A Population-Based Study on Liver Metastases in Women With Newly Diagnosed Ovarian Cancer

Aim: The purpose of this study was to analyze the incidence, clinical characteristics, prognostic factors and survival of ovarian cancer patients with liver metastases upon initial diagnosis. Methods: Patients with ovarian cancer liver metastases upon initial diagnosis between 2010 and 2016 were identified from the Surveillance, Epidemiology, and End Results (SEER) database. Univariate and multivariate logistic regression was performed to identify the predictors of the presence of liver metastases in newly diagnosed ovarian cancer patients. Overall survival (OS) was assessed using the Kaplan-Meier method and log-rank test. Univariate and multivariate Cox regression was conducted to determine the independent prognostic factors for OS. Results: A total of 1,744 ovarian cancer patients with liver metastases was identified from the SEER database, accounting for 6.7% of the entire ovarian cancer patients. As to the unique distant organ provided by SEER, liver was the most common metastatic site of ovarian cancer (4.65%). Age, race, laterality, histology, pathological grade, extrahepatic sites, stage of tumor were the predictors of the presence with liver metastases revealed by multivariable logistic regression model. Median OS for the patients with liver metastases at initial diagnosis of ovarian cancer was 16.0 months. Multivariate Cox regression model confirmed race, histology, extrahepatic metastatic sites, surgery and marital status were independent prognostic factors for OS. Conclusion: The study provided population-based estimates of the incidence and prognosis of newly diagnosed ovary cancer patients with liver metastases, which could be potentially used for the risk assessment and individualized treatment.


INTRODUCTION
According to the latest cancer statistics in 2019, there were about 22,530 patients newly diagnosed with ovarian cancer and 13,980 patients died of ovarian cancer. Ovarian cancer accounts for 2.5% of all cancers in women and ranks the fifth cause of cancer death among women in the United States (1). Owing to the scarcity of early, specific symptoms and effective screening strategies, patients with ovarian cancer were often diagnosed with synchronous distant metastases at an advanced stage, accounting for about 70% of the whole ovarian cancer population (2). The survival of ovarian cancer patients was stage-dependent. The 5 year survival was 92, 75, 29%, respectively, for the localized cases, regional cases and distant cases (1). Studies of the patterns of distant metastases in stage IV ovarian cancer showed liver was the most common distant metastatic organ of ovarian cancer with the proportion of 37-57%, followed by distant lymph nodes, lung, bone and brain (3)(4)(5). Autopsy studies of cases died of primary ovarian cancer showed that the incidence of liver parenchyma was about 48.2-73% (5,6). The median overall survival (OS) was 30 months for patients with a single liver metastasis (4).
At present, data of prevalence and prognostic factors among ovarian cancer patients with liver metastases is limited. The therapy for ovarian cancer patients with liver metastases is controversial. Here, we explored the Surveillance, Epidemiology, and End Results (SEER) database to analyze the characteristics, incidence, risk factors and prognostic factors of ovarian cancer patients with liver metastases upon initial diagnosis.

Study Population
All the primary data were acquired from the SEER database, which collected the information of patient demographics, tumor characteristics, treatment and prognosis accounting for ∼30% of the whole population in the United States. The datasets of this current study are available from SEER * Stat Version 8.3.6 (http: https://seer.cancer.gov/seerstat/). Patients diagnosed with primary and microscopically confirmed ovarian cancer as the only primary malignancy between 1 January 2010 and 31 December 2016 were included in the study. We excluded those patients younger than 18 years old, diagnosed with carcinoma in situ, benign or borderline tumors, with unknown information of liver metastases, diagnosed by autopsy or death certificate. Finally, there were 26,197 patients eligible for incidence analysis. Among these patients, 1,744 patients had liver metastases upon initially diagnosis of ovarian cancer. Next, we removed patients with unknown follow-up, as well as patients who presented with 0 day of survival record, leaving 25,819 patients and 1,718 patients with liver metastases for the survival analysis.

Statistical Analysis
The incidence of different distant metastatic organs including liver, bone, brain, and lung among the total patients with ovarian cancer was calculated in this study. Incidence was defined the number of ovarian cancer patients with corresponding metastases divided by the whole number of patients with ovarian cancer.
We compared the patient characteristics between patients with liver metastases and those without liver metastases by the Pearson chi-square test or Fisher's exact test appropriately. Study variables in the descriptive statistics were included as follows: age at diagnosis, year of diagnosis, race, laterality, histology, primary tumor stage, regional lymph node stage, pathological grade, surgery, numbers of extrahepatic metastatic sites (bone, brain, and lung), marital status and insurance status. In the SEER database, tumor grades were classified into I (well-differentiated), II (moderately differentiated), III (poorly differentiated), and IV (undifferentiated or anaplastic).
Univariable and multivariable logistic regression was conducted to determine the risk factors for the liver metastases at initial diagnosis of ovarian cancer. Variables including age at diagnosis, race, laterality, histology, pathological grade, the extent of extrahepatic metastatic sites, primary tumor stage, lymph node stage, marital status and insurance status were analyzed in this model. Odds ratios (OR) and 95% confidence intervals (CI) were reported from the logistic regression.
Overall survival (OS) was determined as the date of diagnosis of ovarian cancer to the date of death due to any cause or the date of last follow-up. We utilized the univariable and multivariable Cox proportional hazard models to assess the association of study variables with OS by reporting the hazard ratios (HR) and 95% Cis. In our study, the marital status was enrolled in the analysis of prognostic factors for the following reasons. First, epidemiological studies have reported a reduced risk of ovarian cancer among women who have had children. And infertility has been associated with an increasing risk of ovarian cancer (7,8). Second, married patients with cancer may have more support from family members, social services and insurance than unmarried patients. Unmarried patients are at significant higher risk of undertreatment and death from cancer (9). Kaplan-Meier method was performed to obtain the survival estimates and the log-rank test was conducted to analyze the difference between subgroups.
All the statistical analyses were performed by the SPSS statistical software version 22 (IBM Corporation, USA). All P-values calculated were two-sided, and a p < 0.05 was considered statistically significant.

Incidence
Total of 26,197 patients were diagnosed with ovarian cancer from 2010, January to 2016, December reported by the SEER database. Among the total 26,197 patients, there were 13,366 patients with serous ovarian cancer (51.0%), 2,321 patients with endometrioid carcinoma (8.9%), 1,765 patients with clear cell ovarian cancer (6.7%), 1,587 patients with mucinous carcinoma (6.1%) and 7,158 patients with other types such as granular cell cancer and Brenner tumor (27.3%). In view of the small size, we categorized the non-serous ovarian cancer into one group. Figure 1 illustrated the incidence of patients with distant metastases at initial diagnosis of ovarian cancer according to the metastatic sites including liver, bone, lung and brain among the entire cohort. 4.65, 0.39, 4.00, 0.10% ovarian cancer patients had liver metastases, bone metastases, lung metastases, brain metastases only, respectively, at initial diagnosis. As to patients with two distant metastatic sites, patients with liver and lung metastases had the highest proportion, accounting for 1.20% among the entire population.

Patient Characteristics
One thousand seven hundred and forty four patients had liver metastases at initial diagnosis of ovarian cancer, accounting FIGURE 1 | The incidence of ovarian cancer patients with liver, lung, bone, and brain metastases at the time of initial diagnosis. for 6.7% among the total population. Table 1 showed the demographic and clinical characteristics for ovarian cancer patients with and without liver metastases. Compared to patients without liver metastases at initial diagnosis, patients with liver metastases were older (P < 0.001), were more likely to be Black (P < 0.001), had bilateral tumors (P < 0.001), presented with more advanced T stage (P < 0.001), presented with more advanced N stage (P < 0.001), had a higher pathological grade (P < 0.001), had no surgery of primary tumor (P < 0.001), had more extrahepatic metastatic sites to lung, bone and brain (P < 0.001), were more likely to be unmarried (P < 0.001).
The results of univariable and multivariable logistic regression performed among the whole population was displayed in Table 2.

Survival Analysis
One thousand seven hundred and eighteen ovarian cancer patients with liver metastases were enrolled in the survival analysis. There were 854 patients with serous ovarian cancer (49.7%), 44 patients with endometrioid carcinoma (2.6%), 48 patients with clear cell ovarian cancer (2.8%), 46 patients with mucinous carcinoma (2.7%) and 726 patients with other types such as granular cell cancer and Brenner tumor (42.3%). The median OS among patients with liver metastases was 16.0 months, and the interquartile range (IQR) was 3.0-50.0 months (Figure 2A). OS curves stratified by ovarian cancer type of pathology was illustrated ( Figure 2B). Patients with serous ovarian cancer had better survival than those with non-serous ovarian cancer (median OS: 30.0 months vs. 6.0 months, log-rank test, P < 0.001). The impact of the extent of extrahepatic metastatic disease on the survival of patients with liver metastases was estimated ( Figure 2C, log-rank test, P < 0.001). Patients with more numbers of extrahepatic metastatic sites had worse prognosis.
Among the whole population, there were 1,385 patients with extrahepatic metastatic disease. The presence of liver metastases or not on the median survival of these patients stratified by the extent of extrahepatic metastatic sites was concluded in Table 3. Broadly speaking, patients with no baseline of liver metastases had better survival than patients with liver metastases. Specifically, significant difference was shown in patients with bone and liver metastases (median OS, 6.0 months; IQR, 1.0-10.0 months) vs. those with bone metastases only (median OS, 10.0 months; IQR, 3.0-44.0 months) (log-rank test, P = 0.003), patients with lung and liver metastases (median OS, 13.0 months; IQR, 2.0-33.0 months) vs. those with lung metastases only (median OS, 21.0 months; IQR, 6.0-41.0 months) (logrank test, P = 0.001), patients with brain and liver metastases (median OS, 1.0 months; IQR, 1.0-1.0 months) vs. those with brain metastases only (median OS, 7.0 months; IQR, 2.0-NR) (log-rank test, P = 0.016).
Univariate and multivariate Cox proportional hazards models were performed to evaluate the prognostic factors of ovarian cancer patients with liver metastases (  Extrahepatic metastatic sites <0.001* to lung, bone and brain, No.

DISCUSSION
The dissemination types of ovarian cancer were divided into the transcoelomic metastasis, hematogenous and lymphatic spread metastasis, which individually had distinct molecular metastases mechanisms (10). Some research focused on the mechanisms of distant metastases including liver metastases in ovarian cancer. Kim    metastases including the initial diagnosis and later recurrence was 9.4% and the median survival was 5.0 months among the total 255 patients (18). The discordancy may be caused  by the difference of study population. Our study showed that old age, Black race, bilateral tumors, non-serous type, high grade, extrahepatic metastases, advanced stage and lymph nodes involvement were risk factors associated with the presence of liver metastases upon initial diagnosis of ovarian cancer. Previous study found that advanced stage, high grade and lymph node involvement were significant risk factors associated with distant metastases (19). The single-institution study of 244 serous ovarian cancer patients manifested that the increasing age, high grade tumor and advanced stage were risk factors associated with the presence of liver metastases (20). Loizzi et al. analyzed the clinical characteristics and survival of 29 ovarian cancer patients with hepatic metastases. Results indicated that 76% patients presented with papillary serous histology and 62% patients presented with poorly differentiated tumors (21). However, No significant difference was seen when compared the distant metastatic patterns for different histologic variants of ovarian cancer in the study from Rose et al. (22). In our study, the median OS of patients with ovarian cancer was 16 months (IQR, 3-50 months). The subgroup analysis indicated that the patients with non-serous ovarian cancer and more numbers of extrahepatic sites had worse outcome. The multivariate Cox model showed the Black race, non-serous type and extrahepatic metastatic sites were correlated with increased risks of all-cause mortality, which was basically in accordance with previous studies (4,21,23). A study from Loizzi et al. revealed the OS among ovarian cancer patients with liver metastases upon initial diagnosis, with liver metastases as first recurrence, with liver metastases as second relapse was 19 months (IOR: 6-23 months), 24 months (IQR: 3-44 months), 10.0 months (IQR: 1-33 months), respectively, and no significant difference was seen among the three subgroups. The patients with liver metastases only had better survival than those with other metastatic sites (median OS: 25 months vs. 8 months, IQR: 9-44 months vs. 1-20 months, P = 0.033). And patients with serous ovarian cancer had better survival than those with other type of ovarian cancer (median OS: 23 vs. 8 months, IQR: 1-44 vs. 1-15 months, P = 0.005, HR, 2.875, 95% CI, 2.51-3.23) (21).
The treatment for ovarian cancer patients with liver metastases was still uncertain. Our study showed the surgery of primary site reduced the risk of all-cause death (HR, 0.255; 95%  (20). A study from Charlie et al. compared the frequency of visceral metastases between BRCA1/2 deficient ovarian cancer patients and BRCA1/2 proficient ovarian cancer patients, BRCA1/2 deficient ovarian cancer had increased incidence of visceral metastases (BRCA1/2 deficient vs. BRCA1/2 proficient, 58% vs. 5%, P < 0.001) and liver metastases (BRCA1/2 deficient vs. BRCA1/2 proficient, 42% vs. 0%, P < 0.001). BRCA1/2 sequencing should be considered among the ovarian cancer patients for better clinical management (33). Gallotta et al. analyzed the survival of 34 recurrent ovarian cancer patients with liver metastases who underwent liver resection within secondary cytoreductive surgery. Results indicated that patients with BRCA mutation had better survival than those with BRCA wild type (3 year post-liver resection progression free survival: 81.0% vs. 15.2%, P = 0.001). The assessment of BRCA mutational status may be important for risk stratification among the ovarian cancer patients with liver metastases (34). Sood et al. evaluated the status of p53 mutation in 130 ovarian cancer patients. Results revealed patients with a null mutation had a higher incidence of distant metastases than those with missense mutations or wild type p53 (66% vs. 8% vs. 8%, P < 0.001). Twenty five percentage patients with null p53 mutation presented with distant metastases including liver, spleen, brain, and thorax at initial diagnosis of ovarian cancer. It may be important to evaluate the p53 status for clinicians when dealing with the ovarian cancer patients (35).
To our knowledge, this current study was the largest study about ovarian cancer patients with liver metastases. However, several limitations should be acknowledged. First, the information provided by SEER database is insufficient, including the other metastatic sites such as peritoneal metastases of ovarian cancer, the detailed data about size and number of liver metastases and the information of individual treatment. Second, the study population are mainly in the ovarian cancer patients with liver metastases upon initial diagnosis excluding those developed with liver metastases during the recurrence. Third, the SEER database is based on the register in the United States. The study results may cause deviations in the other parts of the whole world.
In conclusion, this present study provided valuable information including incidence, risk factors and prognostic factors for newly diagnosed ovarian cancer patients with liver metastases. These findings assist clinicians make clinical management decisions of prognostic assessment and risk stratification. In the future, basic research and large sample prospective clinical trials are warranted to further evaluate the molecular characteristics and treatment for ovarian cancer patients with liver metastases.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by The Ethical Committee Review Board of Fudan University Shanghai Cancer Centre (Shanghai, China). Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.