higher risk of recurrence than the others [13][14][15]. However, fertility preserving surgery was not shown to compromise overall survival in these patients [16][17][18][19].
Due to excellent prognosis, many patients with BOTs die due to other diseases. Overall survival is the end-point commonly used in previous studies to determine the impact of fertility preserving surgery; however, this may not accurately reflect the outcome of the surgery. Very few studies have investigated the impact of fertility preserving surgery on disease-specific survival [9]. Fertility preserving surgery has been shown age-dependent differences in its impact on recurrence free survival and other clinical outcomes in patients with BOTs [20]. Using a large population from a publicly available database, the objective of this study was to examine the age-specific impact of fertility preserving surgery on disease-specific survival in women with stage I BOTs.

Methods
The data for this study was obtained from the Surveillance, Epidemiology, and End Results (SEER) database maintained by the National Cancer Institute. This database collects information of cancer patients, which covers approximately 28% of the total US population. The SEER program statistical analysis software package (SEER*Stat version 8.3.4) was used to extract data from SEER18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases, Nov 2016 Sub (1973-2014 varying) [21]. BOTs in the SEER database between 1988 and 2000 were identified based upon the following histopathology codes: serous 8442-1, 8451-1 and 8462-1; and mucinous 8472-1, 8473-1 [22,23].
Only patients with stage I BOTs with a record of survival times were included in this study. The status of oophorectomies and hysterectomies were quarried from codes in the site-specific surgery (1983)(1984)(1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997) and RX Summ-Surg Prim Site (1998+) (Additional file 1: Table S1). Fertility preserving surgery refers to preservation of the uterus and at least one side of a functional ovarium. This study thus defined the surgery as removal of the tumor or a unilateral oophorectomy without a hysterectomy. Radical surgery was defined as bilateral salpingo-oophorectomy with or without hysterectomy. Women were excluded if they did not receive surgery, their surgical status or survival time was unknown, or other surgical approaches were performed (Additional file 1: Table S1). The flow chart shows the detailed procedure for selecting patients (Additional file 1: Figure S1).
Variables extracted from the database were patients' demographics (age at diagnosis, ethnicities, marital status), surgery information (oophorectomy, hysterectomy, lymphadenectomy), tumor information (size, histology, stage), follow-up time and disease-specific death. Tumor stages were evaluated based on the American Joint Committee on Cancer (AJCC) 3rd staging classification [24].
To understand age-specific prevalence of fertility preserving surgery, women diagnosed with BOTs in Zhejiang Cancer Hospital during the year 1996-2017 were also included in this study. Tumor stages were evaluated based upon of the International Federation of Gynecology and Obstetrics (FIGO) 2014 classification system [25]. Stage T1 defined in AJCC 3rd is the same as stage I in FIGO 2014, except that stage Ic in FIGO 2014 is further divided into Ic1, Ic2 and Ic3 stages. The inclusion and exclusion criteria for these patients has been described previously [15].
Data were analyzed using SAS software V9.3 (SAS Institute, Inc., Cary, NC.). The ordinal/categorical data were examined using the χ 2 test. Univariate or multivariate Cox proportional hazards models were used to determine the impacts of fertility preserving surgery and other risk factors on disease-specific survival. The Kaplan-Meier survival curves were generated and their significant differences were analyzed by log-rank tests. Two-sided P values less than 0.05 were considered statistically significant.

Results
A total of 6295 women with BOTs were initially identified from the SEER database. Based on our inclusion and exclusion criteria, a total of 2946 cases with stage I BOTs were included in this study. The detailed demographic information and pathoclinical features are listed in Table 1. The mean age of these patients was 47.1 ± 17.0 years with a median age of 45.0 years (range 10-96 years). The median follow-up time was 200 months (range 1-323 months). Within this population, 59.4% (n = 1751) were < 50 years old and 40.6% (n = 1195) were ≥ 50 years. Most patients (85.0%) studied were Caucasian. The majority of BOTs were diagnosed at stage T1a (79.3%). Fertility preserving surgery was performed in 1181 (40.1%) patients. Hysterectomy and recorded lymphadenectomy were performed in 1374 (47.6%) and 341 (11.4%) patients, respectively. At the end of the follow-up year, 70 (2.4%) patients died from this disease.

Discussion
With a sample size of 2946 patients and a median follow-up time of 200 months, this study examined age-specific impact of fertility preserving surgery on disease-specific survival in women with T1 BOTs. The main finding of this study was that fertility preserving surgery was significantly associated with worse disease-specific survival only in patients ≥50 years, but not in overall patients or patients < 50 years. Our results revealed an age-dependent difference in impact of fertility preserving surgery on disease-specific survival in these patients. This finding suggests that while conservative surgery may comprise survival in women ≥50 years, it is safe for patients < 50 years. Future studies with randomized clinical trials are warranted to verify this finding.
Previous studies have consistently shown that fertility preserving surgery may increase the risk of recurrence [11,14,15,26]. Interestingly, the risk of recurrence was higher in younger patients with BOTs [9,11,18,20]. Most of the recurrences showed no malignant transformation and were curable by a single surgery without compromising overall survival [9,[16][17][18]. Invasive carcinoma diagnosed in recurrences [9,11,18,[27][28][29] is the cause of cancer deaths [30]. A sub-analysis of the Arbeitsgemeinschaft Gynaekologische Onkologie (AGO) ROBOT study evaluated data from a total of 950 patients with BOTs. Their results showed that 66.7% of recurrent diseases were invasive carcinoma in patients ≥40 years, which dramatically contrasted with a recurrence of 12% of invasive carcinomas in patients < 40 years [20]. The increased incidence of invasive recurrent ovarian cancer in older patients may account for the reduced disease-specific survival after fertility preserving surgery.
This study is unable to address the molecular mechanism whereby fertility preserving surgery is associated with reduced disease-specific survival in patients ≥50 years. Akeson et al. [7] reported patients > 60 had significantly more aneuploid tumors. Aneuploidy was  [20]. It is speculated that a higher rate of conservative surgery was performed in their patients with stage I BOTs. Comparable to the result from the SEER database, Our study also identified that increased age, a higher stage (T1c) and mucinous histology were significantly associated with decreased disease-specific survival in overall patients or patients ≥50. Using the same database, a previous study revealed that older age (≥ 50), higher stage and mucinous histology were associated with worse disease-specific survival in patients with stage I BOTs [23]. The tumor stage is a known prognostic factor for patients with BOTs [29]. Our results further revealed that higher stage (T1c) was significantly associated with poorer disease-specific survival in BOT patients at the early stage. Patients with mucinous BOTs were reported to have a worse prognosis compared with to patients with serous BOTs [31,35]. The worse survival is partially explained by a higher incidence of invasive recurrent carcinoma in patients with mucinous BOTs. Karlsen et al. [9] found that 6 out of 7 invasive recurrences were patients with mucinous BOTs at FIGO stage I.
An earlier study identified 6017 cases of BOTs from the SEER database. Their results revealed that the lymph node involvement was not significantly associated with disease-specific survival after adjusting with FIGO stages [36]. No impact of lymph node involvement on overall survival in patients with BOT were also observed in other studies [37,38]. Data from our work and the previous study [23] showed that lymphadenectomy were not a risk factor associated with disease-specific survival.
The use of this database has numerous limitations. Patients were included retrospectively and were not randomly assigned to a treatment. Detailed information of fertility preserving surgery is unavailable. Among patients with stage I disease, 41.6% (2118/5094) were excluded from the study due to unclear surgical information. Many important pathological features of the tumors, such as invasive implants, and micropapillary patterns, are unavailable in these patients. Ovarian cancer related blood biomarkers were not recorded in the SEER database. It is unknown whether there have been recurrences and the types of relapses may have occurred in these patients. The location of harvested lymph nodes are not defined and their numbers are missing in some a b c Fig. 1 Kaplan-Meier survival curves for patients ≥50 with stage I borderline ovarian tumors. a Fertility preserving surgery vs radical surgery. b Sub-stages. c Hysterectomy status patients. Many other limitations using the SEER database have been addressed in a previous study [23]. Use of the SEER database in this study had its strength in its relatively large sample size, long follow-up time, and particularly, relatively large number of disease-specific deaths. Using the same database, the previous study identified 4943 cases with stage T1 BOTs from the same database, and reported a total of 159 (3.2%) deaths in a median follow-up time of 187 months [23]. In contrast, the number of disease-specific deaths reported in previous studies was limited. A cohort included 1143 BOT patients with 1005 (87.9%) patients at FIGO stage I. During a median follow-up time of 49.9 months (range 3.5-99 months), only 7 (0.6%) patients I died of this disease [9]. In another study, a total of 151 patients were recruited. Among them, 87 (64.4%) patients were at FIGO stage I, and 113 patients (74.8%) had follow-up information. After a median follow-up time of 86 (range 0.1-432) months, 7 (6.2%) patients died of this disease [39]. A multi-center study included 457 patients with 390 (85.3%) at stage I. During a mean follow-up of 88.3 months, 9 (2%) patients died of this disease [40]. Leake et al. reported 13 (6.5%) disease-specific deaths in a cohort of 200 patients in a median follow-up time of 120 months [41].

Conclusion
It is safe to perform fertility preserving surgery for women of child-bearing age with stage I BOTs. This surgery may increase the risk of disease-specific death for women of older ages (≥ 50 years). A relatively high proportion of patients (≥ 50 years) receive conservative surgery.

Additional file
Additional file 1: Table S1. Codes used to define surgery styles. Table  S2. Features of patients with borderline ovarian tumors from our hospital during 1996-2017. Figure S1. Flowchart of population selection. Figure  S2