Brain Metastases from Ovarian Carcinoma: An Evaluation of Prognostic Factors and Treatment
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.273627
Source of Support: None, Conflict of Interest: None
Keywords: Brain metastases, outcome, ovarian cancer
Anton Wohl and Gil Kimchi contributed equally to this study.
Ovarian cancer accounts for approximately 3% of cancers in the female population, with a lifetime risk of 1.3% (American Cancer Society, 2017). Its most prevalent subtype of serous adenocarcinoma accounts for 90% of malignant ovarian tumors and presents with the extra-ovarian spread at 85% of the primary diagnoses. The incidence of brain metastases from ovarian cancer varies among studies; recent series demonstrated a rate as low as 0.3%, although a rate of 6% has also been reported., Despite the scarcity of the disease, it has long been postulated that the incidence of central nervous system (CNS) metastatic involvement in ovarian cancer is increasing, probably due to improvements in the primary therapeutic options and the consequent prolongation of survival., In several recent studies, the median duration of survival following the diagnosis of CNS involvement ranged from 1–18 months.,,,,
Statistically significant risk factors that are currently established for patients with ovarian brain metastases include distant spread at the time of the diagnosis of CNS involvement, presence of multiple brain lesions,,,, and the diminished time interval between the diagnosis of the primary disease and the occurrence of CNS metastasis., Favorable prognostic factors were established as well; these include a well-controlled primary disease at the time of cerebral involvement, low tumor grade, tumor cells sensitive to platinum, and high-performance status of the patient.
Currently, in addition to systemic chemotherapy and supportive care, two main therapeutic arms are being used to treat brain metastasis: radiation therapy [e.g. whole-brain radiation therapy (WBRT), stereotactic radiosurgery (SRS), and gamma-knife radiosurgery] and surgical resection; either in combined regimens or separately. Although the treatment choice is influenced by multiple factors and should be tailored individually, the superiority of combined therapy (surgery and radiation) has been established in several series of patients., In the absence of Class A guidelines and a sound treatment algorithm, the choice between the different treatments remains subject to the clinician's preference.
In this study, we present the experience of a single institute in the treatment of a cohort of patients with ovarian cancer brain metastases. We evaluated prognostic factors for survival and progression-free survival (PFS) based on different treatment methods.
Following the approval of the Sheba Medical Center ethics committee (approval no. 328-13 SMC), the institutional database of the Sheba Medical Center was searched for all patients diagnosed with “ovarian cancer” and “metastases” using a keyword-based query. Patients with CNS metastases were identified from several hundred entries. Twenty-five women of all age groups, who were initially diagnosed with ovarian cancer in the period between 1995 and 2014, had been consequently diagnosed with CNS metastatic involvement, and treated at the Sheba Medical Center were included in the study. The medical records of these patients were reviewed. We collected data regarding patient demographics and clinical characteristics, including patient age, surgical stage, and histologic grade of the ovarian tumor; treatment, modalities used for primary cancer, and the status of the systemic disease at the time brain metastasis was diagnosed.
The primary tumor cell type was categorized as “serous” or “non-serous” and the International Federation of Gynecology and Obstetrics (FIGO) staging score at the time of initial diagnosis was simplified in our study to the presence or absence of distal organ metastases.
The time interval that had elapsed from the primary ovarian carcinoma diagnosis to CNS involvement was also documented. The age of the patients at the time of the diagnosis of the first CNS involvement was documented. Symptoms and neurologic deficits based on anamnesis and the neurologic examination performed upon admission to the neurosurgical department were documented. Headache, nausea, and vomiting were grouped as symptoms of increased intracranial pressure (ICP). In addition, the patients' general performance status was evaluated using the Karnofsky Performance Status (KPS) score.
Magnetic resonance images were reviewed for the number of brain lesions, their size, and their location. The radiologic response to treatment was also recorded. The number of lesions was evaluated as a prognostic factor for PFS and survival. We divided the patients into two groups: a group having up to three lesions and a group having four or more lesions on the basis of the accepted cut-off point for favoring radiation therapy over surgery. None of the patients, however, had two or three lesions; thus, the cohort comprised patients with either a single lesion or four or more lesions. (Patients with the leptomeningeal disease were included as well). The location of the lesions was categorized as infra-tentorial, supra-tentorial, both supra- and infra-tentorial, or leptomeningeal spread, and it was analyzed as a prognostic factor for PFS and survival.
The patients' medical records did not distinguish between death caused by the progression of brain metastases and death from the progression of systemic ovarian cancer.
Treatment regimens were grouped as follows: “Surgery” indicates a mono-therapy by lesion resection without prior radiation therapy or complementary radiotherapy within the following 6 months post surgery. “Radiation therapy” refers to a mono-therapy with either WBRT, SRS, or combined dual therapy of WBRT and SRS without prior or consequent surgery for a period of 6 months. “Surgery + Radiation therapy” refers to initial treatment by surgery followed by complementary radiation treatment (WBRT and/or SRS) within 6 months.
Both survival analysis and PFS analysis were performed. PFS analysis refers to the interval between the diagnosis of ovarian cancer brain metastasis and the occurrence of progressive disease.
Continuous variables were evaluated for normal distribution through histogram evaluation and presented as the median and interquartile range (IQR) or range. Categorical variables are presented as a frequency and percentage. The Chi-square test, Fisher's exact test, Kruskal Wallis test, and Mann Whitney test were used to compare between types of treatment modalities. Survival analysis in relation to the different treatment modalities was described using the Kaplan Meier estimator and evaluated using a log-rank test. Univariate Cox regression was used to evaluate the association between each predictor and mortality. Multivariate Cox regression was applied to evaluate the association between mortality and variables with P < 0.2 on univariate analysis. When evaluating the modality of treatment, age at the time of primary CNS involvement was adjusted. A P value less than 0.05 was considered statistically significant. All statistical analyses were two-sided. SPSS software (version 22) was used for all statistical analyses.
Twenty-five patients met the inclusion criteria for the study. Median age at the diagnosis of ovarian cancer was 58 years (IQR 53.1–64.8). Median age at the time of diagnosis of the first CNS involvement was 62.7 years (IQR 54.7–68.8). The Median time interval between primary ovarian carcinoma diagnosis and the presence of brain metastases was 42.3 months (IQR 25–49.7) Almost all patients (24/25) received chemotherapy to the primary disease. Eighteen patients (72%) had single brain lesion, whereas seven patients (28%) had 4 lesions or more. Of the 25 patients, 24 (98%) had exhibited progression by the time of the analysis and 22 (88%) were deceased by the time of the analysis. Ten patients received consequent therapeutic regimens, 14 did not, and one patient was progression-free. The median PFS was 22.4 months (IQR 3.7–28.53). Additional characteristics of the study population are shown in [Table 1], which provides the data in relation to the treatment modalities.
The treatment modalities were distributed as follows: 44% of patients were treated with radiation only, 20% were treated with surgery alone, and 36% treated with a combination of surgery and radiation. Of the patients treated with only radiation, 54.5% were treated with only WBRT, 36.4% with only SRS, and 9.1% with both modalities. Of the patient group treated with both surgery and radiation, 44.4% were treated with surgery and complementary WBRT, 22.2% with surgery plus WBRT plus SRS (all within 6 months), and 33.3% were treated with surgery plus SRS.
[Table 2] shows the univariate analysis of survival and PFS. Hazard ratio (HR; with a 95% confidence interval [CI]) and P value are presented. In addition, the percentage of deceased patients is shown against each variable. At the time of the analysis, 12% were still alive. The median survival of the study population (including patients who were alive at the time of the analysis) was 35.83 months (IQR 3.7–62.7). At the time of the analysis, 4.5% of the cohort was progression-free; the median time to progression was 22.4 months (IQR 3.7–28.53). The first progression of the patient's course of disease was distributed as follows: 10 patients (40%) received consequent therapeutic regimens, 14 patients (56%) died (and did not receive additional treatments following the initial treatment), and 1 patient (4%) was free of any progression at the time of the analysis.
Univariate analysis of mortality by log-rank demonstrated that age greater than 62.7 years at the time of CNS involvement (the mean in our study population) was a significant risk factor (P = 0.02, HR 1.064 [CI 1.01–1.12]). Clinical presentation of headache, vomiting, and/or nausea at the time of admission was a statistically significant risk factor (P = 0.027, HR 2.67 [CI 1.12–6.36]). Leptomeningeal disease was a poor prognostic factor in respect to supra-tentorial lesions (P = 0.015, HR 9.08 [CI 1.53–53.76]).
Univariate analysis of progression (PFS) demonstrated that the presence of multiple brain lesions (>4) was a poor prognostic factor (P = 0.009, HR 3.76 [CI 1.39–10.17]).
[Table 3] shows the multivariate analysis of survival and PFS. Multivariate analysis of mortality by Cox regression demonstrated that the presence of multiple brain lesions (>4) was a significant risk factor (P = 0.002, HR 37.957 [CI 3.93–366.9]). Multivariate analysis of progression (PFS) was performed with radiotherapy as monotherapy as a reference. Surgery as mono-therapy was a poor prognostic factor (P = 0.029, HR 6.154 [CI 1.2–31.5]).
Poor Prognostic Factors for Survival (univariate analysis)
Clinical presentation of headache, vomiting and/or nausea [P = 0.027, HR 2.67 (1.12–6.36)].
Age greater than 62.7 years at the time of diagnosis of CNS involvement (P = 0.02, HR 1.064 [CI 1.01–1.12]).
Leptomeningeal disease in reference to supra-tentorial lesions in the univariate analysis of survival (P = 0.015, HR 9.08 [CI 1.53–53.76]) and PFS (P = 0.023, HR 7.456 [CI 1.31–42.22]).
Poor Prognostic Factors for Survival (multivariate analysis)
Multiple brain lesions (>4) (P = 0.002, HR 37.957 [CI 3.93–366.9]).
Prognostic Factors for PFS (multivariate analysis)
Combined treatment with surgery and radiation resulted in longer median periods of progression-free survival than each modality alone, whereas surgery as mono-therapy was a poor prognostic factor (P = 0.029, HR 6.154 [CI 1.2–31.5]).
The Kaplan Meier curves of multivariate analysis of PFS is presented in [Figure 1].
The following variables did not have a statistically significant influence on survival and PFS in our analyses: extracranial metastases, ovarian carcinoma that was active at the time of diagnosis of brain metastatic involvement, primary cell type, distal metastases at the time of diagnosis of ovarian carcinoma, time interval between primary ovarian cancer diagnosis and brain metastatic involvement diagnosis, motor deficit at presentation, dysphasia at presentation, seizure at presentation, gait disturbance/ataxia at presentation, and diameter of the largest metastasis.
We present a retrospective study aimed at identifying the prognostic factors for survival and PFS following the diagnosis and treatment of CNS metastases from ovarian carcinoma. The study was based on a series of 25 patients who were diagnosed between the years 1995 and 2014 and were treated and followed at a single institute. In our cohort of patients, the median age for the diagnosis of CNS metastasis was 62.7 years, which is higher than the age described in the literature (54–56.8 years).,, Our results demonstrate that age older than 62.7 years is a prognostic factor for poor survival [P = 0.02, HR 1.064 (1.01–1.12)]. The increased median age at diagnosis of CNS metastases is probably due to improvements in the primary therapeutic options. It is possible, however, that a delayed diagnosis of brain involvement in some of the cohort accounts for its appearance as a prognostic factor because the disease would be more advanced by the time of detection in that group of patients.
Distant metastases to the liver, lung, brain, and bone are all associated with worse overall survival in patients with metastatic ovarian cancer. In our study, however, the evidence of extracranial metastases and active systemic disease at the time of diagnosis of brain metastatic involvement was not a predictive factor. This may be explained by the small number of patients in this subgroup with active disease in our cohort. These data are consistent with the conclusion of the study of Cohen et al. in a larger cohort of 72 patients that the existence of metastasis to the organs at the time of presentation of brain metastasis is adversely associated with survival, but is not an independent predictive factor of survival.
In our study, clinical signs and symptoms suggesting increased ICP, such as headache, nausea, or vomiting indicated poor prognosis in the univariate analysis, but increased ICP was not an independent poor prognostic factor in our series. A clinical picture of elevated ICP at presentation is not uncommon; its incidence is as high as 50% in some series.
Increased ICP is not as an independent poor prognostic factor in the literature, but in a published review of 13 studies, clinical indicators for increased ICP were correlated with metastatic involvement of the posterior fossa and with multiple brain lesions. These factors are associated with poor prognosis in numerous studies,,,, and in our series as well—both with decreased PFS and increased mortality, in the multivariate analysis. An infra-tentorial site of the lesion, which may cause increased ICP, was not independently associated with poor prognosis in our results, which may be related to the small number of patients. Unsurprisingly, the leptomeningeal spread was significantly associated with poor prognosis (P = 0.015, HR 9.08 [CI 1.53–53.76]).
The multimodal therapeutic approach has repeatedly demonstrated superiority over mono-therapeutic approaches.,, Patients who are treated non-operatively seem to benefit from multimodal regimens as well. Celejewska et al. showed that WBRT followed by SRS significantly improves prognosis in comparison with single-type radiation therapy. According to Lee et al., who compared a group of patients treated with WBRT to a group that was treated with gamma-knife radiosurgery, the gamma-knife group enjoyed a longer median survival. In that group, a trend was found toward the superiority of any combined regimen over any mono-therapeutic approach in terms of survival.
Piura et al. conducted a review of 34 studies that were dedicated to brain metastases of ovarian carcinoma. According to their review, combined treatment with surgery, WBRT, and chemotherapy yield the best survival compared with other treatments, particularly to mono-therapies; the average median survival in the combined treatments group was 20 months—the longest among all treatment groups.
There is limited data in the literature evaluating the role of SRS in the treatment of brain metastases from the gynecologic primary cancer. Johnston et al. reported that ovarian cancer patients have a decreased risk of distant brain failure following SRS, whereas cervical cancer patients had an increased risk of distant brain failure. They concluded that SRS represents a feasible treatment option for patients with brain metastases from gynecologic cancer.
Keller et al. showed that gamma-knife radiosurgery for ovarian brain metastasis has a local failure rate of 8%. Five patients (15%) developed new brain lesions outside the radiation field with a median PFS of 7 (range: 3–9) months. Median overall survival after gamma-knife treatment was 15 months. One-year overall survival from gamma-knife treatment was 72.9%. These data suggest that WBRT may be deferred following SRS for ovarian cancer brain metastasis.
We recently adopted the general approach of five fractions of radiation treatment to the tumor bed at 4–6 weeks after resection of brain metastasis. Additional lesions are treated with SRS. This approach was not performed in the present cohort; we used to perform WBRT following resection vs. conservative follow-up.
In our analysis, we grouped both WBRT and SRS under the definition of “radiation therapy”, due to the small cohort size. We found that the combined approach of surgery and radiation therapy was the most beneficial. Due to the small sample size, however, we were not able to compare the various sub-types of radiation therapies.
The superiority of combined therapy (surgery and radiation) has been established in several studies., The benefit of multimodal therapy on patient survival suggests more personalized decision-making in terms of therapeutic choices. The inconclusive data regarding the role of other factors should be further investigated in large multicenter studies.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Table 1], [Table 2], [Table 3]