Ovarian metastasis from non-gynecologic primary sites: a retrospective analysis of 177 cases and 13-year experience

Metastasis to the ovary is not rare. Generally, the most common non-gynecologic primary site of Ovarian metastasis is gastrointestinal tract. Differential diagnosis between primary malignant ovarian tumor and metastatic ovarian tumor is important, otherwise appropriate treatment cannot be determined. Furthermore, optimal treatment strategy for ovarian metastasis from non-gynecologic primary sites still are needed to explore. This study described the clinicopathologic characteristics of ovarian metastasis from non-gynecologic primary sites and identified the significance of surgical treatment for these patients.

Synchronous ovarian metastasis is easy to be misdiagnosed as primary ovarian cancer. Optimal cytoreductive surgery and postoperative adjuvant treatment can be done to confer survival benefit in selected patients.

Background
As lung and liver, ovary is a frequent site of metastasis for genital and non-gynecologic primary malignancies [1][2][3]. Metastasis to the ovary from non-gynecologic organs accounts for 9% of all ovarian malignancies [4]. Gastrointestinal tract is the most common non-gynecologic primary site metastatic to the ovaries, followed by breast [1,2].
Inappropriate treatment and adverse outcome may occur when ovarian metastasis and primary ovarian cancer are misdiagnosed as each other [5]. An optimum surgical debulking is the most important treatment strategy for ovarian primary cancer because its minimum residual lesion after surgery is associated with a longer term of survival [6]. However, optimal treatment strategy for ovarian metastasis from non-gynecologic primary sites has not been established. First Akhan, S. E. et al. [7] suggested that aggressive surgery should be avoided in patients with extragenital metastatic cancers of the ovary, especially in these complicated with peritoneal metastasis. Because the presence of distant metastasis sites, of course including ovary, usually indicates a dismal prognosis. Then Goere, D. et al. [8] found that ovarian metastasis was less responsive to chemotherapy, so they proposed that ovarian metastasectomy should always be considered, even in the case with nongynecologic metastases.
The purpose of our study was to evaluate the demographic characteristics, disease presentation, role of surgery and postoperative adjuvant treatment, survival and its prognostic indicators.

Study population
One hundred and seventy-seven patients with ovarian metastasis from non-gynecologic primary sites underwent surgery at Peking Union Medical College Hospital (PUMCH) between May 2005 and May 2018. Clinicopathologic information was obtained from medical records and pathology reports. Followup information was obtained from telephone interview. Survival was defined as the date of resection of ovarian metastasis to the date of death or to May 31, 2018. The following data were collected, including: (1) the general conditions; (2) presenting symptoms of ovarian masses related, serum carbohydrate antigen-125(CA-125), preoperative imaging examination; (3) the primary sites, the date of diagnosis of primary cancer, the date of resection of ovarian metastasis, and the interval between the two dates; (4) preoperative diagnosis of ovarian mass; (5) surgical findings, residual lesions size at the completion of surgery, surgical complications; (6) postoperative pathology reports; (7) whether underwent postoperative adjuvant treatment or not; (8) survival time. Ethics approval was granted by the Institutional Review Board of PUMCH.
The chronological sequence of diagnosis was identified as synchronous or metachronous according to the diagnostic date of primary cancer and ovarian metastasis. Namely, the patients with a previous history of non-gynecologic primary cancer before detection of the ovarian metastasis were assigned to the metachronous group; otherwise, they were assigned to the synchronous group. The interval between the two dates in metachronous group is calculated.
All the enrolled patients underwent surgery in our hospital. According to the surgical findings, extraovarian involvement was determined as positive or negative. In consideration of no uniform definition of radical or palliative surgery, and the diverse surgical removal extent, we focus on the residual lesion size at the completion of surgery. Surgery was considered as optimal if the diameter of the largest residual lesion was less than 2 cm.
All ovarian masses had pathological confirmation of the surgical specimens as ovarian metastasis.
According to the pathological results, the degree of differentiation was determined to be well differentiated, moderately differentiated, poorly differentiated and undifferentiated. A majority of primary sites were determined by the previous surgical pathology at our hospital or other hospitals.
The rest of the enrolled patients had neither pathology or diagnostic imaging examination to identify the primary sites. These were classified as the group of unknown primary site.

Statistical analysis
The data were analyzed using SPSS for IOS, version 21. Survival rates were calculated using the life table method, and differences between groups were calculated using the log-rank test. A two-sided P values of 0.05 or less were regarded as statistically significant.   Table 2 Interval of metachronous ovarian metastasis and differentiation of tumor of the 177 patients.  developed metachronous ovarian metastasis after the diagnosis of primary cancers. In metachronous cases, the median interval was 19 months (range 1.5-240 months). (Table 2) The differentiation of tumor can be identified in 166 patients. Of them, patients with well differentiation accounted for 9.6%, moderate differentiation for 35%, low differentiation for 49.2% and undifferentiation for 6.2%. (Table 2) According to the clinical manifestations and previous history, a total of 103 patients (58.2%) were tentatively diagnosed with ovarian metastatic tumor before surgery. Thereinto, 26 patients came from metachronous group and 77 patients came from synchronous group. There was significant difference between the two groups in preoperative diagnostic accuracy (P < 0.0001). (Table 3) Table 3 Preoperative diagnostic accuracy in metachronous and synchronous group.
Tentatively diagnosed with ovarian metastasis Diagnostic accuracy P-value

Survival
The overall 3-year survival rate and 5-year survival rate were 23% and 10%, respectively, and the median survival time was 20 months (95% CI: 16-24 months). (Fig. 1) There is a significant difference in median survival time between different primary sites (P = 0.02).
The median survival time of every primary site was as follows: small intestine 59 months, <link rid="Sec10">appendix</link> 54 months, bladder 27 months, breast 25 months, lung 25 months, colorectum 21 months, stomach 18 months, biliary tract 14 months, pancreas 13 months. The median survival times according to the differentiation of tumor showed significant differences (P = 0.016) and were as follows: well differentiation 34 months, moderate differentiation 21 months, low differentiation or undifferentiation 16 months (Fig. 2). The median survival time was 25 months in patients whose largest residual lesion was less than 2 cm and 14 months in those whose largest residual lesion more than or equal to 2 cm (P = 0.001) (Fig. 3). According to the adjuvant treatment there was a significant difference in survival between patients with postoperative chemotherapy and/or radiation and those without postoperative treatment, with an estimated median survival of 24 months and 8 months, respectively (P < 0.001) (Fig. 4) [4]. And two American series reported that breast occupied the first place [2,10]. In our study, the following primary sites were <link rid="Sec10">appendix</link>, biliary tract, pancreas, breast, small intestine, lung, and bladder. According the other published studies, renal pelvis [10], lymph system [1,4], melanoma of the skin [2], mesothelioma [4,10], thyroid carcinoma [11] can also be the non-gynecologic primary sites of ovarian metastasis. Generally, a variety of non-gynecologic organs can serve as the primary sites of ovarian metastasis, of which the digestive tract is the most common. This is consistent with our results. An awareness on the primary sites of ovarian metastasis help clinicians to differentiate the primary and metastatic ovarian cancer as well as search a possible primary site purposefully.
The mean age of our patients was 48 years when they were diagnosed; approximately 10 to 15 years younger than the age of primary ovarian cancer (late 50 s to early 60 s) [6]. Therefore, this also led that the 59.3% of our patients was premenopausal women. It is precisely because of the higher blood flow to ovary of premenopausal women that ovarian metastasis occurs [12].  [4,9]. One reason for the difficulty in differentiation of primary and metastatic ovarian cancer is that their presenting symptoms can overlap. For primary ovarian tumors, present with 3 to 4 months of abdominal distension or pain is typical [6].
Furthermore, abnormal uterine bleeding occurred in 18.1% of our patients. LI-CHUN LU et al. [13] reported that 14.3% of their patients with Krukenberg tumors occurred menstrual irregularity. One study mentioned above showed that only 5.2% their patients presented abnormal uterine bleeding [4]. Besides, there is a case report on postmenopausal vaginal bleeding as initial presentation of Krukenberg tumor [14]. Abnormal uterine bleeding is considered to be caused by the ovarian metastasis disrupting ovarian function and affecting sex hormone levels.
Among our patients who tested serum CA-125 before surgery, approximately a half of the patients had an abnormally elevated CA-125 of more than 35 U/ml but no more than 200 U/ml. Serum CA-125 is the most common tumor marker for ovarian cancer with a low specificity and sensitivity [15]. Although recent work has suggested the change of CA-125 levels over time is more useful in identifying and monitoring ovarian cancer, but the absolute level of CA-125 may help to differentiate primary and metastatic ovarian cancer. Clinically, epithelial ovarian cancer with CA125 up to several thousand is relatively common, while metastatic ovarian cancer is not.
Although many tumor markers of gastrointestinal cancer, such as CA-19-9, CEA, CA-72-4, are also with a low sensitivity and specificity [16,17], these elevated tumor markers could serve as a hint for clinicians preoperatively. Additionally, tumor markers with differential diagnostic value include CA199, which is often elevated in pancreatic cancer, and CA153, which is often elevated in breast cancer. But, unfortunately, these tumor markers were not tested as common as CA-125 in patients with ovarian masses.  [19,20]. These findings are nearly consistent with ours. However, the reported several non-small lung cancer patients showed an interval of 14 to 20 months [21,22] which seems shorter than ours. Considering that this may be related to the pathological type of lung cancer, we further confirmed that the pathological type of our enrolled lung cancer patients was all small-cell lung cancer. Generally, the non-small lung cancer is with a higher malignancy degree and poorer prognosis than small lung cancer. So that explains why our metachronous lung cancer patients were with a longer interval. In addition, out of the primary sites of our study, Corrado, G. et al. [11] reported a rare case of ovarian metastasis from thyroid carcinoma after 9 years from the diagnosis. Over time after diagnosis of non-gynecologic primary cancer, clinicians have better to keep an improved consciousness in development of ovarian metastasis. It is valuable not just because ovarian metastatic tumor cannot be quite uncommon, but because that familiarity with the probable intervals for different primary cancers is helpful for clinicians to tentatively diagnose ovarian metastasis.
Indeed, 58.2% of our patients were tentatively diagnosed with ovarian metastasis before surgery in which almost all were patients of metachronous group. The preoperative diagnostic accuracy is 95.1% and 27.1% in metachronous and synchronous group, respectively. A significant lower accuracy in synchronous group indicates that synchronous ovarian metastasis misdiagnosed as primary ovarian cancer is quite common. Patients with synchronous ovarian metastasis from <link rid="Sec10">appendix</link>, small intestine, biliary tract, lung, and bladder were all diagnosed before surgery incorrectly. Of synchronous ovarian metastasis, 65% from colorectum and 73.3% from stomach were misdiagnosed as primary ovarian cancer. Furthermore, of particular note is no breast cancer patients occurred synchronous ovarian metastasis in our study, which may be related to early detection of breast cancer. Unfortunately, little research has been conducted in analyzing synchronous ovarian metastasis mimicked by primary ovarian cancer. Taking no account of the chronological sequence of diagnosis, based on the data available in the literature, 32% of mimicking metastasis were from colorectum and 49% of these were from gastrointestinal tract [23,24]. The purpose of surgical intervention is not only to establish a definite diagnosis, but also to alleviate symptoms, minimize residual tumor burden, avoid progression or achieve a cure. Usually, surgeons proceed with surgical intervention depending on the presence of severe symptoms or emergencies, or surgical removal of tumors deemed necessary for asymptomatic patients following a multidisciplinary team consensus. Due to the variety of the tumor size, the depth of invasion, involvement, the surgical removal extent was various in which the minimal one was only unilateral salpingo-oophorectomy and the maximal one can include resection of multiple involved organs and local lymph nodes. Over the past decade, an amount of retrospective studies has proven that the residual lesion size at the completion of surgery is related to a survival benefit for patients with ovarian metastasis from non-gynecologic primary cancers [1,4,9,[25][26][27][28][29][30][31][32]. In our current study, there was a significant survival difference between patients with less than patients with 2 cm of largest residual lesion and more than or equal to 2 cm of largest residual lesion. Median survival time of the two groups were 25 months and 14 months, respectively. The result might indicate us every effort to perform optimal surgery should be made. However, resectability of tumors and underlying surgical complications were required to evaluate before surgery. Surgeons made the decision to proceed with surgery after balancing surgical benefits and risks. In our series, although approximately 10% of patients experienced complications, the fatal complication was only observed in one patient, which is consistent with Seow-En, I.'s results [33]. Generally speaking, surgery for patients of ovarian metastasis from non-gynecologic primary cancers can be performed safely with an acceptable complication rate.
Prognosis of patients with ovarian metastasis from non-gynecologic primary sites is poor. The published study of the largest sample size (158 patients) reported that the 5-year survival rate and median survival time were 7.2% and 15 months, respectively [9]. These results were not different from ours. In our study, the 3-year survival rate, 5-year survival rate, and median survival time were 23%, 10%, and 20 months, respectively. Besides the residual lesion size, our results showed that primary sites, differentiation of ovarian metastasis and postoperative adjuvant treatment were also prognostic indicators.
A considerable amount of studies has reported that survival of patients with ovarian metastasis according to the primary tumor showed significant differences [1,2,4,7,9,25,29,31,34]. Although the primary sites included in each study were different, among the included common non-gynecologic primary cancers, the overall survival time of breast cancer was longer than that of gastrointestinal cancer, and that of colorectal cancer was longer than that of gastric cancer. In our study, the median survival time of breast cancer, colorectal cancer and stomach cancer were 25 months, 21 months and 18 months, respectively; consistent with the results of aforementioned earlier studies. Moreover, for the reason of a more variety of non-gynecologic primary sites included in our study than other studies, some cancers with relatively good prognosis, like <link rid="Sec10">appendix</link> mucinous adenocarcinoma and bladder transitional cell carcinoma, survive longer after resection of ovarian metastasis, whose median survival time were 54 months and 27 months, respectively.
On the contrary, patients of biliary tract and pancreatic cancers with poorer prognosis died one year after surgery for ovarian metastasis in our study. All the 5 enrolled cases of small intestine cancer were adenocarcinoma.
According to the literature, small intestinal adenocarcinoma prognosis seems at an intermediate stage of that of colon and gastric cancers [35]. While the 59 months of median survival time of small intestine cancer after surgery was the longest in our study. This seemingly indicates that prognosis is not entirely consistent with postoperative survival time for ovarian metastasis. Due to the lack of relevant studies, it is not possible to explain this for the time being.
To our knowledge, no study specially analyzed the differentiation of ovarian metastasis as an underlying prognostic indicator. Nevertheless, our results showed that the median survival time of patients with ovarian metastasis well, moderate, low differentiation or undifferentiation were 34 months, 21 months, 16 months, respectively; which was a significant difference.
Postoperative adjuvant treatment is identified as essential. Both an early study [9] and our current studies have proven that the patients who underwent postoperative adjuvant treatment, mainly chemotherapy, survived longer than those who did not.
According to the aforementioned prognostic indicators, patients with ovarian metastasis from non-gynecologic primary sites can be selected to undergo optimal cytoreductive surgery which confers survival benefits with a low rate of surgical complications.

Conclusion
It is worthy of consideration to perform surgery for patients with ovarian metastasis from non-gynecologic primary sites. The decision to proceed with cytoreductive surgery is affected by presence of symptoms, the primary sites, differentiation of tumor. In an experienced hospital, a multidisciplinary team should evaluate resectability of tumors before surgery. Every effort should be made to achieve the an optimal cytoreductive surgery, namely the diameter of largest residual lesion at the completion of surgery are less than 2 cm, followed by postoperative adjuvant treatment.

Availability of data and materials
The dataset used or analysed including in this study are available from the co-corresponding authors on reasonable request.

Competing interests
The authors declare that they have no competing interests.    Survival according to residual lesion (< 2 cm or ≥ 2 cm). The median survival time was 25 months in patients whose largest residual lesion was less than 2 cm and 14 months in those whose largest residual lesion more than or equal to 2 cm (P = 0.001).

Figure 3
Survival according to residual lesion (< 2 cm or ≥ 2 cm). The median survival time was 25 months in patients whose largest residual lesion was less than 2 cm and 14 months in those whose largest residual lesion more than or equal to 2 cm (P = 0.001).