Introduction
Ovarian cancer (OC) represents the most common gynaecological malignancy and has the highest mortality in the developed world [
1]. The vast majority of OCs are of epithelial origin [
2] and the most frequent subtype of epithelial OCs is the serous carcinoma [
3]. More than 70% of OCs are diagnosed in advanced stages with detectable transperitoneal spread [
4]. The 5-year survival rate is approximately 30% [
5]. Despite progress in the diagnosis and treatment of these tumors, the rates of relapse and distant metastasis remain high [
1]. Given the high recurrence rate and poor long-term survival of women with advanced stage disease, there is a strong need to document the unique metastatic patterns of epithelial OC by comparing the differences in genetic profiles between primary and metastatic tissue specimens [
6].
Clinical observation and retrospective clinical studies suggest that serous OCs grow very efficiently within the peritoneal cavity, but rarely metastasize beyond it [
7]. Accordingly, brain metastases (BM) from OC is a very rare phenomenon and a late-stage disease manifestation, usually occurring in the context of widely disseminated disease [
1]. The incidence of BM from epithelial OC ranges from 0.29 to 5% [
8,
9]. However, the incidence of BM seems to be increasing (11.6%) since the introduction of platinum-based chemotherapy [
10] and the more frequent use of sensitive detection methods [
11]. BM are associated with a large burden of neurological symptoms and a poor survival prognosis [
12]. Treatments vary widely, including chemotherapy, steroids, whole brain radiation therapy, surgical resection, and stereotactic radiosurgery [
13].
To date, most research efforts have focused on defining the molecular characteristics of primary OC [
6,
14,
15]. Only very limited data are available on molecular aberrations in BM of OC [
16‐
18]. In this study, we evaluated the mutational profile of OC metastases through Next-Generation Sequencing (NGS) with the aim of identifying potential clinically actionable genetic alterations with options for small molecule targeted therapy.
Discussion
To the best of our knowledge, we present here for the first time results of NGS in BM of OC. Too little is currently known about the genomic makeup of OC BM to offer a potential pathway for targeted therapeutics in this disease. Because of the rarity of BM of OC, the number of studied patients in general as well as in our cohort has been small. With this NGS-based study, we aimed to get some insight for a better understanding of this rare phenomenon. We successfully sequenced eight BM samples of primary OC and detected 37 variants in total, distributed over 22 cancer-related genes (23.4%). Mutations per analyzed BM sample ranged from 3 to 7 with a median of 4.5. The most commonly altered genes were
BRCA1/
2, TP53, ATM, and among others,
CHEK2. Consistently,
TP53, BRCA1, and
BRCA2 were the most frequently altered genes in the TCGA ovarian serous cystadenocarcinoma data. Moreover, in line with other studies on the pattern of somatic mutations in human cancer genomes [
15,
25], we observed a higher rate of transitions than transversions. Even if metastatic tumor spread is a very complex process consisting of many different events, the mutational spectrum of our BM specimens was surprisingly simple, which is consistent with the findings of Beltrame et al., who reported that the genomic architecture of relapsed disease was less heterogenous than that of the primary disease [
26]. Nevertheless, no two BM shared an identical genetic profile, which is similar to published data about metastatic breast cancer [
27,
28].
One of our major findings was the unexpectedly high number of
BRCA1/
2 mutations in BM of OC (7 out of 8). The National Genome Atlas study identified somatic
BRCA1 and
BRCA2 mutations as a significant feature of high grade serous OC [
14], but the characterized mutational frequencies were much lower [
2,
20]. The
BRCA1 gene is involved in DNA repair, cell-cycle checkpoint control, chromatin remodeling, transcriptional regulation and mitogenesis, while the
BRCA2 has an important role in homologous recombination [
29]. In our BM samples, 2/8 revealed a
BRCA2 mutation. Koul and colleagues hypothesized that
BRCA1 might have a possible role in ovarian tumors metastasizing to the brain [
30], which is mostly in line with our mutational results. In our cohort, 5/8 patients showed a
BRCA1 mutation and 2/8 a
BRCA2 mutation. Patient #4 was an interesting case, because we had PT and normal tissue available in addition to BM. A
BRCA1-deletion occurred in part of the PT sample and in the entire BM sample, suggesting that there was a positive selection of
BRCA1 mutation in BM compared to PT. In conclusion, mutated
BRCA1 and
BRCA2 seem to be important in the development of BM.
It was not surprising that
TP53 and
ATM were mutated in our BM samples of primary OC. Previous studies have highlighted that
TP53, that encodes the tumor suppressor protein p53, is the most frequently altered gene in serous OC [
2,
17]. In addition, the majority of our genetic alterations identified in
TP53 were predicted to be deleterious, and 2 of the BM mutations were known hotspot mutations (p.G266V and p.R248Q) (
http://cancerhotspots.org/). Based on these findings one can hypothesize that
TP53 plays a key role in BM specimens of primary OC. In our samples mutations in
TP53 and
ATM (n = 3), and
BRCA1/
2 and
ATM (n = 3) are common combinations. ATM is a major regulator of DNA damage detection and repair [
31]. In response to DNA damage, ATM controls the initial phosphorylation of a wide variety of downstream proteins such as TP53 and BRCA1 [
32]. Moreover, ATM mutations can lead to deficiencies in DNA repair, which in return may give rise to cancer [
33]. Therefore, it can be hypothesized that
ATM does not promote OC metastasizing to the brain isolatedly but through the DNA damage-recognition and repair pathway together with
TP53 and
BRCA1.
The mutation frequencies for
NF1, RB1, and
KIT in our BM samples were not in the range described for ovarian cancer specimens, they were less frequent.
EGFR (2.22%),
APC (2.2%), and
SMARCB1 (1.58%) have to be stressed here as well, because they were mutated in the TCGA dataset, but not in our BM samples. Beltrame and colleagues reported that somatic mutations showed a low rate of concordance between primary and recurrent disease in stage III–IV epithelial ovarian cancer, which may explain both these phenomena, or may be due to an underrepresentation of the mutational burden in the BM samples because of the small sample size [
26].
In this study, we focused on identifying potentially actionable somatic mutations in metastatic OC. In total, 7/8 BM samples analyzed showed
BRCA1/
2 mutations and, moreover, all 8 samples revealed mutations in at least one DNA repair gene (
BRCA1/
2, ATM, CHEK2). It is known that cells which are BRCA deficient and then undergo Poly (adenosine diphosphate-ribose) polymerase (PARP) inhibition (PARPi), suffer cell death [
34]. Accordingly, Mateo and colleagues concluded that, if a cell was lacking homologous repair capacity because of
BRCA1, BRCA2 or
ATM dysfunction or loss, then PARPi would lead to synthetic lethality due to cell cycle arrest and subsequent apoptosis [
35]. Additionally, McCabe et al. showed via in vitro studies that, among others, ATM and CHEK2 abnormalities resulted in sensitivity to PARPi, suggesting that PARPi would be beneficial for a variety of genes involved in the DNA damage response [
36]. The PARPi Olaparib has shown significant clinical activity in BRCA-mutated platinum-sensitive recurrent serous OC and has been approved for clinical use [
37‐
40]. Based on these facts and on our findings, pharmacological PARPi could be one potential targeted therapeutic for brain metastatic OC patients, especially because PARPi Olaparib has shown evidence of crossing the blood–brain barrier [
41]. Multiple PARP inhibitors are already at different stages of clinical development for the management of OC [
42,
43].
Metastatic OC remains largely incurable, and thus a larger population based study and molecular genetic analyses of OCs metastatic to the brain are needed for a better understanding of the role of key genes like
BRCA1/
2, TP53, ATM, and
CHEK2 in this rare phenomenon. Because of the rarity of OC BM our sample size is relatively small. Therefore, the acquisition of sufficiently large cohorts consisting of matched samples of this rare disease will require international collaborations [
2].
Taken together, our NGS study of OC BM revealed a prominent number of BRCA-mutations beside TP53, ATM and CHEK2 mutations. These findings strongly suggest the implication of BRCA and DNA repair malfunction in OC metastasizing to the brain. For any conclusive statement as to whether the DNA damage-recognition and repair pathway plays a key role in this phenotype, the implementation of a similar study with a larger cohort and functional analyses is needed. PARPi represents one of the most promising treatment options for OC patients in general and for OC patients with metastatic BM in particular.